Authors: Christian Fane
Abstract: Diminished reality (DR) refers to the digital removal of real-world objects by compositing background content in their place. This thesis presents a real-time, inpainting-based DR system designed to enable privacy control in shared-space mixed reality (MR) meetings. The system allows a primary headset user to selectively remove personal or sensitive items from their environment, ensuring that those objects are no longer visible to other participants. Removal is achieved through semantic segmentation and precise object selection, followed by real-time inpainting from the viewpoint of a secondary observer, implemented using a mobile ZED 2i depth camera. The solution is designed to be portable and robust, requiring neither a fixed secondary viewpoint nor prior 3D scanning of the environment. The system utilises YOLOv11 for object detection and a modified Decoupled Spatial-Temporal Transformer (DSTT) model for high-quality video inpainting. At 720p resolution, the pipeline sustains frame rates exceeding 20 fps, demonstrating the feasibility of real-time diminished reality for practical privacy-preserving MR applications.
Authors: Alejandra Perez, Chinedu Nwoye, Ramtin Raji Kermani, Omid Mohareri, Muhammad Abdullah Jamal
Abstract: Vision-language pre-training (VLP) offers unique advantages for surgery by aligning language with surgical videos, enabling workflow understanding and transfer across tasks without relying on expert-labeled datasets. However, progress in surgical VLP remains constrained by the limited scale, procedural diversity, semantic quality, and hierarchical structure of existing datasets. In this work, we present SurgLaVi, the largest and most diverse surgical vision-language dataset to date, comprising nearly 240k clip-caption pairs from more than 200 procedures, and comprising hierarchical levels at phase-, step-, and task-level. At the core of SurgLaVi lies a fully automated pipeline that systematically generates fine-grained transcriptions of surgical videos and segments them into coherent procedural units. To ensure high-quality annotations, it applies dual-modality filtering to remove irrelevant and noisy samples. Within this framework, the resulting captions are enriched with contextual detail, producing annotations that are both semantically rich and easy to interpret. To ensure accessibility, we release SurgLaVi-\b{eta}, an open-source derivative of 113k clip-caption pairs constructed entirely from public data, which is over four times larger than existing surgical VLP datasets. To demonstrate the value of SurgLaVi datasets, we introduce SurgCLIP, a CLIP-style video-text contrastive framework with dual encoders, as a representative base model. SurgCLIP achieves consistent improvements across phase, step, action, and tool recognition, surpassing prior state-of-the-art methods, often by large margins. These results validate that large-scale, semantically rich, and hierarchically structured datasets directly translate into stronger and more generalizable representations, establishing SurgLaVi as a key resource for developing surgical foundation models.
Authors: Emily Kaczmarek, Justin Szeto, Brennan Nichyporuk, Tal Arbel
Abstract: 3D structural Magnetic Resonance Imaging (MRI) brain scans are commonly acquired in clinical settings to monitor a wide range of neurological conditions, including neurodegenerative disorders and stroke. While deep learning models have shown promising results analyzing 3D MRI across a number of brain imaging tasks, most are highly tailored for specific tasks with limited labeled data, and are not able to generalize across tasks and/or populations. The development of self-supervised learning (SSL) has enabled the creation of large medical foundation models that leverage diverse, unlabeled datasets ranging from healthy to diseased data, showing significant success in 2D medical imaging applications. However, even the very few foundation models for 3D brain MRI that have been developed remain limited in resolution, scope, or accessibility. In this work, we present a general, high-resolution SimCLR-based SSL foundation model for 3D brain structural MRI, pre-trained on 18,759 patients (44,958 scans) from 11 publicly available datasets spanning diverse neurological diseases. We compare our model to Masked Autoencoders (MAE), as well as two supervised baselines, on four diverse downstream prediction tasks in both in-distribution and out-of-distribution settings. Our fine-tuned SimCLR model outperforms all other models across all tasks. Notably, our model still achieves superior performance when fine-tuned using only 20% of labeled training samples for predicting Alzheimer's disease. We use publicly available code and data, and release our trained model at https://github.com/emilykaczmarek/3D-Neuro-SimCLR, contributing a broadly applicable and accessible foundation model for clinical brain MRI analysis.
Authors: Xiaoyang Ma, Yiyang Chai, Xinran Qu, Hong Sun
Abstract: Reconstructing hyperspectral images (HSIs) from a single RGB image is ill-posed and can become physically inconsistent when the camera spectral sensitivity (CSS) and scene illumination are misspecified. We formulate RGB-to-HSI reconstruction as a physics-grounded inverse problem regularized by a nuclear norm in a learnable transform domain, and we explicitly estimate CSS and illumination to define the forward operator embedded in each iteration, ensuring colorimetric consistency. To avoid the cost and instability of full singular-value decompositions (SVDs) required by singular-value thresholding (SVT), we introduce a data-adaptive low-rank subspace SVT operator. Building on these components, we develop USCTNet, a deep unfolding solver tailored to HSI that couples a parameter estimation module with learnable proximal updates. Extensive experiments on standard benchmarks show consistent improvements over state-of-the-art RGB-based methods in reconstruction accuracy. Code: https://github.com/psykheXX/USCTNet-Code-Implementation.git
URLs: https://github.com/psykheXX/USCTNet-Code-Implementation.git
Authors: Felicia Liu, Jay J. Yoo, Farzad Khalvati
Abstract: Large Language Models (LLMs) have shown strong performance in text-based healthcare tasks. However, their utility in image-based applications remains unexplored. We investigate the effectiveness of LLMs for medical imaging tasks, specifically glioma classification and segmentation, and compare their performance to that of traditional convolutional neural networks (CNNs). Using the BraTS 2020 dataset of multi-modal brain MRIs, we evaluated a general-purpose vision-language LLM (LLaMA 3.2 Instruct) both before and after fine-tuning, and benchmarked its performance against custom 3D CNNs. For glioma classification (Low-Grade vs. High-Grade), the CNN achieved 80% accuracy and balanced precision and recall. The general LLM reached 76% accuracy but suffered from a specificity of only 18%, often misclassifying Low-Grade tumors. Fine-tuning improved specificity to 55%, but overall performance declined (e.g., accuracy dropped to 72%). For segmentation, three methods - center point, bounding box, and polygon extraction, were implemented. CNNs accurately localized gliomas, though small tumors were sometimes missed. In contrast, LLMs consistently clustered predictions near the image center, with no distinction of glioma size, location, or placement. Fine-tuning improved output formatting but failed to meaningfully enhance spatial accuracy. The bounding polygon method yielded random, unstructured outputs. Overall, CNNs outperformed LLMs in both tasks. LLMs showed limited spatial understanding and minimal improvement from fine-tuning, indicating that, in their current form, they are not well-suited for image-based tasks. More rigorous fine-tuning or alternative training strategies may be needed for LLMs to achieve better performance, robustness, and utility in the medical space.
Authors: Hao Zhang, Chun-Han Yao, Simon Donn\'e, Narendra Ahuja, Varun Jampani
Abstract: We present Stable Part Diffusion 4D (SP4D), a framework for generating paired RGB and kinematic part videos from monocular inputs. Unlike conventional part segmentation methods that rely on appearance-based semantic cues, SP4D learns to produce kinematic parts - structural components aligned with object articulation and consistent across views and time. SP4D adopts a dual-branch diffusion model that jointly synthesizes RGB frames and corresponding part segmentation maps. To simplify the architecture and flexibly enable different part counts, we introduce a spatial color encoding scheme that maps part masks to continuous RGB-like images. This encoding allows the segmentation branch to share the latent VAE from the RGB branch, while enabling part segmentation to be recovered via straightforward post-processing. A Bidirectional Diffusion Fusion (BiDiFuse) module enhances cross-branch consistency, supported by a contrastive part consistency loss to promote spatial and temporal alignment of part predictions. We demonstrate that the generated 2D part maps can be lifted to 3D to derive skeletal structures and harmonic skinning weights with few manual adjustments. To train and evaluate SP4D, we construct KinematicParts20K, a curated dataset of over 20K rigged objects selected and processed from Objaverse XL (Deitke et al., 2023), each paired with multi-view RGB and part video sequences. Experiments show that SP4D generalizes strongly to diverse scenarios, including real-world videos, novel generated objects, and rare articulated poses, producing kinematic-aware outputs suitable for downstream animation and motion-related tasks.
Authors: Sven Schreiber, Noha Sarhan, Simone Frintrop, Christian Wilms
Abstract: Isolated Sign Language Recognition (ISLR) approaches primarily rely on RGB data or signer pose information. However, combining these modalities often results in the loss of crucial details, such as hand shape and orientation, due to imprecise representations like bounding boxes. Therefore, we propose the ISLR system SegSLR, which combines RGB and pose information through promptable zero-shot video segmentation. Given the rough localization of the hands and the signer's body from pose information, we segment the respective parts through the video to maintain all relevant shape information. Subsequently, the segmentations focus the processing of the RGB data on the most relevant body parts for ISLR. This effectively combines RGB and pose information. Our evaluation on the complex ChaLearn249 IsoGD dataset shows that SegSLR outperforms state-of-the-art methods. Furthermore, ablation studies indicate that SegSLR strongly benefits from focusing on the signer's body and hands, justifying our design choices.
Authors: Jecia Z. Y. Mao, Francis X Creighton, Russell H Taylor, Manish Sahu
Abstract: Accurate segmentation and tracking of relevant elements of the surgical scene is crucial to enable context-aware intraoperative assistance and decision making. Current solutions remain tethered to domain-specific, supervised models that rely on labeled data and required domain-specific data to adapt to new surgical scenarios and beyond predefined label categories. Recent advances in prompt-driven vision foundation models (VFM) have enabled open-set, zero-shot segmentation across heterogeneous medical images. However, dependence of these models on manual visual or textual cues restricts their deployment in introperative surgical settings. We introduce a speech-guided collaborative perception (SCOPE) framework that integrates reasoning capabilities of large language model (LLM) with perception capabilities of open-set VFMs to support on-the-fly segmentation, labeling and tracking of surgical instruments and anatomy in intraoperative video streams. A key component of this framework is a collaborative perception agent, which generates top candidates of VFM-generated segmentation and incorporates intuitive speech feedback from clinicians to guide the segmentation of surgical instruments in a natural human-machine collaboration paradigm. Afterwards, instruments themselves serve as interactive pointers to label additional elements of the surgical scene. We evaluated our proposed framework on a subset of publicly available Cataract1k dataset and an in-house ex-vivo skull-base dataset to demonstrate its potential to generate on-the-fly segmentation and tracking of surgical scene. Furthermore, we demonstrate its dynamic capabilities through a live mock ex-vivo experiment. This human-AI collaboration paradigm showcase the potential of developing adaptable, hands-free, surgeon-centric tools for dynamic operating-room environments.
Authors: Yi-Ruei Liu, You-Zhe Xie, Yu-Hsiang Hsu, I-Sheng Fang, Yu-Lun Liu, Jun-Cheng Chen
Abstract: Common computer vision systems typically assume ideal pinhole cameras but fail when facing real-world camera effects such as fisheye distortion and rolling shutter, mainly due to the lack of learning from training data with camera effects. Existing data generation approaches suffer from either high costs, sim-to-real gaps or fail to accurately model camera effects. To address this bottleneck, we propose 4D Gaussian Ray Tracing (4D-GRT), a novel two-stage pipeline that combines 4D Gaussian Splatting with physically-based ray tracing for camera effect simulation. Given multi-view videos, 4D-GRT first reconstructs dynamic scenes, then applies ray tracing to generate videos with controllable, physically accurate camera effects. 4D-GRT achieves the fastest rendering speed while performing better or comparable rendering quality compared to existing baselines. Additionally, we construct eight synthetic dynamic scenes in indoor environments across four camera effects as a benchmark to evaluate generated videos with camera effects.
Authors: Marcelo Sandoval-Castaneda, Bryan Russell, Josef Sivic, Gregory Shakhnarovich, Fabian Caba Heilbron
Abstract: Automated tools for video editing and assembly have applications ranging from filmmaking and advertisement to content creation for social media. Previous video editing work has mainly focused on either retrieval or user interfaces, leaving actual editing to the user. In contrast, we propose to automate the core task of video editing, formulating it as sequential decision making process. Ours is a multi-agent approach. We design an Editor agent and a Critic agent. The Editor takes as input a collection of video clips together with natural language instructions and uses tools commonly found in video editing software to produce an edited sequence. On the other hand, the Critic gives natural language feedback to the editor based on the produced sequence or renders it if it is satisfactory. We introduce a learning-based approach for enabling effective communication across specialized agents to address the language-driven video editing task. Finally, we explore an LLM-as-a-judge metric for evaluating the quality of video editing system and compare it with general human preference. We evaluate our system's output video sequences qualitatively and quantitatively through a user study and find that our system vastly outperforms existing approaches in terms of coverage, time constraint satisfaction, and human preference.
Authors: Sajad Amiri, Shahram Taeb, Sara Gharibi, Setareh Dehghanfard, Somayeh Sadat Mehrnia, Mehrdad Oveisi, Ilker Hacihaliloglu, Arman Rahmim, Mohammad R. Salmanpour
Abstract: Gadolinium-based contrast agents (GBCAs) are central to glioma imaging but raise safety, cost, and accessibility concerns. Predicting contrast enhancement from non-contrast MRI using machine learning (ML) offers a safer alternative, as enhancement reflects tumor aggressiveness and informs treatment planning. Yet scanner and cohort variability hinder robust model selection. We propose a stability-aware framework to identify reproducible ML pipelines for multicenter prediction of glioma MRI contrast enhancement. We analyzed 1,446 glioma cases from four TCIA datasets (UCSF-PDGM, UPENN-GB, BRATS-Africa, BRATS-TCGA-LGG). Non-contrast T1WI served as input, with enhancement derived from paired post-contrast T1WI. Using PyRadiomics under IBSI standards, 108 features were extracted and combined with 48 dimensionality reduction methods and 25 classifiers, yielding 1,200 pipelines. Rotational validation was trained on three datasets and tested on the fourth. Cross-validation prediction accuracies ranged from 0.91 to 0.96, with external testing achieving 0.87 (UCSF-PDGM), 0.98 (UPENN-GB), and 0.95 (BRATS-Africa), with an average of 0.93. F1, precision, and recall were stable (0.87 to 0.96), while ROC-AUC varied more widely (0.50 to 0.82), reflecting cohort heterogeneity. The MI linked with ETr pipeline consistently ranked highest, balancing accuracy and stability. This framework demonstrates that stability-aware model selection enables reliable prediction of contrast enhancement from non-contrast glioma MRI, reducing reliance on GBCAs and improving generalizability across centers. It provides a scalable template for reproducible ML in neuro-oncology and beyond.
Authors: Yilun Xiao
Abstract: Dense small objects in UAV imagery are often missed due to long-range viewpoints, occlusion, and clutter[cite: 5]. This paper presents a detector-agnostic post-processing framework that converts overlap-induced redundancy into group evidence[cite: 6]. Overlapping tiling first recovers low-confidence candidates[cite: 7]. A Spatial Gate (DBSCAN on box centroids) and a Semantic Gate (DBSCAN on ResNet-18 embeddings) then validates group evidence[cite: 7]. Validated groups receive controlled confidence reweighting before class-aware NMS fusion[cite: 8]. Experiments on VisDrone show a recall increase from 0.685 to 0.778 (+0.093) and a precision adjustment from 0.801 to 0.595, yielding F1=0.669[cite: 9]. Post-processing latency averages 0.095 s per image[cite: 10]. These results indicate recall-first, precision-trade-off behavior that benefits recall-sensitive applications such as far-field counting and monitoring[cite: 10]. Ablation confirms that tiling exposes missed objects, spatial clustering stabilizes geometry, semantic clustering enforces appearance coherence, and reweighting provides calibrated integration with the baseline[cite: 11]. The framework requires no retraining and integrates with modern detectors[cite: 12]. Future work will reduce semantic gating cost and extend the approach with temporal cues[cite: 13].
Authors: Weipeng Zhong, Peizhou Cao, Yichen Jin, Li Luo, Wenzhe Cai, Jingli Lin, Hanqing Wang, Zhaoyang Lyu, Tai Wang, Bo Dai, Xudong Xu, Jiangmiao Pang
Abstract: The advancement of Embodied AI heavily relies on large-scale, simulatable 3D scene datasets characterized by scene diversity and realistic layouts. However, existing datasets typically suffer from limitations in data scale or diversity, sanitized layouts lacking small items, and severe object collisions. To address these shortcomings, we introduce \textbf{InternScenes}, a novel large-scale simulatable indoor scene dataset comprising approximately 40,000 diverse scenes by integrating three disparate scene sources, real-world scans, procedurally generated scenes, and designer-created scenes, including 1.96M 3D objects and covering 15 common scene types and 288 object classes. We particularly preserve massive small items in the scenes, resulting in realistic and complex layouts with an average of 41.5 objects per region. Our comprehensive data processing pipeline ensures simulatability by creating real-to-sim replicas for real-world scans, enhances interactivity by incorporating interactive objects into these scenes, and resolves object collisions by physical simulations. We demonstrate the value of InternScenes with two benchmark applications: scene layout generation and point-goal navigation. Both show the new challenges posed by the complex and realistic layouts. More importantly, InternScenes paves the way for scaling up the model training for both tasks, making the generation and navigation in such complex scenes possible. We commit to open-sourcing the data, models, and benchmarks to benefit the whole community.
Authors: Robert M. Corless, Deepak Singh Kalhan, Stephen M. Watt
Abstract: Previous work has made use of a parameterized plane curve polynomial representation for mathematical handwriting, with the polynomials represented in a Legendre or Legendre-Sobolev graded basis. This provides a compact geometric representation for the digital ink. Preliminary results have also been shown for Chebyshev and Chebyshev-Sobolev bases. This article explores the trade-offs between basis choice and polynomial degree to achieve accurate modeling with a low computational cost. To do this, we consider the condition number for polynomial evaluation in these bases and bound how the various inner products give norms for the variations between symbols.
Authors: Aghiles Kebaili, Romain Modzelewski, J\'er\^ome Lapuyade-Lahorgue, Maxime Fontanilles, S\'ebastien Thureau, Su Ruan
Abstract: Glioma, an aggressive brain malignancy characterized by rapid progression and its poor prognosis, poses significant challenges for accurate evolution prediction. These challenges are exacerbated by sparse, irregularly acquired longitudinal MRI data in clinical practice, where incomplete follow-up sequences create data imbalances and make reliable modeling difficult. In this paper, we present a multitask diffusion framework for time-agnostic, pixel-wise prediction of glioma progression. The model simultaneously generates future FLAIR sequences at any chosen time point and estimates spatial probabilistic tumor evolution maps derived using signed distance fields (SDFs), allowing uncertainty quantification. To capture temporal dynamics of tumor evolution across arbitrary intervals, we integrate a pretrained deformation module that models inter-scan changes using deformation fields. Regarding the common clinical limitation of data scarcity, we implement a targeted augmentation pipeline that synthesizes complete sequences of three follow-up scans and imputes missing MRI modalities from available patient studies, improving the stability and accuracy of predictive models. Based on merely two follow-up scans at earlier timepoints, our framework produces flexible time-depending probability maps, enabling clinicians to interrogate tumor progression risks at any future temporal milestone. We further introduce a radiotherapy-weighted focal loss term that leverages radiation dose maps, as these highlight regions of greater clinical importance during model training. The proposed method was trained on a public dataset and evaluated on an internal private dataset, achieving promising results in both cases
Authors: Simone Mosco, Daniel Fusaro, Wanmeng Li, Emanuele Menegatti, Alberto Pretto
Abstract: LiDAR point cloud semantic segmentation is essential for interpreting 3D environments in applications such as autonomous driving and robotics. Recent methods achieve strong performance by exploiting different point cloud representations or incorporating data from other sensors, such as cameras or external datasets. However, these approaches often suffer from high computational complexity and require large amounts of training data, limiting their generalization in data-scarce scenarios. In this paper, we improve the performance of point-based methods by effectively learning features from 2D representations through point-plane projections, enabling the extraction of complementary information while relying solely on LiDAR data. Additionally, we introduce a geometry-aware technique for data augmentation that aligns with LiDAR sensor properties and mitigates class imbalance. We implemented and evaluated our method that applies point-plane projections onto multiple informative 2D representations of the point cloud. Experiments demonstrate that this approach leads to significant improvements in limited-data scenarios, while also achieving competitive results on two publicly available standard datasets, as SemanticKITTI and PandaSet. The code of our method is available at https://github.com/SiMoM0/3PNet
Authors: Chongyu Wang, Kunlei Jing, Jihua Zhu, Di Wang
Abstract: Open-vocabulary semantic segmentation enables models to recognize and segment objects from arbitrary natural language descriptions, offering the flexibility to handle novel, fine-grained, or functionally defined categories beyond fixed label sets. While this capability is crucial for large-scale urban point clouds that support applications such as digital twins, smart city management, and urban analytics, it remains largely unexplored in this domain. The main obstacles are the frequent absence of high-quality, well-aligned multi-view imagery in large-scale urban point cloud datasets and the poor generalization of existing three-dimensional (3D) segmentation pipelines across diverse urban environments with substantial variation in geometry, scale, and appearance. To address these challenges, we present OpenUrban3D, the first 3D open-vocabulary semantic segmentation framework for large-scale urban scenes that operates without aligned multi-view images, pre-trained point cloud segmentation networks, or manual annotations. Our approach generates robust semantic features directly from raw point clouds through multi-view, multi-granularity rendering, mask-level vision-language feature extraction, and sample-balanced fusion, followed by distillation into a 3D backbone model. This design enables zero-shot segmentation for arbitrary text queries while capturing both semantic richness and geometric priors. Extensive experiments on large-scale urban benchmarks, including SensatUrban and SUM, show that OpenUrban3D achieves significant improvements in both segmentation accuracy and cross-scene generalization over existing methods, demonstrating its potential as a flexible and scalable solution for 3D urban scene understanding.
Authors: Aryan Kashyap Naveen, Bhuvanesh Singla, Raajan Wankhade, Shreesha M, Ramu S, Ram Mohana Reddy Guddeti
Abstract: The burgeoning of online education has created an urgent need for robust and scalable systems to ensure academic integrity during remote examinations. Traditional human proctoring is often not feasible at scale, while existing automated solutions can be intrusive or fail to detect a wide range of cheating behaviors. This paper introduces AutoOEP (Automated Online Exam Proctoring), a comprehensive, multi-modal framework that leverages computer vision and machine learning to provide effective, automated proctoring. The system utilizes a dual-camera setup to capture both a frontal view of the examinee and a side view of the workspace, minimizing blind spots. Our approach integrates several parallel analyses: the Face Module performs continuous identity verification using ArcFace, along with head pose estimation, gaze tracking, and mouth movement analysis to detect suspicious cues. Concurrently, the Hand Module employs a fine-tuned YOLOv11 model for detecting prohibited items (e.g., mobile phones, notes) and tracks hand proximity to these objects. Features from these modules are aggregated and fed into a Long Short-Term Memory (LSTM) network that analyzes temporal patterns to calculate a real-time cheating probability score. We evaluate AutoOEP on a custom-collected dataset simulating diverse exam conditions. Our system achieves an accuracy of 90.7% in classifying suspicious activities. The object detection component obtains a mean Average Precision (mAP@.5) of 0.57 for prohibited items, and the entire framework processes video streams at approximately 2.4 frames per second without a GPU. The results demonstrate that AutoOEP is an effective and resource-efficient solution for automated proctoring, significantly reducing the need for human intervention and enhancing the integrity of online assessments.
Authors: Weiqiang Zhao, Tianzhu Liu, Yuzhe Gui, Yanfeng Gu
Abstract: Spectral imaging technology has long-faced fundamental challenges in balancing spectral, spatial, and temporal resolutions. While compressive sensing-based Coded Aperture Snapshot Spectral Imaging (CASSI) mitigates this trade-off through optical encoding, high compression ratios result in ill-posed reconstruction problems. Traditional model-based methods exhibit limited performance due to reliance on handcrafted inherent image priors, while deep learning approaches are constrained by their black-box nature, which compromises physical interpretability. To address these limitations, we propose a dual-camera CASSI reconstruction framework that integrates total variation (TV) subgradient theory. By establishing an end-to-end SD-CASSI mathematical model, we reduce the computational complexity of solving the inverse problem and provide a mathematically well-founded framework for analyzing multi-camera systems. A dynamic regularization strategy is introduced, incorporating normalized gradient constraints from RGB/panchromatic-derived reference images, which constructs a TV subgradient similarity function with strict convex optimization guarantees. Leveraging spatial priors from auxiliary cameras, an adaptive reference generation and updating mechanism is designed to provide subgradient guidance. Experimental results demonstrate that the proposed method effectively preserves spatial-spectral structural consistency. The theoretical framework establishes an interpretable mathematical foundation for computational spectral imaging, demonstrating robust performance across diverse reconstruction scenarios. The source code is available at https://github.com/bestwishes43/ADMM-TVDS.
Authors: Mohanad Albughdadi
Abstract: Recent advances in Earth Observation have focused on large-scale foundation models. However, these models are computationally expensive, limiting their accessibility and reuse for downstream tasks. In this work, we investigate compact architectures as a practical pathway toward smaller general-purpose EO models. We propose a Metadata-aware Mixture-of-Experts Masked Autoencoder (MoE-MAE) with only 2.5M parameters. The model combines sparse expert routing with geo-temporal conditioning, incorporating imagery alongside latitude/longitude and seasonal/daily cyclic encodings. We pretrain the MoE-MAE on the BigEarthNet-Landsat dataset and evaluate embeddings from its frozen encoder using linear probes. Despite its small size, the model competes with much larger architectures, demonstrating that metadata-aware pretraining improves transfer and label efficiency. To further assess generalization, we evaluate on the EuroSAT-Landsat dataset, which lacks explicit metadata, and still observe competitive performance compared to models with hundreds of millions of parameters. These results suggest that compact, metadata-aware MoE-MAEs are an efficient and scalable step toward future EO foundation models.
Authors: Farhan Sadik, Christopher L. Newman, Stuart J. Warden, Rachel K. Surowiec
Abstract: Rigid-motion artifacts, such as cortical bone streaking and trabecular smearing, hinder in vivo assessment of bone microstructures in high-resolution peripheral quantitative computed tomography (HR-pQCT). Despite various motion grading techniques, no motion correction methods exist due to the lack of standardized degradation models. We optimize a conventional sinogram-based method to simulate motion artifacts in HR-pQCT images, creating paired datasets of motion-corrupted images and their corresponding ground truth, which enables seamless integration into supervised learning frameworks for motion correction. As such, we propose an Edge-enhanced Self-attention Wasserstein Generative Adversarial Network with Gradient Penalty (ESWGAN-GP) to address motion artifacts in both simulated (source) and real-world (target) datasets. The model incorporates edge-enhancing skip connections to preserve trabecular edges and self-attention mechanisms to capture long-range dependencies, facilitating motion correction. A visual geometry group (VGG)-based perceptual loss is used to reconstruct fine micro-structural features. The ESWGAN-GP achieves a mean signal-to-noise ratio (SNR) of 26.78, structural similarity index measure (SSIM) of 0.81, and visual information fidelity (VIF) of 0.76 for the source dataset, while showing improved performance on the target dataset with an SNR of 29.31, SSIM of 0.87, and VIF of 0.81. The proposed methods address a simplified representation of real-world motion that may not fully capture the complexity of in vivo motion artifacts. Nevertheless, because motion artifacts present one of the foremost challenges to more widespread adoption of this modality, these methods represent an important initial step toward implementing deep learning-based motion correction in HR-pQCT.
Authors: Dillon Lohr, Michael J Proulx, Mehedi Hasan Raju, Oleg V Komogortsev
Abstract: This paper examines the key factors that influence the performance of state-of-the-art gaze-based authentication. Experiments were conducted on a large-scale, in-house dataset comprising 8,849 subjects collected with Meta Quest Pro equivalent hardware running a video oculography-driven gaze estimation pipeline at 72Hz. The state-of-the-art neural network architecture was employed to study the influence of the following factors on authentication performance: eye tracking signal quality, various aspects of eye tracking calibration, and simple filtering on estimated raw gaze. We found that using the same calibration target depth for eye tracking calibration, fusing calibrated and non-calibrated gaze, and improving eye tracking signal quality all enhance authentication performance. We also found that a simple three-sample moving average filter slightly reduces authentication performance in general. While these findings hold true for the most part, some exceptions were noted.
Authors: Haoming Lu
Abstract: Skin tone recognition and generation play important roles in model fairness, healthcare, and generative AI, yet they remain challenging due to the lack of comprehensive datasets and robust methodologies. Compared to other human image analysis tasks, state-of-the-art large multimodal models (LMMs) and image generation models struggle to recognize and synthesize skin tones accurately. To address this, we introduce TrueSkin, a dataset with 7299 images systematically categorized into 6 classes, collected under diverse lighting conditions, camera angles, and capture settings. Using TrueSkin, we benchmark existing recognition and generation approaches, revealing substantial biases: LMMs tend to misclassify intermediate skin tones as lighter ones, whereas generative models struggle to accurately produce specified skin tones when influenced by inherent biases from unrelated attributes in the prompts, such as hairstyle or environmental context. We further demonstrate that training a recognition model on TrueSkin improves classification accuracy by more than 20\% compared to LMMs and conventional approaches, and fine-tuning with TrueSkin significantly improves skin tone fidelity in image generation models. Our findings highlight the need for comprehensive datasets like TrueSkin, which not only serves as a benchmark for evaluating existing models but also provides a valuable training resource to enhance fairness and accuracy in skin tone recognition and generation tasks.
Authors: Nisha Pillai, Aditi Virupakshaiah, Harrison W. Smith, Amanda J. Ashworth, Prasanna Gowda, Phillip R. Owens, Adam R. Rivers, Bindu Nanduri, Mahalingam Ramkumar
Abstract: Animal health monitoring and population management are critical aspects of wildlife conservation and livestock management that increasingly rely on automated detection and tracking systems. While Unmanned Aerial Vehicle (UAV) based systems combined with computer vision offer promising solutions for non-invasive animal monitoring across challenging terrains, limited availability of labeled training data remains an obstacle in developing effective deep learning (DL) models for these applications. Transfer learning has emerged as a potential solution, allowing models trained on large datasets to be adapted for resource-limited scenarios such as those with limited data. However, the vast landscape of pre-trained neural network architectures makes it challenging to select optimal models, particularly for researchers new to the field. In this paper, we propose a reinforcement learning (RL)-based transfer learning framework that employs an upper confidence bound (UCB) algorithm to automatically select the most suitable pre-trained model for animal detection tasks. Our approach systematically evaluates and ranks candidate models based on their performance, streamlining the model selection process. Experimental results demonstrate that our framework achieves a higher detection rate while requiring significantly less computational time compared to traditional methods.
Authors: Abdarahmane Traore, \'Eric Hervet, Andy Couturier
Abstract: The FungiCLEF 2025 competition addresses the challenge of automatic fungal species recognition using realistic, field-collected observational data. Accurate identification tools support both mycologists and citizen scientists, greatly enhancing large-scale biodiversity monitoring. Effective recognition systems in this context must handle highly imbalanced class distributions and provide reliable performance even when very few training samples are available for many species, especially rare and under-documented taxa that are often missing from standard training sets. According to competition organizers, about 20\% of all verified fungi observations, representing nearly 20,000 instances, are associated with these rarely recorded species. To tackle this challenge, we propose a robust deep learning method based on prototypical networks, which enhances prototype representations for few-shot fungal classification. Our prototypical network approach exceeds the competition baseline by more than 30 percentage points in Recall@5 on both the public (PB) and private (PR) leaderboards. This demonstrates strong potential for accurately identifying both common and rare fungal species, supporting the main objectives of FungiCLEF 2025.
Authors: Yuedi Zhang, Zhixiang Xia, Guosheng Yin, Bin Liu
Abstract: Multi-Instance Learning (MIL) is pivotal for analyzing complex, weakly labeled datasets, such as whole-slide images (WSIs) in computational pathology, where bags comprise unordered collections of instances with sparse diagnostic relevance. Traditional MIL approaches, including early statistical methods and recent attention-based frameworks, struggle with instance redundancy and lack explicit mechanisms for discarding non-informative instances, limiting their robustness and interpretability. We propose Cluster-level Sparse MIL (csMIL), a novel framework that integrates global-local instance clustering, within-cluster attention, and cluster-level sparsity induction to address these challenges. Our csMIL first performs global clustering across all bags to establish $K$ cluster centers, followed by local clustering within each bag to assign cluster labels. Attention scores are computed within each cluster, and sparse regularization is applied to cluster weights, enabling the selective retention of diagnostically relevant clusters while discarding irrelevant ones. This approach enhances robustness to noisy instances, improves interpretability by identifying critical regions, and reduces computational complexity. Theoretical analysis demonstrates that csMIL requires $O(s log K)$ bags to recover $s$ relevant clusters, aligning with compressed sensing principles. Empirically, csMIL achieves state-of-the-art performance on two public histopathology benchmarks (CAMELYON16, TCGA-NSCLC).
Authors: Canhui Tang, Sanping Zhou, Haoyue Shi, Le Wang
Abstract: Zero-Shot Video Anomaly Detection (ZS-VAD) requires temporally localizing anomalies without target domain training data, which is a crucial task due to various practical concerns, e.g., data privacy or new surveillance deployments. Skeleton-based approach has inherent generalizable advantages in achieving ZS-VAD as it eliminates domain disparities both in background and human appearance. However, existing methods only learn low-level skeleton representation and rely on the domain-limited normality boundary, which cannot generalize well to new scenes with different normal and abnormal behavior patterns. In this paper, we propose a novel zero-shot video anomaly detection framework, unlocking the potential of skeleton data via action typicality and uniqueness learning. Firstly, we introduce a language-guided semantic typicality modeling module that projects skeleton snippets into action semantic space and distills LLM's knowledge of typical normal and abnormal behaviors during training. Secondly, we propose a test-time context uniqueness analysis module to finely analyze the spatio-temporal differences between skeleton snippets and then derive scene-adaptive boundaries. Without using any training samples from the target domain, our method achieves state-of-the-art results against skeleton-based methods on four large-scale VAD datasets: ShanghaiTech, UBnormal, NWPU, and UCF-Crime, featuring over 100 unseen surveillance scenes.
Authors: Xiaoyu Huang, Lauren M Maxson, Trang Nguyen, Cheng Jack Song, Yuankai Huo
Abstract: Recent advances in organoid models have revolutionized the study of human kidney disease mechanisms and drug discovery by enabling scalable, cost-effective research without the need for animal sacrifice. Here, we present a kidney organoid platform optimized for efficient screening in polycystic kidney disease (PKD). While these systems generate rich spatial-temporal microscopy video datasets, current manual approaches to analysis remain limited to coarse classifications (e.g., hit vs. non-hit), often missing valuable pixel-level and longitudinal information. To help overcome this bottleneck, we developed Organoid Tracker, a graphical user interface (GUI) platform designed with a modular plugin architecture, which empowers researchers to extract detailed, quantitative metrics without programming expertise. Built on the cutting-edge vision foundation model Segment Anything Model 2 (SAM2), Organoid Tracker enables zero-shot segmentation and automated analysis of spatial-temporal microscopy videos. It quantifies key metrics such as cyst formation rate, growth velocity, and morphological changes, while generating comprehensive reports. By providing an extensible, open-source framework, Organoid Tracker offers a powerful solution for improving and accelerating research in kidney development, PKD modeling, and therapeutic discovery. The platform is publicly available as open-source software at https://github.com/hrlblab/OrganoidTracker.
Authors: Jinghan Peng, Jingwen Wang, Xing Yu, Dehui Du
Abstract: This report outlines our approach using vision language model systems for the Driving with Language track of the CVPR 2024 Autonomous Grand Challenge. We have exclusively utilized the DriveLM-nuScenes dataset for training our models. Our systems are built on the LLaVA models, which we enhanced through fine-tuning with the LoRA and DoRA methods. Additionally, we have integrated depth information from open-source depth estimation models to enrich the training and inference processes. For inference, particularly with multiple-choice and yes/no questions, we adopted a Chain-of-Thought reasoning approach to improve the accuracy of the results. This comprehensive methodology enabled us to achieve a top score of 0.7799 on the validation set leaderboard, ranking 1st on the leaderboard.
Authors: Zongwu Xie, Kaijie Yun, Yang Liu, Yiming Ji, Han Li
Abstract: We present a robust multi-modal framework for predicting traversability costmaps for planetary rovers. Our model fuses camera and LiDAR data to produce a bird's-eye-view (BEV) terrain costmap, trained self-supervised using IMU-derived labels. Key updates include a DINOv3-based image encoder, FiLM-based sensor fusion, and an optimization loss combining Huber and smoothness terms. Experimental ablations (removing image color, occluding inputs, adding noise) show only minor changes in MAE/MSE (e.g. MAE increases from ~0.0775 to 0.0915 when LiDAR is sparsified), indicating that geometry dominates the learned cost and the model is highly robust. We attribute the small performance differences to the IMU labeling primarily reflecting terrain geometry rather than semantics and to limited data diversity. Unlike prior work claiming large gains, we emphasize our contributions: (1) a high-fidelity, reproducible simulation environment; (2) a self-supervised IMU-based labeling pipeline; and (3) a strong multi-modal BEV costmap prediction model. We discuss limitations and future work such as domain generalization and dataset expansion.
Authors: Chao Chen, Shunyu Yao, Yuanwu He, Tao Feng, Ruojing Song, Yuliang Guo, Xinyu Huang, Chenxu Wu, Ren Liu, Chen Feng
Abstract: Precise parking requires an end-to-end system where perception adaptively provides policy-relevant details-especially in critical areas where fine control decisions are essential. End-to-end learning offers a unified framework by directly mapping sensor inputs to control actions, but existing approaches lack effective synergy between perception and control. We find that transformer-based self-attention, when used alone, tends to produce unstable and temporally inconsistent spatial attention, which undermines the reliability of downstream policy decisions over time. Instead, we propose CAA-Policy, an end-to-end imitation learning system that allows control signal to guide the learning of visual attention via a novel Control-Aided Attention (CAA) mechanism. For the first time, we train such an attention module in a self-supervised manner, using backpropagated gradients from the control outputs instead of from the training loss. This strategy encourages the attention to focus on visual features that induce high variance in action outputs, rather than merely minimizing the training loss-a shift we demonstrate leads to a more robust and generalizable policy. To further enhance stability, CAA-Policy integrates short-horizon waypoint prediction as an auxiliary task, and introduces a separately trained motion prediction module to robustly track the target spot over time. Extensive experiments in the CARLA simulator show that \titlevariable~consistently surpasses both the end-to-end learning baseline and the modular BEV segmentation + hybrid A* pipeline, achieving superior accuracy, robustness, and interpretability. Code is released at https://github.com/Joechencc/CAAPolicy.
Authors: Zeyu Dong, Yuyang Yin, Yuqi Li, Eric Li, Hao-Xiang Guo, Yikai Wang
Abstract: Generating high-quality 360{\deg} panoramic videos remains a significant challenge due to the fundamental differences between panoramic and traditional perspective-view projections. While perspective videos rely on a single viewpoint with a limited field of view, panoramic content requires rendering the full surrounding environment, making it difficult for standard video generation models to adapt. Existing solutions often introduce complex architectures or large-scale training, leading to inefficiency and suboptimal results. Motivated by the success of Low-Rank Adaptation (LoRA) in style transfer tasks, we propose treating panoramic video generation as an adaptation problem from perspective views. Through theoretical analysis, we demonstrate that LoRA can effectively model the transformation between these projections when its rank exceeds the degrees of freedom in the task. Our approach efficiently fine-tunes a pretrained video diffusion model using only approximately 1,000 videos while achieving high-quality panoramic generation. Experimental results demonstrate that our method maintains proper projection geometry and surpasses previous state-of-the-art approaches in visual quality, left-right consistency, and motion diversity.
Authors: Ruiying Li, Bin Pan, Qiaoying Qu, Xia Xu, Zhenwei Shi
Abstract: The performance of hyperspectral unmixing may be constrained by low spatial resolution, which can be enhanced using super-resolution in a multitask learning way. However, integrating super-resolution and unmixing directly may suffer two challenges: Task affinity is not verified, and the convergence of unmixing is not guaranteed. To address the above issues, in this paper, we provide theoretical analysis and propose super-resolution guided multi-task learning method for hyperspectral unmixing (SMILE). The provided theoretical analysis validates feasibility of multitask learning way and verifies task affinity, which consists of relationship and existence theorems by proving the positive guidance of super-resolution. The proposed framework generalizes positive information from super-resolution to unmixing by learning both shared and specific representations. Moreover, to guarantee the convergence, we provide the accessibility theorem by proving the optimal solution of unmixing. The major contributions of SMILE include providing progressive theoretical support, and designing a new framework for unmixing under the guidance of super-resolution. Our experiments on both synthetic and real datasets have substantiate the usefulness of our work.
Authors: Ruiying Li, Bin Pan, Qiaoying Qu, Xia Xu, Zhenwei Shi
Abstract: Multitemporal hyperspectral unmixing (MTHU) aims to model variable endmembers and dynamical abundances, which emphasizes the critical temporal information. However, existing methods have limitations in modeling temporal dependency, thus fail to capture the dynamical material evolution. Motivated by the ability of copula theory in modeling dependency structure explicitly, in this paper, we propose a copula-guided temporal dependency method (Cog-TD) for multitemporal hyperspectral unmixing. Cog-TD defines new mathematical model, constructs copula-guided framework and provides two key modules with theoretical support. The mathematical model provides explicit formulations for MTHU problem definition, which describes temporal dependency structure by incorporating copula theory. The copula-guided framework is constructed for utilizing copula function, which estimates dynamical endmembers and abundances with temporal dependency. The key modules consist of copula function estimation and temporal dependency guidance, which computes and employs temporal information to guide unmixing process. Moreover, the theoretical support demonstrates that estimated copula function is valid and the represented temporal dependency exists in hyperspectral images. The major contributions of this paper include redefining MTHU problem with temporal dependency, proposing a copula-guided framework, developing two key modules and providing theoretical support. Our experimental results on both synthetic and real-world datasets demonstrate the utility of the proposed method.
Authors: Nhut Le, Ehsan Karimi, Maryam Rahnemoonfar
Abstract: Timely assessment of structural damage is critical for disaster response and recovery. However, most prior work in natural disaster analysis relies on 2D imagery, which lacks depth, suffers from occlusions, and provides limited spatial context. 3D semantic segmentation offers a richer alternative, but existing 3D benchmarks focus mainly on urban or indoor scenes, with little attention to disaster-affected areas. To address this gap, we present 3DAeroRelief--the first 3D benchmark dataset specifically designed for post-disaster assessment. Collected using low-cost unmanned aerial vehicles (UAVs) over hurricane-damaged regions, the dataset features dense 3D point clouds reconstructed via Structure-from-Motion and Multi-View Stereo techniques. Semantic annotations were produced through manual 2D labeling and projected into 3D space. Unlike existing datasets, 3DAeroRelief captures 3D large-scale outdoor environments with fine-grained structural damage in real-world disaster contexts. UAVs enable affordable, flexible, and safe data collection in hazardous areas, making them particularly well-suited for emergency scenarios. To demonstrate the utility of 3DAeroRelief, we evaluate several state-of-the-art 3D segmentation models on the dataset to highlight both the challenges and opportunities of 3D scene understanding in disaster response. Our dataset serves as a valuable resource for advancing robust 3D vision systems in real-world applications for post-disaster scenarios.
Authors: Nhi Kieu, Kien Nguyen, Arnold Wiliem, Clinton Fookes, Sridha Sridharan
Abstract: Multimodal learning has shown significant performance boost compared to ordinary unimodal models across various domains. However, in real-world scenarios, multimodal signals are susceptible to missing because of sensor failures and adverse weather conditions, which drastically deteriorates models' operation and performance. Generative models such as AutoEncoder (AE) and Generative Adversarial Network (GAN) are intuitive solutions aiming to reconstruct missing modality from available ones. Yet, their efficacy in remote sensing semantic segmentation remains underexplored. In this paper, we first examine the limitations of existing generative approaches in handling the heterogeneity of multimodal remote sensing data. They inadequately capture semantic context in complex scenes with large intra-class and small inter-class variation. In addition, traditional generative models are susceptible to heavy dependence on the dominant modality, introducing bias that affects model robustness under missing modality conditions. To tackle these limitations, we propose a novel Generative-Enhanced MultiModal learning Network (GEMMNet) with three key components: (1) Hybrid Feature Extractor (HyFEx) to effectively learn modality-specific representations, (2) Hybrid Fusion with Multiscale Awareness (HyFMA) to capture modality-synergistic semantic context across scales and (3) Complementary Loss (CoLoss) scheme to alleviate the inherent bias by encouraging consistency across modalities and tasks. Our method, GEMMNet, outperforms both generative baselines AE, cGAN (conditional GAN), and state-of-the-art non-generative approaches - mmformer and shaspec - on two challenging semantic segmentation remote sensing datasets (Vaihingen and Potsdam). Source code is made available.
Authors: Yuqiu Liu, Jialin Song, Manolis Savva, Wuyang Chen
Abstract: We propose a pipeline to extract and reconstruct dynamic 3D smoke assets from a single in-the-wild video, and further integrate interactive simulation for smoke design and editing. Recent developments in 3D vision have significantly improved reconstructing and rendering fluid dynamics, supporting realistic and temporally consistent view synthesis. However, current fluid reconstructions rely heavily on carefully controlled clean lab environments, whereas real-world videos captured in the wild are largely underexplored. We pinpoint three key challenges of reconstructing smoke in real-world videos and design targeted techniques, including smoke extraction with background removal, initialization of smoke particles and camera poses, and inferring multi-view videos. Our method not only outperforms previous reconstruction and generation methods with high-quality smoke reconstructions (+2.22 average PSNR on wild videos), but also enables diverse and realistic editing of fluid dynamics by simulating our smoke assets. We provide our models, data, and 4D smoke assets at [https://autumnyq.github.io/WildSmoke](https://autumnyq.github.io/WildSmoke).
URLs: https://autumnyq.github.io/WildSmoke](https://autumnyq.github.io/WildSmoke).
Authors: Ashkan Taghipour, Vahid Naghshin, Benjamin Southwell, Farid Boussaid, Hamid Laga, Mohammed Bennamoun
Abstract: 3D Gaussian Splatting (3DGS) enables fast, high-quality novel view synthesis but typically relies on densification followed by pruning to optimize the number of Gaussians. Existing mask-based pruning, such as MaskGS, regularizes the global mean of the mask, which is misaligned with the local per-pixel (per-ray) reconstruction loss that determines image quality along individual camera rays. This paper introduces SVR-GS, a spatially variant regularizer that renders a per-pixel spatial mask from each Gaussian's effective contribution along the ray, thereby applying sparsity pressure where it matters: on low-importance Gaussians. We explore three spatial-mask aggregation strategies, implement them in CUDA, and conduct a gradient analysis to motivate our final design. Extensive experiments on Tanks\&Temples, Deep Blending, and Mip-NeRF360 datasets demonstrate that, on average across the three datasets, the proposed SVR-GS reduces the number of Gaussians by 1.79\(\times\) compared to MaskGS and 5.63\(\times\) compared to 3DGS, while incurring only 0.50 dB and 0.40 dB PSNR drops, respectively. These gains translate into significantly smaller, faster, and more memory-efficient models, making them well-suited for real-time applications such as robotics, AR/VR, and mobile perception.
Authors: Diego Eustachio Farchione, Ramzi Idoughi, Peter Wonka
Abstract: Effective reef monitoring requires the quantification of coral growth via accurate volumetric and surface area estimates, which is a challenging task due to the complex morphology of corals. We propose a novel, lightweight, and scalable learning framework that addresses this challenge by predicting the 3D volume and surface area of coral-like objects from 2D multi-view RGB images. Our approach utilizes a pre-trained module (VGGT) to extract dense point maps from each view; these maps are merged into a unified point cloud and enriched with per-view confidence scores. The resulting cloud is fed to two parallel DGCNN decoder heads, which jointly output the volume and the surface area of the coral, as well as their corresponding confidence estimate. To enhance prediction stability and provide uncertainty estimates, we introduce a composite loss function based on Gaussian negative log-likelihood in both real and log domains. Our method achieves competitive accuracy and generalizes well to unseen morphologies. This framework paves the way for efficient and scalable coral geometry estimation directly from a sparse set of images, with potential applications in coral growth analysis and reef monitoring.
Authors: Waikit Xiu, Qiang Lu, Xiying Li, Chen Hu, Shengbo Sun
Abstract: As intelligent transportation systems advance, traffic video understanding plays an increasingly pivotal role in comprehensive scene perception and causal analysis. Yet, existing approaches face notable challenges in accurately modeling spatiotemporal causality and integrating domain-specific knowledge, limiting their effectiveness in complex scenarios. To address these limitations, we propose Traffic-MLLM, a multimodal large language model tailored for fine-grained traffic analysis. Built on the Qwen2.5-VL backbone, our model leverages high-quality traffic-specific multimodal datasets and uses Low-Rank Adaptation (LoRA) for lightweight fine-tuning, significantly enhancing its capacity to model continuous spatiotemporal features in video sequences. Furthermore, we introduce an innovative knowledge prompting module fusing Chain-of-Thought (CoT) reasoning with Retrieval-Augmented Generation (RAG), enabling precise injection of detailed traffic regulations and domain knowledge into the inference process. This design markedly boosts the model's logical reasoning and knowledge adaptation capabilities. Experimental results on TrafficQA and DriveQA benchmarks show Traffic-MLLM achieves state-of-the-art performance, validating its superior ability to process multimodal traffic data. It also exhibits remarkable zero-shot reasoning and cross-scenario generalization capabilities.
Authors: Hong Zhang, Fei Guo, Zihan Xie, Dizhao Yao
Abstract: 3D reconstruction technology generates three-dimensional representations of real-world objects, scenes, or environments using sensor data such as 2D images, with extensive applications in robotics, autonomous vehicles, and virtual reality systems. Traditional 3D reconstruction techniques based on 2D images typically relies on RGB spectral information. With advances in sensor technology, additional spectral bands beyond RGB have been increasingly incorporated into 3D reconstruction workflows. Existing methods that integrate these expanded spectral data often suffer from expensive scheme prices, low accuracy and poor geometric features. Three - dimensional reconstruction based on NeRF can effectively address the various issues in current multispectral 3D reconstruction methods, producing high - precision and high - quality reconstruction results. However, currently, NeRF and some improved models such as NeRFacto are trained on three - band data and cannot take into account the multi - band information. To address this problem, we propose Multispectral-NeRF, an enhanced neural architecture derived from NeRF that can effectively integrates multispectral information. Our technical contributions comprise threefold modifications: Expanding hidden layer dimensionality to accommodate 6-band spectral inputs; Redesigning residual functions to optimize spectral discrepancy calculations between reconstructed and reference images; Adapting data compression modules to address the increased bit-depth requirements of multispectral imagery. Experimental results confirm that Multispectral-NeRF successfully processes multi-band spectral features while accurately preserving the original scenes' spectral characteristics.
Authors: Zhiwen Yang, Yuxin Peng
Abstract: Camera-based 3D Semantic Scene Completion (SSC) is a critical task in autonomous driving systems, assessing voxel-level geometry and semantics for holistic scene perception. While existing voxel-based and plane-based SSC methods have achieved considerable progress, they struggle to capture physical regularities for realistic geometric details. On the other hand, neural reconstruction methods like NeRF and 3DGS demonstrate superior physical awareness, but suffer from high computational cost and slow convergence when handling large-scale, complex autonomous driving scenes, leading to inferior semantic accuracy. To address these issues, we propose the Semantic-PHysical Engaged REpresentation (SPHERE) for camera-based SSC, which integrates voxel and Gaussian representations for joint exploitation of semantic and physical information. First, the Semantic-guided Gaussian Initialization (SGI) module leverages dual-branch 3D scene representations to locate focal voxels as anchors to guide efficient Gaussian initialization. Then, the Physical-aware Harmonics Enhancement (PHE) module incorporates semantic spherical harmonics to model physical-aware contextual details and promote semantic-geometry consistency through focal distribution alignment, generating SSC results with realistic details. Extensive experiments and analyses on the popular SemanticKITTI and SSCBench-KITTI-360 benchmarks validate the effectiveness of SPHERE. The code is available at https://github.com/PKU-ICST-MIPL/SPHERE_ACMMM2025.
Authors: Chengde Lin, Xuezhu Gong, Shuxue Ding, Mingzhe Yang, Xijun Lu, Chengjun Mo
Abstract: Image hiding is often referred to as steganography, which aims to hide a secret image in a cover image of the same resolution. Many steganography models are based on genera-tive adversarial networks (GANs) and variational autoencoders (VAEs). However, most existing models suffer from mode collapse. Mode collapse will lead to an information imbalance between the cover and secret images in the stego image and further affect the subsequent extraction. To address these challenges, this paper proposes StegOT, an autoencoder-based steganography model incorporating optimal transport theory. We designed the multiple channel optimal transport (MCOT) module to transform the feature distribution, which exhibits multiple peaks, into a single peak to achieve the trade-off of information. Experiments demonstrate that we not only achieve a trade-off between the cover and secret images but also enhance the quality of both the stego and recovery images. The source code will be released on https://github.com/Rss1124/StegOT.
Authors: Partha Shah, Durva Sankhe, Maariyah Rashid, Zakaa Khaled, Esther Puyol-Ant\'on, Tiarna Lee, Maram Alqarni, Sweta Rai, Andrew P. King
Abstract: Artificial intelligence (AI) models to automatically classify skin lesions from dermatology images have shown promising performance but also susceptibility to bias by skin tone. The most common way of representing skin tone information is the Fitzpatrick Skin Tone (FST) scale. The FST scale has been criticised for having greater granularity in its skin tone categories for lighter-skinned subjects. This paper conducts an investigation of the impact (on performance and bias) on AI classification models of granularity in the FST scale. By training multiple AI models to classify benign vs. malignant lesions using FST-specific data of differing granularity, we show that: (i) when training models using FST-specific data based on three groups (FST 1/2, 3/4 and 5/6), performance is generally better for models trained on FST-specific data compared to a general model trained on FST-balanced data; (ii) reducing the granularity of FST scale information (from 1/2 and 3/4 to 1/2/3/4) can have a detrimental effect on performance. Our results highlight the importance of the granularity of FST groups when training lesion classification models. Given the question marks over possible human biases in the choice of categories in the FST scale, this paper provides evidence for a move away from the FST scale in fair AI research and a transition to an alternative scale that better represents the diversity of human skin tones.
Authors: Yihang She, Andrew Blake, David Coomes, Srinivasan Keshav
Abstract: Accurate tree segmentation is a key step in extracting individual tree metrics from forest laser scans, and is essential to understanding ecosystem functions in carbon cycling and beyond. Over the past decade, tree segmentation algorithms have advanced rapidly due to developments in AI. However existing, public, 3D forest datasets are not large enough to build robust tree segmentation systems. Motivated by the success of synthetic data in other domains such as self-driving, we investigate whether similar approaches can help with tree segmentation. In place of expensive field data collection and annotation, we use synthetic data during pretraining, and then require only minimal, real forest plot annotation for fine-tuning. We have developed a new synthetic data generation pipeline to do this for forest vision tasks, integrating advances in game-engines with physics-based LiDAR simulation. As a result, we have produced a comprehensive, diverse, annotated 3D forest dataset on an unprecedented scale. Extensive experiments with a state-of-the-art tree segmentation algorithm and a popular real dataset show that our synthetic data can substantially reduce the need for labelled real data. After fine-tuning on just a single, real, forest plot of less than 0.1 hectare, the pretrained model achieves segmentations that are competitive with a model trained on the full scale real data. We have also identified critical factors for successful use of synthetic data: physics, diversity, and scale, paving the way for more robust 3D forest vision systems in the future. Our data generation pipeline and the resulting dataset are available at https://github.com/yihshe/CAMP3D.git.
Authors: Yufei Tang, Daiheng Gao, Pingyu Wu, Wenbo Zhou, Bang Zhang, Weiming Zhang
Abstract: In the realm of image generation, the quest for realism and customization has never been more pressing. While existing methods like concept sliders have made strides, they often falter when it comes to no-AIGC images, particularly images captured in real world settings. To bridge this gap, we introduce Beyond Sliders, an innovative framework that integrates GANs and diffusion models to facilitate sophisticated image manipulation across diverse image categories. Improved upon concept sliders, our method refines the image through fine grained guidance both textual and visual in an adversarial manner, leading to a marked enhancement in image quality and realism. Extensive experimental validation confirms the robustness and versatility of Beyond Sliders across a spectrum of applications.
Authors: Johann Schmidt, Sebastian Stober
Abstract: Fine-grained visual classification (FGVC) remains highly sensitive to geometric variability, where objects appear under arbitrary orientations, scales, and perspective distortions. While equivariant architectures address this issue, they typically require substantial computational resources and restrict the hypothesis space. We revisit Spatial Transformer Networks (STNs) as a canonicalization tool for transformer-based vision pipelines, emphasizing their flexibility, backbone-agnostic nature, and lack of architectural constraints. We propose a probabilistic, component-wise extension that improves robustness. Specifically, we decompose affine transformations into rotation, scaling, and shearing, and regress each component under geometric constraints using a shared localization encoder. To capture uncertainty, we model each component with a Gaussian variational posterior and perform sampling-based canonicalization during inference.A novel component-wise alignment loss leverages augmentation parameters to guide spatial alignment. Experiments on challenging moth classification benchmarks demonstrate that our method consistently improves robustness compared to other STNs.
Authors: Rabin Dulal, Lihong Zheng, Ashad Kabir
Abstract: Cattle identification is critical for efficient livestock farming management, currently reliant on radio-frequency identification (RFID) ear tags. However, RFID-based systems are prone to failure due to loss, damage, tampering, and vulnerability to external attacks. As a robust alternative, biometric identification using cattle muzzle patterns similar to human fingerprints has emerged as a promising solution. Deep learning techniques have demonstrated success in leveraging these unique patterns for accurate identification. But deep learning models face significant challenges, including limited data availability, disruptions during data collection, and dynamic herd compositions that require frequent model retraining. To address these limitations, this paper proposes a novel few-shot learning framework for real-time cattle identification using Cooperative Model-Agnostic Meta-Learning (CCoMAML) with Multi-Head Attention Feature Fusion (MHAFF) as a feature extractor model. This model offers great model adaptability to new data through efficient learning from few data samples without retraining. The proposed approach has been rigorously evaluated against current state-of-the-art few-shot learning techniques applied in cattle identification. Comprehensive experimental results demonstrate that our proposed CCoMAML with MHAFF has superior cattle identification performance with 98.46% and 97.91% F1 scores.
Authors: Gao Yu Lee, Tanmoy Dam, Md Meftahul Ferdaus, Daniel Puiu Poenar, Vu N. Duong
Abstract: Few-Shot Learning (FSL), which involves learning to generalize using only a few data samples, has demonstrated promising and superior performances to ordinary CNN methods. While Bayesian based estimation approaches using Kullback-Leibler (KL) divergence have shown improvements, they remain vulnerable to adversarial attacks and natural noises. We introduce ANROT-HELANet, an Adversarially and Naturally RObusT Hellinger Aggregation Network that significantly advances the state-of-the-art in FSL robustness and performance. Our approach implements an adversarially and naturally robust Hellinger distance-based feature class aggregation scheme, demonstrating resilience to adversarial perturbations up to $\epsilon=0.30$ and Gaussian noise up to $\sigma=0.30$. The network achieves substantial improvements across benchmark datasets, including gains of 1.20\% and 1.40\% for 1-shot and 5-shot scenarios on miniImageNet respectively. We introduce a novel Hellinger Similarity contrastive loss function that generalizes cosine similarity contrastive loss for variational few-shot inference scenarios. Our approach also achieves superior image reconstruction quality with a FID score of 2.75, outperforming traditional VAE (3.43) and WAE (3.38) approaches. Extensive experiments conducted on four few-shot benchmarked datasets verify that ANROT-HELANet's combination of Hellinger distance-based feature aggregation, attention mechanisms, and our novel loss function establishes new state-of-the-art performance while maintaining robustness against both adversarial and natural perturbations. Our code repository will be available at https://github.com/GreedYLearner1146/ANROT-HELANet/tree/main.
URLs: https://github.com/GreedYLearner1146/ANROT-HELANet/tree/main.
Authors: Seongwan Park, Jieun Woo, Siheon Yang
Abstract: This paper presents MIS-LSTM, a hybrid framework that joins CNN encoders with an LSTM sequence model for sleep quality and stress prediction at the day level from multimodal lifelog data. Continuous sensor streams are first partitioned into N-hour blocks and rendered as multi-channel images, while sparse discrete events are encoded with a dedicated 1D-CNN. A Convolutional Block Attention Module fuses the two modalities into refined block embeddings, which an LSTM then aggregates to capture long-range temporal dependencies. To further boost robustness, we introduce UALRE, an uncertainty-aware ensemble that overrides lowconfidence majority votes with high-confidence individual predictions. Experiments on the 2025 ETRI Lifelog Challenge dataset show that Our base MISLSTM achieves Macro-F1 0.615; with the UALRE ensemble, the score improves to 0.647, outperforming strong LSTM, 1D-CNN, and CNN baselines. Ablations confirm (i) the superiority of multi-channel over stacked-vertical imaging, (ii) the benefit of a 4-hour block granularity, and (iii) the efficacy of modality-specific discrete encoding.
Authors: Robert Long, Rongxin Jiang, Mingrui Yan
Abstract: Person Re-Identification (ReID) has several challenges in real-world surveillance systems due to clothing changes (CCReID) and the need for maintaining continual learning (LReID). Previous existing methods either develop models specifically for one application, which is mostly a same-cloth (SC) setting or treat CCReID as its own separate sub-problem. In this work, we will introduce the LReID-Hybrid task with the goal of developing a model to achieve both SC and CC while learning in a continual setting. Mismatched representations and forgetting from one task to the next are significant issues, we address this with CMLReID, a CLIP-based framework composed of two novel tasks: (1) Context-Aware Semantic Prompt (CASP) that generates adaptive prompts, and also incorporates context to align richly multi-grained visual cues with semantic text space; and (2) Adaptive Knowledge Fusion and Projection (AKFP) which produces robust SC/CC prototypes through the use of a dual-path learner that aligns features with our Clothing-State-Aware Projection Loss. Experiments performed on a wide range of datasets and illustrate that CMLReID outperforms all state-of-the-art methods with strong robustness and generalization despite clothing variations and a sophisticated process of sequential learning.
Authors: Kerun Mi, Guoliang Kang, Guangyu Li, Lin Zhao, Tao Zhou, Chen Gong
Abstract: Class-Incremental Unsupervised Domain Adaptation (CI-UDA) aims to adapt a model from a labeled source domain to an unlabeled target domain, where the sets of potential target classes appearing at different time steps are disjoint and are subsets of the source classes. The key to solving this problem lies in avoiding catastrophic forgetting of knowledge about previous target classes during continuously mitigating the domain shift. Most previous works cumbersomely combine two technical components. On one hand, they need to store and utilize rehearsal target sample from previous time steps to avoid catastrophic forgetting; on the other hand, they perform alignment only between classes shared across domains at each time step. Consequently, the memory will continuously increase and the asymmetric alignment may inevitably result in knowledge forgetting. In this paper, we propose to mine and preserve domain-invariant and class-agnostic knowledge to facilitate the CI-UDA task. Specifically, via using CLIP, we extract the class-agnostic properties which we name as "attribute". In our framework, we learn a "key-value" pair to represent an attribute, where the key corresponds to the visual prototype and the value is the textual prompt. We maintain two attribute dictionaries, each corresponding to a different domain. Then we perform attribute alignment across domains to mitigate the domain shift, via encouraging visual attention consistency and prediction consistency. Through attribute modeling and cross-domain alignment, we effectively reduce catastrophic knowledge forgetting while mitigating the domain shift, in a rehearsal-free way. Experiments on three CI-UDA benchmarks demonstrate that our method outperforms previous state-of-the-art methods and effectively alleviates catastrophic forgetting. Code is available at https://github.com/RyunMi/VisTA.
Authors: Zhihang Song, Dingyi Yao, Ruibo Ming, Lihui Peng, Danya Yao, Yi Zhang
Abstract: With the rapid advancement of synthetic dataset generation techniques, evaluating the quality of synthetic data has become a critical research focus. Robust evaluation not only drives innovations in data generation methods but also guides researchers in optimizing the utilization of these synthetic resources. However, current evaluation studies for synthetic datasets remain limited, lacking a universally accepted standard framework. To address this, this paper proposes a novel evaluation framework integrating generalized cross-validation experiments and domain transfer learning principles, enabling generalizable and comparable assessments of synthetic dataset quality. The framework involves training task-specific models (e.g., YOLOv5s) on both synthetic datasets and multiple real-world benchmarks (e.g., KITTI, BDD100K), forming a cross-performance matrix. Following normalization, a Generalized Cross-Validation (GCV) Matrix is constructed to quantify domain transferability. The framework introduces two key metrics. One measures the simulation quality by quantifying the similarity between synthetic data and real-world datasets, while another evaluates the transfer quality by assessing the diversity and coverage of synthetic data across various real-world scenarios. Experimental validation on Virtual KITTI demonstrates the effectiveness of our proposed framework and metrics in assessing synthetic data fidelity. This scalable and quantifiable evaluation solution overcomes traditional limitations, providing a principled approach to guide synthetic dataset optimization in artificial intelligence research.
Authors: Lianjun Liao, Chunhui Zhang, Tong Wu, Henglei Lv, Bailin Deng, Lin Gao
Abstract: Image data captured outdoors often exhibit unbounded scenes and unconstrained, varying lighting conditions, making it challenging to decompose them into geometry, reflectance, and illumination. Recent works have focused on achieving this decomposition using Neural Radiance Fields (NeRF) or the 3D Gaussian Splatting (3DGS) representation but remain hindered by two key limitations: the high computational overhead associated with neural networks of NeRF and the use of low-frequency lighting representations, which often result in inefficient rendering and suboptimal relighting accuracy. We propose ROSGS, a two-stage pipeline designed to efficiently reconstruct relightable outdoor scenes using the Gaussian Splatting representation. By leveraging monocular normal priors, ROSGS first reconstructs the scene's geometry with the compact 2D Gaussian Splatting (2DGS) representation, providing an efficient and accurate geometric foundation. Building upon this reconstructed geometry, ROSGS then decomposes the scene's texture and lighting through a hybrid lighting model. This model effectively represents typical outdoor lighting by employing a spherical Gaussian function to capture the directional, high-frequency components of sunlight, while learning a radiance transfer function via Spherical Harmonic coefficients to model the remaining low-frequency skylight comprehensively. Both quantitative metrics and qualitative comparisons demonstrate that ROSGS achieves state-of-the-art performance in relighting outdoor scenes and highlight its ability to deliver superior relighting accuracy and rendering efficiency.
Authors: Yifan Lu, Ziqi Zhang, Chunfeng Yuan, Jun Gao, Congxuan Zhang, Xiaojuan Qi, Bing Li, Weiming Hu
Abstract: Large Vision-Language Models (LVLMs) suffer from serious hallucination problems, where the model-generated responses are inconsistent with the visual inputs. Existing hallucination mitigation methods are mainly based on preference alignment and require external human annotations or auxiliary models for preference data collection, which increase costs and limit sustainable improvement. To tackle these challenges, we propose Autonomous Preference Alignment via Self-Injection (APASI), a novel and generalizable method that mitigates hallucinations without external dependencies. APASI leverages the target LVLM to self-inject hallucinations into a generated response, creating a pair of responses with varying preference levels. During the self-injection process, the dis-preferred response is generated based on three key observations of hallucinations, ensuring it simulates real hallucination patterns. This fidelity offers an accurate learning signal for hallucination mitigation. Moreover, APASI incorporates an iterative alignment training strategy combined with curriculum learning to periodically update the preference data with increasing challenge, enabling stable and continuous enhancement of the LVLM. Extensive experiments across six benchmarks show that APASI not only effectively mitigates hallucinations for three baseline models but also achieves comparable or even superior performance to alignment-based methods with external dependency, thereby demonstrating its effectiveness and generalization capability. The code is available at https://github.com/davidluciolu/APASI.
Authors: Ziling Liu, Ziwei Chen, Mingqi Gao, Jinyu Yang, Feng Zheng
Abstract: Unaligned Scene Change Detection aims to detect scene changes between image pairs captured at different times without assuming viewpoint alignment. To handle viewpoint variations, current methods rely solely on 2D visual cues to establish cross-image correspondence to assist change detection. However, large viewpoint changes can alter visual observations, causing appearance-based matching to drift or fail. Additionally, supervision limited to 2D change masks from small-scale SCD datasets restricts the learning of generalizable multi-view knowledge, making it difficult to reliably identify visual overlaps and handle occlusions. This lack of explicit geometric reasoning represents a critical yet overlooked limitation. In this work, we are the first to leverage geometric priors from a Geometric Foundation Model to address the core challenges of unaligned SCD, including reliable identification of visual overlaps, robust correspondence establishment, and explicit occlusion detection. Building on these priors, we propose a training-free framework that integrates them with the powerful representations of a visual foundation model to enable reliable change detection under viewpoint misalignment. Through extensive evaluation on the PSCD, ChangeSim, and PASLCD datasets, we demonstrate that our approach achieves superior and robust performance. Our code will be released at https://github.com/ZilingLiu/GeoSCD.
Authors: Matthias W\"uest, Francis Engelmann, Ondrej Miksik, Marc Pollefeys, Daniel Barath
Abstract: We propose UnLoc, an efficient data-driven solution for sequential camera localization within floorplans. Floorplan data is readily available, long-term persistent, and robust to changes in visual appearance. We address key limitations of recent methods, such as the lack of uncertainty modeling in depth predictions and the necessity for custom depth networks trained for each environment. We introduce a novel probabilistic model that incorporates uncertainty estimation, modeling depth predictions as explicit probability distributions. By leveraging off-the-shelf pre-trained monocular depth models, we eliminate the need to rely on per-environment-trained depth networks, enhancing generalization to unseen spaces. We evaluate UnLoc on large-scale synthetic and real-world datasets, demonstrating significant improvements over existing methods in terms of accuracy and robustness. Notably, we achieve $2.7$ times higher localization recall on long sequences (100 frames) and $16.7$ times higher on short ones (15 frames) than the state of the art on the challenging LaMAR HGE dataset.
Authors: Jian Song, Wei Mei, Yunfeng Xu, Qiang Fu, Renke Kou, Lina Bu, Yucheng Long
Abstract: Motion estimation is a crucial component in multi-object tracking (MOT). It predicts the trajectory of objects by analyzing the changes in their positions in consecutive frames of images, reducing tracking failures and identity switches. The Kalman filter (KF) based on the linear constant-velocity model is one of the most commonly used methods in MOT. However, it may yield unsatisfactory results when KF's parameters are mismatched and objects move in non-stationary. In this work, we utilize the learning-aided filter to handle the motion estimation of MOT. In particular, we propose a novel method named Semantic-Independent KalmanNet (SIKNet), which encodes the state vector (the input feature) using a Semantic-Independent Encoder (SIE) by two steps. First, the SIE uses a 1D convolution with a kernel size of 1, which convolves along the dimension of homogeneous-semantic elements across different state vectors to encode independent semantic information. Then it employs a fully-connected layer and a nonlinear activation layer to encode nonlinear and cross-dependency information between heterogeneous-semantic elements. To independently evaluate the performance of the motion estimation module in MOT, we constructed a large-scale semi-simulated dataset from several open-source MOT datasets. Experimental results demonstrate that the proposed SIKNet outperforms the traditional KF and achieves superior robustness and accuracy than existing learning-aided filters. The code is available at (https://github.com/SongJgit/filternet and https://github.com/SongJgit/TBDTracker).
URLs: https://github.com/SongJgit/filternet, https://github.com/SongJgit/TBDTracker).
Authors: Mingyuan Meng
Abstract: Medical Image Computing (MIC) is a broad research topic covering both pixel-wise (e.g., segmentation, registration) and image-wise (e.g., classification, regression) vision tasks. Effective analysis demands models that capture both global long-range context and local subtle visual characteristics, necessitating fine-grained long-range visual dependency modeling. Compared to Convolutional Neural Networks (CNNs) that are limited by intrinsic locality, transformers excel at long-range modeling; however, due to the high computational loads of self-attention, transformers typically cannot process high-resolution features (e.g., full-scale image features before downsampling or patch embedding) and thus face difficulties in modeling fine-grained dependency among subtle medical image details. Concurrently, Multi-layer Perceptron (MLP)-based visual models are recognized as computation/memory-efficient alternatives in modeling long-range visual dependency but have yet to be widely investigated in the MIC community. This doctoral research advances deep learning-based MIC by investigating effective long-range visual dependency modeling. It first presents innovative use of transformers for both pixel- and image-wise medical vision tasks. The focus then shifts to MLPs, pioneeringly developing MLP-based visual models to capture fine-grained long-range visual dependency in medical images. Extensive experiments confirm the critical role of long-range dependency modeling in MIC and reveal a key finding: MLPs provide feasibility in modeling finer-grained long-range dependency among higher-resolution medical features containing enriched anatomical/pathological details. This finding establishes MLPs as a superior paradigm over transformers/CNNs, consistently enhancing performance across various medical vision tasks and paving the way for next-generation medical vision backbones.
Authors: Sriram Narayanan, Mani Ramanagopal, Srinivasa G. Narasimhan
Abstract: Long-wave infrared radiation captured by a thermal camera consists of two components: (a) light from the environment reflected or transmitted by a surface, and (b) light emitted by the surface after undergoing heat transport through the object and exchanging heat with the surrounding environment. Separating these components is essential for understanding object properties such as emissivity, temperature, reflectance and shape. Previous thermography studies often assume that only one component is dominant (e.g., in welding) or that the second component is constant and can be subtracted. However, in near-ambient conditions, which are most relevant to computer vision applications, both components are typically comparable in magnitude and vary over time. We introduce the first method that separates reflected and emitted components of light in videos captured by two thermal cameras with different spectral sensitivities. We derive a dual-band thermal image formation model and develop algorithms to estimate the surface's emissivity and its time-varying temperature while isolating a dynamic background. We quantitatively evaluate our approach using carefully calibrated emissivities for a range of materials and show qualitative results on complex everyday scenes, such as a glass filled with hot liquid and people moving in the background.
Authors: Huaiyuan Qin, Muli Yang, Siyuan Hu, Peng Hu, Yu Zhang, Chen Gong, Hongyuan Zhu
Abstract: Self-supervised learning (SSL) conventionally relies on the instance consistency paradigm, assuming that different views of the same image can be treated as positive pairs. However, this assumption breaks down for non-iconic data, where different views may contain distinct objects or semantic information. In this paper, we investigate the effectiveness of SSL when instance consistency is not guaranteed. Through extensive ablation studies, we demonstrate that SSL can still learn meaningful representations even when positive pairs lack strict instance consistency. Furthermore, our analysis further reveals that increasing view diversity, by enforcing zero overlapping or using smaller crop scales, can enhance downstream performance on classification and dense prediction tasks. However, excessive diversity is found to reduce effectiveness, suggesting an optimal range for view diversity. To quantify this, we adopt the Earth Mover's Distance (EMD) as an estimator to measure mutual information between views, finding that moderate EMD values correlate with improved SSL learning, providing insights for future SSL framework design. We validate our findings across a range of settings, highlighting their robustness and applicability on diverse data sources.
Authors: Robin Narsingh Ranabhat, Longwei Wang, Amit Kumar Patel, KC santosh
Abstract: Convolutional Neural Networks (CNNs) excel at image classification but remain vulnerable to common corruptions that humans handle with ease. A key reason for this fragility is their reliance on local texture cues rather than global object shapes -- a stark contrast to human perception. To address this, we propose two complementary regularization strategies designed to encourage shape-biased representations and enhance robustness. The first introduces an auxiliary loss that enforces feature consistency between original and low-frequency filtered inputs, discouraging dependence on high-frequency textures. The second incorporates supervised contrastive learning to structure the feature space around class-consistent, shape-relevant representations. Evaluated on the CIFAR-10-C benchmark, both methods improve corruption robustness without degrading clean accuracy. Our results suggest that loss-level regularization can effectively steer CNNs toward more shape-aware, resilient representations.
Authors: Wan Xu, Feng Zhu, Yihan Zeng, Yuanfan Guo, Ming Liu, Hang Xu, Wangmeng Zuo
Abstract: Video detailed captioning aims to generate comprehensive video descriptions to facilitate video understanding. Recently, most efforts in the video detailed captioning community have been made towards a local-to-global paradigm, which first generates local captions from video clips and then summarizes them into a global caption. However, we find this paradigm leads to less detailed and contextual-inconsistent captions, which can be attributed to (1) no mechanism to ensure fine-grained captions, and (2) weak interaction between local and global captions. To remedy the above two issues, we propose GLaVE-Cap, a Global-Local aligned framework with Vision Expert integration for Captioning, which consists of two core modules: TrackFusion enables comprehensive local caption generation, by leveraging vision experts to acquire cross-frame visual prompts, coupled with a dual-stream structure; while CaptionBridge establishes a local-global interaction, by using global context to guide local captioning, and adaptively summarizing local captions into a coherent global caption. Besides, we construct GLaVE-Bench, a comprehensive video captioning benchmark featuring 5X more queries per video than existing benchmarks, covering diverse visual dimensions to facilitate reliable evaluation. We further provide a training dataset GLaVE-1.2M containing 16K high-quality fine-grained video captions and 1.2M related question-answer pairs. Extensive experiments on four benchmarks show that our GLaVE-Cap achieves state-of-the-art performance. Besides, the ablation studies and student model analyses further validate the effectiveness of the proposed modules and the contribution of GLaVE-1.2M to the video understanding community. The source code, model weights, benchmark, and dataset will be open-sourced.
Authors: Akhil Padmanabha, Arpit Agarwal, Catherine Li, Austin Williams, Dinesh K. Patel, Sankalp Chopkar, Achu Wilson, Ahmet Ozkan, Wenzhen Yuan, Sonal Choudhary, Arash Mostaghimi, Zackory Erickson, Carmel Majidi
Abstract: Three-dimensional (3-D) skin surface reconstruction offers promise for objective and quantitative dermatological assessment, but no portable, high-resolution device exists that has been validated and used for depth reconstruction across various body locations. We present a compact 3-D skin reconstruction probe based on GelSight tactile imaging with a custom elastic gel and a learning-based reconstruction algorithm for micron-level wrinkle height estimation. Our probe, integrated into a handheld probe with force sensing for consistent contact, achieves a mean absolute error of 12.55 micron on wrinkle-like test objects. In a study with 15 participants without skin disorders, we provide the first validated wrinkle depth metrics across multiple body regions. We further demonstrate statistically significant reductions in wrinkle height at three locations following over-the-counter moisturizer application. Our work offers a validated tool for clinical and cosmetic skin analysis, with potential applications in diagnosis, treatment monitoring, and skincare efficacy evaluation.
Authors: Syed Talal Wasim, Hamid Suleman, Olga Zatsarynna, Muzammal Naseer, Juergen Gall
Abstract: We present MixANT, a novel architecture for stochastic long-term dense anticipation of human activities. While recent State Space Models (SSMs) like Mamba have shown promise through input-dependent selectivity on three key parameters, the critical forget-gate ($\textbf{A}$ matrix) controlling temporal memory remains static. We address this limitation by introducing a mixture of experts approach that dynamically selects contextually relevant $\textbf{A}$ matrices based on input features, enhancing representational capacity without sacrificing computational efficiency. Extensive experiments on the 50Salads, Breakfast, and Assembly101 datasets demonstrate that MixANT consistently outperforms state-of-the-art methods across all evaluation settings. Our results highlight the importance of input-dependent forget-gate mechanisms for reliable prediction of human behavior in diverse real-world scenarios.
Authors: Christoph F\"urb\"ock, Paul Weiser, Branko Mitic, Philipp Seeb\"ock, Thomas Helbich, Georg Langs
Abstract: In real world clinical environments, training and applying deep learning models on multi-modal medical imaging data often struggles with partially incomplete data. Standard approaches either discard missing samples, require imputation or repurpose dropout learning schemes, limiting robustness and generalizability. To address this, we propose a hypernetwork-based method that dynamically generates task-specific classification models conditioned on the set of available modalities. Instead of training a fixed model, a hypernetwork learns to predict the parameters of a task model adapted to available modalities, enabling training and inference on all samples, regardless of completeness. We compare this approach with (1) models trained only on complete data, (2) state of the art channel dropout methods, and (3) an imputation-based method, using artificially incomplete datasets to systematically analyze robustness to missing modalities. Results demonstrate superior adaptability of our method, outperforming state of the art approaches with an absolute increase in accuracy of up to 8% when trained on a dataset with 25% completeness (75% of training data with missing modalities). By enabling a single model to generalize across all modality configurations, our approach provides an efficient solution for real-world multi-modal medical data analysis.
Authors: Nikolaos Zioulis, Nikolaos Kotarelas, Georgios Albanis, Spyridon Thermos, Anargyros Chatzitofis
Abstract: Radiance field-based methods have recently been used to reconstruct human avatars, showing that we can significantly downscale the systems needed for creating animated human avatars. Although this progress has been initiated by neural radiance fields, their slow rendering and backward mapping from the observation space to the canonical space have been the main challenges. With Gaussian splatting overcoming both challenges, a new family of approaches has emerged that are faster to train and render, while also straightforward to implement using forward skinning from the canonical to the observation space. However, the linear blend skinning required for the deformation of the Gaussians does not provide valid results for their non-linear rotation properties. To address such artifacts, recent works use mesh properties to rotate the non-linear Gaussian properties or train models to predict corrective offsets. Instead, we propose a weighted rotation blending approach that leverages quaternion averaging. This leads to simpler vertex-based Gaussians that can be efficiently animated and integrated in any engine by only modifying the linear blend skinning technique, and using any Gaussian rasterizer.
Authors: Jeanny Pan, Philipp Seeb\"ock, Christoph F\"urb\"ock, Svitlana Pochepnia, Jennifer Straub, Lucian Beer, Helmut Prosch, Georg Langs
Abstract: Identifying new disease-related patterns in medical imaging data with the help of machine learning enlarges the vocabulary of recognizable findings. This supports diagnostic and prognostic assessment. However, image appearance varies not only due to biological differences, but also due to imaging technology linked to vendors, scanning- or re- construction parameters. The resulting domain shifts impedes data representation learning strategies and the discovery of biologically meaningful cluster appearances. To address these challenges, we introduce an approach to actively learn the domain shift via post-hoc rotation of the data latent space, enabling disentanglement of biological and technical factors. Results on real-world heterogeneous clinical data showcase that the learned disentangled representation leads to stable clusters representing tissue-types across different acquisition settings. Cluster consistency is improved by +19.01% (ARI), +16.85% (NMI), and +12.39% (Dice) compared to the entangled representation, outperforming four state-of-the-art harmonization methods. When using the clusters to quantify tissue composition on idiopathic pulmonary fibrosis patients, the learned profiles enhance Cox survival prediction. This indicates that the proposed label-free framework facilitates biomarker discovery in multi-center routine imaging data. Code is available on GitHub https://github.com/cirmuw/latent-space-rotation-disentanglement.
URLs: https://github.com/cirmuw/latent-space-rotation-disentanglement.
Authors: Ayhan Can Erdur, Christian Beischl, Daniel Scholz, Jiazhen Pan, Benedikt Wiestler, Daniel Rueckert, Jan C Peeken
Abstract: Missing input sequences are common in medical imaging data, posing a challenge for deep learning models reliant on complete input data. In this work, inspired by MultiMAE [2], we develop a masked autoencoder (MAE) paradigm for multi-modal, multi-task learning in 3D medical imaging with brain MRIs. Our method treats each MRI sequence as a separate input modality, leveraging a late-fusion-style transformer encoder to integrate multi-sequence information (multi-modal) and individual decoder streams for each modality for multi-task reconstruction. This pretraining strategy guides the model to learn rich representations per modality while also equipping it to handle missing inputs through cross-sequence reasoning. The result is a flexible and generalizable encoder for brain MRIs that infers missing sequences from available inputs and can be adapted to various downstream applications. We demonstrate the performance and robustness of our method against an MAE-ViT baseline in downstream segmentation and classification tasks, showing absolute improvement of $10.1$ overall Dice score and $0.46$ MCC over the baselines with missing input sequences. Our experiments demonstrate the strength of this pretraining strategy. The implementation is made available.
Authors: BaiChen Fan, Sifan Zhou, Jian Li, Shibo Zhao, Muqing Cao, Qin Wang
Abstract: LiDAR-based 3D single object tracking (3D SOT) is a critical task in robotics and autonomous systems. Existing methods typically follow frame-wise motion estimation or a sequence-based paradigm. However, the two-frame methods are efficient but lack long-term temporal context, making them vulnerable in sparse or occluded scenes, while sequence-based methods that process multiple point clouds gain robustness at a significant computational cost. To resolve this dilemma, we propose a novel trajectory-based paradigm and its instantiation, TrajTrack. TrajTrack is a lightweight framework that enhances a base two-frame tracker by implicitly learning motion continuity from historical bounding box trajectories alone-without requiring additional, costly point cloud inputs. It first generates a fast, explicit motion proposal and then uses an implicit motion modeling module to predict the future trajectory, which in turn refines and corrects the initial proposal. Extensive experiments on the large-scale NuScenes benchmark show that TrajTrack achieves new state-of-the-art performance, dramatically improving tracking precision by 4.48% over a strong baseline while running at 56 FPS. Besides, we also demonstrate the strong generalizability of TrajTrack across different base trackers. Video is available at https://www.bilibili.com/video/BV1ahYgzmEWP.
Authors: Tianyao Sun, Dawei Xiang, Tianqi Ding, Xiang Fang, Yijiashun Qi, Zunduo Zhao
Abstract: Infrared and visible image fusion (IVIF) is a fundamental task in multi-modal perception that aims to integrate complementary structural and textural cues from different spectral domains. In this paper, we propose FusionNet, a novel end-to-end fusion framework that explicitly models inter-modality interaction and enhances task-critical regions. FusionNet introduces a modality-aware attention mechanism that dynamically adjusts the contribution of infrared and visible features based on their discriminative capacity. To achieve fine-grained, interpretable fusion, we further incorporate a pixel-wise alpha blending module, which learns spatially-varying fusion weights in an adaptive and content-aware manner. Moreover, we formulate a target-aware loss that leverages weak ROI supervision to preserve semantic consistency in regions containing important objects (e.g., pedestrians, vehicles). Experiments on the public M3FD dataset demonstrate that FusionNet generates fused images with enhanced semantic preservation, high perceptual quality, and clear interpretability. Our framework provides a general and extensible solution for semantic-aware multi-modal image fusion, with benefits for downstream tasks such as object detection and scene understanding.
Authors: Wenhao Tang, Sheng Huang, Heng Fang, Fengtao Zhou, Bo Liu, Qingshan Liu
Abstract: Digitizing pathological images into gigapixel Whole Slide Images (WSIs) has opened new avenues for Computational Pathology (CPath). As positive tissue comprises only a small fraction of gigapixel WSIs, existing Multiple Instance Learning (MIL) methods typically focus on identifying salient instances via attention mechanisms. However, this leads to a bias towards easy-to-classify instances while neglecting challenging ones. Recent studies have shown that hard examples are crucial for accurately modeling discriminative boundaries. Applying such an idea at the instance level, we elaborate a novel MIL framework with masked hard instance mining (MHIM-MIL), which utilizes a Siamese structure with a consistency constraint to explore the hard instances. Using a class-aware instance probability, MHIM-MIL employs a momentum teacher to mask salient instances and implicitly mine hard instances for training the student model. To obtain diverse, non-redundant hard instances, we adopt large-scale random masking while utilizing a global recycle network to mitigate the risk of losing key features. Furthermore, the student updates the teacher using an exponential moving average, which identifies new hard instances for subsequent training iterations and stabilizes optimization. Experimental results on cancer diagnosis, subtyping, survival analysis tasks, and 12 benchmarks demonstrate that MHIM-MIL outperforms the latest methods in both performance and efficiency. The code is available at: https://github.com/DearCaat/MHIM-MIL.
Authors: Dezhen Wang, Haixiang Zhao, Xiang Shen, Sheng Miao
Abstract: Camouflaged object detection (COD) aims to segment objects that blend into their surroundings. However, most existing studies overlook the semantic differences among textual prompts of different targets as well as fine-grained frequency features. In this work, we propose a novel Semantic and Frequency Guided Network (SFGNet), which incorporates semantic prompts and frequency-domain features to capture camouflaged objects and improve boundary perception. We further design Multi-Band Fourier Module(MBFM) to enhance the ability of the network in handling complex backgrounds and blurred boundaries. In addition, we design an Interactive Structure Enhancement Block (ISEB) to ensure structural integrity and boundary details in the predictions. Extensive experiments conducted on three COD benchmark datasets demonstrate that our method significantly outperforms state-of-the-art approaches. The core code of the model is available at the following link: https://github.com/winter794444/SFGNetICASSP2026.
Authors: Weiming Li, Yan Shao, Jing Yang, Yujing Lu, Ling Zhong, Yuhan Wang, Manni Duan
Abstract: Graphical user interface (GUI) grounding is a fundamental task for building GUI agents. However, general vision-language models (VLMs) struggle with this task due to a lack of specific optimization. We identify a key gap in this paper: while VLMs exhibit significant latent grounding potential, as demonstrated by their performance measured by Pointing Game, they underperform when tasked with outputting explicit coordinates. To address this discrepancy, and bypass the high data and annotation costs of current fine-tuning approaches, we propose three zero-shot auxiliary reasoning methods. By providing explicit spatial cues such as axes, grids and labeled intersections as part of the input image, these methods enable VLMs to articulate their implicit spatial understanding capabilities. We evaluate these methods on four GUI grounding benchmarks across seven open-source and proprietary VLMs. The evaluation results demonstrate that the proposed methods substantially improve the performance of GUI grounding.
Authors: Zhexi Peng, Kun Zhou, Tianjia Shao
Abstract: While recent Gaussian-based SLAM methods achieve photorealistic reconstruction from RGB-D data, their computational performance remains a critical bottleneck. State-of-the-art techniques operate at less than 20 fps, significantly lagging behind geometry-centric approaches like KinectFusion (hundreds of fps). This limitation stems from the heavy computational burden: modeling scenes requires numerous Gaussians and complex iterative optimization to fit RGB-D data, where insufficient Gaussian counts or optimization iterations cause severe quality degradation. To address this, we propose a Gaussian-SDF hybrid representation, combining a colorized Signed Distance Field (SDF) for smooth geometry and appearance with 3D Gaussians to capture underrepresented details. The SDF is efficiently constructed via RGB-D fusion (as in geometry-centric methods), while Gaussians undergo iterative optimization. Our representation enables drastic Gaussian reduction (50% fewer) by avoiding full-scene Gaussian modeling, and efficient Gaussian optimization (75% fewer iterations) through targeted appearance refinement. Building upon this representation, we develop GPS-SLAM (Gaussian-Plus-SDF SLAM), a real-time 3D reconstruction system achieving over 150 fps on real-world Azure Kinect sequences -- delivering an order-of-magnitude speedup over state-of-the-art techniques while maintaining comparable reconstruction quality. We will release the source code and data to facilitate future research.
Authors: Haonan Shi, Yubin Wang, De Cheng, Lingfeng He, Nannan Wang, Xinbo Gao
Abstract: Unsupervised visible-infrared person re-identification (USVI-ReID) aims to learn modality-invariant image features from unlabeled cross-modal person datasets by reducing the modality gap while minimizing reliance on costly manual annotations. Existing methods typically address USVI-ReID using cluster-based contrastive learning, which represents a person by a single cluster center. However, they primarily focus on the commonality of images within each cluster while neglecting the finer-grained differences among them. To address the limitation, we propose a Hierarchical Identity Learning (HIL) framework. Since each cluster may contain several smaller sub-clusters that reflect fine-grained variations among images, we generate multiple memories for each existing coarse-grained cluster via a secondary clustering. Additionally, we propose Multi-Center Contrastive Learning (MCCL) to refine representations for enhancing intra-modal clustering and minimizing cross-modal discrepancies. To further improve cross-modal matching quality, we design a Bidirectional Reverse Selection Transmission (BRST) mechanism, which establishes reliable cross-modal correspondences by performing bidirectional matching of pseudo-labels. Extensive experiments conducted on the SYSU-MM01 and RegDB datasets demonstrate that the proposed method outperforms existing approaches. The source code is available at: https://github.com/haonanshi0125/HIL.
Authors: Mirco Felske, Stefan Stiene
Abstract: We propose BiCDO (Bias-Controlled Class Distribution Optimizer), an iterative, data-centric framework that identifies Pareto optimized class distributions for multi-class image classification. BiCDO enables performance prioritization for specific classes, which is useful in safety-critical scenarios (e.g. prioritizing 'Human' over 'Dog'). Unlike uniform distributions, BiCDO determines the optimal number of images per class to enhance reliability and minimize bias and variance in the objective function. BiCDO can be incorporated into existing training pipelines with minimal code changes and supports any labelled multi-class dataset. We have validated BiCDO using EfficientNet, ResNet and ConvNeXt on CIFAR-10 and iNaturalist21 datasets, demonstrating improved, balanced model performance through optimized data distribution.
Authors: Yanyun Pu, Kehan Li, Zeyi Huang, Zhijie Zhong, Kaixiang Yang
Abstract: With the rapid advancement of video generation models such as Sora, video quality assessment (VQA) is becoming increasingly crucial for selecting high-quality videos from large-scale datasets used in pre-training. Traditional VQA methods, typically producing single numerical scores, often lack comprehensiveness and interpretability. To address these challenges, we introduce MVQA-68K, a novel multi-dimensional VQA dataset comprising over 68,000 carefully annotated videos, covering seven essential quality dimensions: overall aesthetics, camera movement, dynamic degree, texture detail, composition, visual quality, and factual consistency. Each annotation includes detailed chain-of-thought reasoning to facilitate interpretability and comprehensive understanding. Extensive experiments demonstrate that MVQA-68K significantly enhances the performance of various multimodal large language models (MLLMs) on the VQA task, achieving state-of-the-art results not only on our internal test set (Fig.1) but also on public benchmarks including LSVQ-test, LSVQ-1080p, and LIVE-VQC. Meantime, incorporating explicit reasoning process during VQA training substantially boosts the zero-shot generalization. Code and dataset will be available at github: https://github.com/Controller01-ai/MVQA-68K
Authors: Siming Fu, Sijun Dong, Xiaoliang Meng
Abstract: Despite the remarkable success of Self-Supervised Learning (SSL), its generalization is fundamentally hindered by Shortcut Learning, where models exploit superficial features like texture instead of intrinsic structure. We experimentally verify this flaw within the generative paradigm (e.g., MAE) and argue it is a systemic issue also affecting discriminative methods, identifying it as the root cause of their failure on unseen domains. While existing methods often tackle this at a surface level by aligning or separating domain-specific features, they fail to alter the underlying learning mechanism that fosters shortcut dependency. To address this at its core, we propose HyGDL (Hybrid Generative-Discriminative Learning Framework), a hybrid framework that achieves explicit content-style disentanglement. Our approach is guided by the Invariance Pre-training Principle: forcing a model to learn an invariant essence by systematically varying a bias (e.g., style) at the input while keeping the supervision signal constant. HyGDL operates on a single encoder and analytically defines style as the component of a representation that is orthogonal to its style-invariant content, derived via vector projection.
Authors: Seoik Jung, Taekyung Song, Joshua Jordan Daniel, JinYoung Lee, SungJun Lee
Abstract: Video Anomaly Detection (VAD) is critical for surveillance and public safety. However, existing benchmarks are limited to either frame-level or video-level tasks, restricting a holistic view of model generalization. This work first introduces a softmax-based frame allocation strategy that prioritizes anomaly-dense segments while maintaining full-video coverage, enabling balanced sampling across temporal scales. Building on this process, we construct two complementary benchmarks. The image-based benchmark evaluates frame-level reasoning with representative frames, while the video-based benchmark extends to temporally localized segments and incorporates an abnormality scoring task.Experiments on UCF-Crime demonstrate improvements at both the frame and video levels, and ablation studies confirm clear advantages of anomaly-focused sampling over uniform and random baselines.
Authors: Wending Liu, Siyun Liang, Huy H. Nguyen, Isao Echizen
Abstract: We propose a novel 3D deepfake generation framework based on 3D Gaussian Splatting that enables realistic, identity-preserving face swapping and reenactment in a fully controllable 3D space. Compared to conventional 2D deepfake approaches that suffer from geometric inconsistencies and limited generalization to novel view, our method combines a parametric head model with dynamic Gaussian representations to support multi-view consistent rendering, precise expression control, and seamless background integration. To address editing challenges in point-based representations, we explicitly separate the head and background Gaussians and use pre-trained 2D guidance to optimize the facial region across views. We further introduce a repair module to enhance visual consistency under extreme poses and expressions. Experiments on NeRSemble and additional evaluation videos demonstrate that our method achieves comparable performance to state-of-the-art 2D approaches in identity preservation, as well as pose and expression consistency, while significantly outperforming them in multi-view rendering quality and 3D consistency. Our approach bridges the gap between 3D modeling and deepfake synthesis, enabling new directions for scene-aware, controllable, and immersive visual forgeries, revealing the threat that emerging 3D Gaussian Splatting technique could be used for manipulation attacks.
Authors: Yongzhe Lyu, Yu Wu, Yutian Lin, Bo Du
Abstract: Diffusion models have shown promising results in free-form inpainting. Recent studies based on refined diffusion samplers or novel architectural designs led to realistic results and high data consistency. However, random initialization seed (noise) adopted in vanilla diffusion process may introduce mismatched semantic information in masked regions, leading to biased inpainting results, e.g., low consistency and low coherence with the other unmasked area. To address this issue, we propose the Initial Seed refined Diffusion Model (IS-Diff), a completely training-free approach incorporating distributional harmonious seeds to produce harmonious results. Specifically, IS-Diff employs initial seeds sampled from unmasked areas to imitate the masked data distribution, thereby setting a promising direction for the diffusion procedure. Moreover, a dynamic selective refinement mechanism is proposed to detect severe unharmonious inpaintings in intermediate latent and adjust the strength of our initialization prior dynamically. We validate our method on both standard and large-mask inpainting tasks using the CelebA-HQ, ImageNet, and Places2 datasets, demonstrating its effectiveness across all metrics compared to state-of-the-art inpainting methods.
Authors: Qiyuan Guan, Qianfeng Yang, Xiang Chen, Tianyu Song, Guiyue Jin, Jiyu Jin
Abstract: Existing all-in-one image restoration approaches, which aim to handle multiple weather degradations within a single framework, are predominantly trained and evaluated using mixed single-weather synthetic datasets. However, these datasets often differ significantly in resolution, style, and domain characteristics, leading to substantial domain gaps that hinder the development and fair evaluation of unified models. Furthermore, the lack of a large-scale, real-world all-in-one weather restoration dataset remains a critical bottleneck in advancing this field. To address these limitations, we present a real-world all-in-one adverse weather image restoration benchmark dataset, which contains image pairs captured under various weather conditions, including rain, snow, and haze, as well as diverse outdoor scenes and illumination settings. The resulting dataset provides precisely aligned degraded and clean images, enabling supervised learning and rigorous evaluation. We conduct comprehensive experiments by benchmarking a variety of task-specific, task-general, and all-in-one restoration methods on our dataset. Our dataset offers a valuable foundation for advancing robust and practical all-in-one image restoration in real-world scenarios. The dataset has been publicly released and is available at https://github.com/guanqiyuan/WeatherBench.
Authors: Chuang Liu, Nan Guo
Abstract: OCTA is a crucial non-invasive imaging technique for diagnosing and monitoring retinal diseases like diabetic retinopathy, age-related macular degeneration, and glaucoma. Current 2D-based methods for retinal vessel (RV) segmentation offer insufficient accuracy. To address this, we propose RVMamba, a novel architecture integrating multiple feature extraction modules with the Mamba state-space model. Moreover, existing joint segmentation models for OCTA data exhibit performance imbalance between different tasks. To simultaneously improve the segmentation of the foveal avascular zone (FAZ) and mitigate this imbalance, we introduce FAZMamba and a unified Joint-OCTAMamba framework. Experimental results on the OCTA-500 dataset demonstrate that Joint-OCTAMamba outperforms existing models across evaluation metrics.The code is available at https://github.com/lc-sfis/Joint-OCTAMamba.
Authors: Lifei Hao, Yue Cheng, Baoqi Huang, Bing Jia, Xuandong Zhao
Abstract: Intelligent tableware cleaning is a critical application in food safety and smart homes, but existing methods are limited by coarse-grained classification and scarcity of few-shot data, making it difficult to meet industrialization requirements. We propose DTGen, a few-shot data augmentation scheme based on generative diffusion models, specifically designed for fine-grained dirty tableware recognition. DTGen achieves efficient domain specialization through LoRA, generates diverse dirty images via structured prompts, and ensures data quality through CLIP-based cross-modal filtering. Under extremely limited real few-shot conditions, DTGen can synthesize virtually unlimited high-quality samples, significantly improving classifier performance and supporting fine-grained dirty tableware recognition. We further elaborate on lightweight deployment strategies, promising to transfer DTGen's benefits to embedded dishwashers and integrate with cleaning programs to intelligently regulate energy consumption and detergent usage. Research results demonstrate that DTGen not only validates the value of generative AI in few-shot industrial vision but also provides a feasible deployment path for automated tableware cleaning and food safety monitoring.
Authors: Feilong Chen, Yijiang Liu, Yi Huang, Hao Wang, Miren Tian, Ya-Qi Yu, Minghui Liao, Jihao Wu
Abstract: We propose MindVL, a multimodal large langauge model trained on Ascend NPUs. Similar to Qwen2.5-VL, MindVL adopts native-resolution Vision Transformers, which enables it to process images at their original variable resolutions. This design avoids the degradation caused by fixed-resolution tiling while preserving fine-grained details and global layouts, which is crucial for visually dense content such as complex charts and diagrams. To ensure the smooth training of MindVL on Ascend NPUs, we develop Mindspeed-MLLM, a distributed multimodal training framework tailored for Ascend NPUs. To maintain training accuracy, we implement equivalent replacements for certain operators. MindVL undergoes a three-phase training process, namely the warm-up phase, multitask training phase, and supervised instruction tuning phase, to gradually enhance its capabilities. This process starts with basic visual and multimodal pre-training, followed by large-scale multiask trainging and instruction tuning. We also adopt multimodal data packaging and hybrid parallelism techniques, which significantly improve end-to-end training speed. To further boost model performance, we specifically introduce test-time resolution search and model weight averaging. Notably, despite using about 1/10 of the training data required by Qwen2.5-VL, MindVL achieves performance on par with Qwen2.5-VL in evaluations of general multimodal understanding and document/table comprehension. Beyond overall scores, MindVL also delivers leading performance in OCR assessments.
Authors: Bjoern Kremser, Yusuke Matsui
Abstract: Paper maps remain widely used for hiking and sightseeing because they contain curated trails and locally relevant annotations that are often missing from digital navigation applications such as Google Maps. We propose a pipeline to extract navigable trails from scanned maps, enabling their use in GPS-based navigation. Our method combines georeferencing, U-Net-based binary segmentation, graph construction, and an iterative refinement procedure using a routing engine. We evaluate the full end-to-end pipeline as well as individual components, showing that the approach can robustly recover trail networks from diverse map styles and generate GPS routes suitable for practical use.
Authors: Ruimin Ma, Sebastian Zudaire, Zhen Li, Chi Zhang
Abstract: Object 6DoF (6D) pose estimation is essential for robotic perception, especially in industrial settings. It enables robots to interact with the environment and manipulate objects. However, existing benchmarks on object 6D pose estimation primarily use everyday objects with rich textures and low-reflectivity, limiting model generalization to industrial scenarios where objects are often metallic, texture-less, and highly reflective. To address this gap, we propose a novel dataset and benchmark namely \textit{Industrial Metallic Dataset (IMD)}, tailored for industrial applications. Our dataset comprises 45 true-to-scale industrial components, captured with an RGB-D camera under natural indoor lighting and varied object arrangements to replicate real-world conditions. The benchmark supports three tasks, including video object segmentation, 6D pose tracking, and one-shot 6D pose estimation. We evaluate existing state-of-the-art models, including XMem and SAM2 for segmentation, and BundleTrack and BundleSDF for pose estimation, to assess model performance in industrial contexts. Evaluation results show that our industrial dataset is more challenging than existing household object datasets. This benchmark provides the baseline for developing and comparing segmentation and pose estimation algorithms that better generalize to industrial robotics scenarios.
Authors: Jeremiah Fadugba, Petru Manescu, Bolanle Oladejo, Delmiro Fernandez-Reyes, Philipp Berens
Abstract: Uncertainty estimation is critical for reliable medical image segmentation, particularly in retinal vessel analysis, where accurate predictions are essential for diagnostic applications. Deep Ensembles, where multiple networks are trained individually, are widely used to improve medical image segmentation performance. However, training and testing costs increase with the number of ensembles. In this work, we propose Ensemble Distillation as a robust alternative to commonly used uncertainty estimation techniques by distilling the knowledge of multiple ensemble models into a single model. Through extensive experiments on the DRIVE and FIVES datasets, we demonstrate that Ensemble Distillation achieves comparable performance via calibration and segmentation metrics, while significantly reducing computational complexity. These findings suggest that Ensemble distillation provides an efficient and reliable approach for uncertainty estimation in the segmentation of the retinal vessels, making it a promising tool for medical imaging applications.
Authors: Zahra Babaiee, Peyman M. Kiassari, Daniela Rus, Radu Grosu
Abstract: A recent study has proposed the "Master Key Filters Hypothesis" for convolutional neural network filters. This paper extends this hypothesis by radically constraining its scope to a single set of just 8 universal filters that depthwise separable convolutional networks inherently converge to. While conventional DS-CNNs employ thousands of distinct trained filters, our analysis reveals these filters are predominantly linear shifts (ax+b) of our discovered universal set. Through systematic unsupervised search, we extracted these fundamental patterns across different architectures and datasets. Remarkably, networks initialized with these 8 unique frozen filters achieve over 80% ImageNet accuracy, and even outperform models with thousands of trainable parameters when applied to smaller datasets. The identified master key filters closely match Difference of Gaussians (DoGs), Gaussians, and their derivatives, structures that are not only fundamental to classical image processing but also strikingly similar to receptive fields in mammalian visual systems. Our findings provide compelling evidence that depthwise convolutional layers naturally gravitate toward this fundamental set of spatial operators regardless of task or architecture. This work offers new insights for understanding generalization and transfer learning through the universal language of these master key filters.
Authors: Nikolaos Livathinos, Christoph Auer, Ahmed Nassar, Rafael Teixeira de Lima, Maksym Lysak, Brown Ebouky, Cesar Berrospi, Michele Dolfi, Panagiotis Vagenas, Matteo Omenetti, Kasper Dinkla, Yusik Kim, Valery Weber, Lucas Morin, Ingmar Meijer, Viktor Kuropiatnyk, Tim Strohmeyer, A. Said Gurbuz, Peter W. J. Staar
Abstract: This technical report documents the development of novel Layout Analysis models integrated into the Docling document-conversion pipeline. We trained several state-of-the-art object detectors based on the RT-DETR, RT-DETRv2 and DFINE architectures on a heterogeneous corpus of 150,000 documents (both openly available and proprietary). Post-processing steps were applied to the raw detections to make them more applicable to the document conversion task. We evaluated the effectiveness of the layout analysis on various document benchmarks using different methodologies while also measuring the runtime performance across different environments (CPU, Nvidia and Apple GPUs). We introduce five new document layout models achieving 20.6% - 23.9% mAP improvement over Docling's previous baseline, with comparable or better runtime. Our best model, "heron-101", attains 78% mAP with 28 ms/image inference time on a single NVIDIA A100 GPU. Extensive quantitative and qualitative experiments establish best practices for training, evaluating, and deploying document-layout detectors, providing actionable guidance for the document conversion community. All trained checkpoints, code, and documentation are released under a permissive license on HuggingFace.
Authors: Tae Kyeong Jeong, Garam Kim, Juyoun Park
Abstract: Accurate segmentation of thin structures is critical for microsurgical scene understanding but remains challenging due to resolution loss, low contrast, and class imbalance. We propose Microsurgery Instrument Segmentation for Robotic Assistance(MISRA), a segmentation framework that augments RGB input with luminance channels, integrates skip attention to preserve elongated features, and employs an Iterative Feedback Module(IFM) for continuity restoration across multiple passes. In addition, we introduce a dedicated microsurgical dataset with fine-grained annotations of surgical instruments including thin objects, providing a benchmark for robust evaluation Dataset available at https://huggingface.co/datasets/KIST-HARILAB/MISAW-Seg. Experiments demonstrate that MISRA achieves competitive performance, improving the mean class IoU by 5.37% over competing methods, while delivering more stable predictions at instrument contacts and overlaps. These results position MISRA as a promising step toward reliable scene parsing for computer-assisted and robotic microsurgery.
URLs: https://huggingface.co/datasets/KIST-HARILAB/MISAW-Seg.
Authors: Yudong Shen, Wenyu Wu, Jiali Mao, Yixiao Tong, Guoping Liu, Chaoya Wang
Abstract: Trajectory data has become a key resource for automated map in-ference due to its low cost, broad coverage, and continuous availability. However, uneven trajectory density often leads to frag-mented roads in sparse areas and redundant segments in dense regions, posing significant challenges for existing methods. To address these issues, we propose DGMap, a dual-decoding framework with global context awareness, featuring Multi-scale Grid Encoding, Mask-enhanced Keypoint Extraction, and Global Context-aware Relation Prediction. By integrating global semantic context with local geometric features, DGMap improves keypoint detection accuracy to reduce road fragmentation in sparse-trajectory areas. Additionally, the Global Context-aware Relation Prediction module suppresses false connections in dense-trajectory regions by modeling long-range trajectory patterns. Experimental results on three real-world datasets show that DGMap outperforms state-of-the-art methods by 5% in APLS, with notable performance gains on trajectory data from the Didi Chuxing platform
Authors: Hongyuan Zhang, Yuheng Wu, Mingyang Zhao, Zhiwei Chen, Rebecca Li, Fei Zhu, Haohan Zhao, Xiaohua Yuan, Meng Yang, Chunli Qiu, Xiang Cong, Haiyan Chen, Lina Luan, Randolph H. L. Wong, Huai Liao, Colin A Graham, Shi Chang, Guowei Tao, Dong Yi, Zhen Lei, Nassir Navab, Sebastien Ourselin, Jiebo Luo, Hongbin Liu, Gaofeng Meng
Abstract: Artificial intelligence (AI) that can effectively learn ultrasound representations by integrating multi-source data holds significant promise for advancing clinical care. However, the scarcity of large labeled datasets in real-world clinical environments and the limited generalizability of task-specific models have hindered the development of generalizable clinical AI models for ultrasound applications. In this study, we present EchoCare, a novel ultrasound foundation model for generalist clinical use, developed via self-supervised learning on our curated, publicly available, large-scale dataset EchoCareData. EchoCareData comprises 4.5 million ultrasound images, sourced from over 23 countries across 5 continents and acquired via a diverse range of distinct imaging devices, thus encompassing global cohorts that are multi-center, multi-device, and multi-ethnic. Unlike prior studies that adopt off-the-shelf vision foundation model architectures, we introduce a hierarchical classifier into EchoCare to enable joint learning of pixel-level and representation-level features, capturing both global anatomical contexts and local ultrasound characteristics. With minimal training, EchoCare outperforms state-of-the-art comparison models across 10 representative ultrasound benchmarks of varying diagnostic difficulties, spanning disease diagnosis, lesion segmentation, organ detection, landmark prediction, quantitative regression, imaging enhancement and report generation. The code and pretrained model are publicly released, rendering EchoCare accessible for fine-tuning and local adaptation, supporting extensibility to additional applications. EchoCare provides a fully open and generalizable foundation model to boost the development of AI technologies for diverse clinical ultrasound applications.
Authors: Danling Cao
Abstract: Reconstructing 3D face from a single unconstrained image remains a challenging problem due to diverse conditions in unconstrained environments. Recently, learning-based methods have achieved notable results by effectively capturing complex facial structures and details across varying conditions. Consequently, many existing approaches employ projection-based losses between generated and input images to constrain model training. However, learning-based methods for 3D face reconstruction typically require substantial amounts of 3D facial data, which is difficult and costly to obtain. Consequently, to reduce reliance on labeled 3D face datasets, many existing approaches employ projection-based losses between generated and input images to constrain model training. Nonetheless, despite these advancements, existing approaches frequently struggle to capture detailed and multi-scale features under diverse facial attributes and conditions, leading to incomplete or less accurate reconstructions. In this paper, we propose a Multi-Scale Feature Fusion with Multi-Attribute (MSMA) framework for 3D face reconstruction from unconstrained images. Our method integrates multi-scale feature fusion with a focus on multi-attribute learning and leverages a large-kernel attention module to enhance the precision of feature extraction across scales, enabling accurate 3D facial parameter estimation from a single 2D image. Comprehensive experiments on the MICC Florence, Facewarehouse and custom-collect datasets demonstrate that our approach achieves results on par with current state-of-the-art methods, and in some instances, surpasses SOTA performance across challenging conditions.
Authors: Diogo Mendon\c{c}a, Tiago Barros, Cristiano Premebida, Urbano J. Nunes
Abstract: Autonomous systems require robust Multi-Object Tracking (MOT) capabilities to operate reliably in dynamic environments. MOT ensures consistent object identity assignment and precise spatial delineation. Recent advances in foundation models, such as SAM2, have demonstrated strong zero-shot generalization for video segmentation, but their direct application to MOTS (MOT+Segmentation) remains limited by insufficient identity management and memory efficiency. This work introduces Seg2Track-SAM2, a framework that integrates pre-trained object detectors with SAM2 and a novel Seg2Track module to address track initialization, track management, and reinforcement. The proposed approach requires no fine-tuning and remains detector-agnostic. Experimental results on KITTI MOT and KITTI MOTS benchmarks show that Seg2Track-SAM2 achieves state-of-the-art (SOTA) performance, ranking fourth overall in both car and pedestrian classes on KITTI MOTS, while establishing a new benchmark in association accuracy (AssA). Furthermore, a sliding-window memory strategy reduces memory usage by up to 75% with negligible performance degradation, supporting deployment under resource constraints. These results confirm that Seg2Track-SAM2 advances MOTS by combining robust zero-shot tracking, enhanced identity preservation, and efficient memory utilization. The code is available at https://github.com/hcmr-lab/Seg2Track-SAM2
Authors: Changlu Guo, Anders Nymark Christensen, Anders Bjorholm Dahl, Yugen Yi, Morten Rieger Hannemose
Abstract: Retinal vessel segmentation is essential for early diagnosis of diseases such as diabetic retinopathy, hypertension, and neurodegenerative disorders. Although SA-UNet introduces spatial attention in the bottleneck, it underuses attention in skip connections and does not address the severe foreground-background imbalance. We propose SA-UNetv2, a lightweight model that injects cross-scale spatial attention into all skip connections to strengthen multi-scale feature fusion and adopts a weighted Binary Cross-Entropy (BCE) plus Matthews Correlation Coefficient (MCC) loss to improve robustness to class imbalance. On the public DRIVE and STARE datasets, SA-UNetv2 achieves state-of-the-art performance with only 1.2MB memory and 0.26M parameters (less than 50% of SA-UNet), and 1 second CPU inference on 592 x 592 x 3 images, demonstrating strong efficiency and deployability in resource-constrained, CPU-only settings.
Authors: Haodong Chen, Haojian Huang, XinXiang Yin, Dian Shao
Abstract: Video Question Answering (VideoQA) based on Large Language Models (LLMs) has shown potential in general video understanding but faces significant challenges when applied to the inherently complex domain of sports videos. In this work, we propose FineQuest, the first training-free framework that leverages dual-mode reasoning inspired by cognitive science: i) Reactive Reasoning for straightforward sports queries and ii) Deliberative Reasoning for more complex ones. To bridge the knowledge gap between general-purpose models and domain-specific sports understanding, FineQuest incorporates SSGraph, a multimodal sports knowledge scene graph spanning nine sports, which encodes both visual instances and domain-specific terminology to enhance reasoning accuracy. Furthermore, we introduce two new sports VideoQA benchmarks, Gym-QA and Diving-QA, derived from the FineGym and FineDiving datasets, enabling diverse and comprehensive evaluation. FineQuest achieves state-of-the-art performance on these benchmarks as well as the existing SPORTU dataset, while maintains strong general VideoQA capabilities.
Authors: Ang Nan Gu, Michael Tsang, Hooman Vaseli, Purang Abolmaesumi, Teresa Tsang
Abstract: Computer-aided diagnosis systems must make critical decisions from medical images that are often noisy, ambiguous, or conflicting, yet today's models are trained on overly simplistic labels that ignore diagnostic uncertainty. One-hot labels erase inter-rater variability and force models to make overconfident predictions, especially when faced with incomplete or artifact-laden inputs. We address this gap by introducing a novel framework that brings uncertainty back into the label space. Our method leverages neural network training dynamics (NNTD) to assess the inherent difficulty of each training sample. By aggregating and calibrating model predictions during training, we generate uncertainty-aware pseudo-labels that reflect the ambiguity encountered during learning. This label augmentation approach is architecture-agnostic and can be applied to any supervised learning pipeline to enhance uncertainty estimation and robustness. We validate our approach on a challenging echocardiography classification benchmark, demonstrating superior performance over specialized baselines in calibration, selective classification, and multi-view fusion.
Authors: Mehwish Mehmood, Shahzaib Iqbal, Tariq Mahmood Khan, Ivor Spence, Muhammad Fahim
Abstract: Retinal vessel segmentation is critical for the early diagnosis of vision-threatening and systemic diseases, especially in real-world clinical settings with limited computational resources. Although significant improvements have been made in deep learning-based segmentation methods, current models still face challenges in extracting tiny vessels and suffer from high computational costs. In this study, we present LFRA-Net by incorporating focal modulation attention at the encoder-decoder bottleneck and region-aware attention in the selective skip connections. LFRA-Net is a lightweight network optimized for precise and effective retinal vascular segmentation. It enhances feature representation and regional focus by efficiently capturing local and global dependencies. LFRA-Net outperformed many state-of-the-art models while maintaining lightweight characteristics with only 0.17 million parameters, 0.66 MB memory size, and 10.50 GFLOPs. We validated it on three publicly available datasets: DRIVE, STARE, and CHASE\_DB. It performed better in terms of Dice score (84.28\%, 88.44\%, and 85.50\%) and Jaccard index (72.86\%, 79.31\%, and 74.70\%) on the DRIVE, STARE, and CHASE\_DB datasets, respectively. LFRA-Net provides an ideal ratio between segmentation accuracy and computational cost compared to existing deep learning methods, which makes it suitable for real-time clinical applications in areas with limited resources. The code can be found at https://github.com/Mehwish4593/LFRA-Net.
Authors: Haiduo Huang, Fuwei Yang, Zhenhua Liu, Xuanwu Yin, Dong Li, Pengju Ren, Emad Barsoum
Abstract: Speculative decoding is a powerful way to accelerate autoregressive large language models (LLMs), but directly porting it to vision-language models (VLMs) faces unique systems constraints: the prefill stage is dominated by visual tokens whose count scales with image resolution and video length, inflating both compute and memory, especially the key-value (KV) cache. We study speculative decoding for VLMs and introduce SpecVLM, a practical system that (1) establishes a strong EAGLE-2-style baseline, EagleVLM, delivering 1.5--2.3x end-to-end speedups over full autoregressive inference, and (2) further accelerates VLM inference with an elastic visual compressor that adaptively selects among pruning, pooling, convolution, and resampler primitives to balance FLOPs/parameters and accuracy per input. To avoid costly offline distillation corpora, we propose an online-logit distillation protocol that trains the draft model with on-the-fly teacher logits and penultimate features using a combined cross-entropy and Smooth L1 objective, eliminating storage and preprocessing while remaining compute-efficient. This protocol reveals a training-time scaling effect: longer online training monotonically increases the draft model's average accepted length, improving speculative efficiency. Empirically, SpecVLM achieves additional acceleration, culminating in 2.5--2.9x end-to-end speedups within 5 epochs across LLaVA and MMMU, consistently over resolutions and task difficulties, while preserving the target model's output distribution (lossless decoding). Our code is available at https://github.com/haiduo/SpecVLM.
Authors: Liying Wang, Xiaoli Zhang, Chuanmin Jia, Siwei Ma
Abstract: Infrared-visible image fusion methods aim at generating fused images with good visual quality and also facilitate the performance of high-level tasks. Indeed, existing semantic-driven methods have considered semantic information injection for downstream applications. However, none of them investigates the potential for reciprocal promotion between pixel-wise image fusion and cross-modal feature fusion perception tasks from a macroscopic task-level perspective. To address this limitation, we propose a unified network for image fusion and semantic segmentation. MAFS is a parallel structure, containing a fusion sub-network and a segmentation sub-network. On the one hand, We devise a heterogeneous feature fusion strategy to enhance semantic-aware capabilities for image fusion. On the other hand, by cascading the fusion sub-network and a segmentation backbone, segmentation-related knowledge is transferred to promote feature-level fusion-based segmentation. Within the framework, we design a novel multi-stage Transformer decoder to aggregate fine-grained multi-scale fused features efficiently. Additionally, a dynamic factor based on the max-min fairness allocation principle is introduced to generate adaptive weights of two tasks and guarantee smooth training in a multi-task manner. Extensive experiments demonstrate that our approach achieves competitive results compared with state-of-the-art methods. The code is available at https://github.com/Abraham-Einstein/MAFS/.
Authors: Navid Hashemi, Samuel Sasaki, Diego Manzanas Lopez, Ipek Oguz, Meiyi Ma, Taylor T. Johnson
Abstract: Semantic segmentation networks (SSNs) play a critical role in domains such as medical imaging, autonomous driving, and environmental monitoring, where safety hinges on reliable model behavior under uncertainty. Yet, existing probabilistic verification approaches struggle to scale with the complexity and dimensionality of modern segmentation tasks, often yielding guarantees that are too conservative to be practical. We introduce a probabilistic verification framework that is both architecture-agnostic and scalable to high-dimensional outputs. Our approach combines sampling-based reachability analysis with conformal inference (CI) to deliver provable guarantees while avoiding the excessive conservatism of prior methods. To counteract CI's limitations in high-dimensional settings, we propose novel strategies that reduce conservatism without compromising rigor. Empirical evaluation on large-scale segmentation models across CamVid, OCTA-500, Lung Segmentation, and Cityscapes demonstrates that our method provides reliable safety guarantees while substantially tightening bounds compared to SOTA. We also provide a toolbox implementing this technique, available on Github.
Authors: Tim Lebailly, Vijay Veerabadran, Satwik Kottur, Karl Ridgeway, Michael Louis Iuzzolino
Abstract: Generative vision-language models (VLMs) exhibit strong high-level image understanding but lack spatially dense alignment between vision and language modalities, as our findings indicate. Orthogonal to advancements in generative VLMs, another line of research has focused on representation learning for vision-language alignment, targeting zero-shot inference for dense tasks like segmentation. In this work, we bridge these two directions by densely aligning images with synthetic descriptions generated by VLMs. Synthetic captions are inexpensive, scalable, and easy to generate, making them an excellent source of high-level semantic understanding for dense alignment methods. Empirically, our approach outperforms prior work on standard zero-shot open-vocabulary segmentation benchmarks/datasets, while also being more data-efficient.
Authors: Yi-Hsin Li, Thomas Sikora, Sebastian Knorr, M{\aa}arten Sj\"ostr\"om
Abstract: Sparse-view synthesis remains a challenging problem due to the difficulty of recovering accurate geometry and appearance from limited observations. While recent advances in 3D Gaussian Splatting (3DGS) have enabled real-time rendering with competitive quality, existing pipelines often rely on Structure-from-Motion (SfM) for camera pose estimation, an approach that struggles in genuinely sparse-view settings. Moreover, several SfM-free methods replace SfM with multi-view stereo (MVS) models, but generate massive numbers of 3D Gaussians by back-projecting every pixel into 3D space, leading to high memory costs. We propose Segmentation-Driven Initialization for Gaussian Splatting (SDI-GS), a method that mitigates inefficiency by leveraging region-based segmentation to identify and retain only structurally significant regions. This enables selective downsampling of the dense point cloud, preserving scene fidelity while substantially reducing Gaussian count. Experiments across diverse benchmarks show that SDI-GS reduces Gaussian count by up to 50% and achieves comparable or superior rendering quality in PSNR and SSIM, with only marginal degradation in LPIPS. It further enables faster training and lower memory footprint, advancing the practicality of 3DGS for constrained-view scenarios.
Authors: Haodi Ma, Vyom Pathak, Daisy Zhe Wang
Abstract: Video Question Answering (VQA) requires models to reason over spatial, temporal, and causal cues in videos. Recent vision language models (VLMs) achieve strong results but often rely on shallow correlations, leading to weak temporal grounding and limited interpretability. We study symbolic scene graphs (SGs) as intermediate grounding signals for VQA. SGs provide structured object-relation representations that complement VLMs holistic reasoning. We introduce SG-VLM, a modular framework that integrates frozen VLMs with scene graph grounding via prompting and visual localization. Across three benchmarks (NExT-QA, iVQA, ActivityNet-QA) and multiple VLMs (QwenVL, InternVL), SG-VLM improves causal and temporal reasoning and outperforms prior baselines, though gains over strong VLMs are limited. These findings highlight both the promise and current limitations of symbolic grounding, and offer guidance for future hybrid VLM-symbolic approaches in video understanding.
Authors: Meng Luo, Shengqiong Wu, Liqiang Jing, Tianjie Ju, Li Zheng, Jinxiang Lai, Tianlong Wu, Xinya Du, Jian Li, Siyuan Yan, Jiebo Luo, William Yang Wang, Hao Fei, Mong-Li Lee, Wynne Hsu
Abstract: Recent advancements in large video models (LVMs) have significantly enhance video understanding. However, these models continue to suffer from hallucinations, producing content that conflicts with input videos. To address this issue, we propose Dr.V, a hierarchical framework covering perceptive, temporal, and cognitive levels to diagnose video hallucination by fine-grained spatial-temporal grounding. Dr.V comprises of two key components: a benchmark dataset Dr.V-Bench and a satellite video agent Dr.V-Agent. Dr.V-Bench includes 10k instances drawn from 4,974 videos spanning diverse tasks, each enriched with detailed spatial-temporal annotation. Dr.V-Agent detects hallucinations in LVMs by systematically applying fine-grained spatial-temporal grounding at the perceptive and temporal levels, followed by cognitive level reasoning. This step-by-step pipeline mirrors human-like video comprehension and effectively identifies hallucinations. Extensive experiments demonstrate that Dr.V-Agent is effective in diagnosing hallucination while enhancing interpretability and reliability, offering a practical blueprint for robust video understanding in real-world scenarios. All our data and code are available at https://github.com/Eurekaleo/Dr.V.
Authors: Anne Marthe Sophie Ngo Bibinbe, Patrick Gagnon, Jamie Ahloy-Dallaire, Eric R. Paquet
Abstract: Precision livestock farming requires advanced monitoring tools to meet the increasing management needs of the industry. Computer vision systems capable of long-term multi-animal tracking (MAT) are essential for continuous behavioral monitoring in livestock production. MAT, a specialized subset of multi-object tracking (MOT), shares many challenges with MOT, but also faces domain-specific issues including frequent animal occlusion, highly similar appearances among animals, erratic motion patterns, and a wide range of behavior types. While some existing MAT tools are user-friendly and widely adopted, they often underperform compared to state-of-the-art MOT methods, which can result in inaccurate downstream tasks such as behavior analysis, health state estimation, and related applications. In this study, we benchmarked both MAT and MOT approaches for long-term tracking of pigs. We compared tools such as DeepLabCut and idTracker with MOT-based methods including ByteTrack, DeepSORT, cross-input consistency, and newer approaches like Track-Anything and PromptTrack. All methods were evaluated on a 10-minute pig tracking dataset. Our results demonstrate that, overall, MOT approaches outperform traditional MAT tools, even for long-term tracking scenarios. These findings highlight the potential of recent MOT techniques to enhance the accuracy and reliability of automated livestock tracking.
Authors: Sofia Jamil, Kotla Sai Charan, Sriparna Saha, Koustava Goswami, K J Joseph
Abstract: Poetry is an expressive form of art that invites multiple interpretations, as readers often bring their own emotions, experiences, and cultural backgrounds into their understanding of a poem. Recognizing this, we aim to generate images for poems and improve these images in a zero-shot setting, enabling audiences to modify images as per their requirements. To achieve this, we introduce a novel Weighted Prompt Manipulation (WPM) technique, which systematically modifies attention weights and text embeddings within diffusion models. By dynamically adjusting the importance of specific words, WPM enhances or suppresses their influence in the final generated image, leading to semantically richer and more contextually accurate visualizations. Our approach exploits diffusion models and large language models (LLMs) such as GPT in conjunction with existing poetry datasets, ensuring a comprehensive and structured methodology for improved image generation in the literary domain. To the best of our knowledge, this is the first attempt at integrating weighted prompt manipulation for enhancing imagery in poetic language.
Authors: Zhenni Yu, Li Zhao, Guobao Xiao, Xiaoqin Zhang
Abstract: This paper introduces a new Segment Anything Model (SAM) that leverages reverse parameter configuration and test-time training to enhance its performance on Camouflaged Object Detection (COD), named SAM-TTT. While most existing SAM-based COD models primarily focus on enhancing SAM by extracting favorable features and amplifying its advantageous parameters, a crucial gap is identified: insufficient attention to adverse parameters that impair SAM's semantic understanding in downstream tasks. To tackle this issue, the Reverse SAM Parameter Configuration Module is proposed to effectively mitigate the influence of adverse parameters in a train-free manner by configuring SAM's parameters. Building on this foundation, the T-Visioner Module is unveiled to strengthen advantageous parameters by integrating Test-Time Training layers, originally developed for language tasks, into vision tasks. Test-Time Training layers represent a new class of sequence modeling layers characterized by linear complexity and an expressive hidden state. By integrating two modules, SAM-TTT simultaneously suppresses adverse parameters while reinforcing advantageous ones, significantly improving SAM's semantic understanding in COD task. Our experimental results on various COD benchmarks demonstrate that the proposed approach achieves state-of-the-art performance, setting a new benchmark in the field. The code will be available at https://github.com/guobaoxiao/SAM-TTT.
Authors: Francis Xiatian Zhang, Emile Mackute, Mohammadreza Kasaei, Kevin Dhaliwal, Robert Thomson, Mohsen Khadem
Abstract: Monocular depth estimation in bronchoscopy can significantly improve real-time navigation accuracy and enhance the safety of interventions in complex, branching airways. Recent advances in depth foundation models have shown promise for endoscopic scenarios, yet these models often lack anatomical awareness in bronchoscopy, overfitting to local textures rather than capturing the global airway structure, particularly under ambiguous depth cues and poor lighting. To address this, we propose Brea-Depth, a novel framework that integrates airway-specific geometric priors into foundation model adaptation for bronchoscopic depth estimation. Our method introduces a depth-aware CycleGAN, refining the translation between real bronchoscopic images and airway geometries from anatomical data, effectively bridging the domain gap. In addition, we introduce an airway structure awareness loss to enforce depth consistency within the airway lumen while preserving smooth transitions and structural integrity. By incorporating anatomical priors, Brea-Depth enhances model generalization and yields more robust, accurate 3D airway reconstructions. To assess anatomical realism, we introduce Airway Depth Structure Evaluation, a new metric for structural consistency. We validate BREA-Depth on a collected ex vivo human lung dataset and an open bronchoscopic dataset, where it outperforms existing methods in anatomical depth preservation.
Authors: Akito Shinohara, Kohei Fukuda, Hiroaki Aizawa
Abstract: The ability to detect out-of-distribution data is essential not only for ensuring robustness against unknown or unexpected input data but also for improving the generalization performance of the model. Among various out-of-distribution detection methods, Outlier Exposure and Mixture Outlier Exposure are promising approaches that enhance out-of-distribution detection performance by exposing the outlier data during training. However, even with these sophisticated techniques, it remains challenging for models to learn the relationships between classes effectively and to distinguish data sampling from in-distribution and out-of-distribution clearly. Therefore, we focus on the logit space, where the properties between class-wise distributions are distinctly separated from those in the input or feature spaces. Specifically, we propose a linear interpolation technique in the logit space that mixes in-distribution and out-of-distribution data to facilitate smoothing logits between classes and improve the out-of-distribution detection performance, particularly for out-of-distribution data that lie close to the in-distribution data. Additionally, we enforce consistency between the logits obtained through mixing in the logit space and those generated via mixing in the input space. Our experiments demonstrate that our logit-space mixing technique reduces the abrupt fluctuations in the model outputs near the decision boundaries, resulting in smoother and more reliable separation between in-distribution and out-of-distribution data. Furthermore, we evaluate the effectiveness of the proposed method on a fine-grained out-of-distribution detection task.
Authors: Marian Renz, Felix Igelbrink, Martin Atzmueller
Abstract: 3D semantic scene graphs (3DSSG) provide compact structured representations of environments by explicitly modeling objects, attributes, and relationships. While 3DSSGs have shown promise in robotics and embodied AI, many existing methods rely mainly on sensor data, not integrating further information from semantically rich environments. Additionally, most methods assume access to complete scene reconstructions, limiting their applicability in real-world, incremental settings. This paper introduces a novel heterogeneous graph model for incremental 3DSSG prediction that integrates additional, multi-modal information, such as prior observations, directly into the message-passing process. Utilizing multiple layers, the model flexibly incorporates global and local scene representations without requiring specialized modules or full scene reconstructions. We evaluate our approach on the 3DSSG dataset, showing that GNNs enriched with multi-modal information such as semantic embeddings (e.g., CLIP) and prior observations offer a scalable and generalizable solution for complex, real-world environments. The full source code of the presented architecture will be made available at https://github.com/m4renz/incremental-scene-graph-prediction.
URLs: https://github.com/m4renz/incremental-scene-graph-prediction.
Authors: Zilin Li, Weiwei Xu, Xuanqi Zhao, Yiran Zhu
Abstract: Facial emotion recognition (FER) models trained only on pixels often fail to generalize across datasets because facial appearance is an indirect and biased proxy for underlying affect. We present NeuroGaze-Distill, a cross-modal distillation framework that transfers brain-informed priors into an image-only FER student via static Valence/Arousal (V/A) prototypes and a depression-inspired geometric prior (D-Geo). A teacher trained on EEG topographic maps from DREAMER (with MAHNOB-HCI as unlabeled support) produces a consolidated 5x5 V/A prototype grid that is frozen and reused; no EEG-face pairing and no non-visual signals at deployment are required. The student (ResNet-18/50) is trained on FERPlus with conventional CE/KD and two lightweight regularizers: (i) Proto-KD (cosine) aligns student features to the static prototypes; (ii) D-Geo softly shapes the embedding geometry in line with affective findings often reported in depression research (e.g., anhedonia-like contraction in high-valence regions). We evaluate both within-domain (FERPlus validation) and cross-dataset protocols (AffectNet-mini; optional CK+), reporting standard 8-way scores alongside present-only Macro-F1 and balanced accuracy to fairly handle label-set mismatch. Ablations attribute consistent gains to prototypes and D-Geo, and favor 5x5 over denser grids for stability. The method is simple, deployable, and improves robustness without architectural complexity.
Authors: Bo Cao (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), Fan Yu (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), Mengmeng Feng (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), SenHao Zhang (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), Xin Meng (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), Yue Zhang (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China), Zhen Qian (Beijing United Intelligent Imaging Research Institute, China), Jie Lu (Department of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, China)
Abstract: Multimodal learning has attracted much attention in recent years due to its ability to effectively utilize data features from a variety of different modalities. Diagnosing the vulnerability of atherosclerotic plaques directly from carotid 3D MRI images is relatively challenging for both radiologists and conventional 3D vision networks. In clinical practice, radiologists assess patient conditions using a multimodal approach that incorporates various imaging modalities and domain-specific expertise, paving the way for the creation of multimodal diagnostic networks. In this paper, we have developed an effective strategy to leverage radiologists' domain knowledge to automate the diagnosis of carotid plaque vulnerability through Variation inference and Multimodal knowledge Distillation (VMD). This method excels in harnessing cross-modality prior knowledge from limited image annotations and radiology reports within training data, thereby enhancing the diagnostic network's accuracy for unannotated 3D MRI images. We conducted in-depth experiments on the dataset collected in-house and verified the effectiveness of the VMD strategy we proposed.
Authors: Xue Zhang, Bingshuo Hu, Gene Cheung
Abstract: Conventional deep neural nets (DNNs) initialize network parameters at random and then optimize each one via stochastic gradient descent (SGD), resulting in substantial risk of poor-performing local minima.Focusing on the image interpolation problem and leveraging a recent theorem that maps a (pseudo-)linear interpolator {\Theta} to a directed graph filter that is a solution to a MAP problem regularized with a graph shift variation (GSV) prior, we first initialize a directed graph adjacency matrix A based on a known interpolator {\Theta}, establishing a baseline performance.Then, towards further gain, we learn perturbation matrices P and P(2) from data to augment A, whose restoration effects are implemented via Douglas-Rachford (DR) iterations, which we unroll into a lightweight interpretable neural net.Experimental results demonstrate state-of-the-art image interpolation results, while drastically reducing network parameters.
Authors: Mahmoud Z. A. Wahba, Sara Baldoni, Federica Battisti
Abstract: The recent success of immersive applications is pushing the research community to define new approaches to process 360{\deg} images and videos and optimize their transmission. Among these, saliency estimation provides a powerful tool that can be used to identify visually relevant areas and, consequently, adapt processing algorithms. Although saliency estimation has been widely investigated for 2D content, very few algorithms have been proposed for 360{\deg} saliency estimation. Towards this goal, we introduce Sphere-GAN, a saliency detection model for 360{\deg} videos that leverages a Generative Adversarial Network with spherical convolutions. Extensive experiments were conducted using a public 360{\deg} video saliency dataset, and the results demonstrate that Sphere-GAN outperforms state-of-the-art models in accurately predicting saliency maps.
Authors: Debopom Sutradhar, Arefin Ittesafun Abian, Mohaimenul Azam Khan Raiaan, Reem E. Mohamed, Sheikh Izzal Azid, Sami Azam
Abstract: Accurate land cover classification from satellite imagery is crucial in environmental monitoring and sustainable resource management. However, it remains challenging due to the complexity of natural landscapes, the visual similarity between classes, and the significant class imbalance in the available datasets. To address these issues, we propose a dual encoder architecture that independently extracts modality-specific features from optical and Synthetic Aperture Radar (SAR) imagery, which are then fused using a cross-modality attention-fusion module named Cross-modality Land cover segmentation with Attention and Imbalance-aware Reasoning-Enhanced Explanations (CLAIRE). This fusion mechanism highlights complementary spatial and textural features, enabling the network to better capture detailed and diverse land cover patterns. We incorporate a hybrid loss function that utilizes Weighted Focal Loss and Tversky Loss named RIFT (Rare-Instance Focal-Tversky) to address class imbalance and improve segmentation performance across underrepresented categories. Our model achieves competitive performance across multiple benchmarks: a mean Intersection over Union (mIoU) of 56.02% and Overall Accuracy (OA) of 84.56% on the WHU-OPT-SAR dataset; strong generalization with a mIoU of 59.89% and OA of 73.92% on the OpenEarthMap-SAR dataset; and remarkable robustness under cloud-obstructed conditions, achieving an mIoU of 86.86% and OA of 94.58% on the PIE-RGB-SAR dataset. Additionally, we introduce a metric-driven reasoning module generated by a Small Language Model (Phi-3), which generates expert-level, sample-specific justifications for model predictions, thereby enhancing transparency and interpretability.
Authors: Ao Liang, Youquan Liu, Yu Yang, Dongyue Lu, Linfeng Li, Lingdong Kong, Huaici Zhao, Wei Tsang Ooi
Abstract: While generative world models have advanced video and occupancy-based data synthesis, LiDAR generation remains underexplored despite its importance for accurate 3D perception. Extending generation to 4D LiDAR data introduces challenges in controllability, temporal stability, and evaluation. We present LiDARCrafter, a unified framework that converts free-form language into editable LiDAR sequences. Instructions are parsed into ego-centric scene graphs, which a tri-branch diffusion model transforms into object layouts, trajectories, and shapes. A range-image diffusion model generates the initial scan, and an autoregressive module extends it into a temporally coherent sequence. The explicit layout design further supports object-level editing, such as insertion or relocation. To enable fair assessment, we provide EvalSuite, a benchmark spanning scene-, object-, and sequence-level metrics. On nuScenes, LiDARCrafter achieves state-of-the-art fidelity, controllability, and temporal consistency, offering a foundation for LiDAR-based simulation and data augmentation.
Authors: Wenyan Li, Raphael Tang, Chengzu Li, Caiqi Zhang, Ivan Vuli\'c, Anders S{\o}gaard
Abstract: Vision--language models (VLMs) often process visual inputs through a pretrained vision encoder, followed by a projection into the language model's embedding space via a connector component. While crucial for modality fusion, the potential information loss induced by this projection step and its direct impact on model capabilities remain understudied. We introduce two complementary approaches to examine and quantify this loss by analyzing the latent representation space. First, we evaluate semantic information preservation by analyzing changes in k-nearest neighbor relationships between image representations, before and after projection. Second, we directly measure information loss by reconstructing visual embeddings from the projected representation, localizing loss at an image patch level. Experiments reveal that connectors substantially distort the local geometry of visual representations, with k-nearest neighbors diverging by 40--60\% post-projection, correlating with degradation in retrieval performance. The patch-level embedding reconstruction provides interpretable insights for model behavior on visually grounded question-answering tasks, finding that areas of high information loss reliably predict instances where models struggle.
Authors: Zixuan Fu, Yan Ren, Finn Carter, Chenyue Wen, Le Ku, Daheng Yu, Emily Davis, Bo Zhang
Abstract: Diffusion models have achieved unprecedented success in image generation but pose increasing risks in terms of privacy, fairness, and security. A growing demand exists to \emph{erase} sensitive or harmful concepts (e.g., NSFW content, private individuals, artistic styles) from these models while preserving their overall generative capabilities. We introduce \textbf{SCORE} (Secure and Concept-Oriented Robust Erasure), a novel framework for robust concept removal in diffusion models. SCORE formulates concept erasure as an \emph{adversarial independence} problem, theoretically guaranteeing that the model's outputs become statistically independent of the erased concept. Unlike prior heuristic methods, SCORE minimizes the mutual information between a target concept and generated outputs, yielding provable erasure guarantees. We provide formal proofs establishing convergence properties and derive upper bounds on residual concept leakage. Empirically, we evaluate SCORE on Stable Diffusion and FLUX across four challenging benchmarks: object erasure, NSFW removal, celebrity face suppression, and artistic style unlearning. SCORE consistently outperforms state-of-the-art methods including EraseAnything, ANT, MACE, ESD, and UCE, achieving up to \textbf{12.5\%} higher erasure efficacy while maintaining comparable or superior image quality. By integrating adversarial optimization, trajectory consistency, and saliency-driven fine-tuning, SCORE sets a new standard for secure and robust concept erasure in diffusion models.
Authors: Zilong Zhang, Chujie Qin, Chunle Guo, Yong Zhang, Chao Xue, Ming-Ming Cheng, Chongyi Li
Abstract: This work presents Robust Representation Learning via Adaptive Mask (RAM++), a two-stage framework for all-in-one image restoration. RAM++ integrates high-level semantic understanding with low-level texture generation to achieve content-oriented robust restoration. It addresses the limitations of existing degradation-oriented methods in extreme scenarios (e.g., degradations strongly coupled with image structures). RAM++ also mitigates common challenges such as unbalanced performance across tasks, overfitting to seen degradations, and weak generalization to unseen ones through three key designs: 1) Adaptive Semantic-Aware Mask (AdaSAM): a pretraining strategy that applies pixel-level masks to semantically rich and textured regions. This design enables the network to learn both generative priors and image content priors from various degradations. 2) Mask Attribute Conductance (MAC): a selective fine-tuning strategy that adjusts the layers with higher contributions to bridge the integrity gap between masked pretraining and full-image fine-tuning while retaining learned priors. 3) Robust Feature Regularization (RFR): a strategy that leverages DINOv2's semantically consistent and degradation-invariant representations, together with efficient feature fusion, to achieve faithful and semantically coherent restoration. With these designs, RAM++ achieves robust, well-balanced, and state-of-the-art performance across seen, unseen, extreme, and mixed degradations. Our code and model will be released at https://github.com/DragonisCV/RAM
Authors: Bingyu Li, Haocheng Dong, Da Zhang, Zhiyuan Zhao, Junyu Gao, Xuelong Li
Abstract: Open-Vocabulary Remote Sensing Image Segmentation (OVRSIS), an emerging task that adapts Open-Vocabulary Segmentation (OVS) to the remote sensing (RS) domain, remains underexplored due to the absence of a unified evaluation benchmark and the domain gap between natural and RS images. To bridge these gaps, we first establish a standardized OVRSIS benchmark (\textbf{OVRSISBench}) based on widely-used RS segmentation datasets, enabling consistent evaluation across methods. Using this benchmark, we comprehensively evaluate several representative OVS/OVRSIS models and reveal their limitations when directly applied to remote sensing scenarios. Building on these insights, we propose \textbf{RSKT-Seg}, a novel open-vocabulary segmentation framework tailored for remote sensing. RSKT-Seg integrates three key components: (1) a Multi-Directional Cost Map Aggregation (RS-CMA) module that captures rotation-invariant visual cues by computing vision-language cosine similarities across multiple directions; (2) an Efficient Cost Map Fusion (RS-Fusion) transformer, which jointly models spatial and semantic dependencies with a lightweight dimensionality reduction strategy; and (3) a Remote Sensing Knowledge Transfer (RS-Transfer) module that injects pre-trained knowledge and facilitates domain adaptation via enhanced upsampling. Extensive experiments on the benchmark show that RSKT-Seg consistently outperforms strong OVS baselines by +3.8 mIoU and +5.9 mACC, while achieving 2x faster inference through efficient aggregation. Our code is \href{https://github.com/LiBingyu01/RSKT-Seg}{\textcolor{blue}{here}}.
Authors: Zirui Zheng, Takashi Isobe, Tong Shen, Xu Jia, Jianbin Zhao, Xiaomin Li, Mengmeng Ge, Baolu Li, Qinghe Wang, Dong Li, Dong Zhou, Yunzhi Zhuge, Huchuan Lu, Emad Barsoum
Abstract: While autoregressive (AR) models have demonstrated remarkable success in image generation, extending them to layout-conditioned generation remains challenging due to the sparse nature of layout conditions and the risk of feature entanglement. We present Structured Masking for AR-based Layout-to-Image (SMARLI), a novel framework for layoutto-image generation that effectively integrates spatial layout constraints into AR-based image generation. To equip AR model with layout control, a specially designed structured masking strategy is applied to attention computation to govern the interaction among the global prompt, layout, and image tokens. This design prevents mis-association between different regions and their descriptions while enabling sufficient injection of layout constraints into the generation process. To further enhance generation quality and layout accuracy, we incorporate Group Relative Policy Optimization (GRPO) based post-training scheme with specially designed layout reward functions for next-set-based AR models. Experimental results demonstrate that SMARLI is able to seamlessly integrate layout tokens with text and image tokens without compromising generation quality. It achieves superior layoutaware control while maintaining the structural simplicity and generation efficiency of AR models.
Authors: Haiyu Yang, Enhong Liu, Jennifer Sun, Sumit Sharma, Meike van Leerdam, Sebastien Franceschini, Puchun Niu, Miel Hostens
Abstract: Animal behavior analysis plays a crucial role in understanding animal welfare, health status, and productivity in agricultural settings. However, traditional manual observation methods are time-consuming, subjective, and limited in scalability. We present a modular pipeline that leverages open-sourced state-of-the-art computer vision techniques to automate animal behavior analysis in a group housing environment. Our approach combines state-of-the-art models for zero-shot object detection, motion-aware tracking and segmentation, and advanced feature extraction using vision transformers for robust behavior recognition. The pipeline addresses challenges including animal occlusions and group housing scenarios as demonstrated in indoor pig monitoring. We validated our system on the Edinburgh Pig Behavior Video Dataset for multiple behavioral tasks. Our temporal model achieved 94.2% overall accuracy, representing a 21.2 percentage point improvement over existing methods. The pipeline demonstrated robust tracking capabilities with 93.3% identity preservation score and 89.3% object detection precision. The modular design suggests potential for adaptation to other contexts, though further validation across species would be required. The open-source implementation provides a scalable solution for behavior monitoring, contributing to precision pig farming and welfare assessment through automated, objective, and continuous analysis.
Authors: Yuchen Deng, Xiuyang Wu, Hai-Tao Zheng, Suiyang Zhang, Yi He, Yuxing Han
Abstract: Existing talking-head animation approaches based on Generative Adversarial Networks (GANs) or diffusion models often suffer from inter-frame flicker, identity drift, and slow inference. These limitations inherent to their video generation pipelines restrict their suitability for applications. To address this, we introduce AvatarSync, an autoregressive framework on phoneme representations that generates realistic and controllable talking-head animations from a single reference image, driven directly text or audio input. In addition, AvatarSync adopts a two-stage generation strategy, decoupling semantic modeling from visual dynamics, which is a deliberate "Divide and Conquer" design. The first stage, Facial Keyframe Generation (FKG), focuses on phoneme-level semantic representation by leveraging the many-to-one mapping from text or audio to phonemes. A Phoneme-to-Visual Mapping is constructed to anchor abstract phonemes to character-level units. Combined with a customized Text-Frame Causal Attention Mask, the keyframes are generated. The second stage, inter-frame interpolation, emphasizes temporal coherence and visual smoothness. We introduce a timestamp-aware adaptive strategy based on a selective state space model, enabling efficient bidirectional context reasoning. To support deployment, we optimize the inference pipeline to reduce latency without compromising visual fidelity. Extensive experiments show that AvatarSync outperforms existing talking-head animation methods in visual fidelity, temporal consistency, and computational efficiency, providing a scalable and controllable solution.
Authors: Sebastian Diaz, Benjamin Billot, Neel Dey, Molin Zhang, Esra Abaci Turk, P. Ellen Grant, Polina Golland, Elfar Adalsteinsson
Abstract: Fetal motion is a critical indicator of neurological development and intrauterine health, yet its quantification remains challenging, particularly at earlier gestational ages (GA). Current methods track fetal motion by predicting the location of annotated landmarks on 3D echo planar imaging (EPI) time-series, primarily in third-trimester fetuses. The predicted landmarks enable simplification of the fetal body for downstream analysis. While these methods perform well within their training age distribution, they consistently fail to generalize to early GAs due to significant anatomical changes in both mother and fetus across gestation, as well as the difficulty of obtaining annotated early GA EPI data. In this work, we develop a cross-population data augmentation framework that enables pose estimation models to robustly generalize to younger GA clinical cohorts using only annotated images from older GA cohorts. Specifically, we introduce a fetal-specific augmentation strategy that simulates the distinct intrauterine environment and fetal positioning of early GAs. Our experiments find that cross-population augmentation yields reduced variability and significant improvements across both older GA and challenging early GA cases. By enabling more reliable pose estimation across gestation, our work potentially facilitates early clinical detection and intervention in challenging 4D fetal imaging settings. Code is available at https://github.com/sebodiaz/cross-population-pose.
Authors: Farahdiba Zarin, Nicolas Padoy, J\'er\'emy Dana, Vinkle Srivastav
Abstract: The fine-grained surface reconstruction of different organs from 3D medical imaging can provide advanced diagnostic support and improved surgical planning. However, the representation of the organs is often limited by the resolution, with a detailed higher resolution requiring more memory and computing footprint. Implicit representations of objects have been proposed to alleviate this problem in general computer vision by providing compact and differentiable functions to represent the 3D object shapes. However, architectural and data-related differences prevent the direct application of these methods to medical images. This work introduces ImplMORe, an end-to-end deep learning method using implicit surface representations for multi-organ reconstruction from 3D medical images. ImplMORe incorporates local features using a 3D CNN encoder and performs multi-scale interpolation to learn the features in the continuous domain using occupancy functions. We apply our method for single and multiple organ reconstructions using the totalsegmentator dataset. By leveraging the continuous nature of occupancy functions, our approach outperforms the discrete explicit representation based surface reconstruction approaches, providing fine-grained surface details of the organ at a resolution higher than the given input image. The source code will be made publicly available at: https://github.com/CAMMA-public/ImplMORe
Authors: Zhi Qin Tan, Xiatian Zhu, Owen Addison, Yunpeng Li
Abstract: Cone-Beam Computed Tomography (CBCT) is a widely used 3D imaging technique in dentistry, providing volumetric information about the anatomical structures of jaws and teeth. Accurate segmentation of these anatomies is critical for clinical applications such as diagnosis and surgical planning, but remains time-consuming and challenging. In this paper, we present U-Mamba2, a new neural network architecture designed for multi-anatomy CBCT segmentation in the context of the ToothFairy3 challenge. U-Mamba2 integrates the Mamba2 state space models into the U-Net architecture, enforcing stronger structural constraints for higher efficiency without compromising performance. In addition, we integrate interactive click prompts with cross-attention blocks, pre-train U-Mamba2 using self-supervised learning, and incorporate dental domain knowledge into the model design to address key challenges of dental anatomy segmentation in CBCT. Extensive experiments, including independent tests, demonstrate that U-Mamba2 is both effective and efficient, securing top 3 places in both tasks of the Toothfairy3 challenge. In Task 1, U-Mamba2 achieved a mean Dice of 0.792, HD95 of 93.19 with the held-out test data, with an average inference time of XX (TBC during the ODIN workshop). In Task 2, U-Mamba2 achieved the mean Dice of 0.852 and HD95 of 7.39 with the held-out test data. The code is publicly available at https://github.com/zhiqin1998/UMamba2.
Authors: Xiaodong Wang, Ping Wang, Zijun He, Mengjie Qin, Xin Yuan
Abstract: Coded aperture snapshot spectral imaging (CASSI) retrieves a 3D hyperspectral image (HSI) from a single 2D compressed measurement, which is a highly challenging reconstruction task. Recent deep unfolding networks (DUNs), empowered by explicit data-fidelity updates and implicit deep denoisers, have achieved the state of the art in CASSI reconstruction. However, existing unfolding approaches suffer from uncontrollable reconstruction trajectories, leading to abrupt quality jumps and non-gradual refinement across stages. Inspired by diffusion trajectories and flow matching, we propose a novel trajectory-controllable unfolding framework that enforces smooth, continuous optimization paths from noisy initial estimates to high-quality reconstructions. To achieve computational efficiency, we design an efficient spatial-spectral Transformer tailored for hyperspectral reconstruction, along with a frequency-domain fusion module to gurantee feature consistency. Experiments on simulation and real data demonstrate that our method achieves better reconstruction quality and efficiency than prior state-of-the-art approaches.
Authors: Yihong Chen, Jiancheng Yang, Deniz Sayin Mercadier, Hieu Le, Juerg Schwitter, Pascal Fua
Abstract: Reconstructing cardiac motion from cine CMR sequences is critical for diagnosis, prediction, and intervention. Existing methods rely on complete CMR stacks to infer full heart motion, limiting their utility in intra-procedural scenarios where only sparse observations are available. We present TetHeart, the first end-to-end framework that unifies full 4D multi-structure heart mesh recovery from both offline full-stack acquisitions and intra-procedural sparse-slice observations. Our method leverages deep deformable tetrahedra, an explicit-implicit hybrid representation, to capture shape and motion in a coherent space shared across cardiac structures. It is initialized from high-quality pre-procedural or offline-acquired full stacks to build detailed, patient-specific heart meshes, which can then be updated using whatever slices are available, from full stacks down to a single slice. We further incorporate several key innovations: (i) an attentive mechanism for slice-adaptive 2D-3D feature assembly that dynamically integrates information from arbitrary numbers of slices at any position, combined with a distillation strategy from full-slice to sparse-slice settings to ensure accurate reconstruction under extreme sparsity; and (ii) a two-stage weakly supervised motion learning scheme requiring only keyframe (e.g., ED and ES) annotations. Trained and validated on three large public datasets and externally evaluated zero-shot on additional private interventional and public CMR datasets, TetHeart achieves state-of-the-art accuracy and strong generalization in both pre- and intra-procedural settings.
Authors: Bernardo Forni, Gabriele Lombardi, Federico Pozzi, Mirco Planamente
Abstract: Few-shot semantic segmentation has recently attracted great attention. The goal is to develop a model capable of segmenting unseen classes using only a few annotated samples. Most existing approaches adapt a pre-trained model by training from scratch an additional module. Achieving optimal performance with these approaches requires extensive training on large-scale datasets. The Segment Anything Model 2 (SAM2) is a foundational model for zero-shot image and video segmentation with a modular design. In this paper, we propose a Few-Shot segmentation method based on SAM2 (FS-SAM2), where SAM2's video capabilities are directly repurposed for the few-shot task. Moreover, we apply a Low-Rank Adaptation (LoRA) to the original modules in order to handle the diverse images typically found in standard datasets, unlike the temporally connected frames used in SAM2's pre-training. With this approach, only a small number of parameters is meta-trained, which effectively adapts SAM2 while benefiting from its impressive segmentation performance. Our method supports any K-shot configuration. We evaluate FS-SAM2 on the PASCAL-5$^i$, COCO-20$^i$ and FSS-1000 datasets, achieving remarkable results and demonstrating excellent computational efficiency during inference. Code is available at https://github.com/fornib/FS-SAM2
Authors: Ing. Ondrej Valach, Ing. Ivan Gruber
Abstract: Tram-human interaction safety is an important challenge, given that trams frequently operate in densely populated areas, where collisions can range from minor injuries to fatal outcomes. This paper addresses the issue from the perspective of designing a solution leveraging digital image processing, deep learning, and artificial intelligence to improve the safety of pedestrians, drivers, cyclists, pets, and tram passengers. We present RailSafeNet, a real-time framework that fuses semantic segmentation, object detection and a rule-based Distance Assessor to highlight track intrusions. Using only monocular video, the system identifies rails, localises nearby objects and classifies their risk by comparing projected distances with the standard 1435mm rail gauge. Experiments on the diverse RailSem19 dataset show that a class-filtered SegFormer B3 model achieves 65% intersection-over-union (IoU), while a fine-tuned YOLOv8 attains 75.6% mean average precision (mAP) calculated at an intersection over union (IoU) threshold of 0.50. RailSafeNet therefore delivers accurate, annotation-light scene understanding that can warn drivers before dangerous situations escalate. Code available at https://github.com/oValach/RailSafeNet.
Authors: Pu Jian, Junhong Wu, Wei Sun, Chen Wang, Shuo Ren, Jiajun Zhang
Abstract: Recent advances in text-only "slow-thinking" reasoning have prompted efforts to transfer this capability to vision-language models (VLMs), for training visual reasoning models (\textbf{VRMs}). owever, such transfer faces critical challenges: Effective "slow thinking" in VRMs requires \textbf{visual reflection}, the ability to check the reasoning process based on visual information. Through quantitative analysis, we observe that current VRMs exhibit limited visual reflection, as their attention to visual information diminishes rapidly with longer generated responses. To address this challenge, we propose a new VRM \textbf{Reflection-V}, which enhances visual reflection based on reasoning data construction for cold-start and reward design for reinforcement learning (RL). Firstly, we construct vision-centered reasoning data by leveraging an agent that interacts between VLMs and reasoning LLMs, enabling cold-start learning of visual reflection patterns. Secondly, a visual attention based reward model is employed during RL to encourage reasoning based on visual information. Therefore, \textbf{Reflection-V} demonstrates significant improvements across multiple visual reasoning benchmarks. Furthermore, \textbf{Reflection-V} maintains a stronger and more consistent reliance on visual information during visual reasoning, indicating effective enhancement in visual reflection capabilities.
Authors: Nojod M. Alotaibi, Areej M. Alhothali, Manar S. Ali
Abstract: Major depressive disorder (MDD) is a prevalent mental health condition that negatively impacts both individual well-being and global public health. Automated detection of MDD using structural magnetic resonance imaging (sMRI) and deep learning (DL) methods holds increasing promise for improving diagnostic accuracy and enabling early intervention. Most existing methods employ either voxel-level features or handcrafted regional representations built from predefined brain atlases, limiting their ability to capture complex brain patterns. This paper develops a unified pipeline that utilizes Vision Transformers (ViTs) for extracting 3D region embeddings from sMRI data and Graph Neural Network (GNN) for classification. We explore two strategies for defining regions: (1) an atlas-based approach using predefined structural and functional brain atlases, and (2) an cube-based method by which ViTs are trained directly to identify regions from uniformly extracted 3D patches. Further, cosine similarity graphs are generated to model interregional relationships, and guide GNN-based classification. Extensive experiments were conducted using the REST-meta-MDD dataset to demonstrate the effectiveness of our model. With stratified 10-fold cross-validation, the best model obtained 78.98% accuracy, 76.54% sensitivity, 81.58% specificity, 81.58% precision, and 78.98% F1-score. Further, atlas-based models consistently outperformed the cube-based approach, highlighting the importance of using domain-specific anatomical priors for MDD detection.
Authors: Hyolim Kang, Yunsu Park, Youngbeom Yoo, Yeeun Choi, Seon Joo Kim
Abstract: We introduce Hierarchical Streaming Video Understanding, a task that combines online temporal action localization with free-form description generation. Given the scarcity of datasets with hierarchical and fine-grained temporal annotations, we demonstrate that LLMs can effectively group atomic actions into higher-level events, enriching existing datasets. We then propose OpenHOUSE (Open-ended Hierarchical Online Understanding System for Events), which extends streaming action perception beyond action classification. OpenHOUSE features a specialized streaming module that accurately detects boundaries between closely adjacent actions, nearly doubling the performance of direct extensions of existing methods. We envision the future of streaming action perception in the integration of powerful generative models, with OpenHOUSE representing a key step in that direction.
Authors: Nishank Singla, Krisztian Koos, Farzin Haddadpour, Amin Honarmandi Shandiz, Lovish Chum, Xiaojian Xu, Qing Jin, Erhan Bas
Abstract: X-ray imaging is a ubiquitous in radiology, yet most existing AI foundation models are limited to chest anatomy and fail to generalize across broader clinical tasks. In this work, we introduce XR-0, the multi-anatomy X-ray foundation model using self-supervised learning on a large, private dataset of 1.15 million images spanning diverse anatomical regions and evaluated across 12 datasets and 20 downstream tasks, including classification, retrieval, segmentation, localization, visual grounding, and report generation. XR-0 achieves state-of-the-art performance on most multi-anatomy tasks and remains competitive on chest-specific benchmarks. Our results demonstrate that anatomical diversity and supervision are critical for building robust, general-purpose medical vision models, paving the way for scalable and adaptable AI systems in radiology.
Authors: M. Bolhassani, B. Veasey, E. Daugherty, S. Keltner, N. Kumar, N. Dunlap, A. Amini
Abstract: This study investigates the efficacy of Low-Rank Adaptation (LoRA) for fine-tuning large Vision Models, DinoV2 and SwinV2, to diagnose Radiation-Induced Lung Injury (RILI) from X-ray CT scans following Stereotactic Body Radiation Therapy (SBRT). To evaluate the robustness and efficiency of this approach, we compare LoRA with traditional full fine-tuning and inference-only (no fine-tuning) methods. Cropped images of two sizes (50 mm3 and 75 mm3), centered at the treatment isocenter, in addition to different adaptation techniques for adapting the 2D LVMs for 3D data were used to determine the sensitivity of the models to spatial context. Experimental results show that LoRA achieves comparable or superior performance to traditional fine-tuning while significantly reducing computational costs and training times by requiring fewer trainable parameters.
Authors: Johanna Karras, Yingwei Li, Yasamin Jafarian, Ira Kemelmacher-Shlizerman
Abstract: Novel view synthesis (NVS) of in-the-wild garments is a challenging task due significant occlusions, complex human poses, and cloth deformations. Prior methods rely on synthetic 3D training data consisting of mostly unoccluded and static objects, leading to poor generalization on real-world clothing. In this paper, we propose HoloGarment (Hologram-Garment), a method that takes 1-3 images or a continuous video of a person wearing a garment and generates 360{\deg} novel views of the garment in a canonical pose. Our key insight is to bridge the domain gap between real and synthetic data with a novel implicit training paradigm leveraging a combination of large-scale real video data and small-scale synthetic 3D data to optimize a shared garment embedding space. During inference, the shared embedding space further enables dynamic video-to-360{\deg} NVS through the construction of a garment "atlas" representation by finetuning a garment embedding on a specific real-world video. The atlas captures garment-specific geometry and texture across all viewpoints, independent of body pose or motion. Extensive experiments show that HoloGarment achieves state-of-the-art performance on NVS of in-the-wild garments from images and videos. Notably, our method robustly handles challenging real-world artifacts -- such as wrinkling, pose variation, and occlusion -- while maintaining photorealism, view consistency, fine texture details, and accurate geometry. Visit our project page for additional results: https://johannakarras.github.io/HoloGarment
Authors: Felix B. Mueller, Timo Lueddecke, Richard Vogg, Alexander S. Ecker
Abstract: Computer vision for animal behavior offers promising tools to aid research in ecology, cognition, and to support conservation efforts. Video camera traps allow for large-scale data collection, but high labeling costs remain a bottleneck to creating large-scale datasets. We thus need data-efficient learning approaches. In this work, we show that we can utilize self-supervised learning to considerably improve action recognition on primate behavior. On two datasets of great ape behavior (PanAf and ChimpACT), we outperform published state-of-the-art action recognition models by 6.1 %pt. accuracy and 6.3 %pt. mAP, respectively. We achieve this by utilizing a pretrained V-JEPA model and applying domain-adaptive pretraining (DAP), i.e. continuing the pretraining with in-domain data. We show that most of the performance gain stems from the DAP. Our method promises great potential for improving the recognition of animal behavior, as DAP does not require labeled samples. Code is available at https://github.com/ecker-lab/dap-behavior
Authors: Salma Galaaoui, Eduardo Valle, David Picard, Nermin Samet
Abstract: In this paper, we present a comprehensive review of 3D human pose estimation and human mesh recovery from in-the-wild LiDAR point clouds. We compare existing approaches across several key dimensions, and propose a structured taxonomy to classify these methods. Following this taxonomy, we analyze each method's strengths, limitations, and design choices. In addition, (i) we perform a quantitative comparison of the three most widely used datasets, detailing their characteristics; (ii) we compile unified definitions of all evaluation metrics; and (iii) we establish benchmark tables for both tasks on these datasets to enable fair comparisons and promote progress in the field. We also outline open challenges and research directions critical for advancing LiDAR-based 3D human understanding. Moreover, we maintain an accompanying webpage that organizes papers according to our taxonomy and continuously update it with new studies: https://github.com/valeoai/3D-Human-Pose-Shape-Estimation-from-LiDAR
URLs: https://github.com/valeoai/3D-Human-Pose-Shape-Estimation-from-LiDAR
Authors: Yang Zhou, Yifan Wang, Jianjun Zhou, Wenzheng Chang, Haoyu Guo, Zizun Li, Kaijing Ma, Xinyue Li, Yating Wang, Haoyi Zhu, Mingyu Liu, Dingning Liu, Jiange Yang, Zhoujie Fu, Junyi Chen, Chunhua Shen, Jiangmiao Pang, Kaipeng Zhang, Tong He
Abstract: The field of 4D world modeling - aiming to jointly capture spatial geometry and temporal dynamics - has witnessed remarkable progress in recent years, driven by advances in large-scale generative models and multimodal learning. However, the development of truly general 4D world models remains fundamentally constrained by the availability of high-quality data. Existing datasets and benchmarks often lack the dynamic complexity, multi-domain diversity, and spatial-temporal annotations required to support key tasks such as 4D geometric reconstruction, future prediction, and camera-control video generation. To address this gap, we introduce OmniWorld, a large-scale, multi-domain, multi-modal dataset specifically designed for 4D world modeling. OmniWorld consists of a newly collected OmniWorld-Game dataset and several curated public datasets spanning diverse domains. Compared with existing synthetic datasets, OmniWorld-Game provides richer modality coverage, larger scale, and more realistic dynamic interactions. Based on this dataset, we establish a challenging benchmark that exposes the limitations of current state-of-the-art (SOTA) approaches in modeling complex 4D environments. Moreover, fine-tuning existing SOTA methods on OmniWorld leads to significant performance gains across 4D reconstruction and video generation tasks, strongly validating OmniWorld as a powerful resource for training and evaluation. We envision OmniWorld as a catalyst for accelerating the development of general-purpose 4D world models, ultimately advancing machines' holistic understanding of the physical world.
Authors: Zixin Yin, Xili Dai, Duomin Wang, Xianfang Zeng, Lionel M. Ni, Gang Yu, Heung-Yeung Shum
Abstract: The reliance on implicit point matching via attention has become a core bottleneck in drag-based editing, resulting in a fundamental compromise on weakened inversion strength and costly test-time optimization (TTO). This compromise severely limits the generative capabilities of diffusion models, suppressing high-fidelity inpainting and text-guided creation. In this paper, we introduce LazyDrag, the first drag-based image editing method for Multi-Modal Diffusion Transformers, which directly eliminates the reliance on implicit point matching. In concrete terms, our method generates an explicit correspondence map from user drag inputs as a reliable reference to boost the attention control. This reliable reference opens the potential for a stable full-strength inversion process, which is the first in the drag-based editing task. It obviates the necessity for TTO and unlocks the generative capability of models. Therefore, LazyDrag naturally unifies precise geometric control with text guidance, enabling complex edits that were previously out of reach: opening the mouth of a dog and inpainting its interior, generating new objects like a ``tennis ball'', or for ambiguous drags, making context-aware changes like moving a hand into a pocket. Additionally, LazyDrag supports multi-round workflows with simultaneous move and scale operations. Evaluated on the DragBench, our method outperforms baselines in drag accuracy and perceptual quality, as validated by VIEScore and human evaluation. LazyDrag not only establishes new state-of-the-art performance, but also paves a new way to editing paradigms.
Authors: Zhongrui Gui, Junyu Xie, Tengda Han, Weidi Xie, Andrew Zisserman
Abstract: Animated movies are captivating for their unique character designs and imaginative storytelling, yet they pose significant challenges for existing recognition systems. Unlike the consistent visual patterns detected by conventional face recognition methods, animated characters exhibit extreme diversity in their appearance, motion, and deformation. In this work, we propose an audio-visual pipeline to enable automatic and robust animated character recognition, and thereby enhance character-centric understanding of animated movies. Central to our approach is the automatic construction of an audio-visual character bank from online sources. This bank contains both visual exemplars and voice (audio) samples for each character, enabling subsequent multi-modal character recognition despite long-tailed appearance distributions. Building on accurate character recognition, we explore two downstream applications: Audio Description (AD) generation for visually impaired audiences, and character-aware subtitling for the hearing impaired. To support research in this domain, we introduce CMD-AM, a new dataset of 75 animated movies with comprehensive annotations. Our character-centric pipeline demonstrates significant improvements in both accessibility and narrative comprehension for animated content over prior face-detection-based approaches. For the code and dataset, visit https://www.robots.ox.ac.uk/~vgg/research/animated_ad/.
Authors: Friedrich Wolf-Monheim
Abstract: Convolutional neural networks (CNNs) are widely used in computer vision. They can be used not only for conventional digital image material to recognize patterns, but also for feature extraction from digital imagery representing spectral and rhythm features extracted from time-domain digital audio signals for the acoustic classification of sounds. Different spectral and rhythm feature representations like mel-scaled spectrograms, mel-frequency cepstral coefficients (MFCCs), cyclic tempograms, short-time Fourier transform (STFT) chromagrams, constant-Q transform (CQT) chromagrams and chroma energy normalized statistics (CENS) chromagrams are investigated in terms of the audio classification performance using a deep convolutional neural network. It can be clearly shown that the mel-scaled spectrograms and the mel-frequency cepstral coefficients (MFCCs) perform significantly better than the other spectral and rhythm features investigated in this research for audio classification tasks using deep CNNs. The experiments were carried out with the aid of the ESC-50 dataset with 2,000 labeled environmental audio recordings.
Authors: Hemanth Chandravamsi, Dhanush V. Shenoy, Itay Zinn, Shimon Pisnoy, Steven H. Frankel
Abstract: Deep neural networks are known to exhibit a spectral learning bias, wherein low-frequency components are learned early in training, while high-frequency modes emerge more gradually in later epochs. However, when the target signal lacks low-frequency components and is dominated by broadband high frequencies, training suffers from a 'spectral bottleneck', and the model fails to reconstruct the entire signal, including the frequency components that lie within the network's representational capacity. We examine such a scenario in the context of implicit neural representations (INRs) with sinusoidal representation networks (SIRENs), focusing on the challenge of fitting high-frequency-dominant signals that are susceptible to spectral bottleneck. To effectively fit any target signal irrespective of it's frequency content, we propose a generalized target-aware 'weight perturbation scheme' (WINNER - weight initialization with noise for neural representations) for network initialization. The scheme perturbs uniformly initialized weights with Gaussian noise, where the noise scales are adaptively determined by the spectral centroid of the target signal. We show that the noise scales can provide control over the spectra of network activations and the eigenbasis of the empirical neural tangent kernel. This method not only addresses the spectral bottleneck but also yields faster convergence and with improved representation accuracy, outperforming state-of-the-art approaches in audio fitting and achieving notable gains in image fitting and denoising tasks. Beyond signal reconstruction, our approach opens new directions for adaptive weight initialization strategies in computer vision and scientific machine learning.
Authors: Qiuyu Chen (Kevin), Xin Jin (Kevin), Yue Song (Kevin), Xihui Liu (Kevin), Shuai Yang (Kevin), Tao Yang (Kevin), Ziqiang Li (Kevin), Jianguo Huang (Kevin), Yuntao Wei (Kevin), Ba'ao Xie (Kevin), Nicu Sebe (Kevin), Wenjun (Kevin), Zeng, Jooyeol Yun, Davide Abati, Mohamed Omran, Jaegul Choo, Amir Habibian, Auke Wiggers, Masato Kobayashi, Ning Ding, Toru Tamaki, Marzieh Gheisari, Auguste Genovesio, Yuheng Chen, Dingkun Liu, Xinyao Yang, Xinping Xu, Baicheng Chen, Dongrui Wu, Junhao Geng, Lexiang Lv, Jianxin Lin, Hanzhe Liang, Jie Zhou, Xuanxin Chen, Jinbao Wang, Can Gao, Zhangyi Wang, Zongze Li, Bihan Wen, Yixin Gao, Xiaohan Pan, Xin Li, Zhibo Chen, Baorui Peng, Zhongming Chen, Haoran Jin
Abstract: This paper reviews the 1st International Workshop on Disentangled Representation Learning for Controllable Generation (DRL4Real), held in conjunction with ICCV 2025. The workshop aimed to bridge the gap between the theoretical promise of Disentangled Representation Learning (DRL) and its application in realistic scenarios, moving beyond synthetic benchmarks. DRL4Real focused on evaluating DRL methods in practical applications such as controllable generation, exploring advancements in model robustness, interpretability, and generalization. The workshop accepted 9 papers covering a broad range of topics, including the integration of novel inductive biases (e.g., language), the application of diffusion models to DRL, 3D-aware disentanglement, and the expansion of DRL into specialized domains like autonomous driving and EEG analysis. This summary details the workshop's objectives, the themes of the accepted papers, and provides an overview of the methodologies proposed by the authors.
Authors: Mengzheng Yang, Yanfei Ren, David Osei Opoku, Ruochang Li, Peng Ren, Chunxiao Xing
Abstract: Current general-purpose large language models (LLMs) commonly exhibit knowledge hallucination and insufficient domain-specific adaptability in domain-specific tasks, limiting their effectiveness in specialized question answering scenarios. Retrieval-augmented generation (RAG) effectively tackles these challenges by integrating external knowledge to enhance accuracy and relevance. However, traditional RAG still faces limitations in domain knowledge accuracy and context modeling.To enhance domain-specific question answering performance, this work focuses on a graph-based RAG framework, emphasizing the critical role of knowledge graph quality during the generation process. We propose DSRAG (Domain-Specific RAG), a multimodal knowledge graph-driven retrieval-augmented generation framework designed for domain-specific applications. Our approach leverages domain-specific documents as the primary knowledge source, integrating heterogeneous information such as text, images, and tables to construct a multimodal knowledge graph covering both conceptual and instance layers. Building on this foundation, we introduce semantic pruning and structured subgraph retrieval mechanisms, combining knowledge graph context and vector retrieval results to guide the language model towards producing more reliable responses. Evaluations using the Langfuse multidimensional scoring mechanism show that our method excels in domain-specific question answering, validating the efficacy of integrating multimodal knowledge graphs with retrieval-augmented generation.
Authors: Sujatha Kotte, Vangala Govindakrishnan Saipradeep, Vidushi Walia, Dhandapani Nandagopal, Thomas Joseph, Naveen Sivadasan, Bhagat Singh Lali
Abstract: Motivation: Accurate classification of mitotic figures into normal and atypical types is crucial for tumor prognostication in digital pathology. However, developing robust deep learning models for this task is challenging due to the subtle morphological differences, as well as significant class and hardness imbalances in real-world histopathology datasets. Methods: We propose a novel deep learning approach based on a ResNet backbone with specialized classification heads. Our architecture uniquely models both the mitotic figure phenotype and the instance difficulty simultaneously. This method is specifically designed to handle the challenges of diverse tissue types, scanner variability, and imbalanced data. We employed focal loss to effectively mitigate the pronounced class imbalance, and a comprehensive data augmentation pipeline was implemented to enhance the model's robustness and generalizability. Results: Our approach demonstrated strong and consistent performance. In a 5-fold cross-validation on the MIDOG 2025 Track 2 dataset, it achieved a mean balanced accuracy of 0.8744 +/- 0.0093 and an ROC AUC of 0.9505 +/- 0.029. The model showed robust generalization across preliminary leaderboard evaluations, achieving an overall balanced accuracy of 0.8736 +/- 0.0204. Conclusion: The proposed method offers a reliable and generalizable solution for the classification of atypical and normal mitotic figures. By addressing the inherent challenges of real world data, our approach has the potential to support precise prognostic assessments in clinical practice and improve consistency in pathological diagnosis.
Authors: Haolin Yuan, Jingtao Li, Weiming Zhuang, Chen Chen, Lingjuan Lyu
Abstract: Federated Object Detection (FOD) enables clients to collaboratively train a global object detection model without accessing their local data from diverse domains. However, significant variations in environment, weather, and other domain specific factors hinder performance, making cross domain generalization a key challenge. Existing FOD methods often overlook the hardware constraints of edge devices and introduce local training regularizations that incur high computational costs, limiting real-world applicability. In this paper, we propose FEDEXCHANGE, a novel FOD framework that bridges domain gaps without introducing additional local computational overhead. FEDEXCHANGE employs a server side dynamic model exchange strategy that enables each client to gain insights from other clients' domain data without direct data sharing. Specifically, FEDEXCHANGE allows the server to alternate between model aggregation and model exchange. During aggregation rounds, the server aggregates all local models as usual. In exchange rounds, FEDEXCHANGE clusters and exchanges local models based on distance measures, allowing local models to learn from a variety of domains. As all operations are performed on the server side, clients can achieve improved cross domain utility without any additional computational overhead. Extensive evaluations demonstrate that FEDEXCHANGE enhances FOD performance, achieving 1.6X better mean average precision in challenging domains, such as rainy conditions, while requiring only 0.8X the computational resources compared to baseline methods.
Authors: Prajit Sengupta, Islem Rekik
Abstract: Medical image classification requires not only high predictive performance but also interpretability to ensure clinical trust and adoption. Graph Neural Networks (GNNs) offer a powerful framework for modeling relational structures within datasets; however, standard GNNs often operate as black boxes, limiting transparency and usability, particularly in clinical settings. In this work, we present an interpretable graph-based learning framework named FireGNN that integrates trainable fuzzy rules into GNNs for medical image classification. These rules embed topological descriptors - node degree, clustering coefficient, and label agreement - using learnable thresholds and sharpness parameters to enable intrinsic symbolic reasoning. Additionally, we explore auxiliary self-supervised tasks (e.g., homophily prediction, similarity entropy) as a benchmark to evaluate the contribution of topological learning. Our fuzzy-rule-enhanced model achieves strong performance across five MedMNIST benchmarks and the synthetic dataset MorphoMNIST, while also generating interpretable rule-based explanations. To our knowledge, this is the first integration of trainable fuzzy rules within a GNN.
Authors: Kaizhen Tan
Abstract: Air traffic controllers (ATCOs) issue high-intensity voice commands in dense airspace, where accurate workload modeling is critical for safety and efficiency. This paper proposes a multimodal deep learning framework that integrates structured data, trajectory sequences, and image features to estimate two key parameters in the ATCO command lifecycle: the time offset between a command and the resulting aircraft maneuver, and the command duration. A high-quality dataset was constructed, with maneuver points detected using sliding window and histogram-based methods. A CNN-Transformer ensemble model was developed for accurate, generalizable, and interpretable predictions. By linking trajectories to voice commands, this work offers the first model of its kind to support intelligent command generation and provides practical value for workload assessment, staffing, and scheduling.
Authors: Aoi Otani
Abstract: Continual learning -- the ability to acquire knowledge incrementally without forgetting previous skills -- is fundamental to natural intelligence. While the human brain excels at this, artificial neural networks struggle with "catastrophic forgetting," where learning new tasks erases previously acquired knowledge. This challenge is particularly severe for text-to-image diffusion models, which generate images from textual prompts. Additionally, these models face "mode collapse," where their outputs become increasingly repetitive over time. To address these challenges, we apply Latent Replay, a neuroscience-inspired approach, to diffusion models. Traditional replay methods mitigate forgetting by storing and revisiting past examples, typically requiring large collections of images. Latent Replay instead retains only compact, high-level feature representations extracted from the model's internal architecture. This mirrors the hippocampal process of storing neural activity patterns rather than raw sensory inputs, reducing memory usage while preserving critical information. Through experiments with five sequentially learned visual concepts, we demonstrate that Latent Replay significantly outperforms existing methods in maintaining model versatility. After learning all concepts, our approach retained 77.59% Image Alignment (IA) on the earliest concept, 14% higher than baseline methods, while maintaining diverse outputs. Surprisingly, random selection of stored latent examples outperforms similarity-based strategies. Our findings suggest that Latent Replay enables efficient continual learning for generative AI models, paving the way for personalized text-to-image models that evolve with user needs without excessive computational costs.
Authors: Aoife McDonald-Bowyer, Anjana Wijekoon, Ryan Laurance Love, Katie Allan, Scott Colvin, Aleksandra Gentry-Maharaj, Adeola Olaitan, Danail Stoyanov, Agostino Stilli, Sophia Bano
Abstract: Cervical cancer is highly preventable, yet persistent barriers to screening limit progress toward elimination goals. Speculum-free devices that integrate imaging and sampling could improve access, particularly in low-resource settings, but require reliable visual guidance. This study evaluates deep learning methods for real-time segmentation of the cervical os in transvaginal endoscopic images. Five encoder-decoder architectures were compared using 913 frames from 200 cases in the IARC Cervical Image Dataset, annotated by gynaecologists. Performance was assessed using IoU, DICE, detection rate, and distance metrics with ten-fold cross-validation. EndoViT/DPT, a vision transformer pre-trained on surgical video, achieved the highest DICE (0.50 \pm 0.31) and detection rate (0.87 \pm 0.33), outperforming CNN-based approaches. External validation with phantom data demonstrated robust segmentation under variable conditions at 21.5 FPS, supporting real-time feasibility. These results establish a foundation for integrating automated os recognition into speculum-free cervical screening devices to support non-expert use in both high- and low-resource contexts.
Authors: Ali Burak \"Unal, Cem Ata Baykara, Peter Krawitz, Mete Akg\"un
Abstract: Machine learning has shown promise in facial dysmorphology, where characteristic facial features provide diagnostic clues for rare genetic disorders. GestaltMatcher, a leading framework in this field, has demonstrated clinical utility across multiple studies, but its reliance on centralized datasets limits further development, as patient data are siloed across institutions and subject to strict privacy regulations. We introduce a federated GestaltMatcher service based on a cross-silo horizontal federated learning framework, which allows hospitals to collaboratively train a global ensemble feature extractor without sharing patient images. Patient data are mapped into a shared latent space, and a privacy-preserving kernel matrix computation framework enables syndrome inference and discovery while safeguarding confidentiality. New participants can directly benefit from and contribute to the system by adopting the global feature extractor and kernel configuration from previous training rounds. Experiments show that the federated service retains over 90% of centralized performance and remains robust to both varying silo numbers and heterogeneous data distributions.
Authors: Rabeya Tus Sadia, Qiang Cheng
Abstract: Predicting the success of start-up companies, defined as achieving an exit through acquisition or IPO, is a critical problem in entrepreneurship and innovation research. Datasets such as Crunchbase provide both structured information (e.g., funding rounds, industries, investor networks) and unstructured text (e.g., company descriptions), but effectively leveraging this heterogeneous data for prediction remains challenging. Traditional machine learning approaches often rely only on structured features and achieve moderate accuracy, while large language models (LLMs) offer rich reasoning abilities but struggle to adapt directly to domain-specific business data. We present \textbf{CrunchLLM}, a domain-adapted LLM framework for startup success prediction. CrunchLLM integrates structured company attributes with unstructured textual narratives and applies parameter-efficient fine-tuning strategies alongside prompt optimization to specialize foundation models for entrepreneurship data. Our approach achieves accuracy exceeding 80\% on Crunchbase startup success prediction, significantly outperforming traditional classifiers and baseline LLMs. Beyond predictive performance, CrunchLLM provides interpretable reasoning traces that justify its predictions, enhancing transparency and trustworthiness for financial and policy decision makers. This work demonstrates how adapting LLMs with domain-aware fine-tuning and structured--unstructured data fusion can advance predictive modeling of entrepreneurial outcomes. CrunchLLM contributes a methodological framework and a practical tool for data-driven decision making in venture capital and innovation policy.
Authors: Xingchen Wan, Han Zhou, Ruoxi Sun, Hootan Nakhost, Ke Jiang, Rajarishi Sinha, Sercan \"O. Ar{\i}k
Abstract: Text-to-image (T2I) models, while offering immense creative potential, are highly reliant on human intervention, posing significant usability challenges that often necessitate manual, iterative prompt engineering over often underspecified prompts. This paper introduces Maestro, a novel self-evolving image generation system that enables T2I models to autonomously self-improve generated images through iterative evolution of prompts, using only an initial prompt. Maestro incorporates two key innovations: 1) self-critique, where specialized multimodal LLM (MLLM) agents act as 'critics' to identify weaknesses in generated images, correct for under-specification, and provide interpretable edit signals, which are then integrated by a 'verifier' agent while preserving user intent; and 2) self-evolution, utilizing MLLM-as-a-judge for head-to-head comparisons between iteratively generated images, eschewing problematic images, and evolving creative prompt candidates that align with user intents. Extensive experiments on complex T2I tasks using black-box models demonstrate that Maestro significantly improves image quality over initial prompts and state-of-the-art automated methods, with effectiveness scaling with more advanced MLLM components. This work presents a robust, interpretable, and effective pathway towards self-improving T2I generation.
Authors: Jin Yang, Daniel S. Marcus, Aristeidis Sotiras
Abstract: Medical Vision Foundation Models (Med-VFMs) have superior capabilities of interpreting medical images due to the knowledge learned from self-supervised pre-training with extensive unannotated images. To improve their performance on adaptive downstream evaluations, especially segmentation, a few samples from target domains are selected randomly for fine-tuning them. However, there lacks works to explore the way of adapting Med-VFMs to achieve the optimal performance on target domains efficiently. Thus, it is highly demanded to design an efficient way of fine-tuning Med-VFMs by selecting informative samples to maximize their adaptation performance on target domains. To achieve this, we propose an Active Source-Free Domain Adaptation (ASFDA) method to efficiently adapt Med-VFMs to target domains for volumetric medical image segmentation. This ASFDA employs a novel Active Learning (AL) method to select the most informative samples from target domains for fine-tuning Med-VFMs without the access to source pre-training samples, thus maximizing their performance with the minimal selection budget. In this AL method, we design an Active Test Time Sample Query strategy to select samples from the target domains via two query metrics, including Diversified Knowledge Divergence (DKD) and Anatomical Segmentation Difficulty (ASD). DKD is designed to measure the source-target knowledge gap and intra-domain diversity. It utilizes the knowledge of pre-training to guide the querying of source-dissimilar and semantic-diverse samples from the target domains. ASD is designed to evaluate the difficulty in segmentation of anatomical structures by measuring predictive entropy from foreground regions adaptively. Additionally, our ASFDA method employs a Selective Semi-supervised Fine-tuning to improve the performance and efficiency of fine-tuning by identifying samples with high reliability from unqueried ones.
Authors: Mohammadreza Narimani, Alireza Pourreza, Ali Moghimi, Parastoo Farajpoor, Hamid Jafarbiglu, Mohsen Mesgaran
Abstract: Branched broomrape (Phelipanche ramosa (L.) Pomel) is a chlorophyll-deficient parasitic plant that threatens tomato production by extracting nutrients from the host, with reported yield losses up to 80 percent. Its mostly subterranean life cycle and prolific seed production (more than 200,000 seeds per plant, viable for up to 20 years) make early detection essential. We present an end-to-end pipeline that uses Sentinel-2 imagery and time-series analysis to identify broomrape-infested tomato fields in California. Regions of interest were defined from farmer-reported infestations, and images with less than 10 percent cloud cover were retained. We processed 12 spectral bands and sun-sensor geometry, computed 20 vegetation indices (e.g., NDVI, NDMI), and derived five plant traits (Leaf Area Index, Leaf Chlorophyll Content, Canopy Chlorophyll Content, Fraction of Absorbed Photosynthetically Active Radiation, and Fractional Vegetation Cover) using a neural network calibrated with ground-truth and synthetic data. Trends in Canopy Chlorophyll Content delineated transplanting-to-harvest periods, and phenology was aligned using growing degree days. Vegetation pixels were segmented and used to train a Long Short-Term Memory (LSTM) network on 18,874 pixels across 48 growing-degree-day time points. The model achieved 88 percent training accuracy and 87 percent test accuracy, with precision 0.86, recall 0.92, and F1 0.89. Permutation feature importance ranked NDMI, Canopy Chlorophyll Content, FAPAR, and a chlorophyll red-edge index as most informative, consistent with the physiological effects of infestation. Results show the promise of satellite-driven time-series modeling for scalable detection of parasitic stress in tomato farms.
Authors: Qingxiang Liu, Ting Huang, Zeyu Zhang, Hao Tang
Abstract: Embodied navigation requires agents to integrate perception, reasoning, and action for robust interaction in complex 3D environments. Existing approaches often suffer from incoherent and unstable reasoning traces that hinder generalization across diverse environments, and difficulty balancing long-horizon semantic reasoning with low-latency control for real-time navigation. To address these challenges, we propose Nav-R1, an embodied foundation model that unifies reasoning in embodied environments. We first construct Nav-CoT-110K, a large-scale dataset of step-by-step Chains-of-Thought (CoT) for embodied tasks, which enables cold-start initialization with structured reasoning. Building on this foundation, we design a GRPO-based reinforcement learning framework with three complementary rewards: format, understanding, and navigation, to improve structural adherence, semantic grounding, and path fidelity. Furthermore, we introduce a Fast-in-Slow reasoning paradigm, decoupling deliberate semantic reasoning from low-latency reactive control for efficient yet coherent navigation. Extensive evaluations on embodied AI benchmarks demonstrate that Nav-R1 consistently outperforms strong baselines, with over 8% average improvement in reasoning and navigation performance. Real-world deployment on a mobile robot further validates its robustness under limited onboard resources. Code: https://github.com/AIGeeksGroup/Nav-R1. Website: https://aigeeksgroup.github.io/Nav-R1.
URLs: https://github.com/AIGeeksGroup/Nav-R1., https://aigeeksgroup.github.io/Nav-R1.
Authors: Ali Hedayatnia, Mostafa Tavassolipour, Babak Nadjar Araabi, Abdol-Hossein Vahabie
Abstract: Randomized smoothing is a well-established method for achieving certified robustness against l2-adversarial perturbations. By incorporating a denoiser before the base classifier, pretrained classifiers can be seamlessly integrated into randomized smoothing without significant performance degradation. Among existing methods, Diffusion Denoised Smoothing - where a pretrained denoising diffusion model serves as the denoiser - has produced state-of-the-art results. However, we show that employing a denoising diffusion model introduces a covariate shift via misestimation of the added noise, ultimately degrading the smoothed classifier's performance. To address this issue, we propose a novel adversarial objective function focused on the added noise of the denoising diffusion model. This approach is inspired by our understanding of the origin of the covariate shift. Our goal is to train the base classifier to ensure it is robust against the covariate shift introduced by the denoiser. Our method significantly improves certified accuracy across three standard classification benchmarks - MNIST, CIFAR-10, and ImageNet - achieving new state-of-the-art performance in l2-adversarial perturbations. Our implementation is publicly available at https://github.com/ahedayat/Robustifying-DDS-Against-Covariate-Shift
URLs: https://github.com/ahedayat/Robustifying-DDS-Against-Covariate-Shift
Authors: Gurutva Patle, Nilay Girgaonkar, Nagabhushan Somraj, Rajiv Soundararajan
Abstract: 3D Gaussian Splatting (3DGS) has shown impressive results in real-time novel view synthesis. However, it often struggles under sparse-view settings, producing undesirable artifacts such as floaters, inaccurate geometry, and overfitting due to limited observations. We find that a key contributing factor is uncontrolled densification, where adding Gaussian primitives rapidly without guidance can harm geometry and cause artifacts. We propose AD-GS, a novel alternating densification framework that interleaves high and low densification phases. During high densification, the model densifies aggressively, followed by photometric loss based training to capture fine-grained scene details. Low densification then primarily involves aggressive opacity pruning of Gaussians followed by regularizing their geometry through pseudo-view consistency and edge-aware depth smoothness. This alternating approach helps reduce overfitting by carefully controlling model capacity growth while progressively refining the scene representation. Extensive experiments on challenging datasets demonstrate that AD-GS significantly improves rendering quality and geometric consistency compared to existing methods.
Authors: Kevin Valencia, Ziyang Liu, Justin Cui
Abstract: Although numerical weather forecasting methods have dominated the field, recent advances in deep learning methods, such as diffusion models, have shown promise in ensemble weather forecasting. However, such models are typically autoregressive and are thus computationally expensive. This is a challenge in climate science, where data can be limited, costly, or difficult to work with. In this work, we explore the impact of curated data selection on these autoregressive diffusion models. We evaluate several data sampling strategies and show that a simple time stratified sampling approach achieves performance similar to or better than full-data training. Notably, it outperforms the full-data model on certain metrics and performs only slightly worse on others while using only 20% of the training data. Our results demonstrate the feasibility of data-efficient diffusion training, especially for weather forecasting, and motivates future work on adaptive or model-aware sampling methods that go beyond random or purely temporal sampling.
Authors: Thomas Y. Chen
Abstract: We establish the first information-theoretic limits for multimodal retrieval. Casting ranking as lossy source coding, we derive a single-letter rate-distortion function $R(D)$ for reciprocal-rank distortion and prove a converse bound that splits into a modality-balanced term plus a skew penalty $\kappa\,\Delta H$ capturing entropy imbalance and cross-modal redundancy. We then construct an explicit entropy-weighted stochastic quantizer with an adaptive, per-modality temperature decoder; a Blahut-Arimoto argument shows this scheme achieves distortion within $O(n^{-1})$ of $R(D)$ using $n$ training triples. A VC-type analysis yields the first finite-sample excess-risk bound whose complexity scales sub-linearly in both the number of modalities and the entropy gap. Experiments on controlled Gaussian mixtures and Flickr30k confirm that our adaptive codes sit within two percentage points of the theoretical frontier, while fixed-temperature and naive CLIP baselines lag significantly. Taken together, our results give a principled answer to "how many bits per query are necessary" for high-quality multimodal retrieval and provide design guidance for entropy-aware contrastive objectives, continual-learning retrievers, and retrieval-augmented generators.
Authors: Omkar Shailendra Vengurlekar, Adithya Pediredla, Suren Jayasuriya
Abstract: Synthetic aperture sonar (SAS) reconstruction requires recovering both the spatial distribution of acoustic scatterers and their direction-dependent response. Time-domain backprojection is the most common 3D SAS reconstruction algorithm, but it does not model directionality and can suffer from sampling limitations, aliasing, and occlusion. Prior neural volumetric methods applied to synthetic aperture sonar treat each voxel as an isotropic scattering density, not modeling anisotropic returns. We introduce SH-SAS, an implicit neural representation that expresses the complex acoustic scattering field as a set of spherical harmonic (SH) coefficients. A multi-resolution hash encoder feeds a lightweight MLP that outputs complex SH coefficients up to a specified degree L. The zeroth-order coefficient acts as an isotropic scattering field, which also serves as the density term, while higher orders compactly capture directional scattering with minimal parameter overhead. Because the model predicts the complex amplitude for any transmit-receive baseline, training is performed directly from 1-D time-of-flight signals without the need to beamform intermediate images for supervision. Across synthetic and real SAS (both in-air and underwater) benchmarks, results show that SH-SAS performs better in terms of 3D reconstruction quality and geometric metrics than previous methods.
Authors: Zhi Chen
Abstract: Ultrasound imaging is widely used in clinical practice due to its cost-effectiveness, mobility, and safety. However, current AI research often treats disease prediction and tissue segmentation as two separate tasks and their model requires substantial computational overhead. In such a situation, we introduce UltraUPConvNet, a computationally efficient universal framework designed for both ultrasound image classification and segmentation. Trained on a large-scale dataset containing more than 9,700 annotations across seven different anatomical regions, our model achieves state-of-the-art performance on certain datasets with lower computational overhead. Our model weights and codes are available at https://github.com/yyxl123/UltraUPConvNet
Authors: Zheng Li, Pei Qu, Yufei Jia, Shihui Zhou, Haizhou Ge, Jiahang Cao, Jinni Zhou, Guyue Zhou, Jun Ma
Abstract: Deploying visual reinforcement learning (RL) policies in real-world manipulation is often hindered by camera viewpoint changes. A policy trained from a fixed front-facing camera may fail when the camera is shifted--an unavoidable situation in real-world settings where sensor placement is hard to manage appropriately. Existing methods often rely on precise camera calibration or struggle with large perspective changes. To address these limitations, we propose ManiVID-3D, a novel 3D RL architecture designed for robotic manipulation, which learns view-invariant representations through self-supervised disentangled feature learning. The framework incorporates ViewNet, a lightweight yet effective module that automatically aligns point cloud observations from arbitrary viewpoints into a unified spatial coordinate system without the need for extrinsic calibration. Additionally, we develop an efficient GPU-accelerated batch rendering module capable of processing over 5000 frames per second, enabling large-scale training for 3D visual RL at unprecedented speeds. Extensive evaluation across 10 simulated and 5 real-world tasks demonstrates that our approach achieves a 44.7% higher success rate than state-of-the-art methods under viewpoint variations while using 80% fewer parameters. The system's robustness to severe perspective changes and strong sim-to-real performance highlight the effectiveness of learning geometrically consistent representations for scalable robotic manipulation in unstructured environments. Our project website can be found in https://zheng-joe-lee.github.io/manivid3d/.
Authors: Yunheng Wang, Yuetong Fang, Taowen Wang, Yixiao Feng, Yawen Tan, Shuning Zhang, Peiran Liu, Yiding Ji, Renjing Xu
Abstract: Vision-and-Language Navigation in Continuous Environments (VLN-CE), which links language instructions to perception and control in the real world, is a core capability of embodied robots. Recently, large-scale pretrained foundation models have been leveraged as shared priors for perception, reasoning, and action, enabling zero-shot VLN without task-specific training. However, existing zero-shot VLN methods depend on costly perception and passive scene understanding, collapsing control to point-level choices. As a result, they are expensive to deploy, misaligned in action semantics, and short-sighted in planning. To address these issues, we present DreamNav that focuses on the following three aspects: (1) for reducing sensory cost, our EgoView Corrector aligns viewpoints and stabilizes egocentric perception; (2) instead of point-level actions, our Trajectory Predictor favors global trajectory-level planning to better align with instruction semantics; and (3) to enable anticipatory and long-horizon planning, we propose an Imagination Predictor to endow the agent with proactive thinking capability. On VLN-CE and real-world tests, DreamNav sets a new zero-shot state-of-the-art (SOTA), outperforming the strongest egocentric baseline with extra information by up to 7.49\% and 18.15\% in terms of SR and SPL metrics. To our knowledge, this is the first zero-shot VLN method to unify trajectory-level planning and active imagination while using only egocentric inputs.
Authors: Yitong Zhang, Ximo Li, Liyi Cai, Jia Li
Abstract: GUI agents built on LVLMs are increasingly used to interact with websites. However, their exposure to open-world content makes them vulnerable to Environmental Injection Attacks (EIAs) that hijack agent behavior via webpage elements. Many recent studies assume the attacker to be a regular user who can only upload a single trigger image, which is more realistic than earlier assumptions of website-level administrative control. However, these works still fall short of realism: (1) the trigger's position and surrounding context remain largely fixed between training and testing, failing to capture the dynamic nature of real webpages and (2) the trigger often occupies an unrealistically large area, whereas real-world images are typically small. To better reflect real-world scenarios, we introduce a more realistic threat model where the attacker is a regular user and the trigger image is small and embedded within a dynamically changing environment. As a result, existing attacks prove largely ineffective under this threat model. To better expose the vulnerabilities of GUI agents, we propose Chameleon, an attack framework with two main novelties. The first is LLM-Driven Environment Simulation, which automatically generates diverse and high-fidelity webpage simulations. The second is Attention Black Hole, which transforms attention weights into explicit supervisory signals that guide the agent's focus toward the trigger region. We evaluate Chameleon on 6 realistic websites and 4 representative LVLM-powered GUI agents, where it significantly outperforms existing methods. Ablation studies confirm that both novelties are critical to performance. Our findings reveal underexplored vulnerabilities in modern GUI agents and establish a robust foundation for future research on defense in open-world GUI agent systems. The code is publicly available at https://github.com/zhangyitonggg/attack2gui.
Authors: Qiuhao Liu, Ling Li, Yao Lu, Qi Xuan, Zhaowei Zhu, Jiaheng Wei
Abstract: Deep neural networks tend to memorize noisy labels, severely degrading their generalization performance. Although Mixup has demonstrated effectiveness in improving generalization and robustness, existing Mixup-based methods typically perform indiscriminate mixing without principled guidance on sample selection and mixing strategy, inadvertently propagating noisy supervision. To overcome these limitations, we propose SelectMix, a confidence-guided mixing framework explicitly tailored for noisy labels. SelectMix first identifies potentially noisy or ambiguous samples through confidence based mismatch analysis using K-fold cross-validation, then selectively blends identified uncertain samples with confidently predicted peers from their potential classes. Furthermore, SelectMix employs soft labels derived from all classes involved in the mixing process, ensuring the labels accurately represent the composition of the mixed samples, thus aligning supervision signals closely with the actual mixed inputs. Through extensive theoretical analysis and empirical evaluations on multiple synthetic (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100) and real-world benchmark datasets (CIFAR-N, MNIST and Clothing1M), we demonstrate that SelectMix consistently outperforms strong baseline methods, validating its effectiveness and robustness in learning with noisy labels.
Authors: Surajit Das, Pavel Zun
Abstract: This article presents a novel microscopy image analysis framework designed for low-budget labs equipped with a standard CPU desktop. The Python-based program enables cytometric analysis of live, unstained cells in culture through an advanced computer vision and machine learning pipeline. Crucially, the framework operates on label-free data, requiring no manually annotated training data or training phase. It is accessible via a user-friendly, cross-platform GUI that requires no programming skills, while also providing a scripting interface for programmatic control and integration by developers. The end-to-end workflow performs semantic and instance segmentation, feature extraction, analysis, evaluation, and automated report generation. Its modular architecture supports easy maintenance and flexible integration while supporting both single-image and batch processing. Validated on several unstained cell types from the public dataset of livecells, the framework demonstrates superior accuracy and reproducibility compared to contemporary tools like Cellpose and StarDist. Its competitive segmentation speed on a CPU-based platform highlights its significant potential for basic research and clinical applications -- particularly in cell transplantation for personalized medicine and muscle regeneration therapies.
Authors: Loka Li, Wong Yu Kang, Minghao Fu, Guangyi Chen, Zhenhao Chen, Gongxu Luo, Yuewen Sun, Salman Khan, Peter Spirtes, Kun Zhang
Abstract: Understanding human behavior traits is central to applications in human-computer interaction, computational social science, and personalized AI systems. Such understanding often requires integrating multiple modalities to capture nuanced patterns and relationships. However, existing resources rarely provide datasets that combine behavioral descriptors with complementary modalities such as facial attributes and biographical information. To address this gap, we present PersonaX, a curated collection of multimodal datasets designed to enable comprehensive analysis of public traits across modalities. PersonaX consists of (1) CelebPersona, featuring 9444 public figures from diverse occupations, and (2) AthlePersona, covering 4181 professional athletes across 7 major sports leagues. Each dataset includes behavioral trait assessments inferred by three high-performing large language models, alongside facial imagery and structured biographical features. We analyze PersonaX at two complementary levels. First, we abstract high-level trait scores from text descriptions and apply five statistical independence tests to examine their relationships with other modalities. Second, we introduce a novel causal representation learning (CRL) framework tailored to multimodal and multi-measurement data, providing theoretical identifiability guarantees. Experiments on both synthetic and real-world data demonstrate the effectiveness of our approach. By unifying structured and unstructured analysis, PersonaX establishes a foundation for studying LLM-inferred behavioral traits in conjunction with visual and biographical attributes, advancing multimodal trait analysis and causal reasoning.
Authors: Shresth Grover, Akshay Gopalkrishnan, Bo Ai, Henrik I. Christensen, Hao Su, Xuanlin Li
Abstract: Vision-language-action (VLA) models finetuned from vision-language models (VLMs) hold the promise of leveraging rich pretrained representations to build generalist robots across diverse tasks and environments. However, direct fine-tuning on robot data often disrupts these representations and limits generalization. We present a framework that better preserves pretrained features while adapting them for robot manipulation. Our approach introduces three components: (i) a dual-encoder design with one frozen vision encoder to retain pretrained features and another trainable for task adaptation, (ii) a string-based action tokenizer that casts continuous actions into character sequences aligned with the model's pretraining domain, and (iii) a co-training strategy that combines robot demonstrations with vision-language datasets emphasizing spatial reasoning and affordances. Evaluations in simulation and on real robots show that our method improves robustness to visual perturbations, generalization to novel instructions and environments, and overall task success compared to baselines.
Authors: Amir Taherin, Juyi Lin, Arash Akbari, Arman Akbari, Pu Zhao, Weiwei Chen, David Kaeli, Yanzhi Wang
Abstract: Vision-Language-Action (VLA) models have emerged as powerful generalist policies for robotic control, yet their performance scaling across model architectures and hardware platforms, as well as their associated power budgets, remain poorly understood. This work presents an evaluation of five representative VLA models -- spanning state-of-the-art baselines and two newly proposed architectures -- targeting edge and datacenter GPU platforms. Using the LIBERO benchmark, we measure accuracy alongside system-level metrics, including latency, throughput, and peak memory usage, under varying edge power constraints and high-performance datacenter GPU configurations. Our results identify distinct scaling trends: (1) architectural choices, such as action tokenization and model backbone size, strongly influence throughput and memory footprint; (2) power-constrained edge devices exhibit non-linear performance degradation, with some configurations matching or exceeding older datacenter GPUs; and (3) high-throughput variants can be achieved without significant accuracy loss. These findings provide actionable insights when selecting and optimizing VLAs across a range of deployment constraints. Our work challenges current assumptions about the superiority of datacenter hardware for robotic inference.
Authors: Vin\'icius Yu Okubo, Kotaro Shimizu, B. S. Shivaran, Gia-Wei Chern, Hae Yong Kim
Abstract: Labyrinthine stripe patterns are common in many physical systems, yet their lack of long-range order makes quantitative characterization challenging. We investigate the evolution of such patterns in bismuth-doped yttrium iron garnet (Bi:YIG) films subjected to a magnetic field annealing protocol. A U-Net deep learning model, trained with synthetic degradations including additive white Gaussian and Simplex noise, enables robust segmentation of experimental magneto-optical images despite noise and occlusions. Building on this segmentation, we develop a geometric analysis pipeline based on skeletonization, graph mapping, and spline fitting, which quantifies local stripe propagation through length and curvature measurements. Applying this framework to 444 images from 12 annealing protocol trials, we analyze the transition from the "quenched" state to a more parallel and coherent "annealed" state, and identify two distinct evolution modes (Type A and Type B) linked to field polarity. Our results provide a quantitative analysis of geometric and topological properties in magnetic stripe patterns and offer new insights into their local structural evolution, and establish a general tool for analyzing complex labyrinthine systems.
Authors: Jiacheng Liu, Chang Zou, Yuanhuiyi Lyu, Fei Ren, Shaobo Wang, Kaixin Li, Linfeng Zhang
Abstract: Diffusion models have revolutionized high-fidelity image and video synthesis, yet their computational demands remain prohibitive for real-time applications. These models face two fundamental challenges: strict temporal dependencies preventing parallelization, and computationally intensive forward passes required at each denoising step. Drawing inspiration from speculative decoding in large language models, we present SpeCa, a novel 'Forecast-then-verify' acceleration framework that effectively addresses both limitations. SpeCa's core innovation lies in introducing Speculative Sampling to diffusion models, predicting intermediate features for subsequent timesteps based on fully computed reference timesteps. Our approach implements a parameter-free verification mechanism that efficiently evaluates prediction reliability, enabling real-time decisions to accept or reject each prediction while incurring negligible computational overhead. Furthermore, SpeCa introduces sample-adaptive computation allocation that dynamically modulates resources based on generation complexity, allocating reduced computation for simpler samples while preserving intensive processing for complex instances. Experiments demonstrate 6.34x acceleration on FLUX with minimal quality degradation (5.5% drop), 7.3x speedup on DiT while preserving generation fidelity, and 79.84% VBench score at 6.1x acceleration for HunyuanVideo. The verification mechanism incurs minimal overhead (1.67%-3.5% of full inference costs), establishing a new paradigm for efficient diffusion model inference while maintaining generation quality even at aggressive acceleration ratios. Our codes have been released in Github: \textbf{https://github.com/Shenyi-Z/Cache4Diffusion}
Authors: Haisheng Wang, Weiming Zhi
Abstract: This paper formulates the Embodied Questions Answering (EQsA) problem, introduces a corresponding benchmark, and proposes a system to tackle the problem. Classical Embodied Question Answering (EQA) is typically formulated as answering one single question by actively exploring a 3D environment. Real deployments, however, often demand handling multiple questions that may arrive asynchronously and carry different urgencies. We formalize this setting as Embodied Questions Answering (EQsA) and present ParaEQsA, a framework for parallel, urgency-aware scheduling and answering. ParaEQsA leverages a group memory module shared among questions to reduce redundant exploration, and a priority-planning module to dynamically schedule questions. To evaluate this setting, we contribute the Parallel Asynchronous Embodied Questions (PAEQs) benchmark containing 40 indoor scenes and five questions per scene (200 in total), featuring asynchronous follow-up questions and urgency labels. We further propose metrics for EQsA performance: Direct Answer Rate (DAR), and Normalized Urgency-Weighted Latency (NUWL), which jointly measure efficiency and responsiveness of this system. ParaEQsA consistently outperforms strong sequential baselines adapted from recent EQA systems, while reducing exploration and delay. Empirical evaluations investigate the relative contributions of priority, urgency modeling, spatial scope, reward estimation, and dependency reasoning within our framework. Together, these results demonstrate that urgency-aware, parallel scheduling is key to making embodied agents responsive and efficient under realistic, multi-question workloads.
Authors: Wei-Hsin Yeh, Yu-An Su, Chih-Ning Chen, Yi-Hsueh Lin, Calvin Ku, Wen-Hsin Chiu, Min-Chun Hu, Lun-Wei Ku
Abstract: Motion instruction is a crucial task that helps athletes refine their technique by analyzing movements and providing corrective guidance. Although recent advances in multimodal models have improved motion understanding, generating precise and sport-specific instruction remains challenging due to the highly domain-specific nature of sports and the need for informative guidance. We propose CoachMe, a reference-based model that analyzes the differences between a learner's motion and a reference under temporal and physical aspects. This approach enables both domain-knowledge learning and the acquisition of a coach-like thinking process that identifies movement errors effectively and provides feedback to explain how to improve. In this paper, we illustrate how CoachMe adapts well to specific sports such as skating and boxing by learning from general movements and then leveraging limited data. Experiments show that CoachMe provides high-quality instructions instead of directions merely in the tone of a coach but without critical information. CoachMe outperforms GPT-4o by 31.6% in G-Eval on figure skating and by 58.3% on boxing. Analysis further confirms that it elaborates on errors and their corresponding improvement methods in the generated instructions. You can find CoachMe here: https://motionxperts.github.io/
Authors: Wa-Kin Lei, Jun-Cheng Chen, Shang-Tse Chen
Abstract: With the rise of large foundation models, split inference (SI) has emerged as a popular computational paradigm for deploying models across lightweight edge devices and cloud servers, addressing data privacy and computational cost concerns. However, most existing data reconstruction attacks have focused on smaller CNN classification models, leaving the privacy risks of foundation models in SI settings largely unexplored. To address this gap, we propose a novel data reconstruction attack based on guided diffusion, which leverages the rich prior knowledge embedded in a latent diffusion model (LDM) pre-trained on a large-scale dataset. Our method performs iterative reconstruction on the LDM's learned image prior, effectively generating high-fidelity images resembling the original data from their intermediate representations (IR). Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods, both qualitatively and quantitatively, in reconstructing data from deep-layer IRs of the vision foundation model. The results highlight the urgent need for more robust privacy protection mechanisms for large models in SI scenarios. Code is available at: https://github.com/ntuaislab/DRAG.
Authors: Mrinmay Sen, Ankita Das, Sidhant Nair, C Krishna Mohan
Abstract: Federated Domain Adaptation (FDA) is a federated learning (FL) approach that improves model performance at the target client by collaborating with source clients while preserving data privacy. FDA faces two primary challenges: domain shifts between source and target data and limited labeled data at the target. Most existing FDA methods focus on domain shifts, assuming ample target data, yet often neglect the combined challenges of both domain shifts and data scarcity. Moreover, approaches that address both challenges fail to prioritize sharing relevant information from source clients according to the target's objective. In this paper, we propose FedDAF, a novel approach addressing both challenges in FDA. FedDAF uses similarity-based aggregation of the global source model and target model by calculating model functional distance from their mean gradient fields computed on target data. This enables effective model aggregation based on the target objective, constructed using target data, even with limited data. While computing model functional distance between these two models, FedDAF computes the angle between their mean gradient fields and then normalizes with the Gompertz function. To construct the global source model, all the local source models are aggregated using simple average in the server. Experiments on real-world datasets demonstrate FedDAF's superiority over existing FL, PFL, and FDA methods in terms of achieving better test accuracy.
Authors: Jiacheng Liu, Pengxiang Ding, Qihang Zhou, Yuxuan Wu, Da Huang, Zimian Peng, Wei Xiao, Weinan Zhang, Lixin Yang, Cewu Lu, Donglin Wang
Abstract: Imitation learning (IL) enables efficient skill acquisition from demonstrations but often struggles with long-horizon tasks and high-precision control due to compounding errors. Residual policy learning offers a promising, model-agnostic solution by refining a base policy through closed-loop corrections. However, existing approaches primarily focus on local corrections to the base policy, lacking a global understanding of state evolution, which limits robustness and generalization to unseen scenarios. To address this, we propose incorporating global dynamics modeling to guide residual policy updates. Specifically, we leverage Koopman operator theory to impose linear time-invariant structure in a learned latent space, enabling reliable state transitions and improved extrapolation for long-horizon prediction and unseen environments. We introduce KORR (Koopman-guided Online Residual Refinement), a simple yet effective framework that conditions residual corrections on Koopman-predicted latent states, enabling globally informed and stable action refinement. We evaluate KORR on long-horizon, fine-grained robotic furniture assembly tasks under various perturbations. Results demonstrate consistent gains in performance, robustness, and generalization over strong baselines. Our findings further highlight the potential of Koopman-based modeling to bridge modern learning methods with classical control theory. For more details, please refer to https://jiachengliu3.github.io/TrajBooster.
Authors: Marcus Lin, Jennifer Lai
Abstract: A healthy smile plays a significant role in functional as well as esthetic considerations, improving confidence. It is difficult for dental professionals to strike a balance between esthetic requirements and functional requirements. Traditional smile design has had heavy reliance on dentist expertise and used plaster models and hand drawings, raising questions about the outcome for patients. Digital technology, led by Dr. Christian Coachman in 2007, allows photographic and videographic assessments, enabling improved intercommunication among specialists and patients. Advances in artificial intelligence (AI) and big data have supported analysis of facial features and development of personalized smile designs in the last few years. Outputs are, however, susceptible to practitioner bias or limitations of training data, and may be suboptimal for individual users. The study presented here suggests a comprehensive system integrating AI, big data, and recognition technologies to automate the smile design process so that both experienced and inexperienced dentists can generate pleasing aesthetics with ease. The system has a Facial Feature Extraction Module and an Image Generation Module, serving diverse practitioner and patient needs. User data can be incorporated in future research for design optimization and testing of virtual and augmented reality for real-time previewing. Data gathered can also be employed in aesthetic preference analyses, which can enhance our knowledge of smile design in dental practice.
Authors: Mohammadreza Narimani, Alireza Pourreza, Ali Moghimi, Parastoo Farajpoor, Hamid Jafarbiglu, Mohsen B. Mesgaran
Abstract: Branched broomrape (Phelipanche ramosa) is a chlorophyll-deficient parasitic weed that threatens tomato production by extracting nutrients from the host. We investigate early detection using leaf-level spectral reflectance (400-2500 nm) and ensemble machine learning. In a field experiment in Woodland, California, we tracked 300 tomato plants across growth stages defined by growing degree days (GDD). Leaf reflectance was acquired with a portable spectrometer and preprocessed (band denoising, 1 nm interpolation, Savitzky-Golay smoothing, correlation-based band reduction). Clear class differences were observed near 1500 nm and 2000 nm water absorption features, consistent with reduced leaf water content in infected plants at early stages. An ensemble combining Random Forest, XGBoost, SVM with RBF kernel, and Naive Bayes achieved 89% accuracy at 585 GDD, with recalls of 0.86 (infected) and 0.93 (noninfected). Accuracy declined at later stages (e.g., 69% at 1568 GDD), likely due to senescence and weed interference. Despite the small number of infected plants and environmental confounders, results show that proximal sensing with ensemble learning enables timely detection of broomrape before canopy symptoms are visible, supporting targeted interventions and reduced yield losses.
Authors: Thomas A. Buckley, Riccardo Conci, Peter G. Brodeur, Jason Gusdorf, Sourik Beltr\'an, Bita Behrouzi, Byron Crowe, Jacob Dockterman, Muzzammil Muhammad, Sarah Ohnigian, Andrew Sanchez, James A. Diao, Aashna P. Shah, Daniel Restrepo, Eric S. Rosenberg, Andrew S. Lea, Marinka Zitnik, Scott H. Podolsky, Zahir Kanjee, Raja-Elie E. Abdulnour, Jacob M. Koshy, Adam Rodman, Arjun K. Manrai
Abstract: BACKGROUND: For over a century, the New England Journal of Medicine Clinicopathological Conferences (CPCs) have tested the reasoning of expert physicians and, recently, artificial intelligence (AI). However, prior AI evaluations have focused on final diagnoses without addressing the multifaceted reasoning and presentation skills required of expert discussants. METHODS: Using 7102 CPCs (1923-2025) and 1021 Image Challenges (2006-2025), we conducted extensive physician annotation and automated processing to create CPC-Bench, a physician-validated benchmark spanning 10 text-based and multimodal tasks, against which we evaluated leading large language models (LLMs). Then, we developed "Dr. CaBot," an AI discussant designed to produce written and slide-based video presentations using only the case presentation, modeling the role of the human expert in these cases. RESULTS: When challenged with 377 contemporary CPCs, o3 (OpenAI) ranked the final diagnosis first in 60% of cases and within the top ten in 84% of cases, outperforming a 20-physician baseline; next-test selection accuracy reached 98%. Event-level physician annotations quantified AI diagnostic accuracy per unit of information. Performance was lower on literature search and image tasks; o3 and Gemini 2.5 Pro (Google) achieved 67% accuracy on image challenges. In blinded comparisons of CaBot vs. human expert-generated text, physicians misclassified the source of the differential in 46 of 62 (74%) of trials, and scored CaBot more favorably across quality dimensions. To promote research, we are releasing CaBot and CPC-Bench. CONCLUSIONS: LLMs exceed physician performance on complex text-based differential diagnosis and convincingly emulate expert medical presentations, but image interpretation and literature retrieval remain weaker. CPC-Bench and CaBot may enable transparent and continued tracking of progress in medical AI.
Authors: Jie Huang, Wengang Zhou, Qilin Zhang, Houqiang Li, Weiping Li
Abstract: Millions of hearing impaired people around the world routinely use some variants of sign languages to communicate, thus the automatic translation of a sign language is meaningful and important. Currently, there are two sub-problems in Sign Language Recognition (SLR), i.e., isolated SLR that recognizes word by word and continuous SLR that translates entire sentences. Existing continuous SLR methods typically utilize isolated SLRs as building blocks, with an extra layer of preprocessing (temporal segmentation) and another layer of post-processing (sentence synthesis). Unfortunately, temporal segmentation itself is non-trivial and inevitably propagates errors into subsequent steps. Worse still, isolated SLR methods typically require strenuous labeling of each word separately in a sentence, severely limiting the amount of attainable training data. To address these challenges, we propose a novel continuous sign recognition framework, the Hierarchical Attention Network with Latent Space (LS-HAN), which eliminates the preprocessing of temporal segmentation. The proposed LS-HAN consists of three components: a two-stream Convolutional Neural Network (CNN) for video feature representation generation, a Latent Space (LS) for semantic gap bridging, and a Hierarchical Attention Network (HAN) for latent space based recognition. Experiments are carried out on two large scale datasets. Experimental results demonstrate the effectiveness of the proposed framework.
Authors: Tooba Imtiaz, Morgan Kohler, Jared Miller, Zifeng Wang, Masih Eskandar, Mario Sznaier, Octavia Camps, Jennifer Dy
Abstract: Adversarial attacks hamper the decision-making ability of neural networks by perturbing the input signal. The addition of calculated small distortion to images, for instance, can deceive a well-trained image classification network. In this work, we propose a novel attack technique called Sparse Adversarial and Interpretable Attack Framework (SAIF). Specifically, we design imperceptible attacks that contain low-magnitude perturbations at a small number of pixels and leverage these sparse attacks to reveal the vulnerability of classifiers. We use the Frank-Wolfe (conditional gradient) algorithm to simultaneously optimize the attack perturbations for bounded magnitude and sparsity with $O(1/\sqrt{T})$ convergence. Empirical results show that SAIF computes highly imperceptible and interpretable adversarial examples, and outperforms state-of-the-art sparse attack methods on the ImageNet dataset.
Authors: Bingkun Nian, Fenghe Tang, Jianrui Ding, Jie Yang, Zhonglong Zheng, Shaohua Kevin Zhou, Wei Liu
Abstract: Dynamic convolution demonstrates outstanding representation capabilities, which are crucial for natural image segmentation. However, it fails when applied to medical image segmentation (MIS) and infrared small target segmentation (IRSTS) due to limited data and limited fitting capacity. In this paper, we propose a new type of dynamic convolution called dynamic parameter convolution (DPConv) which shows superior fitting capacity, and it can efficiently leverage features from deep layers of encoder in reconstruction tasks to generate DPConv kernels that adapt to input variations.Moreover, we observe that DPConv, built upon deep features derived from reconstruction tasks, significantly enhances downstream segmentation performance. We refer to the segmentation network integrated with DPConv generated from reconstruction network as the siamese reconstruction-segmentation network (SRS). We conduct extensive experiments on seven datasets including five medical datasets and two infrared datasets, and the experimental results demonstrate that our method can show superior performance over several recently proposed methods. Furthermore, the zero-shot segmentation under unseen modality demonstrates the generalization of DPConv. The code is available at: https://github.com/fidshu/SRSNet.
Authors: Yechi Ma, Neehar Peri, Achal Dave, Wei Hua, Deva Ramanan, Shu Kong
Abstract: Contemporary autonomous vehicle (AV) benchmarks have advanced techniques for training 3D detectors. While class labels naturally follow a long-tailed distribution in the real world, existing benchmarks only focus on a few common classes (e.g., pedestrian and car) and neglect many rare but crucial classes (e.g., emergency vehicle and stroller). However, AVs must reliably detect both common and rare classes for safe operation in the open world. We address this challenge by formally studying the problem of Long-Tailed 3D Detection (LT3D), which evaluates all annotated classes, including those in-the-tail. We address LT3D with hierarchical losses that promote feature sharing across classes, and introduce diagnostic metrics that award partial credit to "reasonable" mistakes with respect to the semantic hierarchy. Further, we point out that rare-class accuracy is particularly improved via multi-modal late fusion (MMLF) of independently trained uni-modal LiDAR and RGB detectors. Such an MMLF framework allows us to leverage large-scale uni-modal datasets (with more examples for rare classes) to train better uni-modal detectors. Finally, we examine three critical components of our simple MMLF approach from first principles: whether to train 2D or 3D RGB detectors for fusion, whether to match RGB and LiDAR detections in 3D or the projected 2D image plane, and how to fuse matched detections. Extensive experiments reveal that 2D RGB detectors achieve better recognition accuracy for rare classes than 3D RGB detectors, matching on the 2D image plane mitigates depth estimation errors for better matching, and score calibration and probabilistic fusion notably improves the final performance further. Our MMLF significantly outperforms prior work for LT3D, particularly improving on the six rarest classes from 12.8 to 20.0 mAP! Our code and models are available on our project page.
Authors: Xinzhe Luo, Xin Wang, Linda Shapiro, Chun Yuan, Jianfeng Feng, Xiahai Zhuang
Abstract: This article presents a general Bayesian learning framework for multi-modal groupwise image registration. The method builds on probabilistic modelling of the image generative process, where the underlying common anatomy and geometric variations of the observed images are explicitly disentangled as latent variables. Therefore, groupwise image registration is achieved via hierarchical Bayesian inference. We propose a novel hierarchical variational auto-encoding architecture to realise the inference procedure of the latent variables, where the registration parameters can be explicitly estimated in a mathematically interpretable fashion. Remarkably, this new paradigm learns groupwise image registration in an unsupervised closed-loop self-reconstruction process, sparing the burden of designing complex image-based similarity measures. The computationally efficient disentangled network architecture is also inherently scalable and flexible, allowing for groupwise registration on large-scale image groups with variable sizes. Furthermore, the inferred structural representations from multi-modal images via disentanglement learning are capable of capturing the latent anatomy of the observations with visual semantics. Extensive experiments were conducted to validate the proposed framework, including four different datasets from cardiac, brain, and abdominal medical images. The results have demonstrated the superiority of our method over conventional similarity-based approaches in terms of accuracy, efficiency, scalability, and interpretability.
Authors: Huan-ang Gao, Mingju Gao, Jiaju Li, Wenyi Li, Rong Zhi, Hao Tang, Hao Zhao
Abstract: Semantic image synthesis (SIS) shows good promises for sensor simulation. However, current best practices in this field, based on GANs, have not yet reached the desired level of quality. As latent diffusion models make significant strides in image generation, we are prompted to evaluate ControlNet, a notable method for its dense control capabilities. Our investigation uncovered two primary issues with its results: the presence of weird sub-structures within large semantic areas and the misalignment of content with the semantic mask. Through empirical study, we pinpointed the cause of these problems as a mismatch between the noised training data distribution and the standard normal prior applied at the inference stage. To address this challenge, we developed specific noise priors for SIS, encompassing spatial, categorical, and a novel spatial-categorical joint prior for inference. This approach, which we have named SCP-Diff, has set new state-of-the-art results in SIS on Cityscapes, ADE20K and COCO-Stuff, yielding a FID as low as 10.53 on Cityscapes. The code and models can be accessed via the project page.
Authors: Sahiti Yerramilli, Jayant Sravan Tamarapalli, Tanmay Girish Kulkarni, Jonathan Francis, Eric Nyberg
Abstract: Deep Learning models are incredibly data-hungry and require very large labeled datasets for supervised learning. As a consequence, these models often suffer from overfitting, limiting their ability to generalize to real-world examples. Recent advancements in diffusion models have enabled the generation of photorealistic images based on textual inputs. Leveraging the substantial datasets used to train these diffusion models, we propose a technique to utilize generated images to augment existing datasets. This paper explores various strategies for effective data augmentation to improve the out-of-domain generalization capabilities of deep learning models.
Authors: Jiayin Zhu, Linlin Yang, Angela Yao
Abstract: We present InstructHumans, a novel framework for instruction-driven {animatable} 3D human texture editing. Existing text-based 3D editing methods often directly apply Score Distillation Sampling (SDS). SDS, designed for generation tasks, cannot account for the defining requirement of editing -- maintaining consistency with the source avatar. This work shows that naively using SDS harms editing, as it may destroy consistency. We propose a modified SDS for Editing (SDS-E) that selectively incorporates subterms of SDS across diffusion timesteps. We further enhance SDS-E with spatial smoothness regularization and gradient-based viewpoint sampling for edits with sharp and high-fidelity detailing. Incorporating SDS-E into a 3D human texture editing framework allows us to outperform existing 3D editing methods. Our avatars faithfully reflect the textual edits while remaining consistent with the original avatars. Project page: https://jyzhu.top/instruct-humans/.
Authors: Yang Liu, Jiahua Xiao, Xiang Song, Yu Guo, Peilin Jiang, Haiwei Yang, Fei Wang
Abstract: Effectively modeling global context information in hyperspectral image (HSI) denoising is crucial, but prevailing methods using convolution or transformers still face localized or computational efficiency limitations. Inspired by the emerging Selective State Space Model (Mamba) with nearly linear computational complexity and efficient long-term modeling, we present a novel HSI denoising network named HSIDMamba (HSDM). HSDM is tailored to exploit the capture of potential spatial-spectral dependencies effectively and efficiently for HSI denoising. In particular, HSDM comprises multiple Hyperspectral Continuous Scan Blocks (HCSB) to strengthen spatial-spectral interactions. HCSB links forward and backward scans and enhances information from eight directions through the State Space Model (SSM), strengthening the context representation learning of HSDM and improving denoising performance more effectively. In addition, to enhance the utilization of spectral information and mitigate the degradation problem caused by long-range scanning, spectral attention mechanism. Extensive evaluations against HSI denoising benchmarks validate the superior performance of HSDM, achieving state-of-the-art performance and surpassing the efficiency of the transformer method SERT by 31%.
Authors: Thao Nguyen, Matthew Wallingford, Sebastin Santy, Wei-Chiu Ma, Sewoong Oh, Ludwig Schmidt, Pang Wei Koh, Ranjay Krishna
Abstract: Massive web-crawled image-text datasets lay the foundation for recent progress in multimodal learning. These datasets are designed with the goal of training a model to do well on standard computer vision benchmarks, many of which, however, have been shown to be English-centric (e.g., ImageNet). Consequently, existing data curation techniques gravitate towards using predominantly English image-text pairs and discard many potentially useful non-English samples. Our work questions this practice. Multilingual data is inherently enriching not only because it provides a gateway to learn about culturally salient concepts, but also because it depicts common concepts differently from monolingual data. We thus conduct a systematic study to explore the performance benefits of using more samples of non-English origins with respect to English vision tasks. By translating all multilingual image-text pairs from a raw web crawl to English and re-filtering them, we increase the prevalence of (translated) multilingual data in the resulting training set. Pre-training on this dataset outperforms using English-only or English-dominated datasets on ImageNet, ImageNet distribution shifts, image-English-text retrieval and on average across 38 tasks from the DataComp benchmark. On a geographically diverse task like GeoDE, we also observe improvements across all regions, with the biggest gain coming from Africa. In addition, we quantitatively show that English and non-English data are significantly different in both image and (translated) text space. We hope that our findings motivate future work to be more intentional about including multicultural and multilingual data, not just when non-English or geographically diverse tasks are involved, but to enhance model capabilities at large. All translated captions and metadata (language, CLIP score, etc.) are available on HuggingFace.
Authors: Xu Cao, Yifan Shen, Bolin Lai, Wenqian Ye, Yunsheng Ma, Joerg Heintz, Jintai Chen, Meihuan Huang, Jianguo Cao, Aidong Zhang, James M. Rehg
Abstract: Recently, Multimodal Large Language Models (MLLMs) and Vision Language Models (VLMs) have shown great promise in language-guided perceptual tasks such as recognition, segmentation, and object detection. However, their effectiveness in addressing visual cognition problems that require high-level multi-image reasoning and visual working memory is not well-established. One such challenge is matrix reasoning - the cognitive ability to discern relationships among patterns in a set of images and extrapolate to predict subsequent patterns. This skill is crucial during the early neurodevelopmental stages of children. Inspired by the matrix reasoning tasks in Raven's Progressive Matrices (RPM) and Wechsler Intelligence Scale for Children (WISC), we propose a new dataset MaRs-VQA to evaluate the visual cognition capability of MLLMs and compare their performance with existing human visual cognition studies. Based on the training data of MaRs-VQA, we also finetune a baseline model Qwen2-VCog with multi-stage cognition reasoning annotations. Our comparative experiments with different baselines reveal a gap between MLLMs and human intelligence, highlighting the visual cognitive limitations of current MLLMs. We believe that the public release of MaRs-VQA and the Qwen2-VCog baseline model will drive progress toward the next generation of MLLMs with human-like visual cognition abilities. MaRs-VQA is available at huggingface.co/datasets/IrohXu/VCog-Bench. The training code of Qwen2-VCog is available at github.com/IrohXu/Cognition-MLLM.
Authors: Zhawnen Chen, Tianchun Wang, Yizhou Wang, Michal Kosinski, Xiang Zhang, Yun Fu, Sheng Li
Abstract: Can large multimodal models have a human-like ability for emotional and social reasoning, and if so, how does it work? Recent research has discovered emergent theory-of-mind (ToM) reasoning capabilities in large language models (LLMs). LLMs can reason about people's mental states by solving various text-based ToM tasks that ask questions about the actors' ToM (e.g., human belief, desire, intention). However, human reasoning in the wild is often grounded in dynamic scenes across time. Thus, we consider videos a new medium for examining spatio-temporal ToM reasoning ability. Specifically, we ask explicit probing questions about videos with abundant social and emotional reasoning content. We develop a pipeline for multimodal LLM for ToM reasoning using video and text. We also enable explicit ToM reasoning by retrieving key frames for answering a ToM question, which reveals how multimodal LLMs reason about ToM.
Authors: Youngmin Kim, Saejin Kim, Hoyeon Moon, Youngjae Yu, Junhyug Noh
Abstract: Scalp disorders are highly prevalent worldwide, yet remain underdiagnosed due to limited access to expert evaluation and the high cost of annotation. Although AI-based approaches hold great promise, their practical deployment is hindered by challenges such as severe data imbalance and the absence of pixel-level segmentation labels. To address these issues, we propose ScalpVision, an AI-driven system for the holistic diagnosis of scalp diseases. In ScalpVision, effective hair segmentation is achieved using pseudo image-label pairs and an innovative prompting method in the absence of traditional hair masking labels. Additionally, ScalpVision introduces DiffuseIT-M, a generative model adopted for dataset augmentation while maintaining hair information, facilitating improved predictions of scalp disease severity. Our experimental results affirm ScalpVision's efficiency in diagnosing a variety of scalp conditions, showcasing its potential as a valuable tool in dermatological care. Our code is available at https://github.com/winston1214/ScalpVision.
Authors: Carina I. Hausladen, Manuel Knott, Colin F. Camerer, Pietro Perona
Abstract: We explore social perception of human faces in CLIP, a widely used open-source vision-language model. To this end, we compare the similarity in CLIP embeddings between different textual prompts and a set of face images. Our textual prompts are constructed from well-validated social psychology terms denoting social perception. The face images are synthetic and are systematically and independently varied along six dimensions: the legally protected attributes of age, gender, and race, as well as facial expression, lighting, and pose. Independently and systematically manipulating face attributes allows us to study the effect of each on social perception and avoids confounds that can occur in wild-collected data due to uncontrolled systematic correlations between attributes. Thus, our findings are experimental rather than observational. Our main findings are three. First, while CLIP is trained on the widest variety of images and texts, it is able to make fine-grained human-like social judgments on face images. Second, age, gender, and race do systematically impact CLIP's social perception of faces, suggesting an undesirable bias in CLIP vis-a-vis legally protected attributes. Most strikingly, we find a strong pattern of bias concerning the faces of Black women, where CLIP produces extreme values of social perception across different ages and facial expressions. Third, facial expression impacts social perception more than age and lighting as much as age. The last finding predicts that studies that do not control for unprotected visual attributes may reach the wrong conclusions on bias. Our novel method of investigation, which is founded on the social psychology literature and on the experiments involving the manipulation of individual attributes, yields sharper and more reliable observations than previous observational methods and may be applied to study biases in any vision-language model.
Authors: Zhen Yang, Yanpeng Dong, Jiayu Wang, Heng Wang, Lichao Ma, Zijian Cui, Qi Liu, Haoran Pei, Kexin Zhang, Chao Zhang
Abstract: Multi-sensor fusion significantly enhances the accuracy and robustness of 3D semantic occupancy prediction, which is crucial for autonomous driving and robotics. However, most existing approaches depend on high-resolution images and complex networks to achieve top performance, hindering their deployment in practical scenarios. Moreover, current multi-sensor fusion approaches mainly focus on improving feature fusion while largely neglecting effective supervision strategies for those features. To address these issues, we propose DAOcc, a novel multi-modal occupancy prediction framework that leverages 3D object detection supervision to assist in achieving superior performance, while using a deployment-friendly image backbone and practical input resolution. In addition, we introduce a BEV View Range Extension strategy to mitigate performance degradation caused by lower image resolution. Extensive experiments demonstrate that DAOcc achieves new state-of-the-art results on both the Occ3D-nuScenes and Occ3D-Waymo benchmarks, and outperforms previous state-of-the-art methods by a significant margin using only a ResNet-50 backbone and 256*704 input resolution. With TensorRT optimization, DAOcc reaches 104.9 FPS while maintaining 54.2 mIoU on an NVIDIA RTX 4090 GPU. Code is available at https://github.com/AlphaPlusTT/DAOcc.
Authors: Louis Soum-Fontez, Jean-Emmanuel Deschaud, Fran\c{c}ois Goulette
Abstract: Autonomous systems rely on accurate 3D object detection from LiDAR data, yet most detectors are limited to a predefined set of known classes, making them vulnerable to unexpected out-of-distribution (OOD) objects. In this work, we present HD-OOD3D, a novel two-stage method for detecting unknown objects. We demonstrate the superiority of two-stage approaches over single-stage methods, achieving more robust detection of unknown objects while addressing key challenges in the evaluation protocol. Furthermore, we conduct an in-depth analysis of the standard evaluation protocol for OOD detection, revealing the critical impact of hyperparameter choices. To address the challenge of scaling the learning of unknown objects, we explore unsupervised training strategies to generate pseudo-labels for unknowns. Among the different approaches evaluated, our experiments show that top-5 auto-labelling offers more promising performance compared to simple resizing techniques.
Authors: Huali Xu, Li Liu, Tianpeng Liu, Shuaifeng Zhi, Shuzhou Sun, Ming-Ming Cheng
Abstract: Existing cross-domain few-shot learning (CDFSL) methods, which develop source-domain training strategies to enhance model transferability, face challenges with large-scale pre-trained models (LMs) due to inaccessible source data and training strategies. Moreover, fine-tuning LMs for CDFSL demands substantial computational resources, limiting practicality. This paper addresses the source-free CDFSL (SF-CDFSL) problem, tackling few-shot learning (FSL) in the target domain using only pre-trained models and a few target samples without source data or strategies. To overcome the challenge of inaccessible source data, this paper introduces Step-wise Distribution Alignment Guided Style Prompt Tuning (StepSPT), which implicitly narrows domain gaps through prediction distribution optimization. StepSPT proposes a style prompt to align target samples with the desired distribution and adopts a dual-phase optimization process. In the external process, a step-wise distribution alignment strategy factorizes prediction distribution optimization into a multi-step alignment problem to tune the style prompt. In the internal process, the classifier is updated using standard cross-entropy loss. Evaluations on five datasets demonstrate that StepSPT outperforms existing prompt tuning-based methods and SOTAs. Ablation studies further verify its effectiveness. Code will be made publicly available at https://github.com/xuhuali-mxj/StepSPT.
Authors: Meng Wei, Zhongnian Li, Peng Ying, Xinzheng Xu
Abstract: Recent advances in vision-language models (VLMs) have demonstrated remarkable capabilities in image classification by leveraging predefined sets of labels to construct text prompts for zero-shot reasoning. However, these approaches face significant limitations in undefined domains, where the label space is vocabulary-unknown and composite. We thus introduce Generative Semantic Labels (GSLs), a novel task that aims to predict a comprehensive set of semantic labels for an image without being constrained by a predefined labels set. Unlike traditional zero-shot classification, GSLs generates multiple semantic-level labels, encompassing objects, scenes, attributes, and relationships, thereby providing a richer and more accurate representation of image content. In this paper, we propose Chain-of-Action (CoA), an innovative method designed to tackle the GSLs task. CoA is motivated by the observation that enriched contextual information significantly improves generative performance during inference. Specifically, CoA decomposes the GSLs task into a sequence of detailed actions. Each action extracts and merges key information from the previous step, passing enriched context to the next, ultimately guiding the VLM to generate comprehensive and accurate semantic labels. We evaluate the effectiveness of CoA through extensive experiments on widely-used benchmark datasets. The results demonstrate significant improvements across key performance metrics, validating the capability of CoA to generate accurate and contextually rich semantic labels. Our work not only advances the state-of-the-art in generative semantic labels but also opens new avenues for applying VLMs in open-ended and dynamic real-world scenarios.
Authors: Chenyang Liu, Jiafan Zhang, Keyan Chen, Man Wang, Zhengxia Zou, Zhenwei Shi
Abstract: The interpretation of multi-temporal remote sensing imagery is critical for monitoring Earth's dynamic processes-yet previous change detection methods, which produce binary or semantic masks, fall short of providing human-readable insights into changes. Recent advances in Vision-Language Models (VLMs) have opened a new frontier by fusing visual and linguistic modalities, enabling spatio-temporal vision-language understanding: models that not only capture spatial and temporal dependencies to recognize changes but also provide a richer interactive semantic analysis of temporal images (e.g., generate descriptive captions and answer natural-language queries). In this survey, we present the first comprehensive review of RS-STVLMs. The survey covers the evolution of models from early task-specific models to recent general foundation models that leverage powerful large language models. We discuss progress in representative tasks, such as change captioning, change question answering, and change grounding. Moreover, we systematically dissect the fundamental components and key technologies underlying these models, and review the datasets and evaluation metrics that have driven the field. By synthesizing task-level insights with a deep dive into shared architectural patterns, we aim to illuminate current achievements and chart promising directions for future research in spatio-temporal vision-language understanding for remote sensing. We will keep tracing related works at https://github.com/Chen-Yang-Liu/Awesome-RS-SpatioTemporal-VLMs
URLs: https://github.com/Chen-Yang-Liu/Awesome-RS-SpatioTemporal-VLMs
Authors: Zixian Ma, Jianguo Zhang, Zhiwei Liu, Jieyu Zhang, Juntao Tan, Manli Shu, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Caiming Xiong, Ranjay Krishna, Silvio Savarese
Abstract: While open-source vision-language models perform well on simple question-answering, they still struggle with complex questions that require both perceptual and reasoning capabilities. We propose LATTE, a family of vision-language models that have LeArned to Think wiTh vision spEcialists. By offloading perception to state-of-the-art vision models, our approach enables vision-language models to focus solely on reasoning over high-quality perceptual information. To train LATTE, we synthesize and filter a large dataset of 293K multi-modal reasoning traces over perceptual outputs of vision specialists. LATTE trained on this data achieves significant 4-5% gains over baselines across 6 benchmarks covering both perception and reasoning abilities. Ablation studies reveal that the effectiveness of multi-modal reasoning traces depends on the data sources, formats, and quality of thoughts.
Authors: Will Gao, Dilin Wang, Yuchen Fan, Aljaz Bozic, Tuur Stuyck, Zhengqin Li, Zhao Dong, Rakesh Ranjan, Nikolaos Sarafianos
Abstract: We present a novel approach to shape editing, building on recent progress in 3D reconstruction from multi-view images. We formulate shape editing as a conditional reconstruction problem, where the model must reconstruct the input shape with the exception of a specified 3D region, in which the geometry should be generated from the conditional signal. To this end, we train a conditional Large Reconstruction Model (LRM) for masked reconstruction, using multi-view consistent masks rendered from a randomly generated 3D occlusion, and using one clean viewpoint as the conditional signal. During inference, we manually define a 3D region to edit and provide an edited image from a canonical viewpoint to fill that region. We demonstrate that, in just a single forward pass, our method not only preserves the input geometry in the unmasked region through reconstruction capabilities on par with SoTA, but is also expressive enough to perform a variety of mesh edits from a single image guidance that past works struggle with, while being 2-10x faster than the top-performing prior work.
Authors: Massimiliano Viola, Kevin Qu, Nando Metzger, Bingxin Ke, Alexander Becker, Konrad Schindler, Anton Obukhov
Abstract: Depth completion upgrades sparse depth measurements into dense depth maps guided by a conventional image. Existing methods for this highly ill-posed task operate in tightly constrained settings and tend to struggle when applied to images outside the training domain or when the available depth measurements are sparse, irregularly distributed, or of varying density. Inspired by recent advances in monocular depth estimation, we reframe depth completion as an image-conditional depth map generation guided by sparse measurements. Our method, Marigold-DC, builds on a pretrained latent diffusion model for monocular depth estimation and injects the depth observations as test-time guidance via an optimization scheme that runs in tandem with the iterative inference of denoising diffusion. The method exhibits excellent zero-shot generalization across a diverse range of environments and handles even extremely sparse guidance effectively. Our results suggest that contemporary monocular depth priors greatly robustify depth completion: it may be better to view the task as recovering dense depth from (dense) image pixels, guided by sparse depth; rather than as inpainting (sparse) depth, guided by an image. Project website: https://MarigoldDepthCompletion.github.io/
Authors: Ahmed Alhawwary, Janne Mustaniemi, Phong Nguyen-Ha, Janne Heikkil\"a
Abstract: Portraits or selfie images taken from a close distance typically suffer from perspective distortion. In this paper, we propose an end-to-end deep learning-based rectification pipeline to mitigate the effects of perspective distortion. We learn to predict the facial depth by training a deep CNN. The estimated depth is utilized to adjust the camera-to-subject distance by moving the camera farther, increasing the camera focal length, and reprojecting the 3D image features to the new perspective. The reprojected features are then fed to an inpainting module to fill in the missing pixels. We leverage a differentiable renderer to enable end-to-end training of our depth estimation and feature extraction nets to improve the rectified outputs. To boost the results of the inpainting module, we incorporate an auxiliary module to predict the horizontal movement of the camera which decreases the area that requires hallucination of challenging face parts such as ears. Unlike previous works, we process the full-frame input image at once without cropping the subject's face and processing it separately from the rest of the body, eliminating the need for complex post-processing steps to attach the face back to the subject's body. To train our network, we utilize the popular game engine Unreal Engine to generate a large synthetic face dataset containing various subjects, head poses, expressions, eyewear, clothes, and lighting. Quantitative and qualitative results show that our rectification pipeline outperforms previous methods, and produces comparable results with a time-consuming 3D GAN-based method while being more than 260 times faster.
Authors: Yuanhuiyi Lyu, Xu Zheng, Lutao Jiang, Yibo Yan, Xin Zou, Huiyu Zhou, Linfeng Zhang, Xuming Hu
Abstract: Recent text-to-image generative models, e.g., Stable Diffusion V3 and Flux, have achieved notable progress. However, these models are strongly restricted to their limited knowledge, a.k.a., their own fixed parameters, that are trained with closed datasets. This leads to significant hallucinations or distortions when facing fine-grained and unseen novel real-world objects, e.g., the appearance of the Tesla Cybertruck. To this end, we present the first real-object-based retrieval-augmented generation framework (RealRAG), which augments fine-grained and unseen novel object generation by learning and retrieving real-world images to overcome the knowledge gaps of generative models. Specifically, to integrate missing memory for unseen novel object generation, we train a reflective retriever by self-reflective contrastive learning, which injects the generator's knowledge into the sef-reflective negatives, ensuring that the retrieved augmented images compensate for the model's missing knowledge. Furthermore, the real-object-based framework integrates fine-grained visual knowledge for the generative models, tackling the distortion problem and improving the realism for fine-grained object generation. Our Real-RAG is superior in its modular application to all types of state-of-the-art text-to-image generative models and also delivers remarkable performance boosts with all of them, such as a gain of 16.18% FID score with the auto-regressive model on the Stanford Car benchmark.
Authors: Na Min An, Eunki Kim, Wan Ju Kang, Sangryul Kim, James Thorne, Hyunjung Shim
Abstract: For individuals with blindness or low vision (BLV), navigating complex environments can pose serious risks. Large Vision-Language Models (LVLMs) show promise for generating scene descriptions, but their effectiveness for BLV users remains underexplored. To address this gap, we conducted a user study with eight BLV participants to systematically evaluate preferences for six types of LVLM descriptions. While they helped to reduce fear and improve actionability, user ratings showed wide variation in sufficiency and conciseness. Furthermore, GPT-4o--despite its strong potential to refine descriptions--was not consistently preferred by participants. We use the insights obtained from the user study to build training data for building our new automatic evaluation metric that can capture BLV preferences effectively. Our findings underscore the urgent need for BLV-centered evaluation metrics and human-in-the-loop feedback to advance LVLM description quality for accessibility.
Authors: Jowaria Khan, Alexa Friedman, Sydney Evans, Rachel Klein, Runzi Wang, Katherine E. Manz, Kaley Beins, David Q. Andrews, Elizabeth Bondi-Kelly
Abstract: Per- and polyfluoroalkyl substances (PFAS), chemicals found in products like non-stick cookware, are unfortunately persistent environmental pollutants with severe health risks. Accurately mapping PFAS contamination is crucial for guiding targeted remediation efforts and protecting public and environmental health, yet detection across large regions remains challenging due to the cost of testing and the difficulty of simulating their spread. In this work, we introduce FOCUS, a geospatial deep learning framework with a label noise-aware loss function, to predict PFAS contamination in surface water over large regions. By integrating hydrological flow data, land cover information, and proximity to known PFAS sources, our approach leverages both spatial and environmental context to improve prediction accuracy. We evaluate the performance of our approach through extensive ablation studies, robustness analysis, real-world validation, and comparative analyses against baselines like sparse segmentation, as well as existing scientific methods, including Kriging and pollutant transport simulations. Results and expert feedback highlight our framework's potential for scalable PFAS monitoring.
Authors: Tobias Kirschstein, Javier Romero, Artem Sevastopolsky, Matthias Nie{\ss}ner, Shunsuke Saito
Abstract: Traditionally, creating photo-realistic 3D head avatars requires a studio-level multi-view capture setup and expensive optimization during test-time, limiting the use of digital human doubles to the VFX industry or offline renderings. To address this shortcoming, we present Avat3r, which regresses a high-quality and animatable 3D head avatar from just a few input images, vastly reducing compute requirements during inference. More specifically, we make Large Reconstruction Models animatable and learn a powerful prior over 3D human heads from a large multi-view video dataset. For better 3D head reconstructions, we employ position maps from DUSt3R and generalized feature maps from the human foundation model Sapiens. To animate the 3D head, our key discovery is that simple cross-attention to an expression code is already sufficient. Finally, we increase robustness by feeding input images with different expressions to our model during training, enabling the reconstruction of 3D head avatars from inconsistent inputs, e.g., an imperfect phone capture with accidental movement, or frames from a monocular video. We compare Avat3r with current state-of-the-art methods for few-input and single-input scenarios, and find that our method has a competitive advantage in both tasks. Finally, we demonstrate the wide applicability of our proposed model, creating 3D head avatars from images of different sources, smartphone captures, single images, and even out-of-domain inputs like antique busts. Project website: https://tobias-kirschstein.github.io/avat3r/
Authors: Naifu Xue, Zhaoyang Jia, Jiahao Li, Bin Li, Yuan Zhang, Yan Lu
Abstract: Recent studies in extreme image compression have achieved remarkable performance by compressing the tokens from generative tokenizers. However, these methods often prioritize clustering common semantics within the dataset, while overlooking the diverse details of individual objects. Consequently, this results in suboptimal reconstruction fidelity, especially at low bitrates. To address this issue, we introduce a Dual-generative Latent Fusion (DLF) paradigm. DLF decomposes the latent into semantic and detail elements, compressing them through two distinct branches. The semantic branch clusters high-level information into compact tokens, while the detail branch encodes perceptually critical details to enhance the overall fidelity. Additionally, we propose a cross-branch interactive design to reduce redundancy between the two branches, thereby minimizing the overall bit cost. Experimental results demonstrate the impressive reconstruction quality of DLF even below 0.01 bits per pixel (bpp). On the CLIC2020 test set, our method achieves bitrate savings of up to 27.93% on LPIPS and 53.55% on DISTS compared to MS-ILLM. Furthermore, DLF surpasses recent diffusion-based codecs in visual fidelity while maintaining a comparable level of generative realism. Project: https://dlfcodec.github.io/
Authors: Zhusi Zhong, Yuli Wang, Lulu Bi, Zhuoqi Ma, Sun Ho Ahn, Christopher J. Mullin, Colin F. Greineder, Michael K. Atalay, Scott Collins, Grayson L. Baird, Cheng Ting Lin, Webster Stayman, Todd M. Kolb, Ihab Kamel, Harrison X. Bai, Zhicheng Jiao
Abstract: Medical imaging plays a pivotal role in modern healthcare, with computed tomography pulmonary angiography (CTPA) being a critical tool for diagnosing pulmonary embolism and other thoracic conditions. However, the complexity of interpreting CTPA scans and generating accurate radiology reports remains a significant challenge. This paper introduces Abn-BLIP (Abnormality-aligned Bootstrapping Language-Image Pretraining), an advanced diagnosis model designed to align abnormal findings to generate the accuracy and comprehensiveness of radiology reports. By leveraging learnable queries and cross-modal attention mechanisms, our model demonstrates superior performance in detecting abnormalities, reducing missed findings, and generating structured reports compared to existing methods. Our experiments show that Abn-BLIP outperforms state-of-the-art medical vision-language models and 3D report generation methods in both accuracy and clinical relevance. These results highlight the potential of integrating multimodal learning strategies for improving radiology reporting. The source code is available at https://github.com/zzs95/abn-blip.
Authors: Spyros Kondylatos, Nikolaos Ioannis Bountos, Dimitrios Michail, Xiao Xiang Zhu, Gustau Camps-Valls, Ioannis Papoutsis
Abstract: Recent advances in Computer Vision have introduced the concept of pretrained representation uncertainty, enabling zero-shot uncertainty estimation. This holds significant potential for Earth Observation (EO), where trustworthiness is critical, yet the complexity of EO data poses challenges to uncertainty-aware methods. In this work, we investigate the generalization of representation uncertainty in EO, considering the domain's unique semantic characteristics. We pretrain uncertainties on large EO datasets and propose an evaluation framework to assess their zero-shot performance in multi-label classification and segmentation EO tasks. Our findings reveal that, unlike uncertainties pretrained on natural images, EO-pretraining exhibits strong generalization across unseen EO domains, geographic locations, and target granularities, while maintaining sensitivity to variations in ground sampling distance. We demonstrate the practical utility of pretrained uncertainties showcasing their alignment with task-specific uncertainties in downstream tasks, their sensitivity to real-world EO image noise, and their ability to generate spatial uncertainty estimates out-of-the-box. Initiating the discussion on representation uncertainty in EO, our study provides insights into its strengths and limitations, paving the way for future research in the field. Code and weights are available at: https://github.com/Orion-AI-Lab/EOUncertaintyGeneralization.
URLs: https://github.com/Orion-AI-Lab/EOUncertaintyGeneralization.
Authors: Xinyu Zhang, Haonan Chang, Yuhan Liu, Abdeslam Boularias
Abstract: Gaussian splatting has emerged as a powerful tool for high-fidelity reconstruction of dynamic scenes. However, existing methods primarily rely on implicit motion representations, such as encoding motions into neural networks or per-Gaussian parameters, which makes it difficult to further manipulate the reconstructed motions. This lack of explicit controllability limits existing methods to replaying recorded motions only, which hinders a wider application in robotics. To address this, we propose Motion Blender Gaussian Splatting (MBGS), a novel framework that uses motion graphs as an explicit and sparse motion representation. The motion of a graph's links is propagated to individual Gaussians via dual quaternion skinning, with learnable weight painting functions that determine the influence of each link. The motion graphs and 3D Gaussians are jointly optimized from input videos via differentiable rendering. Experiments show that MBGS achieves state-of-the-art performance on the highly challenging iPhone dataset while being competitive on HyperNeRF. We demonstrate the application potential of our method in animating novel object poses, synthesizing real robot demonstrations, and predicting robot actions through visual planning. The source code, models, video demonstrations can be found at http://mlzxy.github.io/motion-blender-gs.
Authors: Yunxuan Mao, Rong Xiong, Yue Wang, Yiyi Liao
Abstract: Reconstructing and decomposing dynamic urban scenes is crucial for autonomous driving, urban planning, and scene editing. However, existing methods fail to perform instance-aware decomposition without manual annotations, which is crucial for instance-level scene editing.We propose UnIRe, a 3D Gaussian Splatting (3DGS) based approach that decomposes a scene into a static background and individual dynamic instances using only RGB images and LiDAR point clouds. At its core, we introduce 4D superpoints, a novel representation that clusters multi-frame LiDAR points in 4D space, enabling unsupervised instance separation based on spatiotemporal correlations. These 4D superpoints serve as the foundation for our decomposed 4D initialization, i.e., providing spatial and temporal initialization to train a dynamic 3DGS for arbitrary dynamic classes without requiring bounding boxes or object templates.Furthermore, we introduce a smoothness regularization strategy in both 2D and 3D space, further improving the temporal stability.Experiments on benchmark datasets show that our method outperforms existing methods in decomposed dynamic scene reconstruction while enabling accurate and flexible instance-level editing, making it a practical solution for real-world applications.
Authors: Fenglei Hao, Yuliang Yang, Ruiyuan Su, Zhengran Zhao, Yukun Qiao, Mengyu Zhu
Abstract: This paper addresses the challenge of capturing global temporaldependencies in long video sequences for Video Object Segmentation (VOS). Existing architectures often fail to effectively model these dependencies acrossextended temporal horizons. To overcome this limitation, we introduce GISE-TTT, anovel architecture that integrates Temporal Transformer (TTT) layers intotransformer-based frameworks through a co-designed hierarchical approach.The TTTlayer systematically condenses historical temporal information into hidden states thatencode globally coherent contextual representations. By leveraging multi-stagecontextual aggregation through hierarchical concatenation, our frameworkprogressively refines spatiotemporal dependencies across network layers. This designrepresents the first systematic empirical evidence that distributing global informationacross multiple network layers is critical for optimal dependency utilization in videosegmentation tasks.Ablation studies demonstrate that incorporating TTT modules athigh-level feature stages significantly enhances global modeling capabilities, therebyimproving the network's ability to capture long-range temporal relationships. Extensive experiments on DAVIS 2017 show that GISE-TTT achieves a 3.2%improvement in segmentation accuracy over the baseline model, providingcomprehensive evidence that global information should be strategically leveragedthroughout the network architecture.The code will be made available at:https://github.com/uuool/GISE-TTT.
Authors: Junyu Xie, Tengda Han, Max Bain, Arsha Nagrani, Eshika Khandelwal, G\"ul Varol, Weidi Xie, Andrew Zisserman
Abstract: Our objective is the automatic generation of Audio Descriptions (ADs) for edited video material, such as movies and TV series. To achieve this, we propose a two-stage framework that leverages "shots" as the fundamental units of video understanding. This includes extending temporal context to neighbouring shots and incorporating film grammar devices, such as shot scales and thread structures, to guide AD generation. Our method is compatible with both open-source and proprietary Visual-Language Models (VLMs), integrating expert knowledge from add-on modules without requiring additional training of the VLMs. We achieve state-of-the-art performance among all prior training-free approaches and even surpass fine-tuned methods on several benchmarks. To evaluate the quality of predicted ADs, we introduce a new evaluation measure -- an action score -- specifically targeted to assessing this important aspect of AD. Additionally, we propose a novel evaluation protocol that treats automatic frameworks as AD generation assistants and asks them to generate multiple candidate ADs for selection.
Authors: Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma
Abstract: Traffic crash detection in long-form surveillance videos is essential for improving emergency response and infrastructure planning, yet remains difficult due to the brief and infrequent nature of crash events. We present \textbf{HybridMamba}, a novel architecture integrating visual transformers with state-space temporal modeling to achieve high-precision crash time localization. Our approach introduces multi-level token compression and hierarchical temporal processing to maintain computational efficiency without sacrificing temporal resolution. Evaluated on a large-scale dataset from the Iowa Department of Transportation, HybridMamba achieves a mean absolute error of \textbf{1.50 seconds} for 2-minute videos ($p<0.01$ compared to baselines), with \textbf{65.2%} of predictions falling within one second of the ground truth. It outperforms recent video-language models (e.g., TimeChat, VideoLLaMA-2) by up to 3.95 seconds while using significantly fewer parameters (3B vs. 13--72B). Our results demonstrate effective temporal localization across various video durations (2--40 minutes) and diverse environmental conditions, highlighting HybridMamba's potential for fine-grained temporal localization in traffic surveillance while identifying challenges that remain for extended deployment.
Authors: Xiaochen Wei, Weiwei Guo, Wenxian Yu, Feiming Wei, Dongying Li
Abstract: Multimodal remote sensing image registration aligns images from different sensors for data fusion and analysis. However, existing methods often struggle to extract modality-invariant features when faced with large nonlinear radiometric differences, such as those between SAR and optical images. To address these challenges, we propose OSDM-MReg, a novel multimodal image registration framework that bridges the modality gap through image-to-image translation. Specifically, we introduce a one-step unaligned target-guided conditional diffusion model (UTGOS-CDM) to translate source and target images into a unified representation domain. Unlike traditional conditional DDPM that require hundreds of iterative steps for inference, our model incorporates a novel inverse translation objective during training to enable direct prediction of the translated image in a single step at test time, significantly accelerating the registration process. After translation, we design a multimodal multiscale registration network (MM-Reg) that extracts and fuses both unimodal and translated multimodal images using the proposed multimodal fusion strategy, enhancing the robustness and precision of alignment across scales and modalities. Extensive experiments on the OSdataset demonstrate that OSDM-MReg achieves superior registration accuracy compared to state-of-the-art methods.
Authors: Yumeng Shi, Quanyu Long, Wenya Wang
Abstract: Video question answering benefits from the rich information in videos, enabling various applications. However, the large volume of tokens generated from long videos presents challenges to memory efficiency and model performance. To alleviate this, existing works propose to compress video inputs, but often overlook the varying importance of static and dynamic information across different queries, leading to inefficient token usage within limited budgets. We propose a novel token selection strategy, \textsc{explore-then-select}, that adaptively adjusts static and dynamic information based on question requirements. Our framework first explores different token allocations between key frames, which preserve spatial details, and delta frames, which capture temporal changes. Then it employs a query-aware attention-based metric to select the optimal token combination without model updates. Our framework is plug-and-play and can be seamlessly integrated within diverse video language models. Extensive experiments show that our method achieves significant performance improvements (up to 5.8\%) on multiple video question answering benchmarks. Our code is available at https://github.com/ANDgate99/Explore-Then-Select .
Authors: Stefanos Gkikas, Raul Fernandez Rojas, Manolis Tsiknakis
Abstract: Pain is a manifold condition that impacts a significant percentage of the population. Accurate and reliable pain evaluation for the people suffering is crucial to developing effective and advanced pain management protocols. Automatic pain assessment systems provide continuous monitoring and support decision-making processes, ultimately aiming to alleviate distress and prevent functionality decline. This study introduces PainFormer, a vision foundation model based on multi-task learning principles trained simultaneously on 14 tasks/datasets with a total of 10.9 million samples. Functioning as an embedding extractor for various input modalities, the foundation model provides feature representations to the Embedding-Mixer, a transformer-based module that performs the final pain assessment. Extensive experiments employing behavioral modalities - including RGB, synthetic thermal, and estimated depth videos - and physiological modalities such as ECG, EMG, GSR, and fNIRS revealed that PainFormer effectively extracts high-quality embeddings from diverse input modalities. The proposed framework is evaluated on two pain datasets, BioVid and AI4Pain, and directly compared to 75 different methodologies documented in the literature. Experiments conducted in unimodal and multimodal settings demonstrate state-of-the-art performances across modalities and pave the way toward general-purpose models for automatic pain assessment. The foundation model's architecture (code) and weights are available at: https://github.com/GkikasStefanos/PainFormer.
Authors: Xiangru Li, Wei Song, Yingda Huang, Wei Meng, Le Chang, Hongyang Li
Abstract: Gait recognition enables contact-free, long-range person identification that is robust to clothing variations and non-cooperative scenarios. While existing methods perform well in controlled indoor environments, they struggle with cross-vertical view scenarios, where surveillance angles vary significantly in elevation. Our experiments show up to 60\% accuracy degradation in low-to-high vertical view settings due to severe deformations and self-occlusions of key anatomical features. Current CNN and self-attention-based methods fail to effectively handle these challenges, due to their reliance on single-scale convolutions or simplistic attention mechanisms that lack effective multi-frequency feature integration. To tackle this challenge, we propose CVVNet (Cross-Vertical-View Network), a frequency aggregation architecture specifically designed for robust cross-vertical-view gait recognition. CVVNet employs a High-Low Frequency Extraction module (HLFE) that adopts parallel multi-scale convolution/max-pooling path and self-attention path as high- and low-frequency mixers for effective multi-frequency feature extraction from input silhouettes. We also introduce the Dynamic Gated Aggregation (DGA) mechanism to adaptively adjust the fusion ratio of high- and low-frequency features. The integration of our core Multi-Scale Attention Gated Aggregation (MSAGA) module, HLFE and DGA enables CVVNet to effectively handle distortions from view changes, significantly improving the recognition robustness across different vertical views. Experimental results show that our CVVNet achieves state-of-the-art performance, with $8.6\%$ improvement on DroneGait and $2\%$ on Gait3D compared with the best existing methods.
Authors: Yuxuan Mu, Hung Yu Ling, Yi Shi, Ismael Baira Ojeda, Pengcheng Xi, Chang Shu, Fabio Zinno, Xue Bin Peng
Abstract: Motion capture (mocap) data often exhibits visually jarring artifacts due to inaccurate sensors and post-processing. Cleaning this corrupted data can require substantial manual effort from human experts, which can be a costly and time-consuming process. Previous data-driven motion cleanup methods offer the promise of automating this cleanup process, but often require in-domain paired corrupted-to-clean training data. Constructing such paired datasets requires access to high-quality, relatively artifact-free motion clips, which often necessitates laborious manual cleanup. In this work, we present StableMotion, a simple yet effective method for training motion cleanup models directly from unpaired corrupted datasets that need cleanup. The core component of our method is the introduction of motion quality indicators, which can be easily annotated - through manual labeling or heuristic algorithms - and enable training of quality-aware motion generation models on raw motion data with mixed quality. At test time, the model can be prompted to generate high-quality motions using the quality indicators. Our method can be implemented through a simple diffusion-based framework, leading to a unified motion generate-discriminate model, which can be used to both identify and fix corrupted frames. We demonstrate that our proposed method is effective for training motion cleanup models on raw mocap data in production scenarios by applying StableMotion to SoccerMocap, a 245-hour soccer mocap dataset containing real-world motion artifacts. The trained model effectively corrects a wide range of motion artifacts, reducing motion pops and frozen frames by 68% and 81%, respectively. Results and code are available at https://yxmu.foo/stablemotion-page
Authors: Vibhas Vats, Md. Alimoor Reza, David Crandall, Soon-heung Jung
Abstract: Traditional multi-view stereo (MVS) methods primarily depend on photometric and geometric consistency constraints. In contrast, modern learning-based algorithms often rely on the plane sweep algorithm to infer 3D geometry, applying explicit geometric consistency (GC) checks only as a post-processing step, with no impact on the learning process itself. In this work, we introduce GC MVSNet plus plus, a novel approach that actively enforces geometric consistency of reference view depth maps across multiple source views (multi view) and at various scales (multi scale) during the learning phase (see Fig. 1). This integrated GC check significantly accelerates the learning process by directly penalizing geometrically inconsistent pixels, effectively halving the number of training iterations compared to other MVS methods. Furthermore, we introduce a densely connected cost regularization network with two distinct block designs simple and feature dense optimized to harness dense feature connections for enhanced regularization. Extensive experiments demonstrate that our approach achieves a new state of the art on the DTU and BlendedMVS datasets and secures second place on the Tanks and Temples benchmark. To our knowledge, GC MVSNet plus plus is the first method to enforce multi-view, multi-scale supervised geometric consistency during learning. Our code is available.
Authors: Ciyu Ruan, Zihang Gong, Ruishan Guo, Jingao Xu, Xinlei Chen
Abstract: Event cameras provide micro-second latency and broad dynamic range, yet their raw streams are marred by spatial artifacts (e.g., hot pixels) and temporally inconsistent background activity. Existing methods jointly process the entire 4D event volume (x, y, p, t), forcing heavy spatio-temporal attention that inflates parameters, FLOPs, and latency. We introduce EDmamba, a compact event-denoising framework that embraces the key insight that spatial and temporal noise arise from different physical mechanisms and can therefore be suppressed independently. A polarity- and geometry-aware encoder first extracts coarse cues, which are then routed to two lightweight state-space branches: a Spatial-SSM that learns location-conditioned filters to silence persistent artifacts, and a Temporal-SSM that models causal signal dynamics to eliminate bursty background events. This decoupled design distills the network to only 88.9K parameters and 2.27GFLOPs, enabling real-time throughput of 100K events in 68ms on a single GPU, 36x faster than recent Transformer baselines. Despite its economy, EDmamba establishes new state-of-the-art accuracy on four public benchmarks, outscoring the strongest prior model by 2.1 percentage points.
Authors: Zhenglin Hua, Jinghan He, Zijun Yao, Tianxu Han, Haiyun Guo, Yuheng Jia, Junfeng Fang
Abstract: Large vision-language models (LVLMs) have achieved remarkable performance on multimodal tasks. However, they still suffer from hallucinations, generating text inconsistent with visual input, posing significant risks in real-world applications. Existing approaches to address this issue focus on incorporating external knowledge bases, alignment training, or decoding strategies, all of which require substantial computational cost and time. Recent works try to explore more efficient alternatives by adjusting LVLMs' internal representations. Although promising, these methods may cause hallucinations to be insufficiently suppressed or lead to excessive interventions that negatively affect normal semantics. In this work, we leverage sparse autoencoders (SAEs) to identify semantic directions closely associated with faithfulness or hallucination, extracting more precise and disentangled hallucination-related representations. Our analysis demonstrates that interventions along the identified faithful direction can mitigate hallucinations, while those along the hallucinatory direction can exacerbate them. Building on these insights, we propose Steering LVLMs via SAE Latent Directions (SSL), a plug-and-play method based on SAE-derived latent directions to mitigate hallucinations in LVLMs. Extensive experiments demonstrate that SSL significantly outperforms existing decoding approaches in mitigating hallucinations, while maintaining transferability across different model architectures with negligible additional time overhead. The code is available at https://github.com/huazhenglin2003/SSL.
Authors: Zhenglin Huang, Tianxiao Li, Xiangtai Li, Haiquan Wen, Yiwei He, Jiangning Zhang, Hao Fei, Xi Yang, Xiaowei Huang, Bei Peng, Guangliang Cheng
Abstract: Recent advances in AI-powered generative models have enabled the creation of increasingly realistic synthetic images, posing significant risks to information integrity and public trust on social media platforms. While robust detection frameworks and diverse, large-scale datasets are essential to mitigate these risks, existing academic efforts remain limited in scope: current datasets lack the diversity, scale, and realism required for social media contexts, while detection methods struggle with generalization to unseen generative technologies. To bridge this gap, we introduce So-Fake-Set, a comprehensive social media-oriented dataset with over 2 million high-quality images, diverse generative sources, and photorealistic imagery synthesized using 35 state-of-the-art generative models. To rigorously evaluate cross-domain robustness, we establish a novel and large-scale (100K) out-of-domain benchmark (So-Fake-OOD) featuring synthetic imagery from commercial models explicitly excluded from the training distribution, creating a realistic testbed for evaluating real-world performance. Leveraging these resources, we present So-Fake-R1, an advanced vision-language framework that employs reinforcement learning for highly accurate forgery detection, precise localization, and explainable inference through interpretable visual rationales. Extensive experiments show that So-Fake-R1 outperforms the second-best method, with a 1.3% gain in detection accuracy and a 4.5% increase in localization IoU. By integrating a scalable dataset, a challenging OOD benchmark, and an advanced detection framework, this work establishes a new foundation for social media-centric forgery detection research. The code, models, and datasets will be released publicly.
Authors: Lihan Jiang, Yucheng Mao, Linning Xu, Tao Lu, Kerui Ren, Yichen Jin, Xudong Xu, Mulin Yu, Jiangmiao Pang, Feng Zhao, Dahua Lin, Bo Dai
Abstract: We introduce AnySplat, a feed forward network for novel view synthesis from uncalibrated image collections. In contrast to traditional neural rendering pipelines that demand known camera poses and per scene optimization, or recent feed forward methods that buckle under the computational weight of dense views, our model predicts everything in one shot. A single forward pass yields a set of 3D Gaussian primitives encoding both scene geometry and appearance, and the corresponding camera intrinsics and extrinsics for each input image. This unified design scales effortlessly to casually captured, multi view datasets without any pose annotations. In extensive zero shot evaluations, AnySplat matches the quality of pose aware baselines in both sparse and dense view scenarios while surpassing existing pose free approaches. Moreover, it greatly reduce rendering latency compared to optimization based neural fields, bringing real time novel view synthesis within reach for unconstrained capture settings.Project page: https://city-super.github.io/anysplat/
Authors: Yu Huang, Junhao Chen, Shuliang Liu, Hanqian Li, Qi Zheng, Yi R. Fung, Xuming Hu
Abstract: The rapid development of Artificial Intelligence Generated Content (AIGC) has led to significant progress in video generation but also raises serious concerns about intellectual property protection and reliable content tracing. Watermarking is a widely adopted solution to this issue, but existing methods for video generation mainly follow a post-generation paradigm, which introduces additional computational overhead and often fails to effectively balance the trade-off between video quality and watermark extraction. To address these issues, we propose Video Signature (VIDSIG), an in-generation watermarking method for latent video diffusion models, which enables implicit and adaptive watermark integration during generation. Specifically, we achieve this by partially fine-tuning the latent decoder, where Perturbation-Aware Suppression (PAS) pre-identifies and freezes perceptually sensitive layers to preserve visual quality. Beyond spatial fidelity, we further enhance temporal consistency by introducing a lightweight Temporal Alignment module that guides the decoder to generate coherent frame sequences during fine-tuning. Experimental results show that VIDSIG achieves the best overall performance in watermark extraction, visual quality, and generation efficiency. It also demonstrates strong robustness against both spatial and temporal tampering, highlighting its practicality in real-world scenarios. Our code is available at \href{https://github.com/hardenyu21/Video-Signature}{here}
Authors: Yudong Zhang, Ruobing Xie, Yiqing Huang, Jiansheng Chen, Xingwu Sun, Zhanhui Kang, Di Wang, Yu Wang
Abstract: Recent advances in large vision-language models (LVLMs) have showcased their remarkable capabilities across a wide range of multimodal vision-language tasks. However, these models remain vulnerable to visual adversarial attacks, which can substantially compromise their performance. In this paper, we introduce F3, a novel adversarial purification framework that employs a counterintuitive ``fighting fire with fire'' strategy: intentionally introducing simple perturbations to adversarial examples to mitigate their harmful effects. Specifically, F3 leverages cross-modal attentions derived from randomly perturbed adversary examples as reference targets. By injecting noise into these adversarial examples, F3 effectively refines their attention, resulting in cleaner and more reliable model outputs. Remarkably, this seemingly paradoxical approach of employing noise to counteract adversarial attacks yields impressive purification results. Furthermore, F3 offers several distinct advantages: it is training-free and straightforward to implement, and exhibits significant computational efficiency improvements compared to existing purification methods. These attributes render F3 particularly suitable for large-scale industrial applications where both robust performance and operational efficiency are critical priorities. The code is available at https://github.com/btzyd/F3.
Authors: Rynaa Grover, Jayant Sravan Tamarapalli, Sahiti Yerramilli, Nilay Pande
Abstract: Multimodal Large Language Models (MLLMs) excel at high-level visual reasoning, but their performance on nuanced perceptual tasks remains surprisingly limited. We present HueManity, a benchmark designed to assess visual perception in MLLMs. The dataset comprises 83,850 images featuring two-character alphanumeric strings embedded in Ishihara test style dot patterns, challenging models on precise pattern recognition. Our evaluation of nine state-of-the-art MLLMs on HueManity demonstrates a significant performance deficit compared to human and traditional computer vision baselines. The best-performing MLLM achieved a 33.6% accuracy on the numeric `easy' task and a striking 3% on the alphanumeric `hard' task. In contrast, human participants achieved near-perfect scores (100% and 95.6%), and a fine-tuned ResNet50 model reached accuracies of 96.5% and 94.5%. These results highlight a critical gap in the visual capabilities of current MLLMs. Our analysis further explores potential architectural and training-paradigm factors contributing to this perceptual gap in MLLMs. We open-source HueManity dataset and code to foster further research in improving perceptual robustness of MLLMs.
Authors: Ioannis Iakovidis, Zahra Kalantari, Amir Hossein Payberah, Fernando Jaramillo, Francisco Pena Escobar
Abstract: In recent years the wide availability of high-resolution radar satellite images along with the advancement of computer vision models have enabled the remote monitoring of wetland surface areas. However, these models require large amounts of manually annotated satellite images, which are slow and expensive to produce. To overcome this problem we use self-supervised training methods to train a model called DeepAquaCluster to segment radar satellite images into areas that separate water from land without the use of any manual annotations. Our final model outperforms other radar-based water detection techniques in our test dataset, achieving a 0.08 improvement in the Intersection Over Union metric.
Authors: Nikolaos Dionelis, Jente Bosmans, Riccardo Musto, Giancarlo Paoletti, Simone Sarti, Giacomo Cascarano, Casper Fibaek, Luke Camilleri, Bertrand Le Saux, Nicolas Long\'ep\'e
Abstract: Today, Earth Observation (EO) satellites generate massive volumes of data, with the Copernicus Sentinel-2 constellation alone producing approximately 1.6TB per day. To fully exploit this information, it is essential to pretrain EO Foundation Models (FMs) on large unlabeled datasets, enabling efficient fine-tuning for several different downstream tasks with minimal labeled data. In this work, we present the scaling-up of our recently proposed EO Foundation Model, PhilEO Geo-Aware U-Net, on the unlabeled 23TB dataset MajorTOM, which covers the vast majority of the Earth's surface, as well as on the specialized subset FastTOM 2TB that does not include oceans and ice. We develop and study various PhilEO model variants with different numbers of parameters and architectures. We fine-tune the models on the PhilEO Bench for road density estimation, building density pixel-wise regression, and land cover semantic segmentation, and we evaluate the performance. Our results demonstrate that for all n-shots for road density regression, the PhilEO 44M MajorTOM 23TB model outperforms PhilEO Globe 0.5TB 44M. We also show that for most n-shots for road density estimation and building density regression, PhilEO 200M FastTOM outperforms all the other models we examine. The effectiveness of both dataset and model scaling is validated using the PhilEO Bench. We also study the impact of architecture scaling, transitioning from U-Net Convolutional Neural Networks (CNN) to Vision Transformers (ViT).
Authors: Fan Xie, Yang Wang, Yikang Jiao, Zhenyu Yuan, Congxi Chen, Chuanxin Zhao
Abstract: In current practical face authentication systems, most face recognition (FR) algorithms are based on cosine similarity with softmax classification. Despite its reliable classification performance, this method struggles with hard samples. A popular strategy to improve FR performance is incorporating angular or cosine margins. However, it does not take face quality or recognition hardness into account, simply increasing the margin value and thus causing an overly uniform training strategy. To address this problem, a novel loss function is proposed, named Loss function for Hard High-quality Face (LH2Face). Firstly, a similarity measure based on the von Mises-Fisher (vMF) distribution is stated, specifically focusing on the logarithm of the Probability Density Function (PDF), which represents the distance between a probability distribution and a vector. Then, an adaptive margin-based multi-classification method using softmax, called the Uncertainty-Aware Margin Function, is implemented in the article. Furthermore, proxy-based loss functions are used to apply extra constraints between the proxy and sample to optimize their representation space distribution. Finally, a renderer is constructed that optimizes FR through face reconstruction and vice versa. Our LH2Face is superior to similiar schemes on hard high-quality face datasets, achieving 49.39% accuracy on the IJB-B dataset, which surpasses the second-place method by 2.37%.
Authors: Shuaijin Wan
Abstract: Human motion is a continuous physical process in 3D space, governed by complex dynamic and kinematic constraints. Existing methods typically represent the human pose as an abstract graph structure, neglecting the intrinsic physical dependencies between joints, which increases learning difficulty and makes the model prone to generating unrealistic motions. In this paper, we propose GGMotion, a group graph dynamics-kinematics network that models human topology in groups to better leverage dynamics and kinematics priors. To preserve the geometric equivariance in 3D space, we propose a novel radial field for the graph network that captures more comprehensive spatio-temporal dependencies by aggregating joint features through spatial and temporal edges. Inter-group and intra-group interaction modules are employed to capture the dependencies of joints at different scales. Combined with equivariant multilayer perceptrons (MLP), joint position features are updated in each group through parallelized dynamics-kinematics propagation to improve physical plausibility. Meanwhile, we introduce an auxiliary loss to supervise motion priors during training. Extensive experiments on three standard benchmarks, including Human3.6M, CMU-Mocap, and 3DPW, demonstrate the effectiveness and superiority of our approach, achieving a significant performance margin in short-term motion prediction. The code is available at https://github.com/inkcat520/GGMotion.git.
Authors: Hyungjun Doh, Dong In Lee, Seunggeun Chi, Pin-Hao Huang, Kwonjoon Lee, Sangpil Kim, Karthik Ramani
Abstract: We introduce a novel framework for reconstructing dynamic human-object interactions from monocular video that overcomes challenges associated with occlusions and temporal inconsistencies. Traditional 3D reconstruction methods typically assume static objects or full visibility of dynamic subjects, leading to degraded performance when these assumptions are violated-particularly in scenarios where mutual occlusions occur. To address this, our framework leverages amodal completion to infer the complete structure of partially obscured regions. Unlike conventional approaches that operate on individual frames, our method integrates temporal context, enforcing coherence across video sequences to incrementally refine and stabilize reconstructions. This template-free strategy adapts to varying conditions without relying on predefined models, significantly enhancing the recovery of intricate details in dynamic scenes. We validate our approach using 3D Gaussian Splatting on challenging monocular videos, demonstrating superior precision in handling occlusions and maintaining temporal stability compared to existing techniques.
Authors: Yuqing Lan, Chenyang Zhu, Shuaifeng Zhi, Jiazhao Zhang, Zhoufeng Wang, Renjiao Yi, Yijie Wang, Kai Xu
Abstract: The introduction of the neural implicit representation has notably propelled the advancement of online dense reconstruction techniques. Compared to traditional explicit representations, such as TSDF, it improves the mapping completeness and memory efficiency. However, the lack of reconstruction details and the time-consuming learning of neural representations hinder the widespread application of neural-based methods to large-scale online reconstruction. We introduce RemixFusion, a novel residual-based mixed representation for scene reconstruction and camera pose estimation dedicated to high-quality and large-scale online RGB-D reconstruction. In particular, we propose a residual-based map representation comprised of an explicit coarse TSDF grid and an implicit neural module that produces residuals representing fine-grained details to be added to the coarse grid. Such mixed representation allows for detail-rich reconstruction with bounded time and memory budget, contrasting with the overly-smoothed results by the purely implicit representations, thus paving the way for high-quality camera tracking. Furthermore, we extend the residual-based representation to handle multi-frame joint pose optimization via bundle adjustment (BA). In contrast to the existing methods, which optimize poses directly, we opt to optimize pose changes. Combined with a novel technique for adaptive gradient amplification, our method attains better optimization convergence and global optimality. Furthermore, we adopt a local moving volume to factorize the mixed scene representation with a divide-and-conquer design to facilitate efficient online learning in our residual-based framework. Extensive experiments demonstrate that our method surpasses all state-of-the-art ones, including those based either on explicit or implicit representations, in terms of the accuracy of both mapping and tracking on large-scale scenes.
Authors: Joon Hyun Park, Kumju Jo, Sungyong Baik
Abstract: Entrusted with the goal of pixel-level object classification, the semantic segmentation networks entail the laborious preparation of pixel-level annotation masks. To obtain pixel-level annotation masks for a given class without human efforts, recent few works have proposed to generate pairs of images and annotation masks by employing image and text relationships modeled by text-to-image generative models, especially Stable Diffusion. However, these works do not fully exploit the capability of text-guided Diffusion models and thus require a pre-trained segmentation network, careful text prompt tuning, or the training of a segmentation network to generate final annotation masks. In this work, we take a closer look at attention mechanisms of Stable Diffusion, from which we draw connections with classical seeded segmentation approaches. In particular, we show that cross-attention alone provides very coarse object localization, which however can provide initial seeds. Then, akin to region expansion in seeded segmentation, we utilize the semantic-correspondence-modeling capability of self-attention to iteratively spread the attention to the whole class from the seeds using multi-scale self-attention maps. We also observe that a simple-text-guided synthetic image often has a uniform background, which is easier to find correspondences, compared to complex-structured objects. Thus, we further refine a mask using a more accurate background mask. Our proposed method, dubbed SeeDiff, generates high-quality masks off-the-shelf from Stable Diffusion, without additional training procedure, prompt tuning, or a pre-trained segmentation network.
Authors: Shibang Liu, Xuemei Xie, Guangming Shi
Abstract: Recent methods using diffusion models have made significant progress in Human Image Generation (HIG) with various control signals such as pose priors. In HIG, both accurate human poses and coherent visual quality are crucial for image generation. However, most existing methods mainly focus on pose accuracy while neglecting overall image quality, often improving pose alignment at the cost of image quality. To address this, we propose Knowledge-Based Global Guidance and Dynamic pose Masking for human image Generation (KB-DMGen). The Knowledge Base (KB), implemented as a visual codebook, provides coarse, global guidance based on input text-related visual features, improving pose accuracy while maintaining image quality, while the Dynamic pose Mask (DM) offers fine-grained local control to enhance precise pose accuracy. By injecting KB and DM at different stages of the diffusion process, our framework enhances pose accuracy through both global and local control without compromising image quality. Experiments demonstrate the effectiveness of KB-DMGen, achieving new state-of-the-art results in terms of AP and CAP on the HumanArt dataset. The project page and code are available at https://lushbng.github.io/KBDMGen.
Authors: Haiyang Guo, Fei Zhu, Hongbo Zhao, Fanhu Zeng, Wenzhuo Liu, Shijie Ma, Da-Han Wang, Xu-Yao Zhang
Abstract: Continual learning aims to equip AI systems with the ability to continuously acquire and adapt to new knowledge without forgetting previously learned information, similar to human learning. While traditional continual learning methods focusing on unimodal tasks have achieved notable success, the emergence of Multimodal Large Language Models has brought increasing attention to Multimodal Continual Learning tasks involving multiple modalities, such as vision and language. In this setting, models are expected to not only mitigate catastrophic forgetting but also handle the challenges posed by cross-modal interactions and coordination. To facilitate research in this direction, we introduce MCITlib, a comprehensive and constantly evolving code library for continual instruction tuning of Multimodal Large Language Models. In MCITlib, we have currently implemented 8 representative algorithms for Multimodal Continual Instruction Tuning and systematically evaluated them on 2 carefully selected benchmarks. MCITlib will be continuously updated to reflect advances in the Multimodal Continual Learning field. The codebase is released at https://github.com/Ghy0501/MCITlib.
Authors: Kristina Dabrock, Tim Johansson, Anna Donarelli, Mikael Mangold, Noah Pflugradt, Jann Michael Weinand, Jochen Lin{\ss}en
Abstract: Detailed data is required to quantify energy conservation measures in buildings, such as envelop retrofits, without compromising cultural heritage. Novel artificial intelligence tools may improve efficiency in identifying heritage values in buildings compared to costly and time-consuming traditional inventories. In this study, the large language model GPT was used to detect various aspects of cultural heritage value in fa\c{c}ade images. Using this data and building register data as features, machine learning models were trained to classify multi-family and non-residential buildings in Stockholm, Sweden. Validation against an expert-created inventory shows a macro F1-score of 0.71 using a combination of register data and features retrieved from GPT, and a score of 0.60 using only GPT-derived data. The presented methodology can contribute to a higher-quality database and thus support careful energy efficiency measures and integrated consideration of heritage value in large-scale energetic refurbishment scenarios.
Authors: Durgesh Kumar Singh, Qing Cao, Sarina Thomas, Ahc\`ene Boubekki, Robert Jenssen, Michael Kampffmeyer
Abstract: Clinical guidelines recommend performing left ventricular (LV) linear measurements in B-mode echocardiographic images at the basal level -- typically at the mitral valve leaflet tips -- and aligned perpendicular to the LV long axis along a virtual scanline (SL). However, most automated methods estimate landmarks directly from B-mode images for the measurement task, where even small shifts in predicted points along the LV walls can lead to significant measurement errors, reducing their clinical reliability. A recent semi-automatic method, EnLVAM, addresses this limitation by constraining landmark prediction to a clinician-defined SL and training on generated Anatomical Motion Mode (AMM) images to predict LV landmarks along the same. To enable full automation, a contour-aware SL placement approach is proposed in this work, in which the LV contour is estimated using a weakly supervised B-mode landmark detector. SL placement is then performed by inferring the LV long axis and the basal level- mimicking clinical guidelines. Building on this foundation, we introduce \textit{WiseLVAM} -- a novel, fully automated yet manually adaptable framework for automatically placing the SL and then automatically performing the LV linear measurements in the AMM mode. \textit{WiseLVAM} utilizes the structure-awareness from B-mode images and the motion-awareness from AMM mode to enhance robustness and accuracy with the potential to provide a practical solution for the routine clinical application. The source code is publicly available at https://github.com/SFI-Visual-Intelligence/wiselvam.git.
URLs: https://github.com/SFI-Visual-Intelligence/wiselvam.git.
Authors: Jiabo Huang, Chen Chen, Lingjuan Lyu
Abstract: Vision foundation models (VFMs) are predominantly developed using data-centric methods. These methods require training on vast amounts of data usually with high-quality labels, which poses a bottleneck for most institutions that lack both large-scale data and high-end GPUs. On the other hand, many open-source vision models have been pretrained on domain-specific data, enabling them to distill and represent core knowledge in a form that is transferable across diverse applications. Even though these models are highly valuable assets, they remain largely under-explored in empowering the development of a general-purpose VFM. In this paper, we present a new model-driven approach for training VFMs through joint knowledge transfer and preservation. Our method unifies multiple pre-trained teacher models in a shared latent space to mitigate the ``imbalanced transfer'' issue caused by their distributional gaps. Besides, we introduce a knowledge preservation strategy to take a general-purpose teacher as a knowledge base for integrating knowledge from the remaining purpose-specific teachers using an adapter module. By unifying and aggregating existing models, we build a powerful VFM to inherit teachers' expertise without needing to train on a large amount of labeled data. Our model not only provides generalizable visual features, but also inherently supports multiple downstream tasks. Extensive experiments demonstrate that our VFM outperforms existing data-centric models across four fundamental vision tasks, including image classification, object detection, semantic and instance segmentation.
Authors: Wutao Liu, YiDan Wang, Pan Gao
Abstract: Camouflaged object detection (COD) poses a significant challenge in computer vision due to the high similarity between objects and their backgrounds. Existing approaches often rely on heavy training and large computational resources. While foundation models such as the Segment Anything Model (SAM) offer strong generalization, they still struggle to handle COD tasks without fine-tuning and require high-quality prompts to yield good performance. However, generating such prompts manually is costly and inefficient. To address these challenges, we propose \textbf{First RAG, Second SEG (RAG-SEG)}, a training-free paradigm that decouples COD into two stages: Retrieval-Augmented Generation (RAG) for generating coarse masks as prompts, followed by SAM-based segmentation (SEG) for refinement. RAG-SEG constructs a compact retrieval database via unsupervised clustering, enabling fast and effective feature retrieval. During inference, the retrieved features produce pseudo-labels that guide precise mask generation using SAM2. Our method eliminates the need for conventional training while maintaining competitive performance. Extensive experiments on benchmark COD datasets demonstrate that RAG-SEG performs on par with or surpasses state-of-the-art methods. Notably, all experiments are conducted on a \textbf{personal laptop}, highlighting the computational efficiency and practicality of our approach. We present further analysis in the Appendix, covering limitations, salient object detection extension, and possible improvements. \textcolor{blue} {Code: https://github.com/Lwt-diamond/RAG-SEG.}
Authors: Zheyu Fan, Jiateng Liu, Yuji Zhang, Zihan Wang, Yi R. Fung, Manling Li, Heng Ji
Abstract: Humans naturally perform temporal screening by dragging the progress bar and focusing on salient temporal segments, but current Video Large Language Models (Video-LLMs) struggle to capture fine-grained temporal semantics due to sparse frame sampling and insufficient inter-frame reasoning supervision during their training. To address this, Inspired by well-established cognitive science principles, we propose Temporal Visual Screening (TVS), a new task that universally pre-processes video question answering and instruction tuning data by: (1) retaining focus-critical video segments, (2) synchronously reconstructing queries to their most direct form while preserving answer consistency, and (3) keeping the invariance and consistency for any possible answer. TVS is formulated as a modular front-end adapter task that can be seamlessly integrated into both Video Instruction Tuning (training) and Video Question Answering (inference) pipelines. TVS optimizes distribution of reasoning burden and cognitive load; during training, it aligns queries with focus-critical visual information; at inference, it enables query-aware segment focus and streamlined query representations. In particular, we curate the first benchmark for TVS and propose ReSimplifyIt, a baseline outperforming prior approaches on seemingly similar tasks by 0.47 in F-1 score on video trimming while achieving competitive query rewriting performance. Experiments demonstrate that incorporating TVS yields relative gains of 7.33% (training) and 34.6% (inference), demonstrating the effectiveness of temporal information screening for improving video-language understanding.
Authors: Liangjing Shao, Benshuang Chen, Chenkang Du, Xueli Liu, Xinrong Chen
Abstract: Self-supervised monocular depth estimation is a significant task for low-cost and efficient 3D scene perception in endoscopy. In recent years, a series of methods are proposed to address the illumination inconsistency, while certain works also focus on the generalization of the model by efficiently finetuning the foundation models. However, the variety of illumination conditions and scene features is still the primary challenges for depth estimation in endoscopic scenes. In this work, a self-supervised framework is proposed for monocular depth estimation in diverse endoscopy. Firstly, considering the diverse features in endoscopic scenes with different tissues, a novel block-wise mixture of dynamic low-rank experts is proposed to efficiently finetune the foundation model for endoscopic depth estimation. In the proposed module, based on the input feature, different experts with a small amount of trainable parameters are adaptively selected for weighted inference, from low-rank experts which are allocated based on the generalization of each block. Moreover, a novel self-supervised training framework is proposed to jointly cope with brightness inconsistency and reflectance interference. The proposed method outperforms state-of-the-art works on SCARED dataset and SimCol dataset. Furthermore, the proposed network also achieves the best generalization based on zero-shot depth estimation on C3VD, Hamlyn and SERV-CT dataset. The outstanding performance of our model is further demonstrated with 3D reconstruction and ego-motion estimation. The proposed method could contribute to accurate endoscopy for minimally invasive measurement and surgery. The evaluation codes will be released upon acceptance, while the demo videos can be found on: https://endo-gede.netlify.app/.
Authors: Yuan Zhao, Lin Liu
Abstract: Text-to-image diffusion models have achieved remarkable image quality, but they still struggle with complex, multiele ment prompts, and limited stylistic diversity. To address these limitations, we propose a Multi-Expert Planning and Gen eration Framework (MEPG) that synergistically integrates position- and style-aware large language models (LLMs) with spatial-semantic expert modules. The framework comprises two core components: (1) a Position-Style-Aware (PSA) module that utilizes a supervised fine-tuned LLM to decom pose input prompts into precise spatial coordinates and style encoded semantic instructions; and (2) a Multi-Expert Dif fusion (MED) module that implements cross-region genera tion through dynamic expert routing across both local regions and global areas. During the generation process for each lo cal region, specialized models (e.g., realism experts, styliza tion specialists) are selectively activated for each spatial par tition via attention-based gating mechanisms. The architec ture supports lightweight integration and replacement of ex pert models, providing strong extensibility. Additionally, an interactive interface enables real-time spatial layout editing and per-region style selection from a portfolio of experts. Ex periments show that MEPG significantly outperforms base line models with the same backbone in both image quality and style diversity.
Authors: Jeongmin Yu, Susang Kim, Kisu Lee, Taekyoung Kwon, Won-Yong Shin, Ha Young Kim
Abstract: Recent face anti-spoofing (FAS) methods have shown remarkable cross-domain performance by employing vision-language models like CLIP. However, existing CLIP-based FAS models do not fully exploit CLIP's patch embedding tokens, failing to detect critical spoofing clues. Moreover, these models rely on a single text prompt per class (e.g., 'live' or 'fake'), which limits generalization. To address these issues, we propose MVP-FAS, a novel framework incorporating two key modules: Multi-View Slot attention (MVS) and Multi-Text Patch Alignment (MTPA). Both modules utilize multiple paraphrased texts to generate generalized features and reduce dependence on domain-specific text. MVS extracts local detailed spatial features and global context from patch embeddings by leveraging diverse texts with multiple perspectives. MTPA aligns patches with multiple text representations to improve semantic robustness. Extensive experiments demonstrate that MVP-FAS achieves superior generalization performance, outperforming previous state-of-the-art methods on cross-domain datasets. Code: https://github.com/Elune001/MVP-FAS.
Authors: Erez Posner, Ore Shtalrid, Oded Erell, Daniel Noy, Moshe Bouhnik
Abstract: Realistic and parameterized 3D models of human anatomy have become invaluable in research, diagnostics, and surgical planning. However, the development of detailed models for internal organs, such as the stomach, has been limited by data availability and methodological challenges. In this paper, we propose a novel pipeline for the generation of synthetic 3D stomach models, enabling the creation of anatomically diverse morphologies informed by established studies on stomach shape variability. Using this pipeline, we construct a dataset of synthetic stomachs. Building on this dataset, we develop a 3D statistical shape model of the stomach, trained to capture natural anatomical variability in a low-dimensional shape space. The model is further refined using CT meshes derived from publicly available datasets through a semi-supervised alignment process, enhancing its ability to generalize to unseen anatomical variations. We evaluated the model on a held-out test set of real stomach CT scans, demonstrating robust generalization and fit accuracy. We make the statistical shape model along with the synthetic dataset publicly available on GitLab: https://gitlab.com/Erez.Posner/stomach_pytorch to facilitate further research. This work introduces the first statistical 3D shape model of the stomach, with applications ranging from surgical simulation and pre-operative planning to medical education and computational modeling. By combining synthetic data generation, parametric modeling, and real-world validation, our approach represents a significant advancement in organ modeling and opens new possibilities for personalized healthcare solutions.
Authors: Aswini Kumar Patra, Lingaraj Sahoo
Abstract: Plants in their natural habitats endure an array of interacting stresses, both biotic and abiotic, that rarely occur in isolation. Nutrient stress-particularly nitrogen deficiency-becomes even more critical when compounded with drought and weed competition, making it increasingly difficult to distinguish and address its effects. Early detection of nitrogen stress is therefore crucial for protecting plant health and implementing effective management strategies. This study proposes a novel deep learning framework to accurately classify nitrogen stress severity in a combined stress environment. Our model uses a unique blend of four imaging modalities-RGB, multispectral, and two infrared wavelengths-to capture a wide range of physiological plant responses from canopy images. These images, provided as time-series data, document plant health across three levels of nitrogen availability (low, medium, and high) under varying water stress and weed pressures. The core of our approach is a spatio-temporal deep learning pipeline that merges a Convolutional Neural Network (CNN) for extracting spatial features from images with a Long Short-Term Memory (LSTM) network to capture temporal dependencies. We also devised and evaluated a spatial-only CNN pipeline for comparison. Our CNN-LSTM pipeline achieved an impressive accuracy of 98%, impressively surpassing the spatial-only model's 80.45% and other previously reported machine learning method's 76%. These results bring actionable insights based on the power of our CNN-LSTM approach in effectively capturing the subtle and complex interactions between nitrogen deficiency, water stress, and weed pressure. This robust platform offers a promising tool for the timely and proactive identification of nitrogen stress severity, enabling better crop management and improved plant health.
Authors: Waqar Ahmad, Evan Murphy, Vladimir A. Krylov
Abstract: Object re-identification (Re-ID) methods are highly sensitive to label noise, which typically leads to significant performance degradation. We address this challenge by reframing Re-ID as a supervised image similarity task and adopting a Siamese network architecture trained to capture discriminative pairwise relationships. Central to our approach is a novel statistical outlier detection (OD) framework, termed Beta-SOD (Beta mixture Similarity-based Outlier Detection), which models the distribution of cosine similarities between embedding pairs using a two-component Beta distribution mixture model. We establish a novel identifiability result for mixtures of two Beta distributions, ensuring that our learning task is well-posed. The proposed OD step complements the Re-ID architecture combining binary cross-entropy, contrastive, and cosine embedding losses that jointly optimize feature-level similarity learning. We demonstrate the effectiveness of Beta-SOD in de-noising and Re-ID tasks for person Re-ID, on CUHK03 and Market-1501 datasets, and vehicle Re-ID, on VeRi-776 dataset. Our method shows superior performance compared to the state-of-the-art methods across various noise levels (10-30\%), demonstrating both robustness and broad applicability in noisy Re-ID scenarios. The implementation of Beta-SOD is available at: github.com/waqar3411/Beta-SOD
Authors: Andrew Bell, Yan Kit Choi, Steffen E Peterson, Andrew King, Muhummad Sohaib Nazir, Alistair A Young
Abstract: Automatic quantification of intramyocardial motion and strain from tagging MRI remains an important but challenging task. We propose a method using implicit neural representations (INRs), conditioned on learned latent codes, to predict continuous left ventricular (LV) displacement -- without requiring inference-time optimisation. Evaluated on 452 UK Biobank test cases, our method achieved the best tracking accuracy (2.14 mm RMSE) and the lowest combined error in global circumferential (2.86%) and radial (6.42%) strain compared to three deep learning baselines. In addition, our method is $\sim$380$\times$ faster than the most accurate baseline. These results highlight the suitability of INR-based models for accurate and scalable analysis of myocardial strain in large CMR datasets.
Authors: Hesham M. Shehata, Mohammad Abdolrahmani
Abstract: Recent graph convolutional neural networks (GCNs) have shown high performance in the field of human action recognition by using human skeleton poses. However, it fails to detect human-object interaction cases successfully due to the lack of effective representation of the scene information and appropriate learning architectures. In this context, we propose a methodology to utilize human action recognition performance by considering fixed object information in the environment and following a multi-task learning approach. In order to evaluate the proposed method, we collected real data from public environments and prepared our data set, which includes interaction classes of hands-on fixed objects (e.g., ATM ticketing machines, check-in/out machines, etc.) and non-interaction classes of walking and standing. The multi-task learning approach, along with interaction area information, succeeds in recognizing the studied interaction and non-interaction actions with an accuracy of 99.25%, outperforming the accuracy of the base model using only human skeleton poses by 2.75%.
Authors: Jifeng Shen, Haibo Zhan, Xin Zuo, Heng Fan, Xiaohui Yuan, Jun Li, Wankou Yang
Abstract: Current multispectral object detection methods often retain extraneous background or noise during feature fusion, limiting perceptual performance. To address this, we propose an innovative feature fusion framework based on cross-modal feature contrastive and screening strategy, diverging from conventional approaches. The proposed method adaptively enhances salient structures by fusing object-aware complementary cross-modal features while suppressing shared background interference. Our solution centers on two novel, specially designed modules: the Mutual Feature Refinement Module (MFRM) and the Differential Feature Feedback Module (DFFM). The MFRM enhances intra- and inter-modal feature representations by modeling their relationships, thereby improving cross-modal alignment and discriminative power. Inspired by feedback differential amplifiers, the DFFM dynamically computes inter-modal differential features as guidance signals and feeds them back to the MFRM, enabling adaptive fusion of complementary information while suppressing common-mode noise across modalities. To enable robust feature learning, the MFRM and DFFM are integrated into a unified framework, which is formally formulated as an Iterative Relation-Map Differential Guided Feature Fusion mechanism, termed IRDFusion. IRDFusion enables high-quality cross-modal fusion by progressively amplifying salient relational signals through iterative feedback, while suppressing feature noise, leading to significant performance gains. In extensive experiments on FLIR, LLVIP and M$^3$FD datasets, IRDFusion achieves state-of-the-art performance and consistently outperforms existing methods across diverse challenging scenarios, demonstrating its robustness and effectiveness. Code will be available at https://github.com/61s61min/IRDFusion.git.
Authors: Goker Erdogan, Nikhil Parthasarathy, Catalin Ionescu, Drew Hudson, Alexander Lerchner, Andrew Zisserman, Mehdi Sajjadi, Joao Carreira
Abstract: We introduce LayerLock, a simple yet effective approach for self-supervised visual representation learning, that gradually transitions from pixel to latent prediction through progressive layer freezing. First, we make the observation that during training of video masked-autoencoding (MAE) models, ViT layers converge in the order of their depth: shallower layers converge early, deeper layers converge late. We then show that this observation can be exploited to accelerate standard MAE by progressively freezing the model according to an explicit schedule, throughout training. Furthermore, this same schedule can be used in a simple and scalable approach to latent prediction that does not suffer from "representation collapse". We apply our proposed approach, LayerLock, to large models of up to 4B parameters with results surpassing those of non-latent masked prediction on the 4DS perception suite.
Authors: Elias De Smijter, Renaud Detry, Christophe De Vleeschouwer
Abstract: We present the first systematic comparison of implicit and explicit Novel View Synthesis methods for space-based 3D object reconstruction, evaluating the role of appearance embeddings. While embeddings improve photometric fidelity by modeling lighting variation, we show they do not translate into meaningful gains in geometric accuracy - a critical requirement for space robotics applications. Using the SPEED+ dataset, we compare K-Planes, Gaussian Splatting, and Convex Splatting, and demonstrate that embeddings primarily reduce the number of primitives needed for explicit methods rather than enhancing geometric fidelity. Moreover, convex splatting achieves more compact and clutter-free representations than Gaussian splatting, offering advantages for safety-critical applications such as interaction and collision avoidance. Our findings clarify the limits of appearance embeddings for geometry-centric tasks and highlight trade-offs between reconstruction quality and representation efficiency in space scenarios.
Authors: Georgios Pantazopoulos, Eda B. \"Ozyi\u{g}it
Abstract: Visual grounding refers to the ability of a model to identify a region within some visual input that matches a textual description. Consequently, a model equipped with visual grounding capabilities can target a wide range of applications in various domains, including referring expression comprehension, answering questions pertinent to fine-grained details in images or videos, caption visual context by explicitly referring to entities, as well as low and high-level control in simulated and real environments. In this survey paper, we review representative works across the key areas of research on modern general-purpose vision language models (VLMs). We first outline the importance of grounding in VLMs, then delineate the core components of the contemporary paradigm for developing grounded models, and examine their practical applications, including benchmarks and evaluation metrics for grounded multimodal generation. We also discuss the multifaceted interrelations among visual grounding, multimodal chain-of-thought, and reasoning in VLMs. Finally, we analyse the challenges inherent to visual grounding and suggest promising directions for future research.
Authors: Youzhi Liang, Wen Liang
Abstract: The utilization of biometric authentication with pattern images is increasingly popular in compact Internet of Things (IoT) devices. However, the reliability of such systems can be compromised by image quality issues, particularly in the presence of high levels of noise. While state-of-the-art deep learning algorithms designed for generic image denoising have shown promise, their large number of parameters and lack of optimization for unique biometric pattern retrieval make them unsuitable for these devices and scenarios. In response to these challenges, this paper proposes a lightweight and robust deep learning architecture, the Residual Wavelet-Conditioned Convolutional Autoencoder (Res-WCAE) with a Kullback-Leibler divergence (KLD) regularization, designed specifically for fingerprint image denoising. Res-WCAE comprises two encoders - an image encoder and a wavelet encoder - and one decoder. Residual connections between the image encoder and decoder are leveraged to preserve fine-grained spatial features, where the bottleneck layer conditioned on the compressed representation of features obtained from the wavelet encoder using approximation and detail subimages in the wavelet-transform domain. The effectiveness of Res-WCAE is evaluated against several state-of-the-art denoising methods, and the experimental results demonstrate that Res-WCAE outperforms these methods, particularly for heavily degraded fingerprint images in the presence of high levels of noise. Overall, Res-WCAE shows promise as a solution to the challenges faced by biometric authentication systems in compact IoT devices.
Authors: Zahraa Al Sahili, Ioannis Patras, Matthew Purver
Abstract: In the domain of text-to-image generative models, biases inherent in training datasets often propagate into generated content, posing significant ethical challenges, particularly in socially sensitive contexts. We introduce FairCoT, a novel framework that enhances fairness in text to image models through Chain of Thought (CoT) reasoning within multimodal generative large language models. FairCoT employs iterative CoT refinement to systematically mitigate biases, and dynamically adjusts textual prompts in real time, ensuring diverse and equitable representation in generated images. By integrating iterative reasoning processes, FairCoT addresses the limitations of zero shot CoT in sensitive scenarios, balancing creativity with ethical responsibility. Experimental evaluations across popular text-to-image systems including DALLE and various Stable Diffusion variants, demonstrate that FairCoT significantly enhances fairness and diversity without sacrificing image quality or semantic fidelity. By combining robust reasoning, lightweight deployment, and extensibility to multiple models, FairCoT represents a promising step toward more socially responsible and transparent AI driven content generation.
Authors: Uri Berger, Gabriel Stanovsky, Omri Abend, Lea Frermann
Abstract: The task of image captioning has recently been gaining popularity, and with it the complex task of evaluating the quality of image captioning models. In this work, we present the first survey and taxonomy of over 70 different image captioning metrics and their usage in hundreds of papers, specifically designed to help users select the most suitable metric for their needs. We find that despite the diversity of proposed metrics, the vast majority of studies rely on only five popular metrics, which we show to be weakly correlated with human ratings. We hypothesize that combining a diverse set of metrics can enhance correlation with human ratings. As an initial step, we demonstrate that a linear regression-based ensemble method, which we call EnsembEval, trained on one human ratings dataset, achieves improved correlation across five additional datasets, showing there is a lot of room for improvement by leveraging a diverse set of metrics.
Authors: Gen Li, Nikolaos Tsagkas, Jifei Song, Ruaridh Mon-Williams, Sethu Vijayakumar, Kun Shao, Laura Sevilla-Lara
Abstract: Affordance, defined as the potential actions that an object offers, is crucial for embodied AI agents. For example, such knowledge directs an agent to grasp a knife by the handle for cutting or by the blade for safe handover. While existing approaches have made notable progress, affordance research still faces three key challenges: data scarcity, poor generalization, and real-world deployment. Specifically, there is a lack of large-scale affordance datasets with precise segmentation maps, existing models struggle to generalize across different domains or novel object and affordance classes, and little work demonstrates deployability in real-world scenarios. In this work, we address these issues by proposing a complete affordance learning system that (1) takes in egocentric videos and outputs precise affordance annotations without human labeling, (2) leverages geometric information and vision foundation models to improve generalization, and (3) introduces a framework that facilitates affordance-oriented robotic manipulation such as tool grasping and robot-to-human tool handover. Experimental results show that our model surpasses the state-of-the-art by 13.8% in mIoU, and the framework achieves 77.1% successful grasping among 179 trials, including evaluations on seen, unseen classes, and cluttered scenes. Project page: https://reagan1311.github.io/affgrasp.
Authors: Eric Zhu, Mara Levy, Matthew Gwilliam, Abhinav Shrivastava
Abstract: Training a policy that can generalize to unknown objects is a long standing challenge within the field of robotics. The performance of a policy often drops significantly in situations where an object in the scene was not seen during training. To solve this problem, we present NeRF-Aug, a novel method that is capable of teaching a policy to interact with objects that are not present in the dataset. This approach differs from existing approaches by leveraging the speed, photorealism, and 3D consistency of a neural radiance field for augmentation. NeRF-Aug both creates more photorealistic data and runs 63% faster than existing methods. We demonstrate the effectiveness of our method on 5 tasks with 9 novel objects that are not present in the expert demonstrations. We achieve an average performance boost of 55.6% when comparing our method to the next best method. You can see video results at https://nerf-aug.github.io.
Authors: Yiqing Bo, Ansh Soni, Sudhanshu Srivastava, Meenakshi Khosla
Abstract: Neuroscience and artificial intelligence (AI) both face the challenge of interpreting high-dimensional neural data, where the comparative analysis of such data is crucial for revealing shared mechanisms and differences between these complex systems. Despite the widespread use of representational comparisons and the abundance classes of comparison methods, a critical question remains: which metrics are most suitable for these comparisons? While some studies evaluate metrics based on their ability to differentiate models of different origins or constructions (e.g., various architectures), another approach is to assess how well they distinguish models that exhibit distinct behaviors. To investigate this, we examine the degree of alignment between various representational similarity measures and behavioral outcomes, employing group statistics and a comprehensive suite of behavioral metrics for comparison. In our evaluation of eight commonly used representational similarity metrics in the visual domain -- spanning alignment-based, Canonical Correlation Analysis (CCA)-based, inner product kernel-based, and nearest-neighbor methods -- we found that metrics like linear Centered Kernel Alignment (CKA) and Procrustes distance, which emphasize the overall geometric structure or shape of representations, excelled in differentiating trained from untrained models and aligning with behavioral measures, whereas metrics such as linear predictivity, commonly used in neuroscience, demonstrated only moderate alignment with behavior. These insights are crucial for selecting metrics that emphasize behaviorally meaningful comparisons in NeuroAI research.
Authors: Amirhossein Abaskohi, Spandana Gella, Giuseppe Carenini, Issam H. Laradji
Abstract: Multimodal multihop question answering (MMQA) requires reasoning over images and text from multiple sources. Despite advances in visual question answering, this multihop setting remains underexplored due to a lack of quality datasets. Existing methods focus on single-hop, single-modality, or short texts, limiting real-world applications like interpreting educational documents with long, multimodal content. To fill this gap, we introduce FM2DS, the first framework for creating a high-quality dataset for MMQA. Our approach consists of a 5-stage pipeline that involves acquiring relevant multimodal documents from Wikipedia, synthetically generating high-level questions and answers, and validating them through rigorous criteria to ensure data quality. We evaluate our methodology by training models on our synthesized dataset and testing on two benchmarks: MultimodalQA and WebQA. Our results demonstrate that, with an equal sample size, models trained on our synthesized data outperform those trained on human-collected data by 1.9 in exact match (EM) score on average. Additionally, we introduce M2QA-Bench with 1k samples, the first benchmark for MMQA on long documents, generated using FM2DS and refined by human annotators. We believe our data synthesis method will serve as a strong foundation for training and evaluating MMQA models.
Authors: Maja Pavlovic
Abstract: To be considered reliable, a model must be calibrated so that its confidence in each decision closely reflects its true outcome. In this blogpost we'll take a look at the most commonly used definition for calibration and then dive into a frequently used evaluation measure for model calibration. We'll then cover some of the drawbacks of this measure and how these surfaced the need for additional notions of calibration, which require their own new evaluation measures. This post is not intended to be an in-depth dissection of all works on calibration, nor does it focus on how to calibrate models. Instead, it is meant to provide a gentle introduction to the different notions and their evaluation measures as well as to re-highlight some issues with a measure that is still widely used to evaluate calibration.
Authors: Aryan Garg, Yue Jiang, Antti Oulasvirta
Abstract: During the early stages of interface design, designers need to produce multiple sketches to explore a design space. Design tools often fail to support this critical stage, because they insist on specifying more details than necessary. Although recent advances in generative AI have raised hopes of solving this issue, in practice they fail because expressing loose ideas in a prompt is impractical. In this paper, we propose a diffusion-based approach to the low-effort generation of interface sketches. It breaks new ground by allowing flexible control of the generation process via three types of inputs: A) prompts, B) wireframes, and C) visual flows. The designer can provide any combination of these as input at any level of detail, and will get a diverse gallery of low-fidelity solutions in response. The unique benefit is that large design spaces can be explored rapidly with very little effort in input-specification. We present qualitative results for various combinations of input specifications. Additionally, we demonstrate that our model aligns more accurately with these specifications than other models.
Authors: Thomas Sanchez, Vladyslav Zalevskyi, Angeline Mihailov, Gerard Mart\'i-Juan, Elisenda Eixarch, Andras Jakab, Vincent Dunet, M\'eriam Koob, Guillaume Auzias, Meritxell Bach Cuadra
Abstract: Quality control (QC) has long been considered essential to guarantee the reliability of neuroimaging studies. It is particularly important for fetal brain MRI, where acquisitions and image processing techniques are less standardized than in adult imaging. In this work, we focus on automated quality control of super-resolution reconstruction (SRR) volumes of fetal brain MRI, an important processing step where multiple stacks of thick 2D slices are registered together and combined to build a single, isotropic and artifact-free T2 weighted volume. We propose FetMRQC$_{SR}$, a machine-learning method that extracts more than 100 image quality metrics to predict image quality scores using a random forest model. This approach is well suited to a problem that is high dimensional, with highly heterogeneous data and small datasets. We validate FetMRQC$_{SR}$ in an out-of-domain (OOD) setting and report high performance (ROC AUC = 0.89), even when faced with data from an unknown site or SRR method. We also investigate failure cases and show that they occur in $45\%$ of the images due to ambiguous configurations for which the rating from the expert is arguable. These results are encouraging and illustrate how a non deep learning-based method like FetMRQC$_{SR}$ is well suited to this multifaceted problem. Our tool, along with all the code used to generate, train and evaluate the model are available at https://github.com/Medical-Image-Analysis-Laboratory/fetmrqc_sr/ .
URLs: https://github.com/Medical-Image-Analysis-Laboratory/fetmrqc_sr/
Authors: Sidun Liu, Wenyu Li, Peng Qiao, Yong Dou
Abstract: Multi-view 3D reconstruction has remained an essential yet challenging problem in the field of computer vision. While DUSt3R and its successors have achieved breakthroughs in 3D reconstruction from unposed images, these methods exhibit significant limitations when scaling to multi-view scenarios, including high computational cost and cumulative error induced by global alignment. To address these challenges, we propose Regist3R, a novel stereo foundation model tailored for efficient and scalable incremental reconstruction. Regist3R leverages an incremental reconstruction paradigm, enabling large-scale 3D reconstructions from unordered and many-view image collections. We evaluate Regist3R on public datasets for camera pose estimation and 3D reconstruction. Our experiments demonstrate that Regist3R achieves comparable performance with optimization-based methods while significantly improving computational efficiency, and outperforms existing multi-view reconstruction models. Furthermore, to assess its performance in real-world applications, we introduce a challenging oblique aerial dataset which has long spatial spans and hundreds of views. The results highlight the effectiveness of Regist3R. We also demonstrate the first attempt to reconstruct large-scale scenes encompassing over thousands of views through pointmap-based foundation models, showcasing its potential for practical applications in large-scale 3D reconstruction tasks, including urban modeling, aerial mapping, and beyond.
Authors: Ziwen Xu, Shuxun Wang, Kewei Xu, Haoming Xu, Mengru Wang, Xinle Deng, Yunzhi Yao, Guozhou Zheng, Huajun Chen, Ningyu Zhang
Abstract: In this paper, we introduce EasyEdit2, a framework designed to enable plug-and-play adjustability for controlling Large Language Model (LLM) behaviors. EasyEdit2 supports a wide range of test-time interventions, including safety, sentiment, personality, reasoning patterns, factuality, and language features. Unlike its predecessor, EasyEdit2 features a new architecture specifically designed for seamless model steering. It comprises key modules such as the steering vector generator and the steering vector applier, which enable automatic generation and application of steering vectors to influence the model's behavior without modifying its parameters. One of the main advantages of EasyEdit2 is its ease of use-users do not need extensive technical knowledge. With just a single example, they can effectively guide and adjust the model's responses, making precise control both accessible and efficient. Empirically, we report model steering performance across different LLMs, demonstrating the effectiveness of these techniques. We have released the source code on GitHub at https://github.com/zjunlp/EasyEdit along with a demonstration notebook. In addition, we provide a demo video at https://www.youtube.com/watch?v=AkfoiPfp5rQ for a quick introduction.
URLs: https://github.com/zjunlp/EasyEdit, https://www.youtube.com/watch?v=AkfoiPfp5rQ
Authors: Hangyu Li, Qin Zhao, Haoran Xu, Xinyu Jiang, Qingwei Ben, Feiyu Jia, Haoyu Zhao, Liang Xu, Jia Zeng, Hanqing Wang, Bo Dai, Junting Dong, Jiangmiao Pang
Abstract: Teleoperation is a cornerstone of embodied-robot learning, and bimanual dexterous teleoperation in particular provides rich demonstrations that are difficult to obtain with fully autonomous systems. While recent studies have proposed diverse hardware pipelines-ranging from inertial motion-capture gloves to exoskeletons and vision-based interfaces-there is still no unified benchmark that enables fair, reproducible comparison of these systems. In this paper, we introduce TeleOpBench, a simulator-centric benchmark tailored to bimanual dexterous teleoperation. TeleOpBench contains 30 high-fidelity task environments that span pick-and-place, tool use, and collaborative manipulation, covering a broad spectrum of kinematic and force-interaction difficulty. Within this benchmark we implement four representative teleoperation modalities-(i) MoCap, (ii) VR device, (iii) arm-hand exoskeletons, and (iv) monocular vision tracking-and evaluate them with a common protocol and metric suite. To validate that performance in simulation is predictive of real-world behavior, we conduct mirrored experiments on a physical dual-arm platform equipped with two 6-DoF dexterous hands. Across 10 held-out tasks we observe a strong correlation between simulator and hardware performance, confirming the external validity of TeleOpBench. TeleOpBench establishes a common yardstick for teleoperation research and provides an extensible platform for future algorithmic and hardware innovation. Codes is now available at https://github.com/cyjdlhy/TeleOpBench .
Authors: Tianyu Zhang, Xinyu Wang, Lu Li, Zhenghan Tai, Jijun Chi, Jingrui Tian, Hailin He, Suyuchen Wang
Abstract: While diffusion models have revolutionized text-to-image generation with their ability to synthesize realistic and diverse scenes, they continue to struggle to generate consistent and legible text within images. This shortcoming is commonly attributed to the locality bias inherent in diffusion-based generation, which limits their ability to model long-range spatial dependencies. In this paper, we introduce $\textbf{STRICT}$, a benchmark designed to systematically stress-test the ability of diffusion models to render coherent and instruction-aligned text in images. Our benchmark evaluates models across multiple dimensions: (1) the maximum length of readable text that can be generated; (2) the correctness and legibility of the generated text, and (3) the ratio of not following instructions for generating text. We evaluate several state-of-the-art models, including proprietary and open-source variants, and reveal persistent limitations in long-range consistency and instruction-following capabilities. Our findings provide insights into architectural bottlenecks and motivate future research directions in multimodal generative modeling. We release our entire evaluation pipeline at https://github.com/tianyu-z/STRICT-Bench.
Authors: Junyu Liu, R. Kenny Jones, Daniel Ritchie
Abstract: We present PartComposer: a framework for part-level concept learning from single-image examples that enables text-to-image diffusion models to compose novel objects from meaningful components. Existing methods either struggle with effectively learning fine-grained concepts or require a large dataset as input. We propose a dynamic data synthesis pipeline generating diverse part compositions to address one-shot data scarcity. Most importantly, we propose to maximize the mutual information between denoised latents and structured concept codes via a concept predictor, enabling direct regulation on concept disentanglement and re-composition supervision. Our method achieves strong disentanglement and controllable composition, outperforming subject and part-level baselines when mixing concepts from the same, or different, object categories.
Authors: Justin Kerr, Kush Hari, Ethan Weber, Chung Min Kim, Brent Yi, Tyler Bonnen, Ken Goldberg, Angjoo Kanazawa
Abstract: Humans do not passively observe the visual world -- we actively look in order to act. Motivated by this principle, we introduce EyeRobot, a robotic system with gaze behavior that emerges from the need to complete real-world tasks. We develop a mechanical eyeball that can freely rotate to observe its surroundings and train a gaze policy to control it using reinforcement learning. We accomplish this by first collecting teleoperated demonstrations paired with a 360 camera. This data is imported into a simulation environment that supports rendering arbitrary eyeball viewpoints, allowing episode rollouts of eye gaze on top of robot demonstrations. We then introduce a BC-RL loop to train the hand and eye jointly: the hand (BC) agent is trained from rendered eye observations, and the eye (RL) agent is rewarded when the hand produces correct action predictions. In this way, hand-eye coordination emerges as the eye looks towards regions which allow the hand to complete the task. EyeRobot implements a foveal-inspired policy architecture allowing high resolution with a small compute budget, which we find also leads to the emergence of more stable fixation as well as improved ability to track objects and ignore distractors. We evaluate EyeRobot on five panoramic workspace manipulation tasks requiring manipulation in an arc surrounding the robot arm. Our experiments suggest EyeRobot exhibits hand-eye coordination behaviors which effectively facilitate manipulation over large workspaces with a single camera. See project site for videos: https://www.eyerobot.net/
Authors: Xiang Dai, Kyrollos Yanny, Kristina Monakhova, Nicholas Antipa
Abstract: High-dynamic-range (HDR) imaging is an essential technique for overcoming the dynamic range limits of image sensors. The classic method relies on multiple exposures, which slows capture time, resulting in motion artifacts when imaging dynamic scenes. Single-shot HDR imaging alleviates this issue by encoding HDR data into a single exposure, then computationally recovering it. Many established methods use strong image priors to recover improperly exposed image detail. These approaches struggle with extended highlight regions. We utilize the global reset release (GRR) shutter mode of an off-the-shelf sensor. GRR shutter mode applies a longer exposure time to rows closer to the bottom of the sensor. We use optics that relay a randomly permuted (shuffled) image onto the sensor, effectively creating spatially randomized exposures across the scene. The exposure diversity allows us to recover HDR data by solving an optimization problem with a simple total variation image prior. In simulation, we demonstrate that our method outperforms other single-shot methods when many sensor pixels are saturated (10% or more), and is competitive at a modest saturation (1%). Finally, we demonstrate a physical lab prototype that uses an off-the-shelf random fiber bundle for the optical shuffling. The fiber bundle is coupled to a low-cost commercial sensor operating in GRR shutter mode. Our prototype achieves a dynamic range of up to 73dB using an 8-bit sensor with 48dB dynamic range.
Authors: Suhang Hu, Wei Hu, Yuhang Su, Fan Zhang
Abstract: Vision-Language Models (VLMs) struggle with complex image annotation tasks, such as emotion classification and context-driven object detection, which demand sophisticated reasoning. Standard Supervised Fine-Tuning (SFT) focuses solely on annotation outcomes, ignoring underlying rationales, while Visual Reinforcement Fine-Tuning (Visual-RFT) produces inconsistent Chains of Thought (CoTs) due to the absence of high-quality, verified CoTs during pre-training. We introduce RISE (Reason-Inspire-Strengthen-Expertise), a two-stage framework to overcome these limitations. In the Reason stage (RISE-CoT), a reinforcement learning-driven "annotation-reasoning-annotation" closed-loop generates visually grounded, logically consistent CoTs by verifying their ability to reconstruct original annotations without direct leakage. The Inspire and Strengthen stage (RISE-R1) leverages a high-quality CoT subset, filtered by RISE-CoT rewards, for supervised fine-tuning, followed by reinforcement fine-tuning to produce interpretable reasoning and accurate annotations, achieving Expertise in complex visual tasks. Evaluated on complex and simple image annotation tasks, RISE-trained Qwen2-VL-2B outperforms SFT and Visual-RFT, achieving robust performance and enhanced explainability. RISE offers a self-supervised solution for advancing VLM reasoning without requiring manually annotated CoTs.Code and resources are available at: https://github.com/HSH55/RISE.
Authors: Sebastian Ibarra, Javier del Riego, Alessandro Catanese, Julian Cuba, Julian Cardona, Nataly Leon, Jonathan Infante, Karim Lekadir, Oliver Diaz, Richard Osuala
Abstract: Dynamic contrast-enhanced (DCE) MRI is essential for breast cancer diagnosis and treatment. However, its reliance on contrast agents introduces safety concerns, contraindications, increased cost, and workflow complexity. To this end, we present pre-contrast conditioned denoising diffusion probabilistic models to synthesize DCE-MRI, introducing, evaluating, and comparing a total of 22 generative model variants in both single-breast and full breast settings. Towards enhancing lesion fidelity, we introduce both tumor-aware loss functions and explicit tumor segmentation mask conditioning. Using a public multicenter dataset and comparing to respective pre-contrast baselines, we observe that subtraction image-based models consistently outperform post-contrast-based models across five complementary evaluation metrics. Apart from assessing the entire image, we also separately evaluate the region of interest, where both tumor-aware losses and segmentation mask inputs improve evaluation metrics. The latter notably enhance qualitative results capturing contrast uptake, albeit assuming access to tumor localization inputs that are not guaranteed to be available in screening settings. A reader study involving 2 radiologists and 4 MRI technologists confirms the high realism of the synthetic images, indicating an emerging clinical potential of generative contrast-enhancement. We share our codebase at https://github.com/sebastibar/conditional-diffusion-breast-MRI.
URLs: https://github.com/sebastibar/conditional-diffusion-breast-MRI.
Authors: Han Liang, Jiahui Zhou, Zicheng Zhou, Xiaoxi Zhang, Xu Chen
Abstract: The escalating adoption of diffusion models for applications such as image generation demands efficient parallel inference techniques to manage their substantial computational cost. However, existing diffusion parallelism inference schemes often underutilize resources in heterogeneous multi-GPU environments, where varying hardware capabilities or background tasks cause workload imbalance. This paper introduces Spatio-Temporal Adaptive Diffusion Inference (STADI), a novel framework to accelerate diffusion model inference in such settings. At its core is a hybrid scheduler that orchestrates fine-grained parallelism across both temporal and spatial dimensions. Temporally, STADI introduces a novel computation-aware step allocator applied after warmup phases, using a least-common-multiple-minimizing quantization technique to reduce denoising steps on slower GPUs and execution synchronization. To further minimize GPU idle periods, STADI executes an elastic patch parallelism mechanism that allocates variably sized image patches to GPUs according to their computational capability, ensuring balanced workload distribution through a complementary spatial mechanism. Extensive experiments on both load-imbalanced and heterogeneous multi-GPU clusters validate STADI's efficacy, demonstrating improved load balancing and mitigation of performance bottlenecks. Compared to patch parallelism, a state-of-the-art diffusion inference framework, our method significantly reduces end-to-end inference latency by up to 45% and significantly improves resource utilization on heterogeneous GPUs.
Authors: Friedrich Wolf-Monheim
Abstract: Next to decision tree and k-nearest neighbours algorithms deep convolutional neural networks (CNNs) are widely used to classify audio data in many domains like music, speech or environmental sounds. To train a specific CNN various spectral and rhythm features like mel-scaled spectrograms, mel-frequency cepstral coefficients (MFCC), cyclic tempograms, short-time Fourier transform (STFT) chromagrams, constant-Q transform (CQT) chromagrams and chroma energy normalized statistics (CENS) chromagrams can be used as digital image input data for the neural network. The performance of these spectral and rhythm features for audio category level as well as audio class level classification is investigated in detail with a deep CNN and the ESC-50 dataset with 2,000 labeled environmental audio recordings using an end-to-end deep learning pipeline. The evaluated metrics accuracy, precision, recall and F1 score for multiclass classification clearly show that the mel-scaled spectrograms and the mel-frequency cepstral coefficients (MFCC) perform significantly better then the other spectral and rhythm features investigated in this research for audio classification tasks using deep CNNs.