Authors: Masud Ahmed, Zahid Hasan, Syed Arefinul Haque, Abu Zaher Md Faridee, Sanjay Purushotham, Suya You, Nirmalya Roy
Abstract: Traditional transformer-based semantic segmentation relies on quantized embeddings. However, our analysis reveals that autoencoder accuracy on segmentation mask using quantized embeddings (e.g. VQ-VAE) is 8% lower than continuous-valued embeddings (e.g. KL-VAE). Motivated by this, we propose a continuous-valued embedding framework for semantic segmentation. By reformulating semantic mask generation as a continuous image-to-embedding diffusion process, our approach eliminates the need for discrete latent representations while preserving fine-grained spatial and semantic details. Our key contribution includes a diffusion-guided autoregressive transformer that learns a continuous semantic embedding space by modeling long-range dependencies in image features. Our framework contains a unified architecture combining a VAE encoder for continuous feature extraction, a diffusion-guided transformer for conditioned embedding generation, and a VAE decoder for semantic mask reconstruction. Our setting facilitates zero-shot domain adaptation capabilities enabled by the continuity of the embedding space. Experiments across diverse datasets (e.g., Cityscapes and domain-shifted variants) demonstrate state-of-the-art robustness to distribution shifts, including adverse weather (e.g., fog, snow) and viewpoint variations. Our model also exhibits strong noise resilience, achieving robust performance ($\approx$ 95% AP compared to baseline) under gaussian noise, moderate motion blur, and moderate brightness/contrast variations, while experiencing only a moderate impact ($\approx$ 90% AP compared to baseline) from 50% salt and pepper noise, saturation and hue shifts. Code available: https://github.com/mahmed10/CAMSS.git
Authors: Federico Cocchi, Nicholas Moratelli, Davide Caffagni, Sara Sarto, Lorenzo Baraldi, Marcella Cornia, Rita Cucchiara
Abstract: Recent progress in Multimodal Large Language Models (MLLMs) has highlighted the critical roles of both the visual backbone and the underlying language model. While prior work has primarily focused on scaling these components to billions of parameters, the trade-offs between model size, architecture, and performance remain underexplored. Additionally, inconsistencies in training data and evaluation protocols have hindered direct comparisons, making it difficult to derive optimal design choices. In this paper, we introduce LLaVA-MORE, a new family of MLLMs that integrates recent language models with diverse visual backbones. To ensure fair comparisons, we employ a unified training protocol applied consistently across all architectures. Our analysis systematically explores both small- and medium-scale LLMs -- including Phi-4, LLaMA-3.1, and Gemma-2 -- to evaluate multimodal reasoning, generation, and instruction following, while examining the relationship between model size and performance. Beyond evaluating the LLM impact on final results, we conduct a comprehensive study of various visual encoders, ranging from CLIP-based architectures to alternatives such as DINOv2, SigLIP, and SigLIP2. Additional experiments investigate the effects of increased image resolution and variations in pre-training datasets. Overall, our results provide insights into the design of more effective MLLMs, offering a reproducible evaluation framework that facilitates direct comparisons and can guide future model development. Our source code and trained models are publicly available at: https://github.com/aimagelab/LLaVA-MORE.
Authors: Matthew Massey, Abdullah-Al-Zubaer Imran
Abstract: Surficial geologic mapping is essential for understanding Earth surface processes, addressing modern challenges such as climate change and national security, and supporting common applications in engineering and resource management. However, traditional mapping methods are labor-intensive, limiting spatial coverage and introducing potential biases. To address these limitations, we introduce EarthScape, a novel, AI-ready multimodal dataset specifically designed for surficial geologic mapping and Earth surface analysis. EarthScape integrates high-resolution aerial RGB and near-infrared (NIR) imagery, digital elevation models (DEM), multi-scale DEM-derived terrain features, and hydrologic and infrastructure vector data. The dataset provides detailed annotations for seven distinct surficial geologic classes encompassing various geological processes. We present a comprehensive data processing pipeline using open-sourced raw data and establish baseline benchmarks using different spatial modalities to demonstrate the utility of EarthScape. As a living dataset with a vision for expansion, EarthScape bridges the gap between computer vision and Earth sciences, offering a valuable resource for advancing research in multimodal learning, geospatial analysis, and geological mapping. Our code is available at https://github.com/masseygeo/earthscape.
Authors: Am\'elie Royer, Moritz B\"ohle, Gabriel de Marmiesse, Laurent Mazar\'e, Neil Zeghidour, Alexandre D\'efossez, Patrick P\'erez
Abstract: The recent successes of Vision-Language models raise the question of how to equivalently imbue a pretrained speech model with vision understanding, an important milestone towards building a multimodal speech model able to freely converse about images. Building such a conversational Vision-Speech model brings its unique challenges: (i) paired image-speech datasets are much scarcer than their image-text counterparts, (ii) ensuring real-time latency at inference is crucial thus bringing compute and memory constraints, and (iii) the model should preserve prosodic features (e.g., speaker tone) which cannot be inferred from text alone. In this work, we introduce MoshiVis, augmenting a recent dialogue speech LLM, Moshi, with visual inputs through lightweight adaptation modules. An additional dynamic gating mechanism enables the model to more easily switch between the visual inputs and unrelated conversation topics. To reduce training costs, we design a simple one-stage, parameter-efficient fine-tuning pipeline in which we leverage a mixture of image-text (i.e., "speechless") and image-speech samples. We evaluate the model on downstream visual understanding tasks with both audio and text prompts, and report qualitative samples of interactions with MoshiVis. Our inference code will be made available, as well as the image-speech data used for audio evaluation.
Authors: Ritabrata Chakraborty, Shivakumara Palaiahnakote, Umapada Pal, Cheng-Lin Liu
Abstract: Modern scene text recognition systems often depend on large end-to-end architectures that require extensive training and are prohibitively expensive for real-time scenarios. In such cases, the deployment of heavy models becomes impractical due to constraints on memory, computational resources, and latency. To address these challenges, we propose a novel, training-free plug-and-play framework that leverages the strengths of pre-trained text recognizers while minimizing redundant computations. Our approach uses context-based understanding and introduces an attention-based segmentation stage, which refines candidate text regions at the pixel level, improving downstream recognition. Instead of performing traditional text detection that follows a block-level comparison between feature map and source image and harnesses contextual information using pretrained captioners, allowing the framework to generate word predictions directly from scene context.Candidate texts are semantically and lexically evaluated to get a final score. Predictions that meet or exceed a pre-defined confidence threshold bypass the heavier process of end-to-end text STR profiling, ensuring faster inference and cutting down on unnecessary computations. Experiments on public benchmarks demonstrate that our paradigm achieves performance on par with state-of-the-art systems, yet requires substantially fewer resources.
Authors: Jumanh Atoum, Garrison L. H. Johnston, Nabil Simaan, Jie Ying Wu
Abstract: Recognizing surgical gestures in real-time is a stepping stone towards automated activity recognition, skill assessment, intra-operative assistance, and eventually surgical automation. The current robotic surgical systems provide us with rich multi-modal data such as video and kinematics. While some recent works in multi-modal neural networks learn the relationships between vision and kinematics data, current approaches treat kinematics information as independent signals, with no underlying relation between tool-tip poses. However, instrument poses are geometrically related, and the underlying geometry can aid neural networks in learning gesture representation. Therefore, we propose combining motion invariant measures (curvature and torsion) with vision and kinematics data using a relational graph network to capture the underlying relations between different data streams. We show that gesture recognition improves when combining invariant signals with tool position, achieving 90.3\% frame-wise accuracy on the JIGSAWS suturing dataset. Our results show that motion invariant signals coupled with position are better representations of gesture motion compared to traditional position and quaternion representations. Our results highlight the need for geometric-aware modeling of kinematics for gesture recognition.
Authors: Miguel Ure\~na Pliego, Rub\'en Mart\'inez Mar\'in, Nianfang Shi, Takeru Shibayama, Ulrich Leth, Miguel Marchamalo Sacrist\'an
Abstract: This study explores the integration of machine learning into urban aerial image analysis, with a focus on identifying infrastructure surfaces for cars and pedestrians and analyzing historical trends. It emphasizes the transition from convolutional architectures to transformer-based pre-trained models, underscoring their potential in global geospatial analysis. A workflow is presented for automatically generating geospatial datasets, enabling the creation of semantic segmentation datasets from various sources, including WMS/WMTS links, vectorial cartography, and OpenStreetMap (OSM) overpass-turbo requests. The developed code allows a fast dataset generation process for training machine learning models using openly available data without manual labelling. Using aerial imagery and vectorial data from the respective geographical offices of Madrid and Vienna, two datasets were generated for car and pedestrian surface detection. A transformer-based model was trained and evaluated for each city, demonstrating good accuracy values. The historical trend analysis involved applying the trained model to earlier images predating the availability of vectorial data 10 to 20 years, successfully identifying temporal trends in infrastructure for pedestrians and cars across different city areas. This technique is applicable for municipal governments to gather valuable data at a minimal cost.
Authors: Shravan Nayak, Xiangru Jian, Kevin Qinghong Lin, Juan A. Rodriguez, Montek Kalsi, Rabiul Awal, Nicolas Chapados, M. Tamer \"Ozsu, Aishwarya Agrawal, David Vazquez, Christopher Pal, Perouz Taslakian, Spandana Gella, Sai Rajeswar
Abstract: Autonomous agents that navigate Graphical User Interfaces (GUIs) to automate tasks like document editing and file management can greatly enhance computer workflows. While existing research focuses on online settings, desktop environments, critical for many professional and everyday tasks, remain underexplored due to data collection challenges and licensing issues. We introduce UI-Vision, the first comprehensive, license-permissive benchmark for offline, fine-grained evaluation of computer use agents in real-world desktop environments. Unlike online benchmarks, UI-Vision provides: (i) dense, high-quality annotations of human demonstrations, including bounding boxes, UI labels, and action trajectories (clicks, drags, and keyboard inputs) across 83 software applications, and (ii) three fine-to-coarse grained tasks-Element Grounding, Layout Grounding, and Action Prediction-with well-defined metrics to rigorously evaluate agents' performance in desktop environments. Our evaluation reveals critical limitations in state-of-the-art models like UI-TARS-72B, including issues with understanding professional software, spatial reasoning, and complex actions like drag-and-drop. These findings highlight the challenges in developing fully autonomous computer use agents. By releasing UI-Vision as open-source, we aim to advance the development of more capable agents for real-world desktop tasks.
Authors: Kyle Vedder
Abstract: Scene flow estimation is the task of describing 3D motion between temporally successive observations. This thesis aims to build the foundation for building scene flow estimators with two important properties: they are scalable, i.e. they improve with access to more data and computation, and they are flexible, i.e. they work out-of-the-box in a variety of domains and on a variety of motion patterns without requiring significant hyperparameter tuning. In this dissertation we present several concrete contributions towards this. In Chapter 1 we contextualize scene flow and its prior methods. In Chapter 2 we present a blueprint to build and scale feedforward scene flow estimators without requiring expensive human annotations via large scale distillation from pseudolabels provided by strong unsupervised test-time optimization methods. In Chapter 3 we introduce a benchmark to better measure estimate quality across diverse object types, better bringing into focus what we care about and expect from scene flow estimators, and use this benchmark to host a public challenge that produced significant progress. In Chapter 4 we present a state-of-the-art unsupervised scene flow estimator that introduces a new, full sequence problem formulation and exhibits great promise in adjacent domains like 3D point tracking. Finally, in Chapter 5 I philosophize about what's next for scene flow and its potential future broader impacts.
Authors: Yuming Gu, Phong Tran, Yujian Zheng, Hongyi Xu, Heyuan Li, Adilbek Karmanov, Hao Li
Abstract: Generating high-quality 360-degree views of human heads from single-view images is essential for enabling accessible immersive telepresence applications and scalable personalized content creation. While cutting-edge methods for full head generation are limited to modeling realistic human heads, the latest diffusion-based approaches for style-omniscient head synthesis can produce only frontal views and struggle with view consistency, preventing their conversion into true 3D models for rendering from arbitrary angles. We introduce a novel approach that generates fully consistent 360-degree head views, accommodating human, stylized, and anthropomorphic forms, including accessories like glasses and hats. Our method builds on the DiffPortrait3D framework, incorporating a custom ControlNet for back-of-head detail generation and a dual appearance module to ensure global front-back consistency. By training on continuous view sequences and integrating a back reference image, our approach achieves robust, locally continuous view synthesis. Our model can be used to produce high-quality neural radiance fields (NeRFs) for real-time, free-viewpoint rendering, outperforming state-of-the-art methods in object synthesis and 360-degree head generation for very challenging input portraits.
Authors: Arindam Dutta, Meng Zheng, Zhongpai Gao, Benjamin Planche, Anwesha Choudhuri, Terrence Chen, Amit K. Roy-Chowdhury, Ziyan Wu
Abstract: Reconstructing clothed humans from a single image is a fundamental task in computer vision with wide-ranging applications. Although existing monocular clothed human reconstruction solutions have shown promising results, they often rely on the assumption that the human subject is in an occlusion-free environment. Thus, when encountering in-the-wild occluded images, these algorithms produce multiview inconsistent and fragmented reconstructions. Additionally, most algorithms for monocular 3D human reconstruction leverage geometric priors such as SMPL annotations for training and inference, which are extremely challenging to acquire in real-world applications. To address these limitations, we propose CHROME: Clothed Human Reconstruction with Occlusion-Resilience and Multiview-ConsistEncy from a Single Image, a novel pipeline designed to reconstruct occlusion-resilient 3D humans with multiview consistency from a single occluded image, without requiring either ground-truth geometric prior annotations or 3D supervision. Specifically, CHROME leverages a multiview diffusion model to first synthesize occlusion-free human images from the occluded input, compatible with off-the-shelf pose control to explicitly enforce cross-view consistency during synthesis. A 3D reconstruction model is then trained to predict a set of 3D Gaussians conditioned on both the occluded input and synthesized views, aligning cross-view details to produce a cohesive and accurate 3D representation. CHROME achieves significant improvements in terms of both novel view synthesis (upto 3 db PSNR) and geometric reconstruction under challenging conditions.
Authors: William Ljungbergh, Adam Lilja, Adam Tonderski. Arvid Laveno Ling, Carl Lindstr\"om, Willem Verbeke, Junsheng Fu, Christoffer Petersson, Lars Hammarstrand, Michael Felsberg
Abstract: Self-supervised pre-training based on next-token prediction has enabled large language models to capture the underlying structure of text, and has led to unprecedented performance on a large array of tasks when applied at scale. Similarly, autonomous driving generates vast amounts of spatiotemporal data, alluding to the possibility of harnessing scale to learn the underlying geometric and semantic structure of the environment and its evolution over time. In this direction, we propose a geometric and semantic self-supervised pre-training method, GASP, that learns a unified representation by predicting, at any queried future point in spacetime, (1) general occupancy, capturing the evolving structure of the 3D scene; (2) ego occupancy, modeling the ego vehicle path through the environment; and (3) distilled high-level features from a vision foundation model. By modeling geometric and semantic 4D occupancy fields instead of raw sensor measurements, the model learns a structured, generalizable representation of the environment and its evolution through time. We validate GASP on multiple autonomous driving benchmarks, demonstrating significant improvements in semantic occupancy forecasting, online mapping, and ego trajectory prediction. Our results demonstrate that continuous 4D geometric and semantic occupancy prediction provides a scalable and effective pre-training paradigm for autonomous driving. For code and additional visualizations, see \href{https://research.zenseact.com/publications/gasp/.
Authors: C\'edric Vincent, Taehyoung Kim, Henri Mee{\ss}
Abstract: Semantic segmentation from RGB cameras is essential to the perception of autonomous flying vehicles. The stability of predictions through the captured videos is paramount to their reliability and, by extension, to the trustworthiness of the agents. In this paper, we propose a lightweight video semantic segmentation approach-suited to onboard real-time inference-achieving high temporal consistency on aerial data through Semantic Similarity Propagation across frames. SSP temporally propagates the predictions of an efficient image segmentation model with global registration alignment to compensate for camera movements. It combines the current estimation and the prior prediction with linear interpolation using weights computed from the features similarities of the two frames. Because data availability is a challenge in this domain, we propose a consistency-aware Knowledge Distillation training procedure for sparsely labeled datasets with few annotations. Using a large image segmentation model as a teacher to train the efficient SSP, we leverage the strong correlations between labeled and unlabeled frames in the same training videos to obtain high-quality supervision on all frames. KD-SSP obtains a significant temporal consistency increase over the base image segmentation model of 12.5% and 6.7% TC on UAVid and RuralScapes respectively, with higher accuracy and comparable inference speed. On these aerial datasets, KD-SSP provides a superior segmentation quality and inference speed trade-off than other video methods proposed for general applications and shows considerably higher consistency. The code will be made publicly available upon acceptance.
Authors: Benidir Yanis, Gonthier Nicolas, Mallet Clement
Abstract: Bi-temporal change detection at scale based on Very High Resolution (VHR) images is crucial for Earth monitoring. This remains poorly addressed so far: methods either require large volumes of annotated data (semantic case), or are limited to restricted datasets (binary set-ups). Most approaches do not exhibit the versatility required for temporal and spatial adaptation: simplicity in architecture design and pretraining on realistic and comprehensive datasets. Synthetic datasets are the key solution but still fail to handle complex and diverse scenes. In this paper, we present HySCDG a generative pipeline for creating a large hybrid semantic change detection dataset that contains both real VHR images and inpainted ones, along with land cover semantic map at both dates and the change map. Being semantically and spatially guided, HySCDG generates realistic images, leading to a comprehensive and hybrid transfer-proof dataset FSC-180k. We evaluate FSC-180k on five change detection cases (binary and semantic), from zero-shot to mixed and sequential training, and also under low data regime training. Experiments demonstrate that pretraining on our hybrid dataset leads to a significant performance boost, outperforming SyntheWorld, a fully synthetic dataset, in every configuration. All codes, models, and data are available here: $\href{https://yb23.github.io/projects/cywd/}{https://yb23.github.io/projects/cywd/}$.
URLs: https://yb23.github.io/projects/cywd/, https://yb23.github.io/projects/cywd/
Authors: Jiaqi Liu, Jichao Zahng, Paolo Rota, Nicu Sebe
Abstract: The Latent Diffusion Model (LDM) has demonstrated strong capabilities in high-resolution image generation and has been widely employed for Pose-Guided Person Image Synthesis (PGPIS), yielding promising results. However, the compression process of LDM often results in the deterioration of details, particularly in sensitive areas such as facial features and clothing textures. In this paper, we propose a Multi-focal Conditioned Latent Diffusion (MCLD) method to address these limitations by conditioning the model on disentangled, pose-invariant features from these sensitive regions. Our approach utilizes a multi-focal condition aggregation module, which effectively integrates facial identity and texture-specific information, enhancing the model's ability to produce appearance realistic and identity-consistent images. Our method demonstrates consistent identity and appearance generation on the DeepFashion dataset and enables flexible person image editing due to its generation consistency. The code is available at https://github.com/jqliu09/mcld.
Authors: Panagiota Moraiti, Efstathios Karypidis
Abstract: This paper outlines our approach to the 5th CLVision challenge at CVPR, which addresses the Class-Incremental with Repetition (CIR) scenario. In contrast to traditional class incremental learning, this novel setting introduces unique challenges and research opportunities, particularly through the integration of unlabeled data into the training process. In the CIR scenario, encountered classes may reappear in later learning experiences, and each experience may involve only a subset of the overall class distribution. Additionally, the unlabeled data provided during training may include instances of unseen classes, or irrelevant classes which should be ignored. Our approach focuses on retaining previously learned knowledge by utilizing knowledge distillation and pseudo-labeling techniques. The key characteristic of our method is the exploitation of unlabeled data during training, in order to maintain optimal performance on instances of previously encountered categories and reduce the detrimental effects of catastrophic forgetting. Our method achieves an average accuracy of 16.68\% during the pre-selection phase and 21.19% during the final evaluation phase, outperforming the baseline accuracy of 9.39%. We provide the implementation code at https://github.com/panagiotamoraiti/continual-learning-challenge-2024 .
URLs: https://github.com/panagiotamoraiti/continual-learning-challenge-2024
Authors: Neehar Kondapaneni, Oisin Mac Aodha, Pietro Perona
Abstract: How do two deep neural networks differ in how they arrive at a decision? Measuring the similarity of deep networks has been a long-standing open question. Most existing methods provide a single number to measure the similarity of two networks at a given layer, but give no insight into what makes them similar or dissimilar. We introduce an interpretable representational similarity method (RSVC) to compare two networks. We use RSVC to discover shared and unique visual concepts between two models. We show that some aspects of model differences can be attributed to unique concepts discovered by one model that are not well represented in the other. Finally, we conduct extensive evaluation across different vision model architectures and training protocols to demonstrate its effectiveness.
Authors: Sam Narimani, Solveig Roth Hoff, Kathinka Dahli Kurz, Kjell-Inge Gjesdal, Jurgen Geisler, Endre Grovik
Abstract: Purpose: Segmentation of the breast lesion in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an essential step to accurately diagnose and plan treatment and monitor progress. This study aims to highlight the impact of breast region segmentation (BRS) on deep learning-based breast lesion segmentation (BLS) in breast DCE-MRI. Methods Using the Stavanger Dataset containing primarily 59 DCE-MRI scans and UNet++ as deep learning models, four different process were conducted to compare effect of BRS on BLS. These four approaches included the whole volume without BRS and with BRS, BRS with the selected lesion slices and lastly optimal volume with BRS. Preprocessing methods like augmentation and oversampling were used to enhance the small dataset, data shape uniformity and improve model performance. Optimal volume size were investigated by a precise process to ensure that all lesions existed in slices. To evaluate the model, a hybrid loss function including dice, focal and cross entropy along with 5-fold cross validation method were used and lastly a test dataset which was randomly split used to evaluate the model performance on unseen data for each of four mentioned approaches. Results Results demonstrate that using BRS considerably improved model performance and validation. Significant improvement in last approach -- optimal volume with BRS -- compared to the approach without BRS counting around 50 percent demonstrating how effective BRS has been in BLS. Moreover, huge improvement in energy consumption, decreasing up to 450 percent, introduces a green solution toward a more environmentally sustainable approach for future work on large dataset.
Authors: Weiwen Hu, Niccol\`o Parodi, Marcus Zepp, Ingo Feldmann, Oliver Schreer, Peter Eisert
Abstract: Open-vocabulary segmentation, powered by large visual-language models like CLIP, has expanded 2D segmentation capabilities beyond fixed classes predefined by the dataset, enabling zero-shot understanding across diverse scenes. Extending these capabilities to 3D segmentation introduces challenges, as CLIP's image-based embeddings often lack the geometric detail necessary for 3D scene segmentation. Recent methods tend to address this by introducing additional segmentation models or replacing CLIP with variations trained on segmentation data, which lead to redundancy or loss on CLIP's general language capabilities. To overcome this limitation, we introduce SPNeRF, a NeRF based zero-shot 3D segmentation approach that leverages geometric priors. We integrate geometric primitives derived from the 3D scene into NeRF training to produce primitive-wise CLIP features, avoiding the ambiguity of point-wise features. Additionally, we propose a primitive-based merging mechanism enhanced with affinity scores. Without relying on additional segmentation models, our method further explores CLIP's capability for 3D segmentation and achieves notable improvements over original LERF.
Authors: Yuqing Zhang, Qi Han, Ligeng Wang, Kai Cheng, Bo Wang, Kun Zhan
Abstract: Most existing graph-based semi-supervised hyperspectral image classification methods rely on superpixel partitioning techniques. However, they suffer from misclassification of certain pixels due to inaccuracies in superpixel boundaries, \ie, the initial inaccuracies in superpixel partitioning limit overall classification performance. In this paper, we propose a novel graph-weighted contrastive learning approach that avoids the use of superpixel partitioning and directly employs neural networks to learn hyperspectral image representation. Furthermore, while many approaches require all graph nodes to be available during training, our approach supports mini-batch training by processing only a subset of nodes at a time, reducing computational complexity and improving generalization to unseen nodes. Experimental results on three widely-used datasets demonstrate the effectiveness of the proposed approach compared to baselines relying on superpixel partitioning.
Authors: Sarosij Bose, Arindam Dutta, Sayak Nag, Junge Zhang, Jiachen Li, Konstantinos Karydis, Amit K. Roy Chowdhury
Abstract: Reconstructing 3D scenes from a single image is a fundamentally ill-posed task due to the severely under-constrained nature of the problem. Consequently, when the scene is rendered from novel camera views, existing single image to 3D reconstruction methods render incoherent and blurry views. This problem is exacerbated when the unseen regions are far away from the input camera. In this work, we address these inherent limitations in existing single image-to-3D scene feedforward networks. To alleviate the poor performance due to insufficient information beyond the input image's view, we leverage a strong generative prior in the form of a pre-trained latent video diffusion model, for iterative refinement of a coarse scene represented by optimizable Gaussian parameters. To ensure that the style and texture of the generated images align with that of the input image, we incorporate on-the-fly Fourier-style transfer between the generated images and the input image. Additionally, we design a semantic uncertainty quantification module that calculates the per-pixel entropy and yields uncertainty maps used to guide the refinement process from the most confident pixels while discarding the remaining highly uncertain ones. We conduct extensive experiments on real-world scene datasets, including in-domain RealEstate-10K and out-of-domain KITTI-v2, showing that our approach can provide more realistic and high-fidelity novel view synthesis results compared to existing state-of-the-art methods.
Authors: Mir Mohammad Khaleghi, Mehran Safayani, Abdolreza Mirzaei
Abstract: We present GraPLUS (Graph-based Placement Using Semantics), a novel framework for plausible object placement in images that leverages scene graphs and large language models. Our approach uniquely combines graph-structured scene representation with semantic understanding to determine contextually appropriate object positions. The framework employs GPT-2 to transform categorical node and edge labels into rich semantic embeddings that capture both definitional characteristics and typical spatial contexts, enabling nuanced understanding of object relationships and placement patterns. GraPLUS achieves placement accuracy of 92.1% and an FID score of 28.83 on the OPA dataset, outperforming state-of-the-art methods by 8.1% while maintaining competitive visual quality. In human evaluation studies involving 964 samples assessed by 19 participants, our method was preferred in 52.1% of cases, significantly outperforming previous approaches. The framework's key innovations include: (i) leveraging pre-trained scene graph models that transfer knowledge from other domains, (ii) edge-aware graph neural networks that process scene semantics through structured relationships, (iii) a cross-modal attention mechanism that aligns categorical embeddings with enhanced scene features, and (iv) a multiobjective training strategy incorporating semantic consistency constraints.
Authors: Huan Lei
Abstract: Neural surface reconstruction has been dominated by implicit representations with marching cubes for explicit surface extraction. However, those methods typically require high-quality normals for accurate reconstruction. We propose OffsetOPT, a method that reconstructs explicit surfaces directly from 3D point clouds and eliminates the need for point normals. The approach comprises two stages: first, we train a neural network to predict surface triangles based on local point geometry, given uniformly distributed training point clouds. Next, we apply the frozen network to reconstruct surfaces from unseen point clouds by optimizing a per-point offset to maximize the accuracy of triangle predictions. Compared to state-of-the-art methods, OffsetOPT not only excels at reconstructing overall surfaces but also significantly preserves sharp surface features. We demonstrate its accuracy on popular benchmarks, including small-scale shapes and large-scale open surfaces.
Authors: Boshra Khalili, Andrew W. Smyth
Abstract: In autonomous driving, open-ended question answering often suffers from unreliable evaluations because freeform responses require either complex metrics or subjective human judgment. To address this challenge, we introduce AutoDrive-QA, an automatic pipeline that converts existing driving QA datasets (including DriveLM, NuScenes-QA, and LingoQA) into a structured multiple-choice question (MCQ) format. This benchmark systematically assesses perception, prediction, and planning tasks, providing a standardized and objective evaluation framework. AutoDrive-QA employs an automated pipeline that leverages large language models (LLMs) to generate high-quality, contextually relevant distractors based on domain-specific error patterns commonly found in autonomous driving scenarios. To evaluate both general capabilities and generalization performance, we test the benchmark on three public datasets and conduct zero-shot experiments on an unseen dataset. The zero-shot evaluations reveal that GPT-4V leads with 69.57% accuracy -- achieving 74.94% in Perception, 65.33% in Prediction, and 68.45% in Planning -- demonstrating that while all models excel in Perception, they struggle in Prediction. Consequently, AutoDrive-QA establishes a rigorous, unbiased standard for integrating and evaluating different vision-language models across various autonomous driving datasets, thereby improving generalization in this field. We release all the codes in the AutoDrive-QA GitHub Repository.
Authors: Parham Saremi, Amar Kumar, Mohammed Mohammed, Zahra TehraniNasab, Tal Arbel
Abstract: Vision-Language Foundation Models (VLFM) have shown a tremendous increase in performance in terms of generating high-resolution, photorealistic natural images. While VLFMs show a rich understanding of semantic content across modalities, they often struggle with fine-grained alignment tasks that require precise correspondence between image regions and textual descriptions a limitation in medical imaging, where accurate localization and detection of clinical features are essential for diagnosis and analysis. To address this issue, we propose a multi-stage architecture where a pre-trained VLFM provides a cursory semantic understanding, while a reinforcement learning (RL) algorithm refines the alignment through an iterative process that optimizes for understanding semantic context. The reward signal is designed to align the semantic information of the text with synthesized images. We demonstrate the effectiveness of our method on a medical imaging skin dataset where the generated images exhibit improved generation quality and alignment with prompt over the fine-tuned Stable Diffusion. We also show that the synthesized samples could be used to improve disease classifier performance for underrepresented subgroups through augmentation.
Authors: Jingyun Liu, Daiqin Yang, Zhenzhong Chen
Abstract: Recovering high-frequency textures in image demosaicking remains a challenging issue. While existing methods introduced elaborate spatial learning methods, they still exhibit limited performance. To address this issue, a frequency enhancement approach is proposed. Based on the frequency analysis of color filter array (CFA)/demosaicked/ground truth images, we propose Dual-path Frequency Enhancement Network (DFENet), which reconstructs RGB images in a divide-and-conquer manner through fourier-domain frequency selection. In DFENet, two frequency selectors are employed, each selecting a set of frequency components for processing along separate paths. One path focuses on generating missing information through detail refinement in spatial domain, while the other aims at suppressing undesirable frequencies with the guidance of CFA images in frequency domain. Multi-level frequency supervision with a stagewise training strategy is employed to further improve the reconstruction performance. With these designs, the proposed DFENet outperforms other state-of-the-art algorithms on different datasets and demonstrates significant advantages on hard cases. Moreover, to better assess algorithms' ability to reconstruct high-frequency textures, a new dataset, LineSet37, is contributed, which consists of 37 artificially designed and generated images. These images feature complex line patterns and are prone to severe visual artifacts like color moir\'e after demosaicking. Experiments on LineSet37 offer a more targeted evaluation of performance on challenging cases. The code and dataset are available at https://github.com/VelvetReverie/DFENet-demosaicking.
Authors: Abduljaleel Adejumo, Faegheh Yeganli, Clifford Broni-bediako, Aoran Xiao, Naoto Yokoya, Mennatullah Siam
Abstract: Multimodal Large Language Models (MLLMs) have achieved remarkable success in vision-language tasks but their remote sensing (RS) counterpart are relatively under explored. Unlike natural images, RS imagery presents unique challenges that current MLLMs struggle to handle, particularly in visual grounding and spatial reasoning. This study investigates the limitations of CLIP-based MLLMs in RS, highlighting their failure to differentiate visually distinct yet semantically similar RS images. To address this, we introduce a remote sensing multimodal visual patterns (RSMMVP) benchmark. It is designed to evaluate MLLMs in RS tasks by identifying the CLIP-blind pairs, where CLIP-based models incorrectly assign high similarity scores to visually distinct RS images. Through a visual question answering (VQA) evaluation, we analyze the performance of state-of-the-art MLLMs, revealing significant limitations in RS specific representation learning. The results provide valuable insights into the weaknesses of CLIP-based visual encoding and offer a foundation for future research to develop more effective MLLMs tailored for remote sensing applications.
Authors: Haotian Ma, Lin Gu, Siyi Wu, Yingying Zhu
Abstract: 3D point cloud has been widely used in applications such as self-driving cars, robotics, CAD models, etc. To the best of our knowledge, these applications raised the issue of privacy leakage in 3D point clouds, which has not been studied well. Different from the 2D image privacy, which is related to texture and 2D geometric structure, the 3D point cloud is texture-less and only relevant to 3D geometric structure. In this work, we defined the 3D point cloud privacy problem and proposed an efficient privacy-preserving framework named PointFlowGMM that can support downstream classification and segmentation tasks without seeing the original data. Using a flow-based generative model, the point cloud is projected into a latent Gaussian mixture distributed subspace. We further designed a novel angular similarity loss to obfuscate the original geometric structure and reduce the model size from 767MB to 120MB without a decrease in recognition performance. The projected point cloud in the latent space is orthogonally rotated randomly to further protect the original geometric structure, the class-to-class relationship is preserved after rotation, thus, the protected point cloud can support the recognition task. We evaluated our model on multiple datasets and achieved comparable recognition results on encrypted point clouds compared to the original point clouds.
Authors: Zihao Zhang, Haoran Chen, Haoyu Zhao, Guansong Lu, Yanwei Fu, Hang Xu, Zuxuan Wu
Abstract: Handling complex or nonlinear motion patterns has long posed challenges for video frame interpolation. Although recent advances in diffusion-based methods offer improvements over traditional optical flow-based approaches, they still struggle to generate sharp, temporally consistent frames in scenarios with large motion. To address this limitation, we introduce EDEN, an Enhanced Diffusion for high-quality large-motion vidEo frame iNterpolation. Our approach first utilizes a transformer-based tokenizer to produce refined latent representations of the intermediate frames for diffusion models. We then enhance the diffusion transformer with temporal attention across the process and incorporate a start-end frame difference embedding to guide the generation of dynamic motion. Extensive experiments demonstrate that EDEN achieves state-of-the-art results across popular benchmarks, including nearly a 10% LPIPS reduction on DAVIS and SNU-FILM, and an 8% improvement on DAIN-HD.
Authors: Yiren Lu, Yunlai Zhou, Disheng Liu, Tuo Liang, Yu Yin
Abstract: 3D Gaussian Splatting (3DGS) has shown remarkable potential for static scene reconstruction, and recent advancements have extended its application to dynamic scenes. However, the quality of reconstructions depends heavily on high-quality input images and precise camera poses, which are not that trivial to fulfill in real-world scenarios. Capturing dynamic scenes with handheld monocular cameras, for instance, typically involves simultaneous movement of both the camera and objects within a single exposure. This combined motion frequently results in image blur that existing methods cannot adequately handle. To address these challenges, we introduce BARD-GS, a novel approach for robust dynamic scene reconstruction that effectively handles blurry inputs and imprecise camera poses. Our method comprises two main components: 1) camera motion deblurring and 2) object motion deblurring. By explicitly decomposing motion blur into camera motion blur and object motion blur and modeling them separately, we achieve significantly improved rendering results in dynamic regions. In addition, we collect a real-world motion blur dataset of dynamic scenes to evaluate our approach. Extensive experiments demonstrate that BARD-GS effectively reconstructs high-quality dynamic scenes under realistic conditions, significantly outperforming existing methods.
Authors: Xuanming Cui, Jaiminkumar Ashokbhai Bhoi, Chionh Wei Peng, Adriel Kuek, Ser Nam Lim
Abstract: Dynamic Scene Graph Generation (DSGG) for videos is a challenging task in computer vision. While existing approaches often focus on sophisticated architectural design and solely use recall during evaluation, we take a closer look at their predicted scene graphs and discover three critical issues with existing DSGG methods: severe precision-recall trade-off, lack of awareness on triplet importance, and inappropriate evaluation protocols. On the other hand, recent advances of Large Multimodal Models (LMMs) have shown great capabilities in video understanding, yet they have not been tested on fine-grained, frame-wise understanding tasks like DSGG. In this work, we conduct the first systematic analysis of Video LMMs for performing DSGG. Without relying on sophisticated architectural design, we show that LMMs with simple decoder-only structure can be turned into State-of-the-Art scene graph generators that effectively overcome the aforementioned issues, while requiring little finetuning (5-10% training data).
Authors: Zhou Zhenglin, Ma Fan, Fan Hehe, Chua Tat-Seng
Abstract: Animatable head avatar generation typically requires extensive data for training. To reduce the data requirements, a natural solution is to leverage existing data-free static avatar generation methods, such as pre-trained diffusion models with score distillation sampling (SDS), which align avatars with pseudo ground-truth outputs from the diffusion model. However, directly distilling 4D avatars from video diffusion often leads to over-smooth results due to spatial and temporal inconsistencies in the generated video. To address this issue, we propose Zero-1-to-A, a robust method that synthesizes a spatial and temporal consistency dataset for 4D avatar reconstruction using the video diffusion model. Specifically, Zero-1-to-A iteratively constructs video datasets and optimizes animatable avatars in a progressive manner, ensuring that avatar quality increases smoothly and consistently throughout the learning process. This progressive learning involves two stages: (1) Spatial Consistency Learning fixes expressions and learns from front-to-side views, and (2) Temporal Consistency Learning fixes views and learns from relaxed to exaggerated expressions, generating 4D avatars in a simple-to-complex manner. Extensive experiments demonstrate that Zero-1-to-A improves fidelity, animation quality, and rendering speed compared to existing diffusion-based methods, providing a solution for lifelike avatar creation. Code is publicly available at: https://github.com/ZhenglinZhou/Zero-1-to-A.
Authors: Hyojun Go, Byeongjun Park, Hyelin Nam, Byung-Hoon Kim, Hyungjin Chung, Changick Kim
Abstract: We propose VideoRFSplat, a direct text-to-3D model leveraging a video generation model to generate realistic 3D Gaussian Splatting (3DGS) for unbounded real-world scenes. To generate diverse camera poses and unbounded spatial extent of real-world scenes, while ensuring generalization to arbitrary text prompts, previous methods fine-tune 2D generative models to jointly model camera poses and multi-view images. However, these methods suffer from instability when extending 2D generative models to joint modeling due to the modality gap, which necessitates additional models to stabilize training and inference. In this work, we propose an architecture and a sampling strategy to jointly model multi-view images and camera poses when fine-tuning a video generation model. Our core idea is a dual-stream architecture that attaches a dedicated pose generation model alongside a pre-trained video generation model via communication blocks, generating multi-view images and camera poses through separate streams. This design reduces interference between the pose and image modalities. Additionally, we propose an asynchronous sampling strategy that denoises camera poses faster than multi-view images, allowing rapidly denoised poses to condition multi-view generation, reducing mutual ambiguity and enhancing cross-modal consistency. Trained on multiple large-scale real-world datasets (RealEstate10K, MVImgNet, DL3DV-10K, ACID), VideoRFSplat outperforms existing text-to-3D direct generation methods that heavily depend on post-hoc refinement via score distillation sampling, achieving superior results without such refinement.
Authors: Rohit Kundu, Athula Balachandran, Amit K. Roy-Chowdhury
Abstract: Detecting DeepFakes has become a crucial research area as the widespread use of AI image generators enables the effortless creation of face-manipulated and fully synthetic content, yet existing methods are often limited to binary classification (real vs. fake) and lack interpretability. To address these challenges, we propose TruthLens, a novel and highly generalizable framework for DeepFake detection that not only determines whether an image is real or fake but also provides detailed textual reasoning for its predictions. Unlike traditional methods, TruthLens effectively handles both face-manipulated DeepFakes and fully AI-generated content while addressing fine-grained queries such as "Does the eyes/nose/mouth look real or fake?" The architecture of TruthLens combines the global contextual understanding of multimodal large language models like PaliGemma2 with the localized feature extraction capabilities of vision-only models like DINOv2. This hybrid design leverages the complementary strengths of both models, enabling robust detection of subtle manipulations while maintaining interpretability. Extensive experiments on diverse datasets demonstrate that TruthLens outperforms state-of-the-art methods in detection accuracy (by 2-14%) and explainability, in both in-domain and cross-data settings, generalizing effectively across traditional and emerging manipulation techniques.
Authors: Debabrata Mandal, Soumitri Chattopadhyay, Guansen Tong, Praneeth Chakravarthula
Abstract: Image restoration is essential for enhancing degraded images across computer vision tasks. However, most existing methods address only a single type of degradation (e.g., blur, noise, or haze) at a time, limiting their real-world applicability where multiple degradations often occur simultaneously. In this paper, we propose UniCoRN, a unified image restoration approach capable of handling multiple degradation types simultaneously using a multi-head diffusion model. Specifically, we uncover the potential of low-level visual cues extracted from images in guiding a controllable diffusion model for real-world image restoration and we design a multi-head control network adaptable via a mixture-of-experts strategy. We train our model without any prior assumption of specific degradations, through a smartly designed curriculum learning recipe. Additionally, we also introduce MetaRestore, a metalens imaging benchmark containing images with multiple degradations and artifacts. Extensive evaluations on several challenging datasets, including our benchmark, demonstrate that our method achieves significant performance gains and can robustly restore images with severe degradations. Project page: https://codejaeger.github.io/unicorn-gh
Authors: Kyungho Bae, Jinhyung Kim, Sihaeng Lee, Soonyoung Lee, Gunhee Lee, Jinwoo Choi
Abstract: In this work, we tackle action-scene hallucination in Video Large Language Models (Video-LLMs), where models incorrectly predict actions based on the scene context or scenes based on observed actions. We observe that existing Video-LLMs often suffer from action-scene hallucination due to two main factors. First, existing Video-LLMs intermingle spatial and temporal features by applying an attention operation across all tokens. Second, they use the standard Rotary Position Embedding (RoPE), which causes the text tokens to overemphasize certain types of tokens depending on their sequential orders. To address these issues, we introduce MASH-VLM, Mitigating Action-Scene Hallucination in Video-LLMs through disentangled spatial-temporal representations. Our approach includes two key innovations: (1) DST-attention, a novel attention mechanism that disentangles the spatial and temporal tokens within the LLM by using masked attention to restrict direct interactions between the spatial and temporal tokens; (2) Harmonic-RoPE, which extends the dimensionality of the positional IDs, allowing the spatial and temporal tokens to maintain balanced positions relative to the text tokens. To evaluate the action-scene hallucination in Video-LLMs, we introduce the UNSCENE benchmark with 1,320 videos and 4,078 QA pairs. Extensive experiments demonstrate that MASH-VLM achieves state-of-the-art results on the UNSCENE benchmark, as well as on existing video understanding benchmarks.
Authors: Haiguang Wang, Daqi Liu, Hongwei Xie, Haisong Liu, Enhui Ma, Kaicheng Yu, Limin Wang, Bing Wang
Abstract: In recent years, data-driven techniques have greatly advanced autonomous driving systems, but the need for rare and diverse training data remains a challenge, requiring significant investment in equipment and labor. World models, which predict and generate future environmental states, offer a promising solution by synthesizing annotated video data for training. However, existing methods struggle to generate long, consistent videos without accumulating errors, especially in dynamic scenes. To address this, we propose MiLA, a novel framework for generating high-fidelity, long-duration videos up to one minute. MiLA utilizes a Coarse-to-Re(fine) approach to both stabilize video generation and correct distortion of dynamic objects. Additionally, we introduce a Temporal Progressive Denoising Scheduler and Joint Denoising and Correcting Flow modules to improve the quality of generated videos. Extensive experiments on the nuScenes dataset show that MiLA achieves state-of-the-art performance in video generation quality. For more information, visit the project website: https://github.com/xiaomi-mlab/mila.github.io.
Authors: Tiange Xiang, Kai Li, Chengjiang Long, Christian H\"ane, Peihong Guo, Scott Delp, Ehsan Adeli, Li Fei-Fei
Abstract: Recent advances in text-to-image diffusion models have been driven by the increasing availability of paired 2D data. However, the development of 3D diffusion models has been hindered by the scarcity of high-quality 3D data, resulting in less competitive performance compared to their 2D counterparts. To address this challenge, we propose repurposing pre-trained 2D diffusion models for 3D object generation. We introduce Gaussian Atlas, a novel representation that utilizes dense 2D grids, enabling the fine-tuning of 2D diffusion models to generate 3D Gaussians. Our approach demonstrates successful transfer learning from a pre-trained 2D diffusion model to a 2D manifold flattened from 3D structures. To support model training, we compile GaussianVerse, a large-scale dataset comprising 205K high-quality 3D Gaussian fittings of various 3D objects. Our experimental results show that text-to-image diffusion models can be effectively adapted for 3D content generation, bridging the gap between 2D and 3D modeling.
Authors: Hui Liu, Wenya Wang, Kecheng Chen, Jie Liu, Yibing Liu, Tiexin Qin, Peisong He, Xinghao Jiang, Haoliang Li
Abstract: In zero-shot image recognition tasks, humans demonstrate remarkable flexibility in classifying unseen categories by composing known simpler concepts. However, existing vision-language models (VLMs), despite achieving significant progress through large-scale natural language supervision, often underperform in real-world applications because of sub-optimal prompt engineering and the inability to adapt effectively to target classes. To address these issues, we propose a Concept-guided Human-like Bayesian Reasoning (CHBR) framework. Grounded in Bayes' theorem, CHBR models the concept used in human image recognition as latent variables and formulates this task by summing across potential concepts, weighted by a prior distribution and a likelihood function. To tackle the intractable computation over an infinite concept space, we introduce an importance sampling algorithm that iteratively prompts large language models (LLMs) to generate discriminative concepts, emphasizing inter-class differences. We further propose three heuristic approaches involving Average Likelihood, Confidence Likelihood, and Test Time Augmentation (TTA) Likelihood, which dynamically refine the combination of concepts based on the test image. Extensive evaluations across fifteen datasets demonstrate that CHBR consistently outperforms existing state-of-the-art zero-shot generalization methods.
Authors: Haochen Wang, Kai Hu, Liangcai Gao
Abstract: Remote work and online courses have become important methods of knowledge dissemination, leading to a large number of document-based instructional videos. Unlike traditional video datasets, these videos mainly feature rich-text images and audio that are densely packed with information closely tied to the visual content, requiring advanced multimodal understanding capabilities. However, this domain remains underexplored due to dataset availability and its inherent complexity. In this paper, we introduce the DocVideoQA task and dataset for the first time, comprising 1454 videos across 23 categories with a total duration of about 828 hours. The dataset is annotated with 154k question-answer pairs generated manually and via GPT, assessing models' comprehension, temporal awareness, and modality integration capabilities. Initially, we establish a baseline using open-source MLLMs. Recognizing the challenges in modality comprehension for document-centric videos, we present DV-LLaMA, a robust video MLLM baseline. Our method enhances unimodal feature extraction with diverse instruction-tuning data and employs contrastive learning to strengthen modality integration. Through fine-tuning, the LLM is equipped with audio-visual capabilities, leading to significant improvements in document-centric video understanding. Extensive testing on the DocVideoQA dataset shows that DV-LLaMA significantly outperforms existing models. We'll release the code and dataset to facilitate future research.
Authors: Haiyang Yu, Siyang Yi, Ke Niu, Minghan Zhuo, Bin Li
Abstract: With the rapid advancement of deep learning, particularly in the field of medical image analysis, an increasing number of Vision-Language Models (VLMs) are being widely applied to solve complex health and biomedical challenges. However, existing research has primarily focused on specific tasks or single modalities, which limits their applicability and generalization across diverse medical scenarios. To address this challenge, we propose UMIT, a unified multi-modal, multi-task VLM designed specifically for medical imaging tasks. UMIT is able to solve various tasks, including visual question answering, disease detection, and medical report generation. In addition, it is applicable to multiple imaging modalities (e.g., X-ray, CT and PET), covering a wide range of applications from basic diagnostics to complex lesion analysis. Moreover, UMIT supports both English and Chinese, expanding its applicability globally and ensuring accessibility to healthcare services in different linguistic contexts. To enhance the model's adaptability and task-handling capability, we design a unique two-stage training strategy and fine-tune UMIT with designed instruction templates. Through extensive empirical evaluation, UMIT outperforms previous methods in five tasks across multiple datasets. The performance of UMIT indicates that it can significantly enhance diagnostic accuracy and workflow efficiency, thus providing effective solutions for medical imaging applications.
Authors: Jiawei Wang, Kai Hu, Qiang Huo
Abstract: Document structure analysis, aka document layout analysis, is crucial for understanding both the physical layout and logical structure of documents, serving information retrieval, document summarization, knowledge extraction, etc. Hierarchical Document Structure Analysis (HDSA) specifically aims to restore the hierarchical structure of documents created using authoring software with hierarchical schemas. Previous research has primarily followed two approaches: one focuses on tackling specific subtasks of HDSA in isolation, such as table detection or reading order prediction, while the other adopts a unified framework that uses multiple branches or modules, each designed to address a distinct task. In this work, we propose a unified relation prediction approach for HDSA, called UniHDSA, which treats various HDSA sub-tasks as relation prediction problems and consolidates relation prediction labels into a unified label space. This allows a single relation prediction module to handle multiple tasks simultaneously, whether at a page-level or document-level structure analysis. To validate the effectiveness of UniHDSA, we develop a multimodal end-to-end system based on Transformer architectures. Extensive experimental results demonstrate that our approach achieves state-of-the-art performance on a hierarchical document structure analysis benchmark, Comp-HRDoc, and competitive results on a large-scale document layout analysis dataset, DocLayNet, effectively illustrating the superiority of our method across all sub-tasks.
Authors: Junho Kim, Gwangtak Bae, Eun Sun Lee, Young Min Kim
Abstract: Understanding scene contexts is crucial for machines to perform tasks and adapt prior knowledge in unseen or noisy 3D environments. As data-driven learning is intractable to comprehensively encapsulate diverse ranges of layouts and open spaces, we propose teaching machines to identify relational commonalities in 3D spaces. Instead of focusing on point-wise or object-wise representations, we introduce 3D scene analogies, which are smooth maps between 3D scene regions that align spatial relationships. Unlike well-studied single instance-level maps, these scene-level maps smoothly link large scene regions, potentially enabling unique applications in trajectory transfer in AR/VR, long demonstration transfer for imitation learning, and context-aware object rearrangement. To find 3D scene analogies, we propose neural contextual scene maps, which extract descriptor fields summarizing semantic and geometric contexts, and holistically align them in a coarse-to-fine manner for map estimation. This approach reduces reliance on individual feature points, making it robust to input noise or shape variations. Experiments demonstrate the effectiveness of our approach in identifying scene analogies and transferring trajectories or object placements in diverse indoor scenes, indicating its potential for robotics and AR/VR applications.
Authors: Boran Wen, Dingbang Huang, Zichen Zhang, Jiahong Zhou, Jianbin Deng, Jingyu Gong, Yulong Chen, Lizhuang Ma, Yong-Lu Li
Abstract: Reconstructing human-object interactions (HOI) from single images is fundamental in computer vision. Existing methods are primarily trained and tested on indoor scenes due to the lack of 3D data, particularly constrained by the object variety, making it challenging to generalize to real-world scenes with a wide range of objects. The limitations of previous 3D HOI datasets were primarily due to the difficulty in acquiring 3D object assets. However, with the development of 3D reconstruction from single images, recently it has become possible to reconstruct various objects from 2D HOI images. We therefore propose a pipeline for annotating fine-grained 3D humans, objects, and their interactions from single images. We annotated 2.5k+ 3D HOI assets from existing 2D HOI datasets and built the first open-vocabulary in-the-wild 3D HOI dataset Open3DHOI, to serve as a future test set. Moreover, we design a novel Gaussian-HOI optimizer, which efficiently reconstructs the spatial interactions between humans and objects while learning the contact regions. Besides the 3D HOI reconstruction, we also propose several new tasks for 3D HOI understanding to pave the way for future work. Data and code will be publicly available at https://wenboran2002.github.io/3dhoi.
Authors: Jiyuan Wang, Chunyu Lin, Cheng Guan, Lang Nie, Jing He, Haodong Li, Kang Liao, Yao Zhao
Abstract: In this paper, we propose Jasmine, the first Stable Diffusion (SD)-based self-supervised framework for monocular depth estimation, which effectively harnesses SD's visual priors to enhance the sharpness and generalization of unsupervised prediction. Previous SD-based methods are all supervised since adapting diffusion models for dense prediction requires high-precision supervision. In contrast, self-supervised reprojection suffers from inherent challenges (e.g., occlusions, texture-less regions, illumination variance), and the predictions exhibit blurs and artifacts that severely compromise SD's latent priors. To resolve this, we construct a novel surrogate task of hybrid image reconstruction. Without any additional supervision, it preserves the detail priors of SD models by reconstructing the images themselves while preventing depth estimation from degradation. Furthermore, to address the inherent misalignment between SD's scale and shift invariant estimation and self-supervised scale-invariant depth estimation, we build the Scale-Shift GRU. It not only bridges this distribution gap but also isolates the fine-grained texture of SD output against the interference of reprojection loss. Extensive experiments demonstrate that Jasmine achieves SoTA performance on the KITTI benchmark and exhibits superior zero-shot generalization across multiple datasets.
Authors: Jiatong Xia, Libo Sun, Lingqiao Liu
Abstract: Recent methods, such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS), have demonstrated remarkable capabilities in novel view synthesis. However, despite their success in producing high-quality images for viewpoints similar to those seen during training, they struggle when generating detailed images from viewpoints that significantly deviate from the training set, particularly in close-up views. The primary challenge stems from the lack of specific training data for close-up views, leading to the inability of current methods to render these views accurately. To address this issue, we introduce a novel pseudo-label-based learning strategy. This approach leverages pseudo-labels derived from existing training data to provide targeted supervision across a wide range of close-up viewpoints. Recognizing the absence of benchmarks for this specific challenge, we also present a new dataset designed to assess the effectiveness of both current and future methods in this area. Our extensive experiments demonstrate the efficacy of our approach.
Authors: Junsung Park, Hwijeong Lee, Inha Kang, Hyunjung Shim
Abstract: Existing domain generalization methods for LiDAR semantic segmentation under adverse weather struggle to accurately predict "things" categories compared to "stuff" categories. In typical driving scenes, "things" categories can be dynamic and associated with higher collision risks, making them crucial for safe navigation and planning. Recognizing the importance of "things" categories, we identify their performance drop as a serious bottleneck in existing approaches. We observed that adverse weather induces degradation of semantic-level features and both corruption of local features, leading to a misprediction of "things" as "stuff". To mitigate these corruptions, we suggest our method, NTN - segmeNt Things for No-accident. To address semantic-level feature corruption, we bind each point feature to its superclass, preventing the misprediction of things classes into visually dissimilar categories. Additionally, to enhance robustness against local corruption caused by adverse weather, we define each LiDAR beam as a local region and propose a regularization term that aligns the clean data with its corrupted counterpart in feature space. NTN achieves state-of-the-art performance with a +2.6 mIoU gain on the SemanticKITTI-to-SemanticSTF benchmark and +7.9 mIoU on the SemanticPOSS-to-SemanticSTF benchmark. Notably, NTN achieves a +4.8 and +7.9 mIoU improvement on "things" classes, respectively, highlighting its effectiveness.
Authors: Jiayi He, Xu Wang, Ruobei Zhang, Shengeng Tang, Yaxiong Wang, Lechao Cheng
Abstract: We introduce the hfut-lmc team's solution to the SLRTP Sign Production Challenge. The challenge aims to generate semantically aligned sign language pose sequences from text inputs. To this end, we propose a Text-driven Diffusion Model (TDM) framework. During the training phase, TDM utilizes an encoder to encode text sequences and incorporates them into the diffusion model as conditional input to generate sign pose sequences. To guarantee the high quality and accuracy of the generated pose sequences, we utilize two key loss functions. The joint loss function L_{joint} is used to precisely measure and minimize the differences between the joint positions of the generated pose sequences and those of the ground truth. Similarly, the bone orientation loss function L_{bone} is instrumental in ensuring that the orientation of the bones in the generated poses aligns with the actual, correct orientations. In the inference stage, the TDM framework takes on a different yet equally important task. It starts with noisy sequences and, under the strict constraints of the text conditions, gradually refines and generates semantically consistent sign language pose sequences. Our carefully designed framework performs well on the sign language production task, and our solution achieves a BLEU-1 score of 20.17, placing second in the challenge.
Authors: Beilei Cui, Long Bai, Mobarakol Islam, An Wang, Zhiqi Ma, Yiming Huang, Feng Li, Zhen Chen, Zhongliang Jiang, Nassir Navab, Hongliang Ren
Abstract: Accurate 3D scene reconstruction is essential for numerous medical tasks. Given the challenges in obtaining ground truth data, there has been an increasing focus on self-supervised learning (SSL) for endoscopic depth estimation as a basis for scene reconstruction. While foundation models have shown remarkable progress in visual tasks, their direct application to the medical domain often leads to suboptimal results. However, the visual features from these models can still enhance endoscopic tasks, emphasizing the need for efficient adaptation strategies, which still lack exploration currently. In this paper, we introduce Endo3DAC, a unified framework for endoscopic scene reconstruction that efficiently adapts foundation models. We design an integrated network capable of simultaneously estimating depth maps, relative poses, and camera intrinsic parameters. By freezing the backbone foundation model and training only the specially designed Gated Dynamic Vector-Based Low-Rank Adaptation (GDV-LoRA) with separate decoder heads, Endo3DAC achieves superior depth and pose estimation while maintaining training efficiency. Additionally, we propose a 3D scene reconstruction pipeline that optimizes depth maps' scales, shifts, and a few parameters based on our integrated network. Extensive experiments across four endoscopic datasets demonstrate that Endo3DAC significantly outperforms other state-of-the-art methods while requiring fewer trainable parameters. To our knowledge, we are the first to utilize a single network that only requires surgical videos to perform both SSL depth estimation and scene reconstruction tasks. The code will be released upon acceptance.
Authors: Hui Zhang, Tingwei Gao, Jie Shao, Zuxuan Wu
Abstract: Diffusion models have demonstrated impressive generation capabilities, particularly with recent advancements leveraging transformer architectures to improve both visual and artistic quality. However, Diffusion Transformers (DiTs) continue to encounter challenges related to low inference speed, primarily due to the iterative denoising process. To address this issue, we propose BlockDance, a training-free approach that explores feature similarities at adjacent time steps to accelerate DiTs. Unlike previous feature-reuse methods that lack tailored reuse strategies for features at different scales, BlockDance prioritizes the identification of the most structurally similar features, referred to as Structurally Similar Spatio-Temporal (STSS) features. These features are primarily located within the structure-focused blocks of the transformer during the later stages of denoising. BlockDance caches and reuses these highly similar features to mitigate redundant computation, thereby accelerating DiTs while maximizing consistency with the generated results of the original model. Furthermore, considering the diversity of generated content and the varying distributions of redundant features, we introduce BlockDance-Ada, a lightweight decision-making network tailored for instance-specific acceleration. BlockDance-Ada dynamically allocates resources and provides superior content quality. Both BlockDance and BlockDance-Ada have proven effective across various generation tasks and models, achieving accelerations between 25% and 50% while maintaining generation quality.
Authors: Sidi Yang, Binxiao Huang, Yulun Zhang, Dahai Yu, Yujiu Yang, Ngai Wong
Abstract: While deep neural networks have revolutionized image denoising capabilities, their deployment on edge devices remains challenging due to substantial computational and memory requirements. To this end, we present DnLUT, an ultra-efficient lookup table-based framework that achieves high-quality color image denoising with minimal resource consumption. Our key innovation lies in two complementary components: a Pairwise Channel Mixer (PCM) that effectively captures inter-channel correlations and spatial dependencies in parallel, and a novel L-shaped convolution design that maximizes receptive field coverage while minimizing storage overhead. By converting these components into optimized lookup tables post-training, DnLUT achieves remarkable efficiency - requiring only 500KB storage and 0.1% energy consumption compared to its CNN contestant DnCNN, while delivering 20X faster inference. Extensive experiments demonstrate that DnLUT outperforms all existing LUT-based methods by over 1dB in PSNR, establishing a new state-of-the-art in resource-efficient color image denoising. The project is available at https://github.com/Stephen0808/DnLUT.
Authors: Hongda Liu, Longguang Wang, Ye Zhang, Ziru Yu, Yulan Guo
Abstract: Global effective receptive field plays a crucial role for image style transfer (ST) to obtain high-quality stylized results. However, existing ST backbones (e.g., CNNs and Transformers) suffer huge computational complexity to achieve global receptive fields. Recently, the State Space Model (SSM), especially the improved variant Mamba, has shown great potential for long-range dependency modeling with linear complexity, which offers a approach to resolve the above dilemma. In this paper, we develop a Mamba-based style transfer framework, termed SaMam. Specifically, a mamba encoder is designed to efficiently extract content and style information. In addition, a style-aware mamba decoder is developed to flexibly adapt to various styles. Moreover, to address the problems of local pixel forgetting, channel redundancy and spatial discontinuity of existing SSMs, we introduce both local enhancement and zigzag scan. Qualitative and quantitative results demonstrate that our SaMam outperforms state-of-the-art methods in terms of both accuracy and efficiency.
Authors: Yaxiong Chen, Chuang Du, Chunlei Li, Jingliang Hu, Yilei Shi, Shengwu Xiong, Xiao Xiang Zhu, Lichao Mou
Abstract: Automated radiology report generation aims to expedite the tedious and error-prone reporting process for radiologists. While recent works have made progress, learning to align medical images and textual findings remains challenging due to the relative scarcity of labeled medical data. For example, datasets for this task are much smaller than those used for image captioning in computer vision. In this work, we propose to transfer representations from CLIP, a large-scale pre-trained vision-language model, to better capture cross-modal semantics between images and texts. However, directly applying CLIP is suboptimal due to the domain gap between natural images and radiology. To enable efficient adaptation, we introduce UniCrossAdapter, lightweight adapter modules that are incorporated into CLIP and fine-tuned on the target task while keeping base parameters fixed. The adapters are distributed across modalities and their interaction to enhance vision-language alignment. Experiments on two public datasets demonstrate the effectiveness of our approach, advancing state-of-the-art in radiology report generation. The proposed transfer learning framework provides a means of harnessing semantic knowledge from large-scale pre-trained models to tackle data-scarce medical vision-language tasks. Code is available at https://github.com/chauncey-tow/MRG-CLIP.
Authors: Elisei Rykov, Kseniia Petrushina, Kseniia Titova, Alexander Panchenko, Vasily Konovalov
Abstract: Quantifying the realism of images remains a challenging problem in the field of artificial intelligence. For example, an image of Albert Einstein holding a smartphone violates common-sense because modern smartphone were invented after Einstein's death. We introduce a novel method for assessing image realism using Large Vision-Language Models (LVLMs) and Natural Language Inference (NLI). Our approach is based on the premise that LVLMs may generate hallucinations when confronted with images that defy common sense. Using LVLM to extract atomic facts from these images, we obtain a mix of accurate facts and erroneous hallucinations. We proceed by calculating pairwise entailment scores among these facts, subsequently aggregating these values to yield a singular reality score. This process serves to identify contradictions between genuine facts and hallucinatory elements, signaling the presence of images that violate common sense. Our approach has achieved a new state-of-the-art performance in zero-shot mode on the WHOOPS! dataset.
Authors: Yaxiong Chen, Minghong Wei, Zixuan Zheng, Jingliang Hu, Yilei Shi, Shengwu Xiong, Xiao Xiang Zhu, Lichao Mou
Abstract: Referring medical image segmentation targets delineating lesions indicated by textual descriptions. Aligning visual and textual cues is challenging due to their distinct data properties. Inspired by large-scale pre-trained vision-language models, we propose CausalCLIPSeg, an end-to-end framework for referring medical image segmentation that leverages CLIP. Despite not being trained on medical data, we enforce CLIP's rich semantic space onto the medical domain by a tailored cross-modal decoding method to achieve text-to-pixel alignment. Furthermore, to mitigate confounding bias that may cause the model to learn spurious correlations instead of meaningful causal relationships, CausalCLIPSeg introduces a causal intervention module which self-annotates confounders and excavates causal features from inputs for segmentation judgments. We also devise an adversarial min-max game to optimize causal features while penalizing confounding ones. Extensive experiments demonstrate the state-of-the-art performance of our proposed method. Code is available at https://github.com/WUTCM-Lab/CausalCLIPSeg.
Authors: Clive Tinashe Marimo, Benedikt Blumenstiel, Maximilian Nitsche, Johannes Jakubik, Thomas Brunschwiler
Abstract: Vision-language models for Earth observation (EO) typically rely on the visual spectrum of data as the only model input, thus failing to leverage the rich spectral information available in the multispectral channels recorded by satellites. Therefore, in this paper, we introduce Llama3-MS-CLIP, the first vision-language model pre-trained with contrastive learning on a large-scale multispectral dataset and report on the performance gains due to the extended spectral range. Furthermore, we present the largest-to-date image-caption dataset for multispectral data, consisting of one million Sentinel-2 samples and corresponding textual descriptions generated with Llama3-LLaVA-Next and Overture Maps data. We develop a scalable captioning pipeline, which is validated by domain experts. We evaluate Llama3-MS-CLIP on multispectral zero-shot image classification and retrieval using three datasets of varying complexity. Our results demonstrate that Llama3-MS-CLIP significantly outperforms other RGB-based approaches, improving classification accuracy by 6.77% on average and retrieval performance by 4.63% mAP compared to the second-best model. Our results emphasize the relevance of multispectral vision-language learning. We release the image-caption dataset, code, and model weights under an open-source license.
Authors: JunGyu Lee, Kunyoung Lee, Haesol Park, Ig-Jae Kim, Gi Pyo Nam
Abstract: Facial Expression Recognition (FER) plays a crucial role in human affective analysis and has been widely applied in computer vision tasks such as human-computer interaction and psychological assessment. The 8th Affective Behavior Analysis in-the-Wild (ABAW) Challenge aims to assess human emotions using the video-based Aff-Wild2 dataset. This challenge includes various tasks, including the video-based EXPR recognition track, which is our primary focus. In this paper, we demonstrate that addressing label ambiguity and class imbalance, which are known to cause performance degradation, can lead to meaningful performance improvements. Specifically, we propose Video-based Noise-aware Adaptive Weighting (V-NAW), which adaptively assigns importance to each frame in a clip to address label ambiguity and effectively capture temporal variations in facial expressions. Furthermore, we introduce a simple and effective augmentation strategy to reduce redundancy between consecutive frames, which is a primary cause of overfitting. Through extensive experiments, we validate the effectiveness of our approach, demonstrating significant improvements in video-based FER performance.
Authors: Zichen Liu, Kunlun Xu, Bing Su, Xu Zou, Yuxin Peng, Jiahuan Zhou
Abstract: Pre-trained on tremendous image-text pairs, vision-language models like CLIP have demonstrated promising zero-shot generalization across numerous image-based tasks. However, extending these capabilities to video tasks remains challenging due to limited labeled video data and high training costs. Recent video prompting methods attempt to adapt CLIP for video tasks by introducing learnable prompts, but they typically rely on a single static prompt for all video sequences, overlooking the diverse temporal dynamics and spatial variations that exist across frames. This limitation significantly hinders the model's ability to capture essential temporal information for effective video understanding. To address this, we propose an integrated Spatial-TempOral dynamic Prompting (STOP) model which consists of two complementary modules, the intra-frame spatial prompting and inter-frame temporal prompting. Our intra-frame spatial prompts are designed to adaptively highlight discriminative regions within each frame by leveraging intra-frame attention and temporal variation, allowing the model to focus on areas with substantial temporal dynamics and capture fine-grained spatial details. Additionally, to highlight the varying importance of frames for video understanding, we further introduce inter-frame temporal prompts, dynamically inserting prompts between frames with high temporal variance as measured by frame similarity. This enables the model to prioritize key frames and enhances its capacity to understand temporal dependencies across sequences. Extensive experiments on various video benchmarks demonstrate that STOP consistently achieves superior performance against state-of-the-art methods. The code is available at https://github.com/zhoujiahuan1991/CVPR2025-STOP.
Authors: Kendong Liu, Zhiyu Zhu, Hui Liu, Junhui Hou
Abstract: We present Acc3D to tackle the challenge of accelerating the diffusion process to generate 3D models from single images. To derive high-quality reconstructions through few-step inferences, we emphasize the critical issue of regularizing the learning of score function in states of random noise. To this end, we propose edge consistency, i.e., consistent predictions across the high signal-to-noise ratio region, to enhance a pre-trained diffusion model, enabling a distillation-based refinement of the endpoint score function. Building on those distilled diffusion models, we propose an adversarial augmentation strategy to further enrich the generation detail and boost overall generation quality. The two modules complement each other, mutually reinforcing to elevate generative performance. Extensive experiments demonstrate that our Acc3D not only achieves over a $20\times$ increase in computational efficiency but also yields notable quality improvements, compared to the state-of-the-arts.
Authors: Pengyu Liu, Guohua Dong, Dan Guo, Kun Li, Fengling Li, Xun Yang, Meng Wang, Xiaomin Ying
Abstract: In daily life, we encounter diverse external stimuli, such as images, sounds, and videos. As research in multimodal stimuli and neuroscience advances, fMRI-based brain decoding has become a key tool for understanding brain perception and its complex cognitive processes. Decoding brain signals to reconstruct stimuli not only reveals intricate neural mechanisms but also drives progress in AI, disease treatment, and brain-computer interfaces. Recent advancements in neuroimaging and image generation models have significantly improved fMRI-based decoding. While fMRI offers high spatial resolution for precise brain activity mapping, its low temporal resolution and signal noise pose challenges. Meanwhile, techniques like GANs, VAEs, and Diffusion Models have enhanced reconstructed image quality, and multimodal pre-trained models have boosted cross-modal decoding tasks. This survey systematically reviews recent progress in fMRI-based brain decoding, focusing on stimulus reconstruction from passive brain signals. It summarizes datasets, relevant brain regions, and categorizes existing methods by model structure. Additionally, it evaluates model performance and discusses their effectiveness. Finally, it identifies key challenges and proposes future research directions, offering valuable insights for the field. For more information and resources related to this survey, visit https://github.com/LpyNow/BrainDecodingImage.
Authors: Suraj Singh, Anastasia Batsheva, Oleg Y. Rogov, Ahmed Bouridane
Abstract: Contemporary image restoration and super-resolution techniques effectively harness deep neural networks, markedly outperforming traditional methods. However, astrophotography presents unique challenges for deep learning due to limited training data. This work explores hybrid strategies, such as the Deep Image Prior (DIP) model, which facilitates blind training but is susceptible to overfitting, artifact generation, and instability when handling noisy images. We propose enhancements to the DIP model's baseline performance through several advanced techniques. First, we refine the model to process multiple frames concurrently, employing the Back Projection method and the TVNet model. Next, we adopt a Markov approach incorporating Monte Carlo estimation, Langevin dynamics, and a variational input technique to achieve unbiased estimates with minimal variance and counteract overfitting effectively. Collectively, these modifications reduce the likelihood of noise learning and mitigate loss function fluctuations during training, enhancing result stability. We validated our algorithm across multiple image sets of astronomical and celestial objects, achieving performance that not only mitigates limitations of Lucky Imaging, a classical computer vision technique that remains a standard in astronomical image reconstruction but surpasses the original DIP model, state of the art transformer- and diffusion-based models, underscoring the significance of our improvements.
Authors: P. Schulz, T. Hempel, A. Al-Hamadi
Abstract: 3D detection is a critical task to understand spatial characteristics of the environment and is used in a variety of applications including robotics, augmented reality, and image retrieval. Training performant detection models require diverse, precisely annotated, and large scale datasets that involve complex and expensive creation processes. Hence, there are only few public 3D datasets that are additionally limited in their range of classes. In this work, we propose a pipeline for automatic generation of 3D datasets for arbitrary objects. By utilizing the universal 3D representation and rendering capabilities of Radiance Fields, our pipeline generates high quality 3D models for arbitrary objects. These 3D models serve as input for a synthetic dataset generator. Our pipeline is fast, easy to use and has a high degree of automation. Our experiments demonstrate, that 3D pose estimation networks, trained with our generated datasets, archive strong performance in typical application scenarios.
Authors: Haojia Gao, Haohua Que, Hoiian Au, Weihao Shan, Mingkai Liu, Yusen Qin, Lei Mu, Rong Zhao, Xinghua Yang, Qi Wei, Fei Qiao
Abstract: This paper proposes SenseExpo, an efficient autonomous exploration framework based on a lightweight prediction network, which addresses the limitations of traditional methods in computational overhead and environmental generalization. By integrating Generative Adversarial Networks (GANs), Transformer, and Fast Fourier Convolution (FFC), we designed a lightweight prediction model with merely 709k parameters. Our smallest model achieves better performance on the KTH dataset than U-net (24.5M) and LaMa (51M), delivering PSNR 9.026 and SSIM 0.718, particularly representing a 38.7% PSNR improvement over the 51M-parameter LaMa model. Cross-domain testing demonstrates its strong generalization capability, with an FID score of 161.55 on the HouseExpo dataset, significantly outperforming comparable methods. Regarding exploration efficiency, on the KTH dataset,SenseExpo demonstrates approximately a 67.9% time reduction in exploration time compared to MapEx. On the MRPB 1.0 dataset, SenseExpo achieves 77.1% time reduction roughly compared to MapEx. Deployed as a plug-and-play ROS node, the framework seamlessly integrates with existing navigation systems, providing an efficient solution for resource-constrained devices.
Authors: Stijn Groenen, Marzieh Hassanshahi Varposhti, Mahyar Shahsavari
Abstract: This work introduces GazeSCRNN, a novel spiking convolutional recurrent neural network designed for event-based near-eye gaze tracking. Leveraging the high temporal resolution, energy efficiency, and compatibility of Dynamic Vision Sensor (DVS) cameras with event-based systems, GazeSCRNN uses a spiking neural network (SNN) to address the limitations of traditional gaze-tracking systems in capturing dynamic movements. The proposed model processes event streams from DVS cameras using Adaptive Leaky-Integrate-and-Fire (ALIF) neurons and a hybrid architecture optimized for spatio-temporal data. Extensive evaluations on the EV-Eye dataset demonstrate the model's accuracy in predicting gaze vectors. In addition, we conducted ablation studies to reveal the importance of the ALIF neurons, dynamic event framing, and training techniques, such as Forward-Propagation-Through-Time, in enhancing overall system performance. The most accurate model achieved a Mean Angle Error (MAE) of 6.034{\deg} and a Mean Pupil Error (MPE) of 2.094 mm. Consequently, this work is pioneering in demonstrating the feasibility of using SNNs for event-based gaze tracking, while shedding light on critical challenges and opportunities for further improvement.
Authors: Yair Shpitzer, Gal Chechik, Idan Schwartz
Abstract: Personalizing image generation and editing is particularly challenging when we only have a few images of the subject, or even a single image. A common approach to personalization is concept learning, which can integrate the subject into existing models relatively quickly, but produces images whose quality tends to deteriorate quickly when the number of subject images is small. Quality can be improved by pre-training an encoder, but training restricts generation to the training distribution, and is time consuming. It is still an open hard challenge to personalize image generation and editing from a single image without training. Here, we present SISO, a novel, training-free approach based on optimizing a similarity score with an input subject image. More specifically, SISO iteratively generates images and optimizes the model based on loss of similarity with the given subject image until a satisfactory level of similarity is achieved, allowing plug-and-play optimization to any image generator. We evaluated SISO in two tasks, image editing and image generation, using a diverse data set of personal subjects, and demonstrate significant improvements over existing methods in image quality, subject fidelity, and background preservation.
Authors: Sunqi Fan, Meng-Hao Guo, Shuojin Yang
Abstract: Video question answering (VideoQA) enables machines to extract and comprehend key information from videos through natural language interaction, which is a critical step towards achieving intelligence. However, the demand for a thorough understanding of videos and high computational costs still limit the widespread applications of VideoQA. To address it, we propose Agentic Keyframe Search (AKeyS), a simple yet powerful algorithm for identifying keyframes in the VideoQA task. It can effectively distinguish key information from redundant, irrelevant content by leveraging modern language agents to direct classical search algorithms. Specifically, we first segment the video and organize it as a tree structure. Then, AKeyS uses a language agent to estimate heuristics and movement costs while dynamically expanding nodes. Finally, the agent determines if sufficient keyframes have been collected based on termination conditions and provides answers. Extensive experiments on the EgoSchema and NExT-QA datasets show that AKeyS outperforms all previous methods with the highest keyframe searching efficiency, which means it can accurately identify key information and conduct effective visual reasoning with minimal computational overhead. For example, on the EgoSchema subset, it achieves 1.8% higher accuracy while processing only 43.5% of the frames compared to VideoTree. We believe that AKeyS represents a significant step towards building intelligent agents for video understanding. The code is publicly available at https://github.com/fansunqi/AKeyS.
Authors: Zhihang Liu, Chen-Wei Xie, Pandeng Li, Liming Zhao, Longxiang Tang, Yun Zheng, Chuanbin Liu, Hongtao Xie
Abstract: Recent Multi-modal Large Language Models (MLLMs) have been challenged by the computational overhead resulting from massive video frames, often alleviated through compression strategies. However, the visual content is not equally contributed to user instructions, existing strategies (\eg, average pool) inevitably lead to the loss of potentially useful information. To tackle this, we propose the Hybrid-level Instruction Injection Strategy for Conditional Token Compression in MLLMs (HICom), utilizing the instruction as a condition to guide the compression from both local and global levels. This encourages the compression to retain the maximum amount of user-focused information while reducing visual tokens to minimize computational burden. Specifically, the instruction condition is injected into the grouped visual tokens at the local level and the learnable tokens at the global level, and we conduct the attention mechanism to complete the conditional compression. From the hybrid-level compression, the instruction-relevant visual parts are highlighted while the temporal-spatial structure is also preserved for easier understanding of LLMs. To further unleash the potential of HICom, we introduce a new conditional pre-training stage with our proposed dataset HICom-248K. Experiments show that our HICom can obtain distinguished video understanding ability with fewer tokens, increasing the performance by 2.43\% average on three multiple-choice QA benchmarks and saving 78.8\% tokens compared with the SOTA method. The code is available at https://github.com/lntzm/HICom.
Authors: Andrei Jelea, Ahmed Nabil Belbachir, Marius Leordeanu
Abstract: Solving fish segmentation in underwater videos, a real-world problem of great practical value in marine and aquaculture industry, is a challenging task due to the difficulty of the filming environment, poor visibility and limited existing annotated underwater fish data. In order to overcome these obstacles, we introduce a novel two stage unsupervised segmentation approach that requires no human annotations and combines artificially created and real images. Our method generates challenging synthetic training data, by placing virtual fish in real-world underwater habitats, after performing fish transformations such as Thin Plate Spline shape warping and color Histogram Matching, which realistically integrate synthetic fish into the backgrounds, making the generated images increasingly closer to the real world data with every stage of our approach. While we validate our unsupervised method on the popular DeepFish dataset, obtaining a performance close to a fully-supervised SoTA model, we further show its effectiveness on the specific case of salmon segmentation in underwater videos, for which we introduce DeepSalmon, the largest dataset of its kind in the literature (30 GB). Moreover, on both datasets we prove the capability of our approach to boost the performance of the fully-supervised SoTA model.
Authors: Wanshu Fan, Yue Wang, Cong Wang, Yunzhe Zhang, Wei Wang, Dongsheng Zhou
Abstract: Single-Image Super-Resolution (SISR) plays a pivotal role in enhancing the accuracy and reliability of measurement systems, which are integral to various vision-based instrumentation and measurement applications. These systems often require clear and detailed images for precise object detection and recognition. However, images captured by visual measurement tools frequently suffer from degradation, including blurring and loss of detail, which can impede measurement accuracy.As a potential remedy, we in this paper propose a Semantic-Guided Global-Local Collaborative Network (SGGLC-Net) for lightweight SISR. Our SGGLC-Net leverages semantic priors extracted from a pre-trained model to guide the super-resolution process, enhancing image detail quality effectively. Specifically,we propose a Semantic Guidance Module that seamlessly integrates the semantic priors into the super-resolution network, enabling the network to more adeptly capture and utilize semantic priors, thereby enhancing image details. To further explore both local and non-local interactions for improved detail rendition,we propose a Global-Local Collaborative Module, which features three Global and Local Detail Enhancement Modules, as well as a Hybrid Attention Mechanism to work together to efficiently learn more useful features. Our extensive experiments show that SGGLC-Net achieves competitive PSNR and SSIM values across multiple benchmark datasets, demonstrating higher performance with the multi-adds reduction of 12.81G compared to state-of-the-art lightweight super-resolution approaches. These improvements underscore the potential of our approach to enhance the precision and effectiveness of visual measurement systems. Codes are at https://github.com/fanamber831/SGGLC-Net.
Authors: Yike Yuan, Ziyu Wang, Zihao Huang, Defa Zhu, Xun Zhou, Jingyi Yu, Qiyang Min
Abstract: Diffusion models have emerged as mainstream framework in visual generation. Building upon this success, the integration of Mixture of Experts (MoE) methods has shown promise in enhancing model scalability and performance. In this paper, we introduce Race-DiT, a novel MoE model for diffusion transformers with a flexible routing strategy, Expert Race. By allowing tokens and experts to compete together and select the top candidates, the model learns to dynamically assign experts to critical tokens. Additionally, we propose per-layer regularization to address challenges in shallow layer learning, and router similarity loss to prevent mode collapse, ensuring better expert utilization. Extensive experiments on ImageNet validate the effectiveness of our approach, showcasing significant performance gains while promising scaling properties.
Authors: Xu He, Zhen Huang, Qingsong Yao, Xiaoqian Zhou, S. Kevin Zhou
Abstract: Landmark detection plays a crucial role in medical imaging applications such as disease diagnosis, bone age estimation, and therapy planning. However, training models for detecting multiple landmarks simultaneously often encounters the "seesaw phenomenon", where improvements in detecting certain landmarks lead to declines in detecting others. Yet, training a separate model for each landmark increases memory usage and computational overhead. To address these challenges, we propose a novel approach based on the belief that "landmarks are distinct" by training models with pseudo-labels and template data updated continuously during the training process, where each model is dedicated to detecting a single landmark to achieve high accuracy. Furthermore, grounded on the belief that "landmarks are also alike", we introduce an adapter-based fusion model, combining shared weights with landmark-specific weights, to efficiently share model parameters while allowing flexible adaptation to individual landmarks. This approach not only significantly reduces memory and computational resource requirements but also effectively mitigates the seesaw phenomenon in multi-landmark training. Experimental results on publicly available medical image datasets demonstrate that the single-landmark models significantly outperform traditional multi-point joint training models in detecting individual landmarks. Although our adapter-based fusion model shows slightly lower performance compared to the combined results of all single-landmark models, it still surpasses the current state-of-the-art methods while achieving a notable improvement in resource efficiency.
Authors: Qiang Zou, Shuli Cheng, Jiayi Chen
Abstract: Cross-modal hashing is a promising approach for efficient data retrieval and storage optimization. However, contemporary methods exhibit significant limitations in semantic preservation, contextual integrity, and information redundancy, which constrains retrieval efficacy. We present PromptHash, an innovative framework leveraging affinity prompt-aware collaborative learning for adaptive cross-modal hashing. We propose an end-to-end framework for affinity-prompted collaborative hashing, with the following fundamental technical contributions: (i) a text affinity prompt learning mechanism that preserves contextual information while maintaining parameter efficiency, (ii) an adaptive gated selection fusion architecture that synthesizes State Space Model with Transformer network for precise cross-modal feature integration, and (iii) a prompt affinity alignment strategy that bridges modal heterogeneity through hierarchical contrastive learning. To the best of our knowledge, this study presents the first investigation into affinity prompt awareness within collaborative cross-modal adaptive hash learning, establishing a paradigm for enhanced semantic consistency across modalities. Through comprehensive evaluation on three benchmark multi-label datasets, PromptHash demonstrates substantial performance improvements over existing approaches. Notably, on the NUS-WIDE dataset, our method achieves significant gains of 18.22% and 18.65% in image-to-text and text-to-image retrieval tasks, respectively. The code is publicly available at https://github.com/ShiShuMo/PromptHash.
Authors: Yingmao Miao, Zhanpeng Huang, Rui Han, Zibin Wang, Chenhao Lin, Chao Shen
Abstract: While virtual try-on for clothes and shoes with diffusion models has gained attraction, virtual try-on for ornaments, such as bracelets, rings, earrings, and necklaces, remains largely unexplored. Due to the intricate tiny patterns and repeated geometric sub-structures in most ornaments, it is much more difficult to guarantee identity and appearance consistency under large pose and scale variances between ornaments and models. This paper proposes the task of virtual try-on for ornaments and presents a method to improve the geometric and appearance preservation of ornament virtual try-ons. Specifically, we estimate an accurate wearing mask to improve the alignments between ornaments and models in an iterative scheme alongside the denoising process. To preserve structure details, we further regularize attention layers to map the reference ornament mask to the wearing mask in an implicit way. Experimental results demonstrate that our method successfully wears ornaments from reference images onto target models, handling substantial differences in scale and pose while preserving identity and achieving realistic visual effects.
Authors: Tim Seizinger, Florin-Alexandru Vasluianu, Marcos V. Conde, Radu Timofte
Abstract: Bokeh rendering methods play a key role in creating the visually appealing, softly blurred backgrounds seen in professional photography. While recent learning-based approaches show promising results, generating realistic Bokeh with variable strength remains challenging. Existing methods require additional inputs and suffer from unrealistic Bokeh reproduction due to reliance on synthetic data. In this work, we propose Bokehlicious, a highly efficient network that provides intuitive control over Bokeh strength through an Aperture-Aware Attention mechanism, mimicking the physical lens aperture. To further address the lack of high-quality real-world data, we present RealBokeh, a novel dataset featuring 23,000 high-resolution (24-MP) images captured by professional photographers, covering diverse scenes with varied aperture and focal length settings. Evaluations on both our new RealBokeh and established Bokeh rendering benchmarks show that Bokehlicious consistently outperforms SOTA methods while significantly reducing computational cost and exhibiting strong zero-shot generalization. Our method and dataset further extend to defocus deblurring, achieving competitive results on the RealDOF benchmark. Our code and data can be found at https://github.com/TimSeizinger/Bokehlicious
Authors: Longbin Ji, Lei Zhong, Pengfei Wei, Changjian Li
Abstract: Recent advancements in trajectory-guided video generation have achieved notable progress. However, existing models still face challenges in generating object motions with potentially changing 6D poses under wide-range rotations, due to limited 3D understanding. To address this problem, we introduce PoseTraj, a pose-aware video dragging model for generating 3D-aligned motion from 2D trajectories. Our method adopts a novel two-stage pose-aware pretraining framework, improving 3D understanding across diverse trajectories. Specifically, we propose a large-scale synthetic dataset PoseTraj-10K, containing 10k videos of objects following rotational trajectories, and enhance the model perception of object pose changes by incorporating 3D bounding boxes as intermediate supervision signals. Following this, we fine-tune the trajectory-controlling module on real-world videos, applying an additional camera-disentanglement module to further refine motion accuracy. Experiments on various benchmark datasets demonstrate that our method not only excels in 3D pose-aligned dragging for rotational trajectories but also outperforms existing baselines in trajectory accuracy and video quality.
Authors: Aniek Eijpe, Soufyan Lakbir, Melis Erdal Cesur, Sara P. Oliveira, Sanne Abeln, Wilson Silva
Abstract: To improve the prediction of cancer survival using whole-slide images and transcriptomics data, it is crucial to capture both modality-shared and modality-specific information. However, multimodal frameworks often entangle these representations, limiting interpretability and potentially suppressing discriminative features. To address this, we propose Disentangled and Interpretable Multimodal Attention Fusion (DIMAF), a multimodal framework that separates the intra- and inter-modal interactions within an attention-based fusion mechanism to learn distinct modality-specific and modality-shared representations. We introduce a loss based on Distance Correlation to promote disentanglement between these representations and integrate Shapley additive explanations to assess their relative contributions to survival prediction. We evaluate DIMAF on four public cancer survival datasets, achieving a relative average improvement of 1.85% in performance and 23.7% in disentanglement compared to current state-of-the-art multimodal models. Beyond improved performance, our interpretable framework enables a deeper exploration of the underlying interactions between and within modalities in cancer biology.
Authors: Gabriela Ghimpeteanu, Hayat Rajani, Josep Quintana, Rafael Garcia
Abstract: Ensuring food safety and quality is critical in the food processing industry, where the detection of contaminants remains a persistent challenge. This study presents an automated solution for detecting foreign objects on pork belly meat using hyperspectral imaging (HSI). A hyperspectral camera was used to capture data across various bands in the near-infrared (NIR) spectrum (900-1700 nm), enabling accurate identification of contaminants that are often undetectable through traditional visual inspection methods. The proposed solution combines pre-processing techniques with a segmentation approach based on a lightweight Vision Transformer (ViT) to distinguish contaminants from meat, fat, and conveyor belt materials. The adopted strategy demonstrates high detection accuracy and training efficiency, while also addressing key industrial challenges such as inherent noise, temperature variations, and spectral similarity between contaminants and pork belly. Experimental results validate the effectiveness of hyperspectral imaging in enhancing food safety, highlighting its potential for broad real-time applications in automated quality control processes.
Authors: Lucas Morin, Val\'ery Weber, Ahmed Nassar, Gerhard Ingmar Meijer, Luc Van Gool, Yawei Li, Peter Staar
Abstract: The automated analysis of chemical literature holds promise to accelerate discovery in fields such as material science and drug development. In particular, search capabilities for chemical structures and Markush structures (chemical structure templates) within patent documents are valuable, e.g., for prior-art search. Advancements have been made in the automatic extraction of chemical structures from text and images, yet the Markush structures remain largely unexplored due to their complex multi-modal nature. In this work, we present MarkushGrapher, a multi-modal approach for recognizing Markush structures in documents. Our method jointly encodes text, image, and layout information through a Vision-Text-Layout encoder and an Optical Chemical Structure Recognition vision encoder. These representations are merged and used to auto-regressively generate a sequential graph representation of the Markush structure along with a table defining its variable groups. To overcome the lack of real-world training data, we propose a synthetic data generation pipeline that produces a wide range of realistic Markush structures. Additionally, we present M2S, the first annotated benchmark of real-world Markush structures, to advance research on this challenging task. Extensive experiments demonstrate that our approach outperforms state-of-the-art chemistry-specific and general-purpose vision-language models in most evaluation settings. Code, models, and datasets will be available.
Authors: Mohamad Hassan N C, Divyam Gupta, Mainak Singha, Sai Bhargav Rongali, Ankit Jha, Muhammad Haris Khan, Biplab Banerjee
Abstract: We introduce Low-Shot Open-Set Domain Generalization (LSOSDG), a novel paradigm unifying low-shot learning with open-set domain generalization (ODG). While prompt-based methods using models like CLIP have advanced DG, they falter in low-data regimes (e.g., 1-shot) and lack precision in detecting open-set samples with fine-grained semantics related to training classes. To address these challenges, we propose OSLOPROMPT, an advanced prompt-learning framework for CLIP with two core innovations. First, to manage limited supervision across source domains and improve DG, we introduce a domain-agnostic prompt-learning mechanism that integrates adaptable domain-specific cues and visually guided semantic attributes through a novel cross-attention module, besides being supported by learnable domain- and class-generic visual prompts to enhance cross-modal adaptability. Second, to improve outlier rejection during inference, we classify unfamiliar samples as "unknown" and train specialized prompts with systematically synthesized pseudo-open samples that maintain fine-grained relationships to known classes, generated through a targeted query strategy with off-the-shelf foundation models. This strategy enhances feature learning, enabling our model to detect open samples with varied granularity more effectively. Extensive evaluations across five benchmarks demonstrate that OSLOPROMPT establishes a new state-of-the-art in LSOSDG, significantly outperforming existing methods.
Authors: Jiyong Rao, Brian Nlong Zhao, Yu Wang
Abstract: Multi-species animal pose estimation has emerged as a challenging yet critical task, hindered by substantial visual diversity and uncertainty. This paper challenges the problem by efficient prompt learning for Vision-Language Pretrained (VLP) models, \textit{e.g.} CLIP, aiming to resolve the cross-species generalization problem. At the core of the solution lies in the prompt designing, probabilistic prompt modeling and cross-modal adaptation, thereby enabling prompts to compensate for cross-modal information and effectively overcome large data variances under unbalanced data distribution. To this end, we propose a novel probabilistic prompting approach to fully explore textual descriptions, which could alleviate the diversity issues caused by long-tail property and increase the adaptability of prompts on unseen category instance. Specifically, we first introduce a set of learnable prompts and propose a diversity loss to maintain distinctiveness among prompts, thus representing diverse image attributes. Diverse textual probabilistic representations are sampled and used as the guidance for the pose estimation. Subsequently, we explore three different cross-modal fusion strategies at spatial level to alleviate the adverse impacts of visual uncertainty. Extensive experiments on multi-species animal pose benchmarks show that our method achieves the state-of-the-art performance under both supervised and zero-shot settings. The code is available at https://github.com/Raojiyong/PPAP.
Authors: Jiangyi Wang, Na Zhao
Abstract: Active learning has emerged as a promising approach to reduce the substantial annotation burden in 3D object detection tasks, spurring several initiatives in outdoor environments. However, its application in indoor environments remains unexplored. Compared to outdoor 3D datasets, indoor datasets face significant challenges, including fewer training samples per class, a greater number of classes, more severe class imbalance, and more diverse scene types and intra-class variances. This paper presents the first study on active learning for indoor 3D object detection, where we propose a novel framework tailored for this task. Our method incorporates two key criteria - uncertainty and diversity - to actively select the most ambiguous and informative unlabeled samples for annotation. The uncertainty criterion accounts for both inaccurate detections and undetected objects, ensuring that the most ambiguous samples are prioritized. Meanwhile, the diversity criterion is formulated as a joint optimization problem that maximizes the diversity of both object class distributions and scene types, using a new Class-aware Adaptive Prototype (CAP) bank. The CAP bank dynamically allocates representative prototypes to each class, helping to capture varying intra-class diversity across different categories. We evaluate our method on SUN RGB-D and ScanNetV2, where it outperforms baselines by a significant margin, achieving over 85% of fully-supervised performance with just 10% of the annotation budget.
Authors: Benedykt Pawlus, Bogdan Smolka, Jolanta Kawulok, Michal Kawulok
Abstract: Assessing smile genuineness from video sequences is a vital topic concerned with recognizing facial expression and linking them with the underlying emotional states. There have been a number of techniques proposed underpinned with handcrafted features, as well as those that rely on deep learning to elaborate the useful features. As both of these approaches have certain benefits and limitations, in this work we propose to combine the features learned by a long short-term memory network with the features handcrafted to capture the dynamics of facial action units. The results of our experiments indicate that the proposed solution is more effective than the baseline techniques and it allows for assessing the smile genuineness from video sequences in real-time.
Authors: Shiyang Zhou, Haijin Zeng, Yunfan Lu, Tong Shao, Ke Tang, Yongyong Chen, Jie Liu, Jingyong Su
Abstract: Quad Bayer demosaicing is the central challenge for enabling the widespread application of Hybrid Event-based Vision Sensors (HybridEVS). Although existing learning-based methods that leverage long-range dependency modeling have achieved promising results, their complexity severely limits deployment on mobile devices for real-world applications. To address these limitations, we propose a lightweight Mamba-based binary neural network designed for efficient and high-performing demosaicing of HybridEVS RAW images. First, to effectively capture both global and local dependencies, we introduce a hybrid Binarized Mamba-Transformer architecture that combines the strengths of the Mamba and Swin Transformer architectures. Next, to significantly reduce computational complexity, we propose a binarized Mamba (Bi-Mamba), which binarizes all projections while retaining the core Selective Scan in full precision. Bi-Mamba also incorporates additional global visual information to enhance global context and mitigate precision loss. We conduct quantitative and qualitative experiments to demonstrate the effectiveness of BMTNet in both performance and computational efficiency, providing a lightweight demosaicing solution suited for real-world edge devices. Our codes and models are available at https://github.com/Clausy9/BMTNet.
Authors: Tianyi Wei, Yifan Zhou, Dongdong Chen, Xingang Pan
Abstract: The integration of Rotary Position Embedding (RoPE) in Multimodal Diffusion Transformer (MMDiT) has significantly enhanced text-to-image generation quality. However, the fundamental reliance of self-attention layers on positional embedding versus query-key similarity during generation remains an intriguing question. We present the first mechanistic analysis of RoPE-based MMDiT models (e.g., FLUX), introducing an automated probing strategy that disentangles positional information versus content dependencies by strategically manipulating RoPE during generation. Our analysis reveals distinct dependency patterns that do not straightforwardly correlate with depth, offering new insights into the layer-specific roles in RoPE-based MMDiT. Based on these findings, we propose a training-free, task-specific image editing framework that categorizes editing tasks into three types: position-dependent editing (e.g., object addition), content similarity-dependent editing (e.g., non-rigid editing), and region-preserved editing (e.g., background replacement). For each type, we design tailored key-value injection strategies based on the characteristics of the editing task. Extensive qualitative and quantitative evaluations demonstrate that our method outperforms state-of-the-art approaches, particularly in preserving original semantic content and achieving seamless modifications.
Authors: Xiangyu Li, Wanshu Fan, Yue Shen, Cong Wang, Wei Wang, Xin Yang, Qiang Zhang, Dongsheng Zhou
Abstract: High-fidelity imaging is crucial for the successful safety supervision and intelligent deployment of vision-based measurement systems (VBMS). It ensures high-quality imaging in VBMS, which is fundamental for reliable visual measurement and analysis. However, imaging quality can be significantly impaired by adverse weather conditions, particularly rain, leading to blurred images and reduced contrast. Such impairments increase the risk of inaccurate evaluations and misinterpretations in VBMS. To address these limitations, we propose an Expectation Maximization Reconstruction Transformer (EMResformer) for single image rain streak removal. The EMResformer retains the key self-attention values for feature aggregation, enhancing local features to produce superior image reconstruction. Specifically, we propose an Expectation Maximization Block seamlessly integrated into the single image rain streak removal network, enhancing its ability to eliminate superfluous information and restore a cleaner background image. Additionally, to further enhance local information for improved detail rendition, we introduce a Local Model Residual Block, which integrates two local model blocks along with a sequence of convolutions and activation functions. This integration synergistically facilitates the extraction of more pertinent features for enhanced single image rain streak removal. Extensive experiments validate that our proposed EMResformer surpasses current state-of-the-art single image rain streak removal methods on both synthetic and real-world datasets, achieving an improved balance between model complexity and single image deraining performance. Furthermore, we evaluate the effectiveness of our method in VBMS scenarios, demonstrating that high-quality imaging significantly improves the accuracy and reliability of VBMS tasks.
Authors: Soham Roy, Abhishek Mishra, Shirish Karande, Murari Mandal
Abstract: Modern text-to-image generative models can inadvertently reproduce copyrighted content memorized in their training data, raising serious concerns about potential copyright infringement. We introduce Guardians of Generation, a model agnostic inference time framework for dynamic copyright shielding in AI image generation. Our approach requires no retraining or modification of the generative model weights, instead integrating seamlessly with existing diffusion pipelines. It augments the generation process with an adaptive guidance mechanism comprising three components: a detection module, a prompt rewriting module, and a guidance adjustment module. The detection module monitors user prompts and intermediate generation steps to identify features indicative of copyrighted content before they manifest in the final output. If such content is detected, the prompt rewriting mechanism dynamically transforms the user's prompt by sanitizing or replacing references that could trigger copyrighted material while preserving the prompt's intended semantics. The adaptive guidance module adaptively steers the diffusion process away from flagged content by modulating the model's sampling trajectory. Together, these components form a robust shield that enables a tunable balance between preserving creative fidelity and ensuring copyright compliance. We validate our method on a variety of generative models such as Stable Diffusion, SDXL, and Flux, demonstrating substantial reductions in copyrighted content generation with negligible impact on output fidelity or alignment with user intent. This work provides a practical, plug-and-play safeguard for generative image models, enabling more responsible deployment under real-world copyright constraints. Source code is available at: https://respailab.github.io/gog
Authors: Fatemeh Amerehi, Patrick Healy
Abstract: Efforts to address declining accuracy as a result of data shifts often involve various data-augmentation strategies. Adversarial training is one such method, designed to improve robustness to worst-case distribution shifts caused by adversarial examples. While this method can improve robustness, it may also hinder generalization to clean examples and exacerbate performance imbalances across different classes. This paper explores the impact of adversarial training on both overall and class-specific performance, as well as its spill-over effects. We observe that enhanced labeling during training boosts adversarial robustness by 53.50% and mitigates class imbalances by 5.73%, leading to improved accuracy in both clean and adversarial settings compared to standard adversarial training.
Authors: Andrea Maracani, Savas Ozkan, Sijun Cho, Hyowon Kim, Eunchung Noh, Jeongwon Min, Cho Jung Min, Dookun Park, Mete Ozay
Abstract: Scaling architectures have been proven effective for improving Scene Text Recognition (STR), but the individual contribution of vision encoder and text decoder scaling remain under-explored. In this work, we present an in-depth empirical analysis and demonstrate that, contrary to previous observations, scaling the decoder yields significant performance gains, always exceeding those achieved by encoder scaling alone. We also identify label noise as a key challenge in STR, particularly in real-world data, which can limit the effectiveness of STR models. To address this, we propose Cloze Self-Distillation (CSD), a method that mitigates label noise by distilling a student model from context-aware soft predictions and pseudolabels generated by a teacher model. Additionally, we enhance the decoder architecture by introducing differential cross-attention for STR. Our methodology achieves state-of-the-art performance on 10 out of 11 benchmarks using only real data, while significantly reducing the parameter size and computational costs.
Authors: Peihao Wu, Yongxiang Yao, Wenfei Zhang, Dong Wei, Yi Wan, Yansheng Li, Yongjun Zhang
Abstract: Multimodal remote sensing image (MRSI) matching is pivotal for cross-modal fusion, localization, and object detection, but it faces severe challenges due to geometric, radiometric, and viewpoint discrepancies across imaging modalities. Existing unimodal datasets lack scale and diversity, limiting deep learning solutions. This paper proposes MapGlue, a universal MRSI matching framework, and MapData, a large-scale multimodal dataset addressing these gaps. Our contributions are twofold. MapData, a globally diverse dataset spanning 233 sampling points, offers original images (7,000x5,000 to 20,000x15,000 pixels). After rigorous cleaning, it provides 121,781 aligned electronic map-visible image pairs (512x512 pixels) with hybrid manual-automated ground truth, addressing the scarcity of scalable multimodal benchmarks. MapGlue integrates semantic context with a dual graph-guided mechanism to extract cross-modal invariant features. This structure enables global-to-local interaction, enhancing descriptor robustness against modality-specific distortions. Extensive evaluations on MapData and five public datasets demonstrate MapGlue's superiority in matching accuracy under complex conditions, outperforming state-of-the-art methods. Notably, MapGlue generalizes effectively to unseen modalities without retraining, highlighting its adaptability. This work addresses longstanding challenges in MRSI matching by combining scalable dataset construction with a robust, semantics-driven framework. Furthermore, MapGlue shows strong generalization capabilities on other modality matching tasks for which it was not specifically trained. The dataset and code are available at https://github.com/PeihaoWu/MapGlue.
Authors: Ming Li, Shitian Zhao, Jike Zhong, Yuxiang Lai, Kaipeng Zhang
Abstract: Classification is a core task in machine learning. Recent research has shown that although Multimodal Large Language Models (MLLMs) are initially poor at image classification, fine-tuning them with an adequate amount of data can significantly enhance their performance, making them comparable to SOTA classification models. However, acquiring large-scale labeled data is expensive. In this paper, we explore few-shot MLLM classification fine-tuning. We found that SFT can cause severe overfitting issues and may even degrade performance over the zero-shot approach. To address this challenge, inspired by the recent successes in rule-based reinforcement learning, we propose CLS-RL, which uses verifiable signals as reward to fine-tune MLLMs. We discovered that CLS-RL outperforms SFT in most datasets and has a much higher average accuracy on both base-to-new and few-shot learning setting. Moreover, we observed a free-lunch phenomenon for CLS-RL; when models are fine-tuned on a particular dataset, their performance on other distinct datasets may also improve over zero-shot models, even if those datasets differ in distribution and class names. This suggests that RL-based methods effectively teach models the fundamentals of classification. Lastly, inspired by recent works in inference time thinking, we re-examine the `thinking process' during fine-tuning, a critical aspect of RL-based methods, in the context of visual classification. We question whether such tasks require extensive thinking process during fine-tuning, proposing that this may actually detract from performance. Based on this premise, we introduce the No-Thinking-CLS-RL method, which minimizes thinking processes during training by setting an equality accuracy reward. Our findings indicate that, with much less fine-tuning time, No-Thinking-CLS-RL method achieves superior in-domain performance and generalization capabilities than CLS-RL.
Authors: Ziyao Guo, Kaipeng Zhang, Michael Qizhe Shieh
Abstract: Autoregressive models have shown remarkable success in image generation by adapting sequential prediction techniques from language modeling. However, applying these approaches to images requires discretizing continuous pixel data through vector quantization methods like VQ-VAE. To alleviate the quantization errors that existed in VQ-VAE, recent works tend to use larger codebooks. However, this will accordingly expand vocabulary size, complicating the autoregressive modeling task. This paper aims to find a way to enjoy the benefits of large codebooks without making autoregressive modeling more difficult. Through empirical investigation, we discover that tokens with similar codeword representations produce similar effects on the final generated image, revealing significant redundancy in large codebooks. Based on this insight, we propose to predict tokens from coarse to fine (CTF), realized by assigning the same coarse label for similar tokens. Our framework consists of two stages: (1) an autoregressive model that sequentially predicts coarse labels for each token in the sequence, and (2) an auxiliary model that simultaneously predicts fine-grained labels for all tokens conditioned on their coarse labels. Experiments on ImageNet demonstrate our method's superior performance, achieving an average improvement of 59 points in Inception Score compared to baselines. Notably, despite adding an inference step, our approach achieves faster sampling speeds.
Authors: Chia-Yi Hsu, Jia-You Chen, Yu-Lin Tsai, Chih-Hsun Lin, Pin-Yu Chen, Chia-Mu Yu, Chun-Ying Huang
Abstract: Differentially private (DP) synthetic data has become the de facto standard for releasing sensitive data. However, many DP generative models suffer from the low utility of synthetic data, especially for high-resolution images. On the other hand, one of the emerging techniques in parameter efficient fine-tuning (PEFT) is visual prompting (VP), which allows well-trained existing models to be reused for the purpose of adapting to subsequent downstream tasks. In this work, we explore such a phenomenon in constructing captivating generative models with DP constraints. We show that VP in conjunction with DP-NTK, a DP generator that exploits the power of the neural tangent kernel (NTK) in training DP generative models, achieves a significant performance boost, particularly for high-resolution image datasets, with accuracy improving from 0.644$\pm$0.044 to 0.769. Lastly, we perform ablation studies on the effect of different parameters that influence the overall performance of VP-NTK. Our work demonstrates a promising step forward in improving the utility of DP synthetic data, particularly for high-resolution images.
Authors: Yu Cao, Zengqun Zhao, Ioannis Patras, Shaogang Gong
Abstract: Visual artifacts remain a persistent challenge in diffusion models, even with training on massive datasets. Current solutions primarily rely on supervised detectors, yet lack understanding of why these artifacts occur in the first place. In our analysis, we identify three distinct phases in the diffusion generative process: Profiling, Mutation, and Refinement. Artifacts typically emerge during the Mutation phase, where certain regions exhibit anomalous score dynamics over time, causing abrupt disruptions in the normal evolution pattern. This temporal nature explains why existing methods focusing only on spatial uncertainty of the final output fail at effective artifact localization. Based on these insights, we propose ASCED (Abnormal Score Correction for Enhancing Diffusion), that detects artifacts by monitoring abnormal score dynamics during the diffusion process, with a trajectory-aware on-the-fly mitigation strategy that appropriate generation of noise in the detected areas. Unlike most existing methods that apply post hoc corrections, \eg, by applying a noising-denoising scheme after generation, our mitigation strategy operates seamlessly within the existing diffusion process. Extensive experiments demonstrate that our proposed approach effectively reduces artifacts across diverse domains, matching or surpassing existing supervised methods without additional training.
Authors: Max Gutbrod, David Rauber, Danilo Weber Nunes, Christoph Palm
Abstract: The growing reliance on Artificial Intelligence (AI) in critical domains such as healthcare demands robust mechanisms to ensure the trustworthiness of these systems, especially when faced with unexpected or anomalous inputs. This paper introduces the Open Medical Imaging Benchmarks for Out-Of-Distribution Detection (OpenMIBOOD), a comprehensive framework for evaluating out-of-distribution (OOD) detection methods specifically in medical imaging contexts. OpenMIBOOD includes three benchmarks from diverse medical domains, encompassing 14 datasets divided into covariate-shifted in-distribution, near-OOD, and far-OOD categories. We evaluate 24 post-hoc methods across these benchmarks, providing a standardized reference to advance the development and fair comparison of OOD detection methods. Results reveal that findings from broad-scale OOD benchmarks in natural image domains do not translate to medical applications, underscoring the critical need for such benchmarks in the medical field. By mitigating the risk of exposing AI models to inputs outside their training distribution, OpenMIBOOD aims to support the advancement of reliable and trustworthy AI systems in healthcare. The repository is available at https://github.com/remic-othr/OpenMIBOOD.
Authors: Markus Karmann, Peng-Tao Jiang, Bo Li, Onay Urfalioglu
Abstract: We present Markov Map Nearest Neighbor V2 (M2N2V2), a novel and simple, yet effective approach which leverages depth guidance and attention maps for unsupervised and training-free point-prompt-based interactive segmentation. Following recent trends in supervised multimodal approaches, we carefully integrate depth as an additional modality to create novel depth-guided Markov-maps. Furthermore, we observe occasional segment size fluctuations in M2N2 during the interactive process, which can decrease the overall mIoU's. To mitigate this problem, we model the prompting as a sequential process and propose a novel adaptive score function which considers the previous segmentation and the current prompt point in order to prevent unreasonable segment size changes. Using Stable Diffusion 2 and Depth Anything V2 as backbones, we empirically show that our proposed M2N2V2 significantly improves the Number of Clicks (NoC) and mIoU compared to M2N2 in all datasets except those from the medical domain. Interestingly, our unsupervised approach achieves competitive results compared to supervised methods like SAM and SimpleClick in the more challenging DAVIS and HQSeg44K datasets in the NoC metric, reducing the gap between supervised and unsupervised methods.
Authors: Keda Tao, Haoxuan You, Yang Sui, Can Qin, Huan Wang
Abstract: Video large language models (VideoLLMs) have demonstrated the capability to process longer video inputs and enable complex reasoning and analysis. However, due to the thousands of visual tokens from the video frames, key-value (KV) cache can significantly increase memory requirements, becoming a bottleneck for inference speed and memory usage. KV cache quantization is a widely used approach to address this problem. In this paper, we find that 2-bit KV quantization of VideoLLMs can hardly hurt the model performance, while the limit of KV cache quantization in even lower bits has not been investigated. To bridge this gap, we introduce VidKV, a plug-and-play KV cache quantization method to compress the KV cache to lower than 2 bits. Specifically, (1) for key, we propose a mixed-precision quantization strategy in the channel dimension, where we perform 2-bit quantization for anomalous channels and 1-bit quantization combined with FFT for normal channels; (2) for value, we implement 1.58-bit quantization while selectively filtering semantically salient visual tokens for targeted preservation, for a better trade-off between precision and model performance. Importantly, our findings suggest that the value cache of VideoLLMs should be quantized in a per-channel fashion instead of the per-token fashion proposed by prior KV cache quantization works for LLMs. Empirically, extensive results with LLaVA-OV-7B and Qwen2.5-VL-7B on six benchmarks show that VidKV effectively compresses the KV cache to 1.5-bit and 1.58-bit precision with almost no performance drop compared to the FP16 counterparts.
Authors: Zijian Li, Jingjing Fu, Lei Song, Jiang Bian, Jun Zhang, Rui Wang
Abstract: Visual reasoning is crucial for multimodal large language models (MLLMs) to address complex chart queries, yet high-quality rationale data remains scarce. Existing methods leveraged (M)LLMs for data generation, but direct prompting often yields limited precision and diversity. In this paper, we propose \textit{Chain of Functions (CoF)}, a novel programmatic reasoning data generation pipeline that utilizes freely-explored reasoning paths as supervision to ensure data precision and diversity. Specifically, it starts with human-free exploration among the atomic functions (e.g., maximum data and arithmetic operations) to generate diverse function chains, which are then translated into linguistic rationales and questions with only a moderate open-sourced LLM. \textit{CoF} provides multiple benefits: 1) Precision: function-governed generation reduces hallucinations compared to freeform generation; 2) Diversity: enumerating function chains enables varied question taxonomies; 3) Explainability: function chains serve as built-in rationales, allowing fine-grained evaluation beyond overall accuracy; 4) Practicality: eliminating reliance on extremely large models. Employing \textit{CoF}, we construct the \textit{ChartCoF} dataset, with 1.4k complex reasoning Q\&A for fine-grained analysis and 50k Q\&A for reasoning enhancement. The fine-grained evaluation on \textit{ChartCoF} reveals varying performance across question taxonomies for each MLLM, and the experiments also show that finetuning with \textit{ChartCoF} achieves state-of-the-art performance among same-scale MLLMs on widely used benchmarks. Furthermore, the novel paradigm of function-governed rationale generation in \textit{CoF} could inspire broader applications beyond charts.
Authors: Ayberk Acar, Mariana Smith, Lidia Al-Zogbi, Tanner Watts, Fangjie Li, Hao Li, Nural Yilmaz, Paul Maria Scheikl, Jesse F. d'Almeida, Susheela Sharma, Lauren Branscombe, Tayfun Efe Ertop, Robert J. Webster III, Ipek Oguz, Alan Kuntz, Axel Krieger, Jie Ying Wu
Abstract: Surgical automation requires precise guidance and understanding of the scene. Current methods in the literature rely on bulky depth cameras to create maps of the anatomy, however this does not translate well to space-limited clinical applications. Monocular cameras are small and allow minimally invasive surgeries in tight spaces but additional processing is required to generate 3D scene understanding. We propose a 3D mapping pipeline that uses only RGB images to create segmented point clouds of the target anatomy. To ensure the most precise reconstruction, we compare different structure from motion algorithms' performance on mapping the central airway obstructions, and test the pipeline on a downstream task of tumor resection. In several metrics, including post-procedure tissue model evaluation, our pipeline performs comparably to RGB-D cameras and, in some cases, even surpasses their performance. These promising results demonstrate that automation guidance can be achieved in minimally invasive procedures with monocular cameras. This study is a step toward the complete autonomy of surgical robots.
Authors: Zhaochong An, Guolei Sun, Yun Liu, Runjia Li, Junlin Han, Ender Konukoglu, Serge Belongie
Abstract: Generalized few-shot 3D point cloud segmentation (GFS-PCS) adapts models to new classes with few support samples while retaining base class segmentation. Existing GFS-PCS methods enhance prototypes via interacting with support or query features but remain limited by sparse knowledge from few-shot samples. Meanwhile, 3D vision-language models (3D VLMs), generalizing across open-world novel classes, contain rich but noisy novel class knowledge. In this work, we introduce a GFS-PCS framework that synergizes dense but noisy pseudo-labels from 3D VLMs with precise yet sparse few-shot samples to maximize the strengths of both, named GFS-VL. Specifically, we present a prototype-guided pseudo-label selection to filter low-quality regions, followed by an adaptive infilling strategy that combines knowledge from pseudo-label contexts and few-shot samples to adaptively label the filtered, unlabeled areas. Additionally, we design a novel-base mix strategy to embed few-shot samples into training scenes, preserving essential context for improved novel class learning. Moreover, recognizing the limited diversity in current GFS-PCS benchmarks, we introduce two challenging benchmarks with diverse novel classes for comprehensive generalization evaluation. Experiments validate the effectiveness of our framework across models and datasets. Our approach and benchmarks provide a solid foundation for advancing GFS-PCS in the real world. The code is at https://github.com/ZhaochongAn/GFS-VL
Authors: Sharon Peled, Yosef E. Maruvka, Moti Freiman
Abstract: Whole Slide Images (WSIs) are high-resolution digital scans widely used in medical diagnostics. WSI classification is typically approached using Multiple Instance Learning (MIL), where the slide is partitioned into tiles treated as interconnected instances. While attention-based MIL methods aim to identify the most informative tiles, they often fail to fully exploit the spatial relationships among them, potentially overlooking intricate tissue structures crucial for accurate diagnosis. To address this limitation, we propose Probabilistic Spatial Attention MIL (PSA-MIL), a novel attention-based MIL framework that integrates spatial context into the attention mechanism through learnable distance-decayed priors, formulated within a probabilistic interpretation of self-attention as a posterior distribution. This formulation enables a dynamic inference of spatial relationships during training, eliminating the need for predefined assumptions often imposed by previous approaches. Additionally, we suggest a spatial pruning strategy for the posterior, effectively reducing self-attention's quadratic complexity. To further enhance spatial modeling, we introduce a diversity loss that encourages variation among attention heads, ensuring each captures distinct spatial representations. Together, PSA-MIL enables a more data-driven and adaptive integration of spatial context, moving beyond predefined constraints. We achieve state-of-the-art performance across both contextual and non-contextual baselines, while significantly reducing computational costs.
Authors: Inwoo Hwang, Bing Zhou, Young Min Kim, Jian Wang, Chuan Guo
Abstract: Modeling human-scene interactions (HSI) is essential for understanding and simulating everyday human behaviors. Recent approaches utilizing generative modeling have made progress in this domain; however, they are limited in controllability and flexibility for real-world applications. To address these challenges, we propose reformulating the HSI modeling problem as Scene-aware Motion In-betweening -- a more tractable and practical task. We introduce SceneMI, a framework that supports several practical applications, including keyframe-guided character animation in 3D scenes and enhancing the motion quality of imperfect HSI data. SceneMI employs dual scene descriptors to comprehensively encode global and local scene context. Furthermore, our framework leverages the inherent denoising nature of diffusion models to generalize on noisy keyframes. Experimental results demonstrate SceneMI's effectiveness in scene-aware keyframe in-betweening and generalization to the real-world GIMO dataset, where motions and scenes are acquired by noisy IMU sensors and smartphones. We further showcase SceneMI's applicability in HSI reconstruction from monocular videos.
Authors: Zeqiang Lai, Yunfei Zhao, Zibo Zhao, Haolin Liu, Fuyun Wang, Huiwen Shi, Xianghui Yang, Qinxiang Lin, Jinwei Huang, Yuhong Liu, Jie Jiang, Chunchao Guo, Xiangyu Yue
Abstract: 3D shape generation has greatly flourished through the development of so-called "native" 3D diffusion, particularly through the Vecset Diffusion Model (VDM). While recent advancements have shown promising results in generating high-resolution 3D shapes, VDM still struggles with high-speed generation. Challenges exist because of difficulties not only in accelerating diffusion sampling but also VAE decoding in VDM, areas under-explored in previous works. To address these challenges, we present FlashVDM, a systematic framework for accelerating both VAE and DiT in VDM. For DiT, FlashVDM enables flexible diffusion sampling with as few as 5 inference steps and comparable quality, which is made possible by stabilizing consistency distillation with our newly introduced Progressive Flow Distillation. For VAE, we introduce a lightning vecset decoder equipped with Adaptive KV Selection, Hierarchical Volume Decoding, and Efficient Network Design. By exploiting the locality of the vecset and the sparsity of shape surface in the volume, our decoder drastically lowers FLOPs, minimizing the overall decoding overhead. We apply FlashVDM to Hunyuan3D-2 to obtain Hunyuan3D-2 Turbo. Through systematic evaluation, we show that our model significantly outperforms existing fast 3D generation methods, achieving comparable performance to the state-of-the-art while reducing inference time by over 45x for reconstruction and 32x for generation. Code and models are available at https://github.com/Tencent/FlashVDM.
Authors: Edgar Sucar, Zihang Lai, Eldar Insafutdinov, Andrea Vedaldi
Abstract: DUSt3R has recently shown that one can reduce many tasks in multi-view geometry, including estimating camera intrinsics and extrinsics, reconstructing the scene in 3D, and establishing image correspondences, to the prediction of a pair of viewpoint-invariant point maps, i.e., pixel-aligned point clouds defined in a common reference frame. This formulation is elegant and powerful, but unable to tackle dynamic scenes. To address this challenge, we introduce the concept of Dynamic Point Maps (DPM), extending standard point maps to support 4D tasks such as motion segmentation, scene flow estimation, 3D object tracking, and 2D correspondence. Our key intuition is that, when time is introduced, there are several possible spatial and time references that can be used to define the point maps. We identify a minimal subset of such combinations that can be regressed by a network to solve the sub tasks mentioned above. We train a DPM predictor on a mixture of synthetic and real data and evaluate it across diverse benchmarks for video depth prediction, dynamic point cloud reconstruction, 3D scene flow and object pose tracking, achieving state-of-the-art performance. Code, models and additional results are available at https://www.robots.ox.ac.uk/~vgg/research/dynamic-point-maps/.
Authors: Ruonan Yu, Songhua Liu, Zhenxiong Tan, Xinchao Wang
Abstract: Text-to-image diffusion models have achieved remarkable progress in recent years. However, training models for high-resolution image generation remains challenging, particularly when training data and computational resources are limited. In this paper, we explore this practical problem from two key perspectives: data and parameter efficiency, and propose a set of key guidelines for ultra-resolution adaptation termed \emph{URAE}. For data efficiency, we theoretically and empirically demonstrate that synthetic data generated by some teacher models can significantly promote training convergence. For parameter efficiency, we find that tuning minor components of the weight matrices outperforms widely-used low-rank adapters when synthetic data are unavailable, offering substantial performance gains while maintaining efficiency. Additionally, for models leveraging guidance distillation, such as FLUX, we show that disabling classifier-free guidance, \textit{i.e.}, setting the guidance scale to 1 during adaptation, is crucial for satisfactory performance. Extensive experiments validate that URAE achieves comparable 2K-generation performance to state-of-the-art closed-source models like FLUX1.1 [Pro] Ultra with only 3K samples and 2K iterations, while setting new benchmarks for 4K-resolution generation. Codes are available \href{https://github.com/Huage001/URAE}{here}.
Authors: Shengjun Zhang, Xin Fei, Fangfu Liu, Haixu Song, Yueqi Duan
Abstract: 3D Gaussian Splatting (3DGS) has demonstrated impressive novel view synthesis performance. While conventional methods require per-scene optimization, more recently several feed-forward methods have been proposed to generate pixel-aligned Gaussian representations with a learnable network, which are generalizable to different scenes. However, these methods simply combine pixel-aligned Gaussians from multiple views as scene representations, thereby leading to artifacts and extra memory cost without fully capturing the relations of Gaussians from different images. In this paper, we propose Gaussian Graph Network (GGN) to generate efficient and generalizable Gaussian representations. Specifically, we construct Gaussian Graphs to model the relations of Gaussian groups from different views. To support message passing at Gaussian level, we reformulate the basic graph operations over Gaussian representations, enabling each Gaussian to benefit from its connected Gaussian groups with Gaussian feature fusion. Furthermore, we design a Gaussian pooling layer to aggregate various Gaussian groups for efficient representations. We conduct experiments on the large-scale RealEstate10K and ACID datasets to demonstrate the efficiency and generalization of our method. Compared to the state-of-the-art methods, our model uses fewer Gaussians and achieves better image quality with higher rendering speed.
Authors: Tao Feng, Yifan Xie, Xun Guan, Jiyuan Song, Zhou Liu, Fei Ma, Fei Yu
Abstract: Precise audio-visual synchronization in speech videos is crucial for content quality and viewer comprehension. Existing methods have made significant strides in addressing this challenge through rule-based approaches and end-to-end learning techniques. However, these methods often rely on limited audio-visual representations and suboptimal learning strategies, potentially constraining their effectiveness in more complex scenarios. To address these limitations, we present UniSync, a novel approach for evaluating audio-visual synchronization using embedding similarities. UniSync offers broad compatibility with various audio representations (e.g., Mel spectrograms, HuBERT) and visual representations (e.g., RGB images, face parsing maps, facial landmarks, 3DMM), effectively handling their significant dimensional differences. We enhance the contrastive learning framework with a margin-based loss component and cross-speaker unsynchronized pairs, improving discriminative capabilities. UniSync outperforms existing methods on standard datasets and demonstrates versatility across diverse audio-visual representations. Its integration into talking face generation frameworks enhances synchronization quality in both natural and AI-generated content.
Authors: Muyao Li, Zihao Wang, Kaichen He, Xiaojian Ma, Yitao Liang
Abstract: Recently, action-based decision-making in open-world environments has gained significant attention. Visual Language Action (VLA) models, pretrained on large-scale web datasets, have shown promise in decision-making tasks. However, previous work has primarily focused on action post-training, often neglecting enhancements to the foundational model itself. In response, we introduce a novel approach, Act from Visual Language Post-Training, which refines Visual Language Models (VLMs) through visual and linguistic guidance in a self-supervised manner. This enhancement improves the models' capabilities in world knowledge, visual recognition, and spatial grounding in open-world environments. Following the above post-training paradigms, we obtain the first VLA models in Minecraft that can follow human instructions on over 1k different atomic tasks, including crafting, smelting, cooking, mining, and killing. Our experiments demonstrate that post-training on non-trajectory tasks leads to a significant 40% improvement over the best agent baseline on a diverse set of atomic tasks. Furthermore, we demonstrate that our approach surpasses traditional imitation learning-based policies in Minecraft, achieving state-of-the-art performance. We have open-sourced the code, models, and datasets to foster further research. The project page can be found in https://craftjarvis.github.io/JarvisVLA.
Authors: Han-Hung Lee, Qinghong Han, Angel X. Chang
Abstract: In this paper, we explore the task of generating expansive outdoor scenes, ranging from castles to high-rises. Unlike indoor scene generation, which has been a primary focus of prior work, outdoor scene generation presents unique challenges, including wide variations in scene heights and the need for a method capable of rapidly producing large landscapes. To address this, we propose an efficient approach that encodes scene chunks as uniform vector sets, offering better compression and performance than the spatially structured latents used in prior methods. Furthermore, we train an explicit outpainting model for unbounded generation, which improves coherence compared to prior resampling-based inpainting schemes while also speeding up generation by eliminating extra diffusion steps. To facilitate this task, we curate NuiScene43, a small but high-quality set of scenes, preprocessed for joint training. Notably, when trained on scenes of varying styles, our model can blend different environments, such as rural houses and city skyscrapers, within the same scene, highlighting the potential of our curation process to leverage heterogeneous scenes for joint training.
Authors: Leyang Wang, Joice Lin
Abstract: The success of modern machine learning, particularly in facial translation networks, is highly dependent on the availability of high-quality, paired, large-scale datasets. However, acquiring sufficient data is often challenging and costly. Inspired by the recent success of diffusion models in high-quality image synthesis and advancements in Large Language Models (LLMs), we propose a novel framework called LLM-assisted Paired Image Generation (LaPIG). This framework enables the construction of comprehensive, high-quality paired visible and thermal images using captions generated by LLMs. Our method encompasses three parts: visible image synthesis with ArcFace embedding, thermal image translation using Latent Diffusion Models (LDMs), and caption generation with LLMs. Our approach not only generates multi-view paired visible and thermal images to increase data diversity but also produces high-quality paired data while maintaining their identity information. We evaluate our method on public datasets by comparing it with existing methods, demonstrating the superiority of LaPIG.
Authors: Tzu-Yun Tseng, Alexey Nekrasov, Malcolm Burdorf, Bastian Leibe, Julie Stephany Berrio, Mao Shan, Stewart Worrall
Abstract: Existing autonomous driving datasets are predominantly oriented towards well-structured urban settings and favorable weather conditions, leaving the complexities of rural environments and adverse weather conditions largely unaddressed. Although some datasets encompass variations in weather and lighting, bad weather scenarios do not appear often. Rainfall can significantly impair sensor functionality, introducing noise and reflections in LiDAR and camera data and reducing the system's capabilities for reliable environmental perception and safe navigation. We introduce the Panoptic-CUDAL dataset, a novel dataset purpose-built for panoptic segmentation in rural areas subject to rain. By recording high-resolution LiDAR, camera, and pose data, Panoptic-CUDAL offers a diverse, information-rich dataset in a challenging scenario. We present analysis of the recorded data and provide baseline results for panoptic and semantic segmentation methods on LiDAR point clouds. The dataset can be found here: https://robotics.sydney.edu.au/our-research/intelligent-transportation-systems/
URLs: https://robotics.sydney.edu.au/our-research/intelligent-transportation-systems/
Authors: Akhil Perincherry, Jacob Krantz, Stefan Lee
Abstract: Vision-and-Language Navigation (VLN) agents are tasked with navigating an unseen environment using natural language instructions. In this work, we study if visual representations of sub-goals implied by the instructions can serve as navigational cues and lead to increased navigation performance. To synthesize these visual representations or imaginations, we leverage a text-to-image diffusion model on landmark references contained in segmented instructions. These imaginations are provided to VLN agents as an added modality to act as landmark cues and an auxiliary loss is added to explicitly encourage relating these with their corresponding referring expressions. Our findings reveal an increase in success rate (SR) of around 1 point and up to 0.5 points in success scaled by inverse path length (SPL) across agents. These results suggest that the proposed approach reinforces visual understanding compared to relying on language instructions alone. Code and data for our work can be found at https://www.akhilperincherry.com/VLN-Imagine-website/.
URLs: https://www.akhilperincherry.com/VLN-Imagine-website/.
Authors: Chun-Han Yao, Yiming Xie, Vikram Voleti, Huaizu Jiang, Varun Jampani
Abstract: We present Stable Video 4D 2.0 (SV4D 2.0), a multi-view video diffusion model for dynamic 3D asset generation. Compared to its predecessor SV4D, SV4D 2.0 is more robust to occlusions and large motion, generalizes better to real-world videos, and produces higher-quality outputs in terms of detail sharpness and spatio-temporal consistency. We achieve this by introducing key improvements in multiple aspects: 1) network architecture: eliminating the dependency of reference multi-views and designing blending mechanism for 3D and frame attention, 2) data: enhancing quality and quantity of training data, 3) training strategy: adopting progressive 3D-4D training for better generalization, and 4) 4D optimization: handling 3D inconsistency and large motion via 2-stage refinement and progressive frame sampling. Extensive experiments demonstrate significant performance gain by SV4D 2.0 both visually and quantitatively, achieving better detail (-14\% LPIPS) and 4D consistency (-44\% FV4D) in novel-view video synthesis and 4D optimization (-12\% LPIPS and -24\% FV4D) compared to SV4D. Project page: https://sv4d2.0.github.io.
Authors: Nikita Starodubcev, Denis Kuznedelev, Artem Babenko, Dmitry Baranchuk
Abstract: We present SwD, a scale-wise distillation framework for diffusion models (DMs), which effectively employs next-scale prediction ideas for diffusion-based few-step generators. In more detail, SwD is inspired by the recent insights relating diffusion processes to the implicit spectral autoregression. We suppose that DMs can initiate generation at lower data resolutions and gradually upscale the samples at each denoising step without loss in performance while significantly reducing computational costs. SwD naturally integrates this idea into existing diffusion distillation methods based on distribution matching. Also, we enrich the family of distribution matching approaches by introducing a novel patch loss enforcing finer-grained similarity to the target distribution. When applied to state-of-the-art text-to-image diffusion models, SwD approaches the inference times of two full resolution steps and significantly outperforms the counterparts under the same computation budget, as evidenced by automated metrics and human preference studies.
Authors: Chen Chen, Zhirui Wang, Taowei Sheng, Yi Jiang, Yundu Li, Peirui Cheng, Luning Zhang, Kaiqiang Chen, Yanfeng Hu, Xue Yang, Xian Sun
Abstract: Existing vision-based 3D occupancy prediction methods are inherently limited in accuracy due to their exclusive reliance on street-view imagery, neglecting the potential benefits of incorporating satellite views. We propose SA-Occ, the first Satellite-Assisted 3D occupancy prediction model, which leverages GPS & IMU to integrate historical yet readily available satellite imagery into real-time applications, effectively mitigating limitations of ego-vehicle perceptions, involving occlusions and degraded performance in distant regions. To address the core challenges of cross-view perception, we propose: 1) Dynamic-Decoupling Fusion, which resolves inconsistencies in dynamic regions caused by the temporal asynchrony between satellite and street views; 2) 3D-Proj Guidance, a module that enhances 3D feature extraction from inherently 2D satellite imagery; and 3) Uniform Sampling Alignment, which aligns the sampling density between street and satellite views. Evaluated on Occ3D-nuScenes, SA-Occ achieves state-of-the-art performance, especially among single-frame methods, with a 39.05% mIoU (a 6.97% improvement), while incurring only 6.93 ms of additional latency per frame. Our code and newly curated dataset are available at https://github.com/chenchen235/SA-Occ.
Authors: Ananta R. Bhattarai, Xingzhe He, Alla Sheffer, Helge Rhodin
Abstract: DreamFusion established a new paradigm for unsupervised 3D reconstruction from virtual views by combining advances in generative models and differentiable rendering. However, the underlying multi-view rendering, along with supervision from large-scale generative models, is computationally expensive and under-constrained. We propose DreamTexture, a novel Shape-from-Virtual-Texture approach that leverages monocular depth cues to reconstruct 3D objects. Our method textures an input image by aligning a virtual texture with the real depth cues in the input, exploiting the inherent understanding of monocular geometry encoded in modern diffusion models. We then reconstruct depth from the virtual texture deformation with a new conformal map optimization, which alleviates memory-intensive volumetric representations. Our experiments reveal that generative models possess an understanding of monocular shape cues, which can be extracted by augmenting and aligning texture cues -- a novel monocular reconstruction paradigm that we call Analysis by Augmentation.
Authors: Xueyan Zou, Yuchen Song, Ri-Zhao Qiu, Xuanbin Peng, Jianglong Ye, Sifei Liu, Xiaolong Wang
Abstract: We present 3D Spatial MultiModal Memory (M3), a multimodal memory system designed to retain information about medium-sized static scenes through video sources for visual perception. By integrating 3D Gaussian Splatting techniques with foundation models, M3 builds a multimodal memory capable of rendering feature representations across granularities, encompassing a wide range of knowledge. In our exploration, we identify two key challenges in previous works on feature splatting: (1) computational constraints in storing high-dimensional features for each Gaussian primitive, and (2) misalignment or information loss between distilled features and foundation model features. To address these challenges, we propose M3 with key components of principal scene components and Gaussian memory attention, enabling efficient training and inference. To validate M3, we conduct comprehensive quantitative evaluations of feature similarity and downstream tasks, as well as qualitative visualizations to highlight the pixel trace of Gaussian memory attention. Our approach encompasses a diverse range of foundation models, including vision-language models (VLMs), perception models, and large multimodal and language models (LMMs/LLMs). Furthermore, to demonstrate real-world applicability, we deploy M3's feature field in indoor scenes on a quadruped robot. Notably, we claim that M3 is the first work to address the core compression challenges in 3D feature distillation.
Authors: Liming Jiang, Qing Yan, Yumin Jia, Zichuan Liu, Hao Kang, Xin Lu
Abstract: Achieving flexible and high-fidelity identity-preserved image generation remains formidable, particularly with advanced Diffusion Transformers (DiTs) like FLUX. We introduce InfiniteYou (InfU), one of the earliest robust frameworks leveraging DiTs for this task. InfU addresses significant issues of existing methods, such as insufficient identity similarity, poor text-image alignment, and low generation quality and aesthetics. Central to InfU is InfuseNet, a component that injects identity features into the DiT base model via residual connections, enhancing identity similarity while maintaining generation capabilities. A multi-stage training strategy, including pretraining and supervised fine-tuning (SFT) with synthetic single-person-multiple-sample (SPMS) data, further improves text-image alignment, ameliorates image quality, and alleviates face copy-pasting. Extensive experiments demonstrate that InfU achieves state-of-the-art performance, surpassing existing baselines. In addition, the plug-and-play design of InfU ensures compatibility with various existing methods, offering a valuable contribution to the broader community.
Authors: Paul Engstler, Aleksandar Shtedritski, Iro Laina, Christian Rupprecht, Andrea Vedaldi
Abstract: We address the challenge of generating 3D worlds from textual descriptions. We propose SynCity, a training- and optimization-free approach, which leverages the geometric precision of pre-trained 3D generative models and the artistic versatility of 2D image generators to create large, high-quality 3D spaces. While most 3D generative models are object-centric and cannot generate large-scale worlds, we show how 3D and 2D generators can be combined to generate ever-expanding scenes. Through a tile-based approach, we allow fine-grained control over the layout and the appearance of scenes. The world is generated tile-by-tile, and each new tile is generated within its world-context and then fused with the scene. SynCity generates compelling and immersive scenes that are rich in detail and diversity.
Authors: Quanhao Li, Zhen Xing, Rui Wang, Hui Zhang, Qi Dai, Zuxuan Wu
Abstract: Recent advances in video generation have led to remarkable improvements in visual quality and temporal coherence. Upon this, trajectory-controllable video generation has emerged to enable precise object motion control through explicitly defined spatial paths. However, existing methods struggle with complex object movements and multi-object motion control, resulting in imprecise trajectory adherence, poor object consistency, and compromised visual quality. Furthermore, these methods only support trajectory control in a single format, limiting their applicability in diverse scenarios. Additionally, there is no publicly available dataset or benchmark specifically tailored for trajectory-controllable video generation, hindering robust training and systematic evaluation. To address these challenges, we introduce MagicMotion, a novel image-to-video generation framework that enables trajectory control through three levels of conditions from dense to sparse: masks, bounding boxes, and sparse boxes. Given an input image and trajectories, MagicMotion seamlessly animates objects along defined trajectories while maintaining object consistency and visual quality. Furthermore, we present MagicData, a large-scale trajectory-controlled video dataset, along with an automated pipeline for annotation and filtering. We also introduce MagicBench, a comprehensive benchmark that assesses both video quality and trajectory control accuracy across different numbers of objects. Extensive experiments demonstrate that MagicMotion outperforms previous methods across various metrics. Our project page are publicly available at https://quanhaol.github.io/magicmotion-site.
Authors: Yuheng Yuan, Qiuhong Shen, Xingyi Yang, Xinchao Wang
Abstract: 4D Gaussian Splatting (4DGS) has recently gained considerable attention as a method for reconstructing dynamic scenes. Despite achieving superior quality, 4DGS typically requires substantial storage and suffers from slow rendering speed. In this work, we delve into these issues and identify two key sources of temporal redundancy. (Q1) \textbf{Short-Lifespan Gaussians}: 4DGS uses a large portion of Gaussians with short temporal span to represent scene dynamics, leading to an excessive number of Gaussians. (Q2) \textbf{Inactive Gaussians}: When rendering, only a small subset of Gaussians contributes to each frame. Despite this, all Gaussians are processed during rasterization, resulting in redundant computation overhead. To address these redundancies, we present \textbf{4DGS-1K}, which runs at over 1000 FPS on modern GPUs. For Q1, we introduce the Spatial-Temporal Variation Score, a new pruning criterion that effectively removes short-lifespan Gaussians while encouraging 4DGS to capture scene dynamics using Gaussians with longer temporal spans. For Q2, we store a mask for active Gaussians across consecutive frames, significantly reducing redundant computations in rendering. Compared to vanilla 4DGS, our method achieves a $41\times$ reduction in storage and $9\times$ faster rasterization speed on complex dynamic scenes, while maintaining comparable visual quality. Please see our project page at https://4DGS-1K.github.io.
Authors: Ron Campos, Ashmal Vayani, Parth Parag Kulkarni, Rohit Gupta, Aritra Dutta, Mubarak Shah
Abstract: Image geolocalization, in which, traditionally, an AI model predicts the precise GPS coordinates of an image is a challenging task with many downstream applications. However, the user cannot utilize the model to further their knowledge other than the GPS coordinate; the model lacks an understanding of the location and the conversational ability to communicate with the user. In recent days, with tremendous progress of large multimodal models (LMMs) proprietary and open-source researchers have attempted to geolocalize images via LMMs. However, the issues remain unaddressed; beyond general tasks, for more specialized downstream tasks, one of which is geolocalization, LMMs struggle. In this work, we propose to solve this problem by introducing a conversational model GAEA that can provide information regarding the location of an image, as required by a user. No large-scale dataset enabling the training of such a model exists. Thus we propose a comprehensive dataset GAEA with 800K images and around 1.6M question answer pairs constructed by leveraging OpenStreetMap (OSM) attributes and geographical context clues. For quantitative evaluation, we propose a diverse benchmark comprising 4K image-text pairs to evaluate conversational capabilities equipped with diverse question types. We consider 11 state-of-the-art open-source and proprietary LMMs and demonstrate that GAEA significantly outperforms the best open-source model, LLaVA-OneVision by 25.69% and the best proprietary model, GPT-4o by 8.28%. Our dataset, model and codes are available
Authors: Zigang Geng, Mengde Xu, Han Hu, Shuyang Gu
Abstract: This paper proposes a fundamentally new paradigm for image generation through set-based tokenization and distribution modeling. Unlike conventional methods that serialize images into fixed-position latent codes with a uniform compression ratio, we introduce an unordered token set representation to dynamically allocate coding capacity based on regional semantic complexity. This TokenSet enhances global context aggregation and improves robustness against local perturbations. To address the critical challenge of modeling discrete sets, we devise a dual transformation mechanism that bijectively converts sets into fixed-length integer sequences with summation constraints. Further, we propose Fixed-Sum Discrete Diffusion--the first framework to simultaneously handle discrete values, fixed sequence length, and summation invariance--enabling effective set distribution modeling. Experiments demonstrate our method's superiority in semantic-aware representation and generation quality. Our innovations, spanning novel representation and modeling strategies, advance visual generation beyond traditional sequential token paradigms. Our code and models are publicly available at https://github.com/Gengzigang/TokenSet.
Authors: Keyan Chen, Chenyang Liu, Bowen Chen, Wenyuan Li, Zhengxia Zou, Zhenwei Shi
Abstract: The advancement of remote sensing technology has improved the spatial resolution of satellite imagery, facilitating more detailed visual representations for diverse interpretations. However, existing methods exhibit limited generalization capabilities across varied applications. While some contemporary foundation models demonstrate potential, they are hindered by insufficient cross-task adaptability and primarily process low-resolution imagery of restricted sizes, thus failing to fully exploit high-resolution data or leverage comprehensive large-scene semantics. Crucially, remote sensing imagery differs fundamentally from natural images, as key foreground targets (eg., maritime objects, artificial structures) often occupy minimal spatial proportions (~1%) and exhibit sparse distributions. Efficiently modeling cross-task generalizable knowledge from lengthy 2D tokens (~100,000) poses a significant challenge yet remains critical for remote sensing image understanding. Motivated by the selective attention mechanisms inherent to the human visual system, we propose DynamicVis, a dynamic visual perception foundation model for remote sensing imagery. The framework integrates a novel dynamic region perception backbone based on the selective state space model, which strategically balances localized detail extraction with global contextual integration, enabling computationally efficient encoding of large-scale data while maintaining architectural scalability. To enhance cross-task knowledge transferring, we introduce a multi-instance learning paradigm utilizing meta-embedding representations, trained on million-scale region-level annotations. Evaluations across nine downstream tasks demonstrate the model's versatility. DynamicVis achieves multi-level feature modeling with exceptional efficiency, processing (2048x2048) pixels with 97 ms latency (6% of ViT's) and 833 MB GPU memory (3% of ViT's).
Authors: Xiaoyang Wu, Daniel DeTone, Duncan Frost, Tianwei Shen, Chris Xie, Nan Yang, Jakob Engel, Richard Newcombe, Hengshuang Zhao, Julian Straub
Abstract: In this paper, we question whether we have a reliable self-supervised point cloud model that can be used for diverse 3D tasks via simple linear probing, even with limited data and minimal computation. We find that existing 3D self-supervised learning approaches fall short when evaluated on representation quality through linear probing. We hypothesize that this is due to what we term the "geometric shortcut", which causes representations to collapse to low-level spatial features. This challenge is unique to 3D and arises from the sparse nature of point cloud data. We address it through two key strategies: obscuring spatial information and enhancing the reliance on input features, ultimately composing a Sonata of 140k point clouds through self-distillation. Sonata is simple and intuitive, yet its learned representations are strong and reliable: zero-shot visualizations demonstrate semantic grouping, alongside strong spatial reasoning through nearest-neighbor relationships. Sonata demonstrates exceptional parameter and data efficiency, tripling linear probing accuracy (from 21.8% to 72.5%) on ScanNet and nearly doubling performance with only 1% of the data compared to previous approaches. Full fine-tuning further advances SOTA across both 3D indoor and outdoor perception tasks.
Authors: Yuqing Wang, Zhijie Lin, Yao Teng, Yuanzhi Zhu, Shuhuai Ren, Jiashi Feng, Xihui Liu
Abstract: Autoregressive visual generation models typically rely on tokenizers to compress images into tokens that can be predicted sequentially. A fundamental dilemma exists in token representation: discrete tokens enable straightforward modeling with standard cross-entropy loss, but suffer from information loss and tokenizer training instability; continuous tokens better preserve visual details, but require complex distribution modeling, complicating the generation pipeline. In this paper, we propose TokenBridge, which bridges this gap by maintaining the strong representation capacity of continuous tokens while preserving the modeling simplicity of discrete tokens. To achieve this, we decouple discretization from the tokenizer training process through post-training quantization that directly obtains discrete tokens from continuous representations. Specifically, we introduce a dimension-wise quantization strategy that independently discretizes each feature dimension, paired with a lightweight autoregressive prediction mechanism that efficiently model the resulting large token space. Extensive experiments show that our approach achieves reconstruction and generation quality on par with continuous methods while using standard categorical prediction. This work demonstrates that bridging discrete and continuous paradigms can effectively harness the strengths of both approaches, providing a promising direction for high-quality visual generation with simple autoregressive modeling. Project page: https://yuqingwang1029.github.io/TokenBridge.
Authors: Matthew Wilchek, Linhan Wang, Sally Dickinson, Erica Feuerbacher, Kurt Luther, Feras A. Batarseh
Abstract: In urban search and rescue (USAR) operations, communication between handlers and specially trained canines is crucial but often complicated by challenging environments and the specific behaviors canines are trained to exhibit when detecting a person. Since a USAR canine often works out of sight of the handler, the handler lacks awareness of the canine's location and situation, known as the 'sensemaking gap.' In this paper, we propose KHAIT, a novel approach to close the sensemaking gap and enhance USAR effectiveness by integrating object detection-based Artificial Intelligence (AI) and Augmented Reality (AR). Equipped with AI-powered cameras, edge computing, and AR headsets, KHAIT enables precise and rapid object detection from a canine's perspective, improving survivor localization. We evaluate this approach in a real-world USAR environment, demonstrating an average survival allocation time decrease of 22%, enhancing the speed and accuracy of operations.
Authors: Inwoo Hwang, Jinseok Bae, Donggeun Lim, Young Min Kim
Abstract: Creating expressive character animations is labor-intensive, requiring intricate manual adjustment of animators across space and time. Previous works on controllable motion generation often rely on a predefined set of dense spatio-temporal specifications (e.g., dense pelvis trajectories with exact per-frame timing), limiting practicality for animators. To process high-level intent and intuitive control in diverse scenarios, we propose a practical controllable motions synthesis framework that respects sparse and flexible keyjoint signals. Our approach employs a decomposed diffusion-based motion synthesis framework that first synthesizes keyjoint movements from sparse input control signals and then synthesizes full-body motion based on the completed keyjoint trajectories. The low-dimensional keyjoint movements can easily adapt to various control signal types, such as end-effector position for diverse goal-driven motion synthesis, or incorporate functional constraints on a subset of keyjoints. Additionally, we introduce a time-agnostic control formulation, eliminating the need for frame-specific timing annotations and enhancing control flexibility. Then, the shared second stage can synthesize a natural whole-body motion that precisely satisfies the task requirement from dense keyjoint movements. We demonstrate the effectiveness of sparse and flexible keyjoint control through comprehensive experiments on diverse datasets and scenarios.
Authors: NVIDIA, :, Alisson Azzolini, Hannah Brandon, Prithvijit Chattopadhyay, Huayu Chen, Jinju Chu, Yin Cui, Jenna Diamond, Yifan Ding, Francesco Ferroni, Rama Govindaraju, Jinwei Gu, Siddharth Gururani, Imad El Hanafi, Zekun Hao, Jacob Huffman, Jingyi Jin, Brendan Johnson, Rizwan Khan, George Kurian, Elena Lantz, Nayeon Lee, Zhaoshuo Li, Xuan Li, Tsung-Yi Lin, Yen-Chen Lin, Ming-Yu Liu, Andrew Mathau, Yun Ni, Lindsey Pavao, Wei Ping, David W. Romero, Misha Smelyanskiy, Shuran Song, Lyne Tchapmi, Andrew Z. Wang, Boxin Wang, Haoxiang Wang, Fangyin Wei, Jiashu Xu, Yao Xu, Xiaodong Yang, Zhuolin Yang, Xiaohui Zeng, Zhe Zhang
Abstract: Physical AI systems need to perceive, understand, and perform complex actions in the physical world. In this paper, we present the Cosmos-Reason1 models that can understand the physical world and generate appropriate embodied decisions (e.g., next step action) in natural language through long chain-of-thought reasoning processes. We begin by defining key capabilities for Physical AI reasoning, with a focus on physical common sense and embodied reasoning. To represent physical common sense, we use a hierarchical ontology that captures fundamental knowledge about space, time, and physics. For embodied reasoning, we rely on a two-dimensional ontology that generalizes across different physical embodiments. Building on these capabilities, we develop two multimodal large language models, Cosmos-Reason1-8B and Cosmos-Reason1-56B. We curate data and train our models in four stages: vision pre-training, general supervised fine-tuning (SFT), Physical AI SFT, and Physical AI reinforcement learning (RL) as the post-training. To evaluate our models, we build comprehensive benchmarks for physical common sense and embodied reasoning according to our ontologies. Evaluation results show that Physical AI SFT and reinforcement learning bring significant improvements. To facilitate the development of Physical AI, we will make our code and pre-trained models available under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-reason1.
Authors: Nicol\'as Laverde, Melissa Robles, Johan Rodr\'iguez
Abstract: We present Shap-MeD, a text-to-3D object generative model specialized in the biomedical domain. The objective of this study is to develop an assistant that facilitates the 3D modeling of medical objects, thereby reducing development time. 3D modeling in medicine has various applications, including surgical procedure simulation and planning, the design of personalized prosthetic implants, medical education, the creation of anatomical models, and the development of research prototypes. To achieve this, we leverage Shap-e, an open-source text-to-3D generative model developed by OpenAI, and fine-tune it using a dataset of biomedical objects. Our model achieved a mean squared error (MSE) of 0.089 in latent generation on the evaluation set, compared to Shap-e's MSE of 0.147. Additionally, we conducted a qualitative evaluation, comparing our model with others in the generation of biomedical objects. Our results indicate that Shap-MeD demonstrates higher structural accuracy in biomedical object generation.
Authors: Alba M\'arquez-Rodr\'iguez, Miguel \'Angel Mohedano-Munoz, Manuel J. Mar\'in-Jim\'enez, Eduardo Santamar\'ia-Garc\'ia, Giulia Bastianelli, Pedro Jordano, Irene Mendoza
Abstract: Passive Acoustic Monitoring with automatic recorders is essential for ecosystem conservation but generates vast unsupervised audio data, posing challenges for extracting meaningful information. Deep Learning techniques offer a promising solution. BirdNET, a widely used model for bird identification, has shown success in many study systems but is limited in some regions due to biases in its training data. A key challenge in bird species detection is that many recordings either lack target species or contain overlapping vocalizations. To overcome these problems, we developed a multi-stage pipeline for automatic bird vocalization identification in Do\~nana National Park (SW Spain), a region facing significant conservation threats. Our approach included a Bird Song Detector to isolate vocalizations and custom classifiers trained with BirdNET embeddings. We manually annotated 461 minutes of audio from three habitats across nine locations, yielding 3,749 annotations for 34 classes. Spectrograms facilitated the use of image processing techniques. Applying the Bird Song Detector before classification improved species identification, as all classification models performed better when analyzing only the segments where birds were detected. Specifically, the combination of the Bird Song Detector and fine-tuned BirdNET compared to the baseline without the Bird Song Detector. Our approach demonstrated the effectiveness of integrating a Bird Song Detector with fine-tuned classification models for bird identification at local soundscapes. These findings highlight the need to adapt general-purpose tools for specific ecological challenges, as demonstrated in Do\~nana. Automatically detecting bird species serves for tracking the health status of this threatened ecosystem, given the sensitivity of birds to environmental changes, and helps in the design of conservation measures for reducing biodiversity loss
Authors: Zeqi Gu, Difan Liu, Timothy Langlois, Matthew Fisher, Abe Davis
Abstract: Recent diffusion-based methods have achieved impressive results on animating images of human subjects. However, most of that success has built on human-specific body pose representations and extensive training with labeled real videos. In this work, we extend the ability of such models to animate images of characters with more diverse skeletal topologies. Given a small number (3-5) of example frames showing the character in different poses with corresponding skeletal information, our model quickly infers a rig for that character that can generate images corresponding to new skeleton poses. We propose a procedural data generation pipeline that efficiently samples training data with diverse topologies on the fly. We use it, along with a novel skeleton representation, to train our model on articulated shapes spanning a large space of textures and topologies. Then during fine-tuning, our model rapidly adapts to unseen target characters and generalizes well to rendering new poses, both for realistic and more stylized cartoon appearances. To better evaluate performance on this novel and challenging task, we create the first 2D video dataset that contains both humanoid and non-humanoid subjects with per-frame keypoint annotations. With extensive experiments, we demonstrate the superior quality of our results. Project page: https://traindragondiffusion.github.io/
Authors: Ragendhu Sp, Tony Thomas, Sabu Emmanuel
Abstract: Cancelable biometric schemes are designed to extract an identity-preserving, non-invertible as well as revocable pseudo-identifier from biometric data. Recognition systems need to store only this pseudo-identifier, to avoid tampering and/or stealing of original biometric data during the recognition process. State-of-the-art cancelable schemes generate pseudo-identifiers by transforming the original template using either user-specific salting or many-to-one transformations. In addition to the performance concerns, most of such schemes are modality-specific and prone to reconstruction attacks as there are chances for unauthorized access to security-critical transformation keys. A novel, modality-independent cancelable biometric scheme is proposed to overcome these limitations. In this scheme, a cancelable template (pseudo identifier) is generated as a distance vector between multiple random transformations of the biometric feature vector. These transformations were done by grouping feature vector components based on a set of user-specific random vectors. The proposed scheme nullifies the possibility of template reconstruction as the generated cancelable template contains only the distance values between the different random transformations of the feature vector and it does not store any details of the biometric template. The recognition performance of the proposed scheme is evaluated for face and fingerprint modalities. Equal Error Rate (EER) of 1.5 is obtained for face and 1.7 is obtained for the fingerprint in the worst case.
Authors: Bingxuan Li, Jiahao Wu, Yuan Xu, Yunxiang Zhang, Zezheng Zhu, Nanfang Yu, Qi Sun
Abstract: Depth imaging is a foundational building block for broad applications, such as autonomous driving and virtual/augmented reality. Traditionally, depth cameras have relied on time-of-flight sensors or multi-lens systems to achieve physical depth measurements. However, these systems often face a trade-off between a bulky form factor and imprecise approximations, limiting their suitability for spatially constrained scenarios. Inspired by the emerging advancements of nano-optics, we present Nano-3D, a metasurface-based neural depth imaging solution with an ultra-compact footprint. Nano-3D integrates our custom-fabricated 700 nm thick TiO2 metasurface with a multi-module deep neural network to extract precise metric depth information from monocular metasurface-polarized imagery. We demonstrate the effectiveness of Nano-3D with both simulated and physical experiments. We hope the exhibited success paves the way for the community to bridge future graphics systems with emerging nanomaterial technologies through novel computational approaches.
Authors: Yuci Han, Charles Toth, Alper Yilmaz
Abstract: The aim of this work is to develop an approach that enables Unmanned Aerial System (UAS) to efficiently learn to navigate in large-scale urban environments and transfer their acquired expertise to novel environments. To achieve this, we propose a meta-curriculum training scheme. First, meta-training allows the agent to learn a master policy to generalize across tasks. The resulting model is then fine-tuned on the downstream tasks. We organize the training curriculum in a hierarchical manner such that the agent is guided from coarse to fine towards the target task. In addition, we introduce Incremental Self-Adaptive Reinforcement learning (ISAR), an algorithm that combines the ideas of incremental learning and meta-reinforcement learning (MRL). In contrast to traditional reinforcement learning (RL), which focuses on acquiring a policy for a specific task, MRL aims to learn a policy with fast transfer ability to novel tasks. However, the MRL training process is time consuming, whereas our proposed ISAR algorithm achieves faster convergence than the conventional MRL algorithm. We evaluate the proposed methodologies in simulated environments and demonstrate that using this training philosophy in conjunction with the ISAR algorithm significantly improves the convergence speed for navigation in large-scale cities and the adaptation proficiency in novel environments.
Authors: Xuan Gao, Jingtao Zhou, Dongyu Liu, Yuqi Zhou, Juyong Zhang
Abstract: Recent advances in diffusion models have made significant progress in digital human generation. However, most existing models still struggle to maintain 3D consistency, temporal coherence, and motion accuracy. A key reason for these shortcomings is the limited representation ability of commonly used control signals(e.g., landmarks, depth maps, etc.). In addition, the lack of diversity in identity and pose variations in public datasets further hinders progress in this area. In this paper, we analyze the shortcomings of current control signals and introduce a novel control signal representation that is optimizable, dense, expressive, and 3D consistent. Our method embeds a learnable neural Gaussian onto a parametric head surface, which greatly enhances the consistency and expressiveness of diffusion-based head models. Regarding the dataset, we synthesize a large-scale dataset with multiple poses and identities. In addition, we use real/synthetic labels to effectively distinguish real and synthetic data, minimizing the impact of imperfections in synthetic data on the generated head images. Extensive experiments show that our model outperforms existing methods in terms of realism, expressiveness, and 3D consistency. Our code, synthetic datasets, and pre-trained models will be released in our project page: https://ustc3dv.github.io/Learn2Control/
Authors: Jie Gan, Zhuonan Liang, Jianan Fan, Lisa Mcguire, Caterina Watson, Jacqueline Spurway, Jillian Clarke, Weidong Cai
Abstract: The intrapartum ultrasound guideline established by ISUOG highlights the Angle of Progression (AoP) and Head Symphysis Distance (HSD) as pivotal metrics for assessing fetal head descent and predicting delivery outcomes. Accurate measurement of the AoP and HSD requires a structured process. This begins with identifying standardized ultrasound planes, followed by the detection of specific anatomical landmarks within the regions of the pubic symphysis and fetal head that correlate with the delivery parameters AoP and HSD. Finally, these measurements are derived based on the identified anatomical landmarks. Addressing the clinical demands and standard operation process outlined in the ISUOG guideline, we introduce the Sequential Spatial-Temporal Network (SSTN), the first interpretable model specifically designed for the video of intrapartum ultrasound analysis. The SSTN operates by first identifying ultrasound planes, then segmenting anatomical structures such as the pubic symphysis and fetal head, and finally detecting key landmarks for precise measurement of HSD and AoP. Furthermore, the cohesive framework leverages task-related information to improve accuracy and reliability. Experimental evaluations on clinical datasets demonstrate that SSTN significantly surpasses existing models, reducing the mean absolute error by 18% for AoP and 22% for HSD.
Authors: Zeqi Zheng, Yanchen Huang, Yingchao Yu, Zizheng Zhu, Junfeng Tang, Zhaofei Yu, Yaochu Jin
Abstract: Spiking Neural Networks (SNNs) based on Transformers have garnered significant attention due to their superior performance and high energy efficiency. However, the spiking attention modules of most existing Transformer-based SNNs are adapted from those of analog Transformers, failing to fully address the issue of over-allocating attention to irrelevant contexts. To fix this fundamental yet overlooked issue, we propose a Lateral Inhibition-inspired Spiking Transformer (SpiLiFormer). It emulates the brain's lateral inhibition mechanism, guiding the model to enhance attention to relevant tokens while suppressing attention to irrelevant ones. Our model achieves state-of-the-art (SOTA) performance across multiple datasets, including CIFAR-10 (+0.45%), CIFAR-100 (+0.48%), CIFAR10-DVS (+2.70%), N-Caltech101 (+1.94%), and ImageNet-1K (+1.6%). Notably, on the ImageNet-1K dataset, SpiLiFormer (69.9M parameters, 4 time steps, 384 resolution) outperforms E-SpikeFormer (173.0M parameters, 8 time steps, 384 resolution), a SOTA spiking Transformer, by 0.46% using only 39% of the parameters and half the time steps. Our code and training checkpoints will be released upon acceptance.
Authors: Marc Bened\'i San Mill\'an, Angela Dai, Matthias Nie{\ss}ner
Abstract: Animation of humanoid characters is essential in various graphics applications, but requires significant time and cost to create realistic animations. We propose an approach to synthesize 4D animated sequences of input static 3D humanoid meshes, leveraging strong generalized motion priors from generative video models -- as such video models contain powerful motion information covering a wide variety of human motions. From an input static 3D humanoid mesh and a text prompt describing the desired animation, we synthesize a corresponding video conditioned on a rendered image of the 3D mesh. We then employ an underlying SMPL representation to animate the corresponding 3D mesh according to the video-generated motion, based on our motion optimization. This enables a cost-effective and accessible solution to enable the synthesis of diverse and realistic 4D animations.
Authors: Xiaomeng Chu, Jiajun Deng, Guoliang You, Wei Liu, Xingchen Li, Jianmin Ji, Yanyong Zhang
Abstract: Flexible instruction-guided 6-DoF grasping is a significant yet challenging task for real-world robotic systems. Existing methods utilize the contextual understanding capabilities of the large language models (LLMs) to establish mappings between expressions and targets, allowing robots to comprehend users' intentions in the instructions. However, the LLM's knowledge about objects' physical properties remains underexplored despite its tight relevance to grasping. In this work, we propose GraspCoT, a 6-DoF grasp detection framework that integrates a Chain-of-Thought (CoT) reasoning mechanism oriented to physical properties, guided by auxiliary question-answering (QA) tasks. Particularly, we design a set of QA templates to enable hierarchical reasoning that includes three stages: target parsing, physical property analysis, and grasp action selection. Moreover, GraspCoT presents a unified multimodal LLM architecture, which encodes multi-view observations of 3D scenes into 3D-aware visual tokens, and then jointly embeds these visual tokens with CoT-derived textual tokens within LLMs to generate grasp pose predictions. Furthermore, we present IntentGrasp, a large-scale benchmark that fills the gap in public datasets for multi-object grasp detection under diverse and indirect verbal commands. Extensive experiments on IntentGrasp demonstrate the superiority of our method, with additional validation in real-world robotic applications confirming its practicality. Codes and data will be released.
Authors: Abdelrahman Elsayed, Sarim Hashmi, Mohammed Elseiagy, Hu Wang, Mohammad Yaqub, Ibrahim Almakky
Abstract: The complex nature of medical image segmentation calls for models that are specifically designed to capture detailed, domain-specific features. Large foundation models offer considerable flexibility, yet the cost of fine-tuning these models remains a significant barrier. Parameter-Efficient Fine-Tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA), efficiently update model weights with low-rank matrices but may suffer from underfitting when the chosen rank is insufficient to capture domain-specific nuances. Conversely, full-rank Singular Value Decomposition (SVD) based methods provide comprehensive updates by modifying all singular values, yet they often lack flexibility and exhibit variable performance across datasets. We propose SALT (Singular Value Adaptation with Low-Rank Transformation), a method that selectively adapts the most influential singular values using trainable scale and shift parameters while complementing this with a low-rank update for the remaining subspace. This hybrid approach harnesses the advantages of both LoRA and SVD, enabling effective adaptation without relying on increasing model size or depth. Evaluated on 5 challenging medical datasets, ranging from as few as 20 samples to 1000, SALT outperforms state-of-the-art PEFT (LoRA and SVD) by 2% to 5% in Dice with only 3.9% trainable parameters, demonstrating robust adaptation even in low-resource settings. The code for SALT is available at: https://github.com/BioMedIA-MBZUAI/SALT
Authors: Marek Wodzinski, Henning M\"uller
Abstract: Lightsheet microscopy is a powerful 3-D imaging technique that addresses limitations of traditional optical and confocal microscopy but suffers from a low penetration depth and reduced image quality at greater depths. Multiview lightsheet microscopy improves 3-D resolution by combining multiple views but simultaneously increasing the complexity and the photon budget, leading to potential photobleaching and phototoxicity. The FuseMyCells challenge, organized in conjunction with the IEEE ISBI 2025 conference, aims to benchmark deep learning-based solutions for fusing high-quality 3-D volumes from single 3-D views, potentially simplifying procedures and conserving the photon budget. In this work, we propose a contribution to the FuseMyCells challenge based on a two-step procedure. The first step processes a downsampled version of the image to capture the entire region of interest, while the second step uses a patch-based approach for high-resolution inference, incorporating adversarial loss to enhance visual outcomes. This method addresses challenges related to high data resolution, the necessity of global context, and the preservation of high-frequency details. Experimental results demonstrate the effectiveness of our approach, highlighting its potential to improve 3-D image fusion quality and extend the capabilities of lightsheet microscopy. The average SSIM for the nucleus and membranes is greater than 0.85 and 0.91, respectively.
Authors: Dong Chen, Boyue Zhao, Yi Zhang, Meng Zhao
Abstract: Efficient modal feature fusion strategy is the key to achieve accurate segmentation of brain glioma. However, due to the specificity of different MRI modes, it is difficult to carry out cross-modal fusion with large differences in modal features, resulting in the model ignoring rich feature information. On the other hand, the problem of multi-modal feature redundancy interaction occurs in parallel networks due to the proliferation of feature dimensions, further increase the difficulty of multi-modal feature fusion at the bottom end. In order to solve the above problems, we propose a noval complementary feature compression interaction network (CFCI-Net), which realizes the complementary fusion and compression interaction of multi-modal feature information with an efficient mode fusion strategy. Firstly, we propose a selective complementary feature fusion (SCFF) module, which adaptively fuses rich cross-modal feature information by complementary soft selection weights. Secondly, a modal feature compression interaction (MFCI) transformer is proposed to deal with the multi-mode fusion redundancy problem when the feature dimension surges. The MFCI transformer is composed of modal feature compression (MFC) and modal feature interaction (MFI) to realize redundancy feature compression and multi-mode feature interactive learning. %In MFI, we propose a hierarchical interactive attention mechanism based on multi-head attention. Evaluations on the BraTS2019 and BraTS2020 datasets demonstrate that CFCI-Net achieves superior results compared to state-of-the-art models. Code: https://github.com/CDmm0/CFCI-Net
Authors: Shiyong Liu, Xiao Tang, Zhihao Li, Yingfan He, Chongjie Ye, Jianzhuang Liu, Binxiao Huang, Shunbo Zhou, Xiaofei Wu
Abstract: In large-scale scene reconstruction using 3D Gaussian splatting, it is common to partition the scene into multiple smaller regions and reconstruct them individually. However, existing division methods are occlusion-agnostic, meaning that each region may contain areas with severe occlusions. As a result, the cameras within those regions are less correlated, leading to a low average contribution to the overall reconstruction. In this paper, we propose an occlusion-aware scene division strategy that clusters training cameras based on their positions and co-visibilities to acquire multiple regions. Cameras in such regions exhibit stronger correlations and a higher average contribution, facilitating high-quality scene reconstruction. We further propose a region-based rendering technique to accelerate large scene rendering, which culls Gaussians invisible to the region where the viewpoint is located. Such a technique significantly speeds up the rendering without compromising quality. Extensive experiments on multiple large scenes show that our method achieves superior reconstruction results with faster rendering speed compared to existing state-of-the-art approaches. Project page: https://occlugaussian.github.io.
Authors: Teresa Klatzer, Savvas Melidonis, Marcelo Pereyra, Konstantinos C. Zygalakis
Abstract: This paper introduces a novel plug-and-play (PnP) Langevin sampling methodology for Bayesian inference in low-photon Poisson imaging problems, a challenging class of problems with significant applications in astronomy, medicine, and biology. PnP Langevin sampling algorithms offer a powerful framework for Bayesian image restoration, enabling accurate point estimation as well as advanced inference tasks, including uncertainty quantification and visualization analyses, and empirical Bayesian inference for automatic model parameter tuning. However, existing PnP Langevin algorithms are not well-suited for low-photon Poisson imaging due to high solution uncertainty and poor regularity properties, such as exploding gradients and non-negativity constraints. To address these challenges, we propose two strategies for extending Langevin PnP sampling to Poisson imaging models: (i) an accelerated PnP Langevin method that incorporates boundary reflections and a Poisson likelihood approximation and (ii) a mirror sampling algorithm that leverages a Riemannian geometry to handle the constraints and the poor regularity of the likelihood without approximations. The effectiveness of these approaches is demonstrated through extensive numerical experiments and comparisons with state-of-the-art methods.
Authors: Dawood Wasif, Terrence J. Moore, Jin-Hee Cho
Abstract: Autonomous vehicles (AVs) increasingly rely on Federated Learning (FL) to enhance perception models while preserving privacy. However, existing FL frameworks struggle to balance privacy, fairness, and robustness, leading to performance disparities across demographic groups. Privacy-preserving techniques like differential privacy mitigate data leakage risks but worsen fairness by restricting access to sensitive attributes needed for bias correction. This work explores the trade-off between privacy and fairness in FL-based object detection for AVs and introduces RESFL, an integrated solution optimizing both. RESFL incorporates adversarial privacy disentanglement and uncertainty-guided fairness-aware aggregation. The adversarial component uses a gradient reversal layer to remove sensitive attributes, reducing privacy risks while maintaining fairness. The uncertainty-aware aggregation employs an evidential neural network to weight client updates adaptively, prioritizing contributions with lower fairness disparities and higher confidence. This ensures robust and equitable FL model updates. We evaluate RESFL on the FACET dataset and CARLA simulator, assessing accuracy, fairness, privacy resilience, and robustness under varying conditions. RESFL improves detection accuracy, reduces fairness disparities, and lowers privacy attack success rates while demonstrating superior robustness to adversarial conditions compared to other approaches.
Authors: Dounia Hammou, Yancheng Cai, Pavan Madhusudanarao, Christos G. Bampis, Rafa{\l} K. Mantiuk
Abstract: Image and video quality metrics, such as SSIM, LPIPS, and VMAF, are aimed to predict the perceived quality of the evaluated content and are often claimed to be "perceptual". Yet, few metrics directly model human visual perception, and most rely on hand-crafted formulas or training datasets to achieve alignment with perceptual data. In this paper, we propose a set of tests for full-reference quality metrics that examine their ability to model several aspects of low-level human vision: contrast sensitivity, contrast masking, and contrast matching. The tests are meant to provide additional scrutiny for newly proposed metrics. We use our tests to analyze 33 existing image and video quality metrics and find their strengths and weaknesses, such as the ability of LPIPS and MS-SSIM to predict contrast masking and poor performance of VMAF in this task. We further find that the popular SSIM metric overemphasizes differences in high spatial frequencies, but its multi-scale counterpart, MS-SSIM, addresses this shortcoming. Such findings cannot be easily made using existing evaluation protocols.
Authors: Vivek Gopalakrishnan, Neel Dey, David-Dimitris Chlorogiannis, Andrew Abumoussa, Anna M. Larson, Darren B. Orbach, Sarah Frisken, Polina Golland
Abstract: The integration of artificial intelligence in image-guided interventions holds transformative potential, promising to extract 3D geometric and quantitative information from conventional 2D imaging modalities during complex procedures. Achieving this requires the rapid and precise alignment of 2D intraoperative images (e.g., X-ray) with 3D preoperative volumes (e.g., CT, MRI). However, current 2D/3D registration methods fail across the broad spectrum of procedures dependent on X-ray guidance: traditional optimization techniques require custom parameter tuning for each subject, whereas neural networks trained on small datasets do not generalize to new patients or require labor-intensive manual annotations, increasing clinical burden and precluding application to new anatomical targets. To address these challenges, we present xvr, a fully automated framework for training patient-specific neural networks for 2D/3D registration. xvr uses physics-based simulation to generate abundant high-quality training data from a patient's own preoperative volumetric imaging, thereby overcoming the inherently limited ability of supervised models to generalize to new patients and procedures. Furthermore, xvr requires only 5 minutes of training per patient, making it suitable for emergency interventions as well as planned procedures. We perform the largest evaluation of a 2D/3D registration algorithm on real X-ray data to date and find that xvr robustly generalizes across a diverse dataset comprising multiple anatomical structures, imaging modalities, and hospitals. Across surgical tasks, xvr achieves submillimeter-accurate registration at intraoperative speeds, improving upon existing methods by an order of magnitude. xvr is released as open-source software freely available at https://github.com/eigenvivek/xvr.
Authors: Yunzhi Yao, Jizhan Fang, Jia-Chen Gu, Ningyu Zhang, Shumin Deng, Huajun Chen, Nanyun Peng
Abstract: Knowledge Editing (KE) enables the modification of outdated or incorrect information in large language models (LLMs). While existing KE methods can update isolated facts, they struggle to generalize these updates to multi-hop reasoning tasks that depend on the modified knowledge. Through an analysis of reasoning circuits -- the neural pathways LLMs use for knowledge-based inference, we observe that current layer-localized KE approaches, such as MEMIT and WISE, which edit only single or a few model layers, struggle to effectively incorporate updated information into these reasoning pathways. To address this limitation, we propose CaKE (Circuit-aware Knowledge Editing), a novel method that enables more effective integration of updated knowledge in LLMs. CaKE leverages strategically curated data, guided by our circuits-based analysis, that enforces the model to utilize the modified knowledge, stimulating the model to develop appropriate reasoning circuits for newly integrated knowledge. Experimental results show that CaKE enables more accurate and consistent use of updated knowledge across related reasoning tasks, leading to an average of 20% improvement in multi-hop reasoning accuracy on MQuAKE dataset compared to existing KE methods. We release the code and data in https://github.com/zjunlp/CaKE.
Authors: Kristin Qi, Xinhan Di
Abstract: Retinal Optical Coherence Tomography (OCT) segmentation is essential for diagnosing pathology. Traditional methods focus on either spatial or spectral domains, overlooking their combined dependencies. We propose a triple-encoder network that integrates CNNs for spatial features, Fast Fourier Convolution (FFC) for spectral features, and attention mechanisms to capture global relationships across both domains. Attention fusion modules integrate convolution and cross-attention to further enhance features. Our method achieves an average Dice score improvement from 0.855 to 0.864, outperforming prior work.
Authors: SeungJu Cha, Kwanyoung Lee, Ye-Chan Kim, Hyunwoo Oh, Dong-Jin Kim
Abstract: Recent large-scale text-to-image diffusion models generate photorealistic images but often struggle to accurately depict interactions between humans and objects due to their limited ability to differentiate various interaction words. In this work, we propose VerbDiff to address the challenge of capturing nuanced interactions within text-to-image diffusion models. VerbDiff is a novel text-to-image generation model that weakens the bias between interaction words and objects, enhancing the understanding of interactions. Specifically, we disentangle various interaction words from frequency-based anchor words and leverage localized interaction regions from generated images to help the model better capture semantics in distinctive words without extra conditions. Our approach enables the model to accurately understand the intended interaction between humans and objects, producing high-quality images with accurate interactions aligned with specified verbs. Extensive experiments on the HICO-DET dataset demonstrate the effectiveness of our method compared to previous approaches.
Authors: Yiran Qin, Li Kang, Xiufeng Song, Zhenfei Yin, Xiaohong Liu, Xihui Liu, Ruimao Zhang, Lei Bai
Abstract: Designing effective embodied multi-agent systems is critical for solving complex real-world tasks across domains. Due to the complexity of multi-agent embodied systems, existing methods fail to automatically generate safe and efficient training data for such systems. To this end, we propose the concept of compositional constraints for embodied multi-agent systems, addressing the challenges arising from collaboration among embodied agents. We design various interfaces tailored to different types of constraints, enabling seamless interaction with the physical world. Leveraging compositional constraints and specifically designed interfaces, we develop an automated data collection framework for embodied multi-agent systems and introduce the first benchmark for embodied multi-agent manipulation, RoboFactory. Based on RoboFactory benchmark, we adapt and evaluate the method of imitation learning and analyzed its performance in different difficulty agent tasks. Furthermore, we explore the architectures and training strategies for multi-agent imitation learning, aiming to build safe and efficient embodied multi-agent systems.
Authors: Xi Liu, Chaoyi Zhou, Nanxuan Zhao, Siyu Huang
Abstract: Differentiable vector graphics (VGs) are widely used in image vectorization and vector synthesis, while existing representations are costly to optimize and struggle to achieve high-quality rendering results for high-resolution images. This work introduces a new differentiable VG representation, dubbed B\'ezier splatting, that enables fast yet high-fidelity VG rasterization. B\'ezier splatting samples 2D Gaussians along B\'ezier curves, which naturally provide positional gradients at object boundaries. Thanks to the efficient splatting-based differentiable rasterizer, B\'ezier splatting achieves over 20x and 150x faster per forward and backward rasterization step for open curves compared to DiffVG. Additionally, we introduce an adaptive pruning and densification strategy that dynamically adjusts the spatial distribution of curves to escape local minima, further improving VG quality. Experimental results show that B\'ezier splatting significantly outperforms existing methods with better visual fidelity and 10x faster optimization speed.
Authors: Ruyi Xu, Guangxuan Xiao, Haofeng Huang, Junxian Guo, Song Han
Abstract: Long-Context Transformer Models (LCTMs) are vital for real-world applications but suffer high computational costs due to attention's quadratic complexity. Block-sparse attention mitigates this by focusing computation on critical regions, yet existing methods struggle with balancing accuracy and efficiency due to costly block importance measurements. In this paper, we introduce XAttention, a plug-and-play framework that dramatically accelerates long-context inference in Transformers models using sparse attention. XAttention's key innovation is the insight that the sum of antidiagonal values (i.e., from the lower-left to upper-right) in the attention matrix provides a powerful proxy for block importance. This allows for precise identification and pruning of non-essential blocks, resulting in high sparsity and dramatically accelerated inference. Across comprehensive evaluations on demanding long-context benchmarks-including RULER and LongBench for language, VideoMME for video understanding, and VBench for video generation. XAttention achieves accuracy comparable to full attention while delivering substantial computational gains. We demonstrate up to 13.5x acceleration in attention computation. These results underscore XAttention's ability to unlock the practical potential of block sparse attention, paving the way for scalable and efficient deployment of LCTMs in real-world applications. Code is available at https://github.com/mit-han-lab/x-attention.
Authors: Jiayi Su, Xingbin Tu, Fengzhong Qu, Yan Wei
Abstract: Forward-Looking Sonar (FLS) has started to gain attention in the field of near-bottom close-range underwater inspection because of its high resolution and high framerate features. Although Automatic Target Recognition (ATR) algorithms have been applied tentatively for object-searching tasks, human supervision is still indispensable, especially when involving critical areas. A clear FLS mosaic containing all suspicious information is in demand to help experts deal with tremendous perception data. However, previous work only considered that FLS is working in an ideal system configuration, which assumes an appropriate sonar imaging setup and the availability of accurate positioning data. Without those promises, the intra-frame and inter-frame artifacts will appear and degrade the quality of the final mosaic by making the information of interest invisible. In this paper, we propose a novel blending method for FLS mosaicing which can preserve interested information. A Long-Short Time Sliding Window (LST-SW) is designed to rectify the local statistics of raw sonar images. The statistics are then utilized to construct a Global Variance Map (GVM). The GVM helps to emphasize the useful information contained in images in the blending phase by classifying the informative and featureless pixels, thereby enhancing the quality of final mosaic. The method is verified using data collected in the real environment. The results show that our method can preserve more details in FLS mosaics for human inspection purposes in practice.
Authors: Jiamian Wang, Zongliang Wu, Yulun Zhang, Xin Yuan, Tao Lin, Zhiqiang Tao
Abstract: Existing reconstruction models in snapshot compressive imaging systems (SCI) are trained with a single well-calibrated hardware instance, making their performance vulnerable to hardware shifts and limited in adapting to multiple hardware configurations. To facilitate cross-hardware learning, previous efforts attempt to directly collect multi-hardware data and perform centralized training, which is impractical due to severe user data privacy concerns and hardware heterogeneity across different platforms/institutions. In this study, we explicitly consider data privacy and heterogeneity in cooperatively optimizing SCI systems by proposing a Federated Hardware-Prompt learning (FedHP) framework. Rather than mitigating the client drift by rectifying the gradients, which only takes effect on the learning manifold but fails to solve the heterogeneity rooted in the input data space, FedHP learns a hardware-conditioned prompter to align inconsistent data distribution across clients, serving as an indicator of the data inconsistency among different hardware (e.g., coded apertures). Extensive experimental results demonstrate that the proposed FedHP coordinates the pre-trained model to multiple hardware configurations, outperforming prevalent FL frameworks for 0.35dB under challenging heterogeneous settings. Moreover, a Snapshot Spectral Heterogeneous Dataset has been built upon multiple practical SCI systems. Data and code are aveilable at https://github.com/Jiamian-Wang/FedHP-Snapshot-Compressive-Imaging
URLs: https://github.com/Jiamian-Wang/FedHP-Snapshot-Compressive-Imaging
Authors: Pu Cao, Feng Zhou, Lu Yang, Tianrui Huang, Qing Song
Abstract: In-domain generation aims to perform a variety of tasks within a specific domain, such as unconditional generation, text-to-image, image editing, 3D generation, and more. Early research typically required training specialized generators for each unique task and domain, often relying on fully-labeled data. Motivated by the powerful generative capabilities and broad applications of diffusion models, we are driven to explore leveraging label-free data to empower these models for in-domain generation. Fine-tuning a pre-trained generative model on domain data is an intuitive but challenging way and often requires complex manual hyper-parameter adjustments since the limited diversity of the training data can easily disrupt the model's original generative capabilities. To address this challenge, we propose a guidance-decoupled prior preservation mechanism to achieve high generative quality and controllability by image-only data, inspired by preserving the pre-trained model from a denoising guidance perspective. We decouple domain-related guidance from the conditional guidance used in classifier-free guidance mechanisms to preserve open-world control guidance and unconditional guidance from the pre-trained model. We further propose an efficient domain knowledge learning technique to train an additional text-free UNet copy to predict domain guidance. Besides, we theoretically illustrate a multi-guidance in-domain generation pipeline for a variety of generative tasks, leveraging multiple guidances from distinct diffusion models and conditions. Extensive experiments demonstrate the superiority of our method in domain-specific synthesis and its compatibility with various diffusion-based control methods and applications.
Authors: Shoubin Yu, Jaehong Yoon, Mohit Bansal
Abstract: Despite impressive advancements in recent multimodal reasoning approaches, they are still limited in flexibility and efficiency, as these models typically process only a few fixed modality inputs and require updates to numerous parameters. This paper tackles these critical challenges and proposes CREMA, a generalizable, highly efficient, and modular modality-fusion framework that can incorporate any new modality to enhance video reasoning. We first augment multiple informative modalities (such as optical flow, 3D point cloud, audio, thermal heatmap, and touch map) from given videos without extra human annotation by leveraging sensors or existing pre-trained models. Next, we introduce a query transformer with multiple parameter-efficient modules associated with each accessible modality. It projects diverse modality features to the LLM token embedding space, allowing the model to integrate different data types for response generation. Furthermore, we propose a novel progressive multimodal fusion design supported by a lightweight fusion module and modality-sequential training strategy. It helps compress information across various assisting modalities, maintaining computational efficiency in the LLM while improving performance. We validate our method on 7 video-language reasoning tasks assisted by diverse modalities, including conventional VideoQA and Video-Audio/3D/Touch/Thermal QA, and achieve better/equivalent performance against strong multimodal LLMs, including OneLLM, BLIP-2, and SeViLA while reducing over 90% trainable parameters. We provide extensive analyses of CREMA, including the impact of each modality on reasoning domains, the design of the fusion module, and example visualizations.
Authors: Yongzhi Huang, Fengjun Xi, Liyun Tu, Jinxin Zhu, Haseeb Hassan, Liyilei Su, Yun Peng, Jingyu Li, Jun Ma, Bingding Huang
Abstract: Accurate segmentation of multiple organs in Computed Tomography (CT) images plays a vital role in computer-aided diagnosis systems. While various supervised learning approaches have been proposed recently, these methods heavily depend on a large amount of high-quality labeled data, which are expensive to obtain in practice. To address this challenge, we propose a label-efficient framework using knowledge transfer from a pre-trained diffusion model for CT multi-organ segmentation. Specifically, we first pre-train a denoising diffusion model on 207,029 unlabeled 2D CT slices to capture anatomical patterns. Then, the model backbone is transferred to the downstream multi-organ segmentation task, followed by fine-tuning with few labeled data. In fine-tuning, two fine-tuning strategies, linear classification and fine-tuning decoder, are employed to enhance segmentation performance while preserving learned representations. Quantitative results show that the pre-trained diffusion model is capable of generating diverse and realistic 256x256 CT images (Fr\'echet inception distance (FID): 11.32, spatial Fr\'echet inception distance (sFID): 46.93, F1-score: 73.1%). Compared to state-of-the-art methods for multi-organ segmentation, our method achieves competitive performance on the FLARE 2022 dataset, particularly in limited labeled data scenarios. After fine-tuning with 1% and 10% labeled data, our method achieves dice similarity coefficients (DSCs) of 71.56% and 78.51%, respectively. Remarkably, the method achieves a DSC score of 51.81% using only four labeled CT slices. These results demonstrate the efficacy of our approach in overcoming the limitations of supervised learning approaches that is highly dependent on large-scale labeled data.
Authors: Jiayi Liu, Manolis Savva, Ali Mahdavi-Amiri
Abstract: 3D modeling of articulated objects is a research problem within computer vision, graphics, and robotics. Its objective is to understand the shape and motion of the articulated components, represent the geometry and mobility of object parts, and create realistic models that reflect articulated objects in the real world. This survey provides a comprehensive overview of the current state-of-the-art in 3D modeling of articulated objects, with a specific focus on the task of articulated part perception and articulated object creation (reconstruction and generation). We systematically review and discuss the relevant literature from two perspectives: geometry modeling (i.e., structure and shape of articulated parts) and articulation modeling (i.e., dynamics and motion of parts). Through this survey, we highlight the substantial progress made in these areas, outline the ongoing challenges, and identify gaps for future research. Our survey aims to serve as a foundational reference for researchers and practitioners in computer vision and graphics, offering insights into the complexities of articulated object modeling.
Authors: Navve Wasserman, Noam Rotstein, Roy Ganz, Ron Kimmel
Abstract: Image editing has advanced significantly with the introduction of text-conditioned diffusion models. Despite this progress, seamlessly adding objects to images based on textual instructions without requiring user-provided input masks remains a challenge. We address this by leveraging the insight that removing objects (Inpaint) is significantly simpler than its inverse process of adding them (Paint), attributed to inpainting models that benefit from segmentation mask guidance. Capitalizing on this realization, by implementing an automated and extensive pipeline, we curate a filtered large-scale image dataset containing pairs of images and their corresponding object-removed versions. Using these pairs, we train a diffusion model to inverse the inpainting process, effectively adding objects into images. Unlike other editing datasets, ours features natural target images instead of synthetic ones while ensuring source-target consistency by construction. Additionally, we utilize a large Vision-Language Model to provide detailed descriptions of the removed objects and a Large Language Model to convert these descriptions into diverse, natural-language instructions. Our quantitative and qualitative results show that the trained model surpasses existing models in both object addition and general editing tasks. Visit our project page for the released dataset and trained models at https://rotsteinnoam.github.io/Paint-by-Inpaint.
Authors: Yachan Guo, Yi Xiao, Danna Xue, Jose Luis Gomez Zurita, Antonio M. Lopez
Abstract: Instance segmentation is crucial for autonomous driving but is hindered by the lack of annotated real-world data due to expensive labeling costs. Unsupervised Domain Adaptation (UDA) offers a solution by transferring knowledge from labeled synthetic data to unlabeled real-world data. While UDA methods for synthetic to real-world domains (synth-to-real) show remarkable performance in tasks such as semantic segmentation and object detection, very few have been proposed for instance segmentation in vision-based autonomous driving. Moreover, existing methods rely on suboptimal baselines, which severely limits performance. We introduce \textbf{UDA4Inst}, a powerful framework for synth-to-real UDA in instance segmentation. Our framework enhances instance segmentation through \textit{Semantic Category Training} and \textit{Bidirectional Mixing Training}. With the Semantic Category Training method, semantically related classes are grouped and trained separately, enabling the generation of higher-quality pseudo-labels and improved segmentation performance. We further propose a bidirectional cross-domain data mixing strategy that combines instance-wise and patch-wise mixing techniques to effectively utilize data from both source and target domains, producing realistic composite images that improve the model's generalization performance. Extensive experiments demonstrate the effectiveness of our methods. Our approach establishes a new state-of-the-art on the SYNTHIA->Cityscapes benchmark with mAP 31.3. Notably, we are the first to report results on multiple novel synth-to-real instance segmentation datasets, using UrbanSyn and Synscapes as source domains while Cityscapes and KITTI360 serve as target domains. Our code will be released soon.
Authors: Erik Landolsi, Fredrik Kahl
Abstract: Few-shot learning deals with problems such as image classification using very few training examples. Recent vision foundation models show excellent few-shot transfer abilities, but are large and slow at inference. Using knowledge distillation, the capabilities of high-performing but slow models can be transferred to tiny, efficient models. However, common distillation methods require a large set of unlabeled data, which is not available in the few-shot setting. To overcome this lack of data, there has been a recent interest in using synthetic data. We expand on this line of research by presenting a novel diffusion model inversion technique (TINT) combining the diversity of textual inversion with the specificity of null-text inversion. Using this method in a few-shot distillation pipeline leads to state-of-the-art accuracy among small student models on popular benchmarks, while being significantly faster than prior work. Popular few-shot benchmarks involve evaluation over a large number of episodes, which is computationally cumbersome for methods involving synthetic data generation. We also present a theoretical analysis on how the accuracy estimator variance depends on the number of episodes and query examples, and use these results to lower the computational effort required for method evaluation. Finally, to further motivate the use of generative models in few-shot distillation, we demonstrate that our method outperforms training on real data mined from the dataset used in the original diffusion model training. Source code is available at https://github.com/pixwse/tiny2.
Authors: Han-Hung Lee, Yiming Zhang, Angel X. Chang
Abstract: We introduce Duoduo CLIP, a model for 3D representation learning that learns shape encodings from multi-view images instead of point clouds. The choice of multi-view images allows us to leverage 2D priors from off-the-shelf CLIP models to facilitate fine-tuning with 3D data. Our approach not only shows better generalization compared to existing point cloud methods, but also reduces GPU requirements and training time. In addition, the model is modified with cross-view attention to leverage information across multiple frames of the object which further boosts performance. Notably, our model is permutation invariant to the order of multi-view images while being pose-free. Compared to the current SOTA point cloud method that requires 480 A100 hours to train 1 billion model parameters we only require 57 A5000 hours and 87 million parameters. Multi-view images also provide more flexibility including being able to encode objects with a variable number of images, and performance scales when more views are used. In contrast, point cloud based methods require an entire scan or model of the object. We showcase this flexibility with benchmarks from images of real-world objects. Our model also achieves better performance in more fine-grained text to shape retrieval, demonstrating better text-and-shape alignment than point cloud based models.
Authors: Roman Beliy, Navve Wasserman, Amit Zalcher, Michal Irani
Abstract: Image-to-fMRI encoding is important for both neuroscience research and practical applications. However, such "Brain-Encoders" have been typically trained per-subject and per fMRI-dataset, thus restricted to very limited training data. In this paper we propose a Universal Brain-Encoder, which can be trained jointly on data from many different subjects/datasets/machines. What makes this possible is our new voxel-centric Encoder architecture, which learns a unique "voxel-embedding" per brain-voxel. Our Encoder trains to predict the response of each brain-voxel on every image, by directly computing the cross-attention between the brain-voxel embedding and multi-level deep image features. This voxel-centric architecture allows the functional role of each brain-voxel to naturally emerge from the voxel-image cross-attention. We show the power of this approach to (i) combine data from multiple different subjects (a "Crowd of Brains") to improve each individual brain-encoding, (ii) quick & effective Transfer-Learning across subjects, datasets, and machines (e.g., 3-Tesla, 7-Tesla), with few training examples, and (iii) use the learned voxel-embeddings as a powerful tool to explore brain functionality (e.g., what is encoded where in the brain).
Authors: Lijie Hu, Tianhao Huang, Huanyi Xie, Xilin Gong, Chenyang Ren, Zhengyu Hu, Lu Yu, Ping Ma, Di Wang
Abstract: Concept Bottleneck Models (CBMs) have garnered increasing attention due to their ability to provide concept-based explanations for black-box deep learning models while achieving high final prediction accuracy using human-like concepts. However, the training of current CBMs is heavily dependent on the precision and richness of the annotated concepts in the dataset. These concept labels are typically provided by experts, which can be costly and require significant resources and effort. Additionally, concept saliency maps frequently misalign with input saliency maps, causing concept predictions to correspond to irrelevant input features - an issue related to annotation alignment. To address these limitations, we propose a new framework called SSCBM (Semi-supervised Concept Bottleneck Model). Our SSCBM is suitable for practical situations where annotated data is scarce. By leveraging joint training on both labeled and unlabeled data and aligning the unlabeled data at the concept level, we effectively solve these issues. We proposed a strategy to generate pseudo labels and an alignment loss. Experiments demonstrate that our SSCBM is both effective and efficient. With only 10% labeled data, our model's concept and task accuracy on average across four datasets is only 2.44% and 3.93% lower, respectively, compared to the best baseline in the fully supervised learning setting.
Authors: Ankan Bhunia, Changjian Li, Hakan Bilen
Abstract: This paper introduces a novel anomaly detection (AD) problem aimed at identifying `odd-looking' objects within a scene by comparing them to other objects present. Unlike traditional AD benchmarks with fixed anomaly criteria, our task detects anomalies specific to each scene by inferring a reference group of regular objects. To address occlusions, we use multiple views of each scene as input, construct 3D object-centric models for each instance from 2D views, enhancing these models with geometrically consistent part-aware representations. Anomalous objects are then detected through cross-instance comparison. We also introduce two new benchmarks, ToysAD-8K and PartsAD-15K as testbeds for future research in this task. We provide a comprehensive analysis of our method quantitatively and qualitatively on these benchmarks.
Authors: Yusu Qian, Hanrong Ye, Jean-Philippe Fauconnier, Peter Grasch, Yinfei Yang, Zhe Gan
Abstract: We introduce MIA-Bench, a new benchmark designed to evaluate multimodal large language models (MLLMs) on their ability to strictly adhere to complex instructions. Our benchmark comprises a diverse set of 400 image-prompt pairs, each crafted to challenge the models' compliance with layered instructions in generating accurate responses that satisfy specific requested patterns. Evaluation results from a wide array of state-of-the-art MLLMs reveal significant variations in performance, highlighting areas for improvement in instruction fidelity. Additionally, we create extra training data and explore supervised fine-tuning to enhance the models' ability to strictly follow instructions without compromising performance on other tasks. We hope this benchmark not only serves as a tool for measuring MLLM adherence to instructions, but also guides future developments in MLLM training methods.
Authors: Wojciech Zielonka, Timo Bolkart, Thabo Beeler, Justus Thies
Abstract: Current personalized neural head avatars face a trade-off: lightweight models lack detail and realism, while high-quality, animatable avatars require significant computational resources, making them unsuitable for commodity devices. To address this gap, we introduce Gaussian Eigen Models (GEM), which provide high-quality, lightweight, and easily controllable head avatars. GEM utilizes 3D Gaussian primitives for representing the appearance combined with Gaussian splatting for rendering. Building on the success of mesh-based 3D morphable face models (3DMM), we define GEM as an ensemble of linear eigenbases for representing the head appearance of a specific subject. In particular, we construct linear bases to represent the position, scale, rotation, and opacity of the 3D Gaussians. This allows us to efficiently generate Gaussian primitives of a specific head shape by a linear combination of the basis vectors, only requiring a low-dimensional parameter vector that contains the respective coefficients. We propose to construct these linear bases (GEM) by distilling high-quality compute-intense CNN-based Gaussian avatar models that can generate expression-dependent appearance changes like wrinkles. These high-quality models are trained on multi-view videos of a subject and are distilled using a series of principal component analyses. Once we have obtained the bases that represent the animatable appearance space of a specific human, we learn a regressor that takes a single RGB image as input and predicts the low-dimensional parameter vector that corresponds to the shown facial expression. In a series of experiments, we compare GEM's self-reenactment and cross-person reenactment results to state-of-the-art 3D avatar methods, demonstrating GEM's higher visual quality and better generalization to new expressions.
Authors: Junseo Park, Hyeryung Jang
Abstract: Large-scale diffusion models have made significant advances in image generation, particularly through cross-attention mechanisms. While cross-attention has been well-studied in text-to-image tasks, their interpretability in image-to-image (I2I) diffusion models remains underexplored. This paper introduces Image-to-Image Attribution Maps (I2AM), a method that enhances the interpretability of I2I models by visualizing bidirectional attribution maps, from the reference image to the generated image and vice versa. I2AM aggregates cross-attention scores across time steps, attention heads, and layers, offering insights into how critical features are transferred between images. We demonstrate the effectiveness of I2AM across object detection, inpainting, and super-resolution tasks. Our results demonstrate that I2AM successfully identifies key regions responsible for generating the output, even in complex scenes. Additionally, we introduce the Inpainting Mask Attention Consistency Score (IMACS) as a novel evaluation metric to assess the alignment between attribution maps and inpainting masks, which correlates strongly with existing performance metrics. Through extensive experiments, we show that I2AM enables model debugging and refinement, providing practical tools for improving I2I model's performance and interpretability.
Authors: Ahmed Abdelkawy, Asem Ali, Aly Farag
Abstract: Existing multimodal-based human action recognition approaches are computationally intensive, limiting their deployment in real-time applications. In this work, we present a novel and efficient pose-driven attention-guided multimodal network (EPAM-Net) for action recognition in videos. Specifically, we propose eXpand temporal Shift (X-ShiftNet) convolutional architectures for RGB and pose streams to capture spatio-temporal features from RGB videos and their skeleton sequences. The X-ShiftNet tackles the high computational cost of the 3D CNNs by integrating the Temporal Shift Module (TSM) into an efficient 2D CNN, enabling efficient spatiotemporal learning. Then skeleton features are utilized to guide the visual network stream, focusing on keyframes and their salient spatial regions using the proposed spatial-temporal attention block. Finally, the predictions of the two streams are fused for final classification. The experimental results show that our method, with a significant reduction in floating-point operations (FLOPs), outperforms and competes with the state-of-the-art methods on NTU RGB-D 60, NTU RGB-D 120, PKU-MMD, and Toyota SmartHome datasets. The proposed EPAM-Net provides up to a 72.8x reduction in FLOPs and up to a 48.6x reduction in the number of network parameters. The code will be available at https://github.com/ahmed-nady/Multimodal-Action-Recognition.
URLs: https://github.com/ahmed-nady/Multimodal-Action-Recognition.
Authors: Zhenyu Lu, Hao Tang
Abstract: Data-Free Class Incremental Learning (DFCIL) aims to enable models to continuously learn new classes while retraining knowledge of old classes, even when the training data for old classes is unavailable. Although explored primarily with image datasets by researchers, this study focuses on investigating DFCIL for skeleton-based gesture classification due to its significant real-world implications, particularly considering the growing prevalence of VR/AR headsets where gestures serve as the primary means of control and interaction. In this work, we made an intriguing observation: skeleton models trained with base classes(even very limited) demonstrate strong generalization capabilities to unseen classes without requiring additional training. Building on this insight, we developed Synthetic Feature Replay (SFR) that can sample synthetic features from class prototypes to replay for old classes and augment for new classes (under a few-shot setting). Our proposed method showcases significant advancements over the state-of-the-art, achieving up to 15% enhancements in mean accuracy across all steps and largely mitigating the accuracy imbalance between base classes and new classes.
Authors: Jingyu Liu, Minquan Wang, Ye Ma, Bo Wang, Aozhu Chen, Quan Chen, Peng Jiang, Xirong Li
Abstract: Videos showcasing specific products are increasingly important for E-commerce. Key moments naturally exist as the first appearance of a specific product, presentation of its distinctive features, the presence of a buying link, etc. Adding proper sound effects (SFX) to these key moments, or video decoration with SFX (VDSFX), is crucial for enhancing the user engaging experience. Previous studies about adding SFX to videos perform video to SFX matching at a holistic level, lacking the ability of adding SFX to a specific moment. Meanwhile, previous studies on video highlight detection or video moment retrieval consider only moment localization, leaving moment to SFX matching untouched. By contrast, we propose in this paper D&M, a unified method that accomplishes key moment detection and moment to SFX matching simultaneously. Moreover, for the new VDSFX task we build a large-scale dataset SFX-Moment from an E-commerce platform. For a fair comparison, we build competitive baselines by extending a number of current video moment detection methods to the new task. Extensive experiments on SFX-Moment show the superior performance of the proposed method over the baselines.
Authors: Armin Danesh Pazho, Shanle Yao, Ghazal Alinezhad Noghre, Babak Rahimi Ardabili, Vinit Katariya, Hamed Tabkhi
Abstract: Human-centric Video Anomaly Detection (VAD) aims to identify human behaviors that deviate from normal. At its core, human-centric VAD faces substantial challenges, such as the complexity of diverse human behaviors, the rarity of anomalies, and ethical constraints. These challenges limit access to high-quality datasets and highlight the need for a dataset and framework supporting continual learning. Moving towards adaptive human-centric VAD, we introduce the HuVAD (Human-centric privacy-enhanced Video Anomaly Detection) dataset and a novel Unsupervised Continual Anomaly Learning (UCAL) framework. UCAL enables incremental learning, allowing models to adapt over time, bridging traditional training and real-world deployment. HuVAD prioritizes privacy by providing de-identified annotations and includes seven indoor/outdoor scenes, offering over 5x more pose-annotated frames than previous datasets. Our standard and continual benchmarks, utilize a comprehensive set of metrics, demonstrating that UCAL-enhanced models achieve superior performance in 82.14% of cases, setting a new state-of-the-art (SOTA). The dataset can be accessed at https://github.com/TeCSAR-UNCC/HuVAD.
Authors: Asifullah Khan, Anabia Sohail, Mustansar Fiaz, Mehdi Hassan, Tariq Habib Afridi, Sibghat Ullah Marwat, Farzeen Munir, Safdar Ali, Hannan Naseem, Muhammad Zaigham Zaheer, Kamran Ali, Tangina Sultana, Ziaurrehman Tanoli, Naeem Akhter
Abstract: Deep supervised learning models require high volume of labeled data to attain sufficiently good results. Although, the practice of gathering and annotating such big data is costly and laborious. Recently, the application of self supervised learning (SSL) in vision tasks has gained significant attention. The intuition behind SSL is to exploit the synchronous relationships within the data as a form of self-supervision, which can be versatile. In the current big data era, most of the data is unlabeled, and the success of SSL thus relies in finding ways to utilize this vast amount of unlabeled data available. Thus it is better for deep learning algorithms to reduce reliance on human supervision and instead focus on self-supervision based on the inherent relationships within the data. With the advent of ViTs, which have achieved remarkable results in computer vision, it is crucial to explore and understand the various SSL mechanisms employed for training these models specifically in scenarios where there is limited labelled data available. In this survey, we develop a comprehensive taxonomy of systematically classifying the SSL techniques based upon their representations and pre-training tasks being applied. Additionally, we discuss the motivations behind SSL, review popular pre-training tasks, and highlight the challenges and advancements in this field. Furthermore, we present a comparative analysis of different SSL methods, evaluate their strengths and limitations, and identify potential avenues for future research.
Authors: Xinheng Xie, Kureha Yamaguchi, Margaux Leblanc, Simon Malzard, Varun Chhabra, Victoria Nockles, Yue Wu
Abstract: The rapid advancement of machine learning technologies raises questions about the security of machine learning models, with respect to both training-time (poisoning) and test-time (evasion, impersonation, and inversion) attacks. Models performing image-related tasks, e.g. detection, and classification, are vulnerable to adversarial attacks that can degrade their performance and produce undesirable outcomes. This paper introduces a novel technique for anomaly detection in images called 2DSig-Detect, which uses a 2D-signature-embedded semi-supervised framework rooted in rough path theory. We demonstrate our method in adversarial settings for training-time and test-time attacks, and benchmark our framework against other state of the art methods. Using 2DSig-Detect for anomaly detection, we show both superior performance and a reduction in the computation time to detect the presence of adversarial perturbations in images.
Authors: Gonzalo Martin Garcia, Karim Abou Zeid, Christian Schmidt, Daan de Geus, Alexander Hermans, Bastian Leibe
Abstract: Recent work showed that large diffusion models can be reused as highly precise monocular depth estimators by casting depth estimation as an image-conditional image generation task. While the proposed model achieved state-of-the-art results, high computational demands due to multi-step inference limited its use in many scenarios. In this paper, we show that the perceived inefficiency was caused by a flaw in the inference pipeline that has so far gone unnoticed. The fixed model performs comparably to the best previously reported configuration while being more than 200$\times$ faster. To optimize for downstream task performance, we perform end-to-end fine-tuning on top of the single-step model with task-specific losses and get a deterministic model that outperforms all other diffusion-based depth and normal estimation models on common zero-shot benchmarks. We surprisingly find that this fine-tuning protocol also works directly on Stable Diffusion and achieves comparable performance to current state-of-the-art diffusion-based depth and normal estimation models, calling into question some of the conclusions drawn from prior works.
Authors: Gennady Sidorov, Malik Mohrat, Denis Gridusov, Ruslan Rakhimov, Sergey Kolyubin
Abstract: Although various visual localization approaches exist, such as scene coordinate regression and camera pose regression, these methods often struggle with optimization complexity or limited accuracy. To address these challenges, we explore the use of novel view synthesis techniques, particularly 3D Gaussian Splatting (3DGS), which enables the compact encoding of both 3D geometry and scene appearance. We propose a two-stage procedure that integrates dense and robust keypoint descriptors from the lightweight XFeat feature extractor into 3DGS, enhancing performance in both indoor and outdoor environments. The coarse pose estimates are directly obtained via 2D-3D correspondences between the 3DGS representation and query image descriptors. In the second stage, the initial pose estimate is refined by minimizing the rendering-based photometric warp loss. Benchmarking on widely used indoor and outdoor datasets demonstrates improvements over recent neural rendering-based localization methods, such as NeRFMatch and PNeRFLoc.
Authors: Junchen Yu, Si-Yuan Cao, Runmin Zhang, Chenghao Zhang, Zhu Yu, Shujie Chen, Bailin Yang, Hui-liang Shen
Abstract: We propose a novel unsupervised cross-modal homography estimation learning framework, named Split Supervised Homography estimation Network (SSHNet). SSHNet reformulates the unsupervised cross-modal homography estimation into two supervised sub-problems, each addressed by its specialized network: a homography estimation network and a modality transfer network. To realize stable training, we introduce an effective split optimization strategy to train each network separately within its respective sub-problem. We also formulate an extra homography feature space supervision to enhance feature consistency, further boosting the estimation accuracy. Moreover, we employ a simple yet effective distillation training technique to reduce model parameters and improve cross-domain generalization ability while maintaining comparable performance. The training stability of SSHNet enables its cooperation with various homography estimation architectures. Experiments reveal that the SSHNet using IHN as homography estimation network, namely SSHNet-IHN, outperforms previous unsupervised approaches by a significant margin. Even compared to supervised approaches MHN and LocalTrans, SSHNet-IHN achieves 47.4% and 85.8% mean average corner errors (MACEs) reduction on the challenging OPT-SAR dataset.
Authors: Kai Chen, Yunhao Gou, Runhui Huang, Zhili Liu, Daxin Tan, Jing Xu, Chunwei Wang, Yi Zhu, Yihan Zeng, Kuo Yang, Dingdong Wang, Kun Xiang, Haoyuan Li, Haoli Bai, Jianhua Han, Xiaohui Li, Weike Jin, Nian Xie, Yu Zhang, James T. Kwok, Hengshuang Zhao, Xiaodan Liang, Dit-Yan Yeung, Xiao Chen, Zhenguo Li, Wei Zhang, Qun Liu, Jun Yao, Lanqing Hong, Lu Hou, Hang Xu
Abstract: GPT-4o, an omni-modal model that enables vocal conversations with diverse emotions and tones, marks a milestone for omni-modal foundation models. However, empowering Large Language Models to perceive and generate images, texts, and speeches end-to-end with publicly available data remains challenging for the open-source community. Existing vision-language models rely on external tools for speech processing, while speech-language models still suffer from limited or totally without vision-understanding capabilities. To address this gap, we propose the EMOVA (EMotionally Omni-present Voice Assistant), to enable Large Language Models with end-to-end speech abilities while maintaining the leading vision-language performance. With a semantic-acoustic disentangled speech tokenizer, we surprisingly notice that omni-modal alignment can further enhance vision-language and speech abilities compared with the bi-modal aligned counterparts. Moreover, a lightweight style module is introduced for the flexible speech style controls including emotions and pitches. For the first time, EMOVA achieves state-of-the-art performance on both the vision-language and speech benchmarks, and meanwhile, supporting omni-modal spoken dialogue with vivid emotions.
Authors: Yifan Duan, Jian Zhao, pengcheng, Junyuan Mao, Hao Wu, Jingyu Xu, Shilong Wang, Caoyuan Ma, Kai Wang, Kun Wang, Xuelong Li
Abstract: Spatio-temporal (ST) prediction has garnered a De facto attention in earth sciences, such as meteorological prediction, human mobility perception. However, the scarcity of data coupled with the high expenses involved in sensor deployment results in notable data imbalances. Furthermore, models that are excessively customized and devoid of causal connections further undermine the generalizability and interpretability. To this end, we establish a causal framework for ST predictions, termed CaPaint, which targets to identify causal regions in data and endow model with causal reasoning ability in a two-stage process. Going beyond this process, we utilize the back-door adjustment to specifically address the sub-regions identified as non-causal in the upstream phase. Specifically, we employ a novel image inpainting technique. By using a fine-tuned unconditional Diffusion Probabilistic Model (DDPM) as the generative prior, we in-fill the masks defined as environmental parts, offering the possibility of reliable extrapolation for potential data distributions. CaPaint overcomes the high complexity dilemma of optimal ST causal discovery models by reducing the data generation complexity from exponential to quasi-linear levels. Extensive experiments conducted on five real-world ST benchmarks demonstrate that integrating the CaPaint concept allows models to achieve improvements ranging from 4.3% to 77.3%. Moreover, compared to traditional mainstream ST augmenters, CaPaint underscores the potential of diffusion models in ST enhancement, offering a novel paradigm for this field. Our project is available at https://anonymous.4open.science/r/12345-DFCC.
Authors: Wenbo Hu, Jia-Chen Gu, Zi-Yi Dou, Mohsen Fayyaz, Pan Lu, Kai-Wei Chang, Nanyun Peng
Abstract: Existing multimodal retrieval benchmarks primarily focus on evaluating whether models can retrieve and utilize external textual knowledge for question answering. However, there are scenarios where retrieving visual information is either more beneficial or easier to access than textual data. In this paper, we introduce a multimodal retrieval-augmented generation benchmark, MRAG-Bench, in which we systematically identify and categorize scenarios where visually augmented knowledge is better than textual knowledge, for instance, more images from varying viewpoints. MRAG-Bench consists of 16,130 images and 1,353 human-annotated multiple-choice questions across 9 distinct scenarios. With MRAG-Bench, we conduct an evaluation of 10 open-source and 4 proprietary large vision-language models (LVLMs). Our results show that all LVLMs exhibit greater improvements when augmented with images compared to textual knowledge, confirming that MRAG-Bench is vision-centric. Additionally, we conduct extensive analysis with MRAG-Bench, which offers valuable insights into retrieval-augmented LVLMs. Notably, the top-performing model, GPT-4o, faces challenges in effectively leveraging retrieved knowledge, achieving only a 5.82% improvement with ground-truth information, in contrast to a 33.16% improvement observed in human participants. These findings highlight the importance of MRAG-Bench in encouraging the community to enhance LVLMs' ability to utilize retrieved visual knowledge more effectively.
Authors: Aritra Bhowmik, Mohammad Mahdi Derakhshani, Dennis Koelma, Yuki M. Asano, Martin R. Oswald, Cees G. M. Snoek
Abstract: Spatial awareness is key to enable embodied multimodal AI systems. Yet, without vast amounts of spatial supervision, current Multimodal Large Language Models (MLLMs) struggle at this task. In this paper, we introduce TWIST & SCOUT, a framework that equips pre-trained MLLMs with visual grounding ability without forgetting their existing image and language understanding skills. To this end, we propose TWIST, a twin-expert stepwise tuning module that modifies the decoder of the language model using one frozen module pre-trained on image understanding tasks and another learnable one for visual grounding tasks. This allows the MLLM to retain previously learned knowledge and skills, while acquiring what is missing. To fine-tune the model effectively, we generate a high-quality synthetic dataset we call SCOUT, which mimics human reasoning in visual grounding. This dataset provides rich supervision signals, describing a step-by-step multimodal reasoning process, thereby simplifying the task of visual grounding. We evaluate our approach on several standard benchmark datasets, encompassing grounded image captioning, zero-shot localization, and visual grounding tasks. Our method consistently delivers strong performance across all tasks, while retaining the pre-trained image understanding capabilities.
Authors: Yoonjeon Kim, Soohyun Ryu, Yeonsung Jung, Hyunkoo Lee, Joowon Kim, June Yong Yang, Jaeryong Hwang, Eunho Yang
Abstract: The development of vision-language and generative models has significantly advanced text-guided image editing, which seeks the preservation of core elements in the source image while implementing modifications based on the target text. However, existing metrics have a context-blindness problem, indiscriminately applying the same evaluation criteria on completely different pairs of source image and target text, biasing towards either modification or preservation. Directional CLIP similarity, the only metric that considers both source image and target text, is also biased towards modification aspects and attends to irrelevant editing regions of the image. We propose AugCLIP, a context-aware metric that adaptively coordinates preservation and modification aspects, depending on the specific context of a given source image and target text. This is done by deriving the CLIP representation of an ideally edited image, that preserves the source image with necessary modifications to align with target text. More specifically, using a multi-modal large language model, AugCLIP augments the textual descriptions of the source and target, then calculates a modification vector through a hyperplane that separates source and target attributes in CLIP space. Extensive experiments on five benchmark datasets, encompassing a diverse range of editing scenarios, show that AugCLIP aligns remarkably well with human evaluation standards, outperforming existing metrics. The code is available at https://github.com/augclip/augclip_eval.
Authors: Guangda Ji, Silvan Weder, Francis Engelmann, Marc Pollefeys, Hermann Blum
Abstract: Neural network performance scales with both model size and data volume, as shown in both language and image processing. This requires scaling-friendly architectures and large datasets. While transformers have been adapted for 3D vision, a `GPT-moment' remains elusive due to limited training data. We introduce ARKit LabelMaker, a large-scale real-world 3D dataset with dense semantic annotation that is more than three times larger than prior largest dataset. Specifically, we extend ARKitScenes with automatically generated dense 3D labels using an extended LabelMaker pipeline, tailored for large-scale pre-training. Training on our dataset improves accuracy across architectures, achieving state-of-the-art 3D semantic segmentation scores on ScanNet and ScanNet200, with notable gains on tail classes. Our code is available at https://labelmaker.org and our dataset at https://huggingface.co/datasets/labelmaker/arkit_labelmaker.
URLs: https://labelmaker.org, https://huggingface.co/datasets/labelmaker/arkit_labelmaker.
Authors: Jiayi Liu, Denys Iliash, Angel X. Chang, Manolis Savva, Ali Mahdavi-Amiri
Abstract: We address the challenge of creating 3D assets for household articulated objects from a single image. Prior work on articulated object creation either requires multi-view multi-state input, or only allows coarse control over the generation process. These limitations hinder the scalability and practicality for articulated object modeling. In this work, we propose a method to generate articulated objects from a single image. Observing the object in resting state from an arbitrary view, our method generates an articulated object that is visually consistent with the input image. To capture the ambiguity in part shape and motion posed by a single view of the object, we design a diffusion model that learns the plausible variations of objects in terms of geometry and kinematics. To tackle the complexity of generating structured data with attributes in multiple domains, we design a pipeline that produces articulated objects from high-level structure to geometric details in a coarse-to-fine manner, where we use a part connectivity graph and part abstraction as proxies. Our experiments show that our method outperforms the state-of-the-art in articulated object creation by a large margin in terms of the generated object realism, resemblance to the input image, and reconstruction quality.
Authors: Shaofei Cai, Zihao Wang, Kewei Lian, Zhancun Mu, Xiaojian Ma, Anji Liu, Yitao Liang
Abstract: Vision-language models (VLMs) have excelled in multimodal tasks, but adapting them to embodied decision-making in open-world environments presents challenges. One critical issue is bridging the gap between discrete entities in low-level observations and the abstract concepts required for effective planning. A common solution is building hierarchical agents, where VLMs serve as high-level reasoners that break down tasks into executable sub-tasks, typically specified using language. However, language suffers from the inability to communicate detailed spatial information. We propose visual-temporal context prompting, a novel communication protocol between VLMs and policy models. This protocol leverages object segmentation from past observations to guide policy-environment interactions. Using this approach, we train ROCKET-1, a low-level policy that predicts actions based on concatenated visual observations and segmentation masks, supported by real-time object tracking from SAM-2. Our method unlocks the potential of VLMs, enabling them to tackle complex tasks that demand spatial reasoning. Experiments in Minecraft show that our approach enables agents to achieve previously unattainable tasks, with a $\mathbf{76}\%$ absolute improvement in open-world interaction performance. Codes and demos are now available on the project page: https://craftjarvis.github.io/ROCKET-1.
Authors: Amit Misra, Kevin White, Simone Fobi Nsutezo, William Straka, Juan Lavista
Abstract: Floods cause extensive global damage annually, making effective monitoring essential. While satellite observations have proven invaluable for flood detection and tracking, comprehensive global flood datasets spanning extended time periods remain scarce. In this study, we introduce a novel deep learning flood detection model that leverages the cloud-penetrating capabilities of Sentinel-1 Synthetic Aperture Radar (SAR) satellite imagery, enabling consistent flood extent mapping in through cloud cover and in both day and night conditions. By applying this model to 10 years of SAR data, we create a unique, longitudinal global flood extent dataset with predictions unaffected by cloud coverage, offering comprehensive and consistent insights into historically flood-prone areas over the past decade. We use our model predictions to identify historically flood-prone areas in Ethiopia and demonstrate real-time disaster response capabilities during the May 2024 floods in Kenya. Additionally, our longitudinal analysis reveals potential increasing trends in global flood extent over time, although further validation is required to explore links to climate change. To maximize impact, we provide public access to both our model predictions and a code repository, empowering researchers and practitioners worldwide to advance flood monitoring and enhance disaster response strategies.
Authors: Xin Gu, Ming Li, Libo Zhang, Fan Chen, Longyin Wen, Tiejian Luo, Sijie Zhu
Abstract: High-quality training triplets (instruction, original image, edited image) are essential for instruction-based image editing. Predominant training datasets (e.g., InsPix2Pix) are created using text-to-image generative models (e.g., Stable Diffusion, DALL-E) which are not trained for image editing. Accordingly, these datasets suffer from inaccurate instruction following, poor detail preserving, and generation artifacts. In this paper, we propose to address the training data quality issue with multi-perspective reward data instead of refining the ground-truth image quality. 1) we first design a quantitative metric system based on best-in-class LVLM (Large Vision Language Model), i.e., GPT-4o in our case, to evaluate the generation quality from 3 perspectives, namely, instruction following, detail preserving, and generation quality. For each perspective, we collected quantitative score in $0\sim 5$ and text descriptive feedback on the specific failure points in ground-truth edited images, resulting in a high-quality editing reward dataset, i.e., RewardEdit20K. 2) We further proposed a novel training framework to seamlessly integrate the metric output, regarded as multi-reward, into editing models to learn from the imperfect training triplets. During training, the reward scores and text descriptions are encoded as embeddings and fed into both the latent space and the U-Net of the editing models as auxiliary conditions. 3) We also build a challenging evaluation benchmark with real-world images/photos and diverse editing instructions, named Real-Edit. Experiments indicate that our multi-reward conditioned model outperforms its no-reward counterpart on two popular editing pipelines, i.e., InsPix2Pix and SmartEdit. Code is released at https://github.com/bytedance/Multi-Reward-Editing.
Authors: Xun Huang, Jinlong Wang, Qiming Xia, Siheng Chen, Bisheng Yang, Xin Li, Cheng Wang, Chenglu Wen
Abstract: Current Vehicle-to-Everything (V2X) systems have significantly enhanced 3D object detection using LiDAR and camera data. However, these methods suffer from performance degradation in adverse weather conditions. The weather-robust 4D radar provides Doppler and additional geometric information, raising the possibility of addressing this challenge. To this end, we present V2X-R, the first simulated V2X dataset incorporating LiDAR, camera, and 4D radar. V2X-R contains 12,079 scenarios with 37,727 frames of LiDAR and 4D radar point clouds, 150,908 images, and 170,859 annotated 3D vehicle bounding boxes. Subsequently, we propose a novel cooperative LiDAR-4D radar fusion pipeline for 3D object detection and implement it with various fusion strategies. To achieve weather-robust detection, we additionally propose a Multi-modal Denoising Diffusion (MDD) module in our fusion pipeline. MDD utilizes weather-robust 4D radar feature as a condition to prompt the diffusion model to denoise noisy LiDAR features. Experiments show that our LiDAR-4D radar fusion pipeline demonstrates superior performance in the V2X-R dataset. Over and above this, our MDD module further improved the performance of basic fusion model by up to 5.73%/6.70% in foggy/snowy conditions with barely disrupting normal performance. The dataset and code will be publicly available at: https://github.com/ylwhxht/V2X-R.
Authors: Markus Karmann, Onay Urfalioglu
Abstract: Recent progress in interactive point prompt based Image Segmentation allows to significantly reduce the manual effort to obtain high quality semantic labels. State-of-the-art unsupervised methods use self-supervised pre-trained models to obtain pseudo-labels which are used in training a prompt-based segmentation model. In this paper, we propose a novel unsupervised and training-free approach based solely on the self-attention of Stable Diffusion. We interpret the self-attention tensor as a Markov transition operator, which enables us to iteratively construct a Markov chain. Pixel-wise counting of the required number of iterations along the Markov chain to reach a relative probability threshold yields a Markov-iteration-map, which we simply call a Markov-map. Compared to the raw attention maps, we show that our proposed Markov-map has less noise, sharper semantic boundaries and more uniform values within semantically similar regions. We integrate the Markov-map in a simple yet effective truncated nearest neighbor framework to obtain interactive point prompt based segmentation. Despite being training-free, we experimentally show that our approach yields excellent results in terms of Number of Clicks (NoC), even outperforming state-of-the-art training based unsupervised methods in most of the datasets. Code is available at https://github.com/mkarmann/m2n2.
Authors: Vipula Rawte, Sarthak Jain, Aarush Sinha, Garv Kaushik, Aman Bansal, Prathiksha Rumale Vishwanath, Samyak Rajesh Jain, Aishwarya Naresh Reganti, Vinija Jain, Aman Chadha, Amit P. Sheth, Amitava Das
Abstract: Recent advances in Large Multimodal Models (LMMs) have expanded their capabilities to video understanding, with Text-to-Video (T2V) models excelling in generating videos from textual prompts. However, they still frequently produce hallucinated content, revealing AI-generated inconsistencies. We introduce ViBe (https://vibe-t2v-bench.github.io/): a large-scale dataset of hallucinated videos from open-source T2V models. We identify five major hallucination types: Vanishing Subject, Omission Error, Numeric Variability, Subject Dysmorphia, and Visual Incongruity. Using ten T2V models, we generated and manually annotated 3,782 videos from 837 diverse MS COCO captions. Our proposed benchmark includes a dataset of hallucinated videos and a classification framework using video embeddings. ViBe serves as a critical resource for evaluating T2V reliability and advancing hallucination detection. We establish classification as a baseline, with the TimeSFormer + CNN ensemble achieving the best performance (0.345 accuracy, 0.342 F1 score). While initial baselines proposed achieve modest accuracy, this highlights the difficulty of automated hallucination detection and the need for improved methods. Our research aims to drive the development of more robust T2V models and evaluate their outputs based on user preferences.
Authors: Tingyu Fan, Yueyu Hu, Ran Gong, Yao Wang
Abstract: Point cloud video (PCV) is a versatile 3D representation of dynamic scenes with emerging applications. This paper introduces U-Motion, a learning-based compression scheme for both PCV geometry and attributes. We propose a U-Structured inter-frame prediction framework, U-Inter, which performs explicit motion estimation and compensation (ME/MC) at different scales with varying levels of detail. It integrates Top-Down (Fine-to-Coarse) Motion Propagation, Bottom-Up Motion Predictive Coding and Multi-scale Group Motion Compensation to enable accurate motion estimation and efficient motion compression at each scale. In addition, we design a multi-scale spatial-temporal predictive coding module to capture the cross-scale spatial redundancy remaining after U-Inter prediction. We conduct experiments following the MPEG Common Test Condition for dense dynamic point clouds and demonstrate that U-Motion can achieve significant gains over MPEG G-PCC-GesTM v3.0 and recently published learning-based methods for both geometry and attribute compression.
Authors: Zhengrui Guo, Conghao Xiong, Jiabo Ma, Qichen Sun, Lishuang Feng, Jinzhuo Wang, Hao Chen
Abstract: Few-shot learning presents a critical solution for cancer diagnosis in computational pathology (CPath), addressing fundamental limitations in data availability, particularly the scarcity of expert annotations and patient privacy constraints. A key challenge in this paradigm stems from the inherent disparity between the limited training set of whole slide images (WSIs) and the enormous number of contained patches, where a significant portion of these patches lacks diagnostically relevant information, potentially diluting the model's ability to learn and focus on critical diagnostic features. While recent works attempt to address this by incorporating additional knowledge, several crucial gaps hinder further progress: (1) despite the emergence of powerful pathology foundation models (FMs), their potential remains largely untapped, with most approaches limiting their use to basic feature extraction; (2) current language guidance mechanisms attempt to align text prompts with vast numbers of WSI patches all at once, struggling to leverage rich pathological semantic information. To this end, we introduce the knowledge-enhanced adaptive visual compression framework, dubbed FOCUS, which uniquely combines pathology FMs with language prior knowledge to enable a focused analysis of diagnostically relevant regions by prioritizing discriminative WSI patches. Our approach implements a progressive three-stage compression strategy: we first leverage FMs for global visual redundancy elimination, and integrate compressed features with language prompts for semantic relevance assessment, then perform neighbor-aware visual token filtering while preserving spatial coherence. Extensive experiments on pathological datasets spanning breast, lung, and ovarian cancers demonstrate its superior performance in few-shot pathology diagnosis. Codes are available at https://github.com/dddavid4real/FOCUS.
Authors: Jooyoung Choi, Chaehun Shin, Yeongtak Oh, Heeseung Kim, Jungbeom Lee, Sungroh Yoon
Abstract: Recent text-to-image diffusion models generate high-quality images but struggle to learn new, personalized styles, which limits the creation of unique style templates. In style-driven generation, users typically supply reference images exemplifying the desired style, together with text prompts that specify desired stylistic attributes. Previous approaches popularly rely on fine-tuning, yet it often blindly utilizes objectives and noise level distributions from pre-training without adaptation. We discover that stylistic features predominantly emerge at higher noise levels, leading current fine-tuning methods to exhibit suboptimal style alignment. We propose the Style-friendly SNR sampler, which aggressively shifts the signal-to-noise ratio (SNR) distribution toward higher noise levels during fine-tuning to focus on noise levels where stylistic features emerge. This enhances models' ability to capture novel styles indicated by reference images and text prompts. We demonstrate improved generation of novel styles that cannot be adequately described solely with a text prompt, enabling the creation of new style templates for personalized content creation.
Authors: Daeun Lee, Jaehong Yoon, Jaemin Cho, Mohit Bansal
Abstract: Recent text-to-video (T2V) diffusion models have demonstrated impressive generation capabilities across various domains. However, these models often generate videos that have misalignments with text prompts, especially when the prompts describe complex scenes with multiple objects and attributes. To address this, we introduce VideoRepair, a novel model-agnostic, training-free video refinement framework that automatically identifies fine-grained text-video misalignments and generates explicit spatial and textual feedback, enabling a T2V diffusion model to perform targeted, localized refinements. VideoRepair consists of two stages: In (1) video refinement planning, we first detect misalignments by generating fine-grained evaluation questions and answering them using an MLLM. Based on video evaluation outputs, we identify accurately generated objects and construct localized prompts to precisely refine misaligned regions. In (2) localized refinement, we enhance video alignment by 'repairing' the misaligned regions from the original video while preserving the correctly generated areas. This is achieved by frame-wise region decomposition using our Region-Preserving Segmentation (RPS) module. On two popular video generation benchmarks (EvalCrafter and T2V-CompBench), VideoRepair substantially outperforms recent baselines across various text-video alignment metrics. We provide a comprehensive analysis of VideoRepair components and qualitative examples.
Authors: Noam Rotstein, Gal Yona, Daniel Silver, Roy Velich, David Bensa\"id, Ron Kimmel
Abstract: Recent advances in image editing, driven by image diffusion models, have shown remarkable progress. However, significant challenges remain, as these models often struggle to follow complex edit instructions accurately and frequently compromise fidelity by altering key elements of the original image. Simultaneously, video generation has made remarkable strides, with models that effectively function as consistent and continuous world simulators. In this paper, we propose merging these two fields by utilizing image-to-video models for image editing. We reformulate image editing as a temporal process, using pretrained video models to create smooth transitions from the original image to the desired edit. This approach traverses the image manifold continuously, ensuring consistent edits while preserving the original image's key aspects. Our approach achieves state-of-the-art results on text-based image editing, demonstrating significant improvements in both edit accuracy and image preservation. Visit our project page at https://rotsteinnoam.github.io/Frame2Frame.
Authors: Shuangqi Li, Hieu Le, Jingyi Xu, Mathieu Salzmann
Abstract: Text-to-image diffusion models have demonstrated remarkable capability in generating realistic images from arbitrary text prompts. However, they often produce inconsistent results for compositional prompts such as "two dogs" or "a penguin on the right of a bowl". Understanding these inconsistencies is crucial for reliable image generation. In this paper, we highlight the significant role of initial noise in these inconsistencies, where certain noise patterns are more reliable for compositional prompts than others. Our analyses reveal that different initial random seeds tend to guide the model to place objects in distinct image areas, potentially adhering to specific patterns of camera angles and image composition associated with the seed. To improve the model's compositional ability, we propose a method for mining these reliable cases, resulting in a curated training set of generated images without requiring any manual annotation. By fine-tuning text-to-image models on these generated images, we significantly enhance their compositional capabilities. For numerical composition, we observe relative increases of 29.3% and 19.5% for Stable Diffusion and PixArt-{\alpha}, respectively. Spatial composition sees even larger gains, with 60.7% for Stable Diffusion and 21.1% for PixArt-{\alpha}.
Authors: Hongda Liu, Yunfan Liu, Min Ren, Hao Wang, Yunlong Wang, Zhenan Sun
Abstract: In skeleton-based action recognition, a key challenge is distinguishing between actions with similar trajectories of joints due to the lack of image-level details in skeletal representations. Recognizing that the differentiation of similar actions relies on subtle motion details in specific body parts, we direct our approach to focus on the fine-grained motion of local skeleton components. To this end, we introduce ProtoGCN, a Graph Convolutional Network (GCN)-based model that breaks down the dynamics of entire skeleton sequences into a combination of learnable prototypes representing core motion patterns of action units. By contrasting the reconstruction of prototypes, ProtoGCN can effectively identify and enhance the discriminative representation of similar actions. Without bells and whistles, ProtoGCN achieves state-of-the-art performance on multiple benchmark datasets, including NTU RGB+D, NTU RGB+D 120, Kinetics-Skeleton, and FineGYM, which demonstrates the effectiveness of the proposed method. The code is available at https://github.com/firework8/ProtoGCN.
Authors: Tiantian Geng, Jinrui Zhang, Qingni Wang, Teng Wang, Jinming Duan, Feng Zheng
Abstract: Despite impressive advancements in video understanding, most efforts remain limited to coarse-grained or visual-only video tasks. However, real-world videos encompass omni-modal information (vision, audio, and speech) with a series of events forming a cohesive storyline. The lack of multi-modal video data with fine-grained event annotations and the high cost of manual labeling are major obstacles to comprehensive omni-modality video perception. To address this gap, we propose an automatic pipeline consisting of high-quality multi-modal video filtering, semantically coherent omni-modal event boundary detection, and cross-modal correlation-aware event captioning. In this way, we present LongVALE, the first-ever Vision-Audio-Language Event understanding benchmark comprising 105K omni-modal events with precise temporal boundaries and detailed relation-aware captions within 8.4K high-quality long videos. Further, we build a baseline that leverages LongVALE to enable video large language models (LLMs) for omni-modality fine-grained temporal video understanding for the first time. Extensive experiments demonstrate the effectiveness and great potential of LongVALE in advancing comprehensive multi-modal video understanding.
Authors: Hang Ye, Xiaoxuan Ma, Hai Ci, Wentao Zhu, Yizhou Wang
Abstract: Achieving realistic animated human avatars requires accurate modeling of pose-dependent clothing deformations. Existing learning-based methods heavily rely on the Linear Blend Skinning (LBS) of minimally-clothed human models like SMPL to model deformation. However, they struggle to handle loose clothing, such as long dresses, where the canonicalization process becomes ill-defined when the clothing is far from the body, leading to disjointed and fragmented results. To overcome this limitation, we propose FreeCloth, a novel hybrid framework to model challenging clothed humans. Our core idea is to use dedicated strategies to model different regions, depending on whether they are close to or distant from the body. Specifically, we segment the human body into three categories: unclothed, deformed, and generated. We simply replicate unclothed regions that require no deformation. For deformed regions close to the body, we leverage LBS to handle the deformation. As for the generated regions, which correspond to loose clothing areas, we introduce a novel free-form, part-aware generator to model them, as they are less affected by movements. This free-form generation paradigm brings enhanced flexibility and expressiveness to our hybrid framework, enabling it to capture the intricate geometric details of challenging loose clothing, such as skirts and dresses. Experimental results on the benchmark dataset featuring loose clothing demonstrate that FreeCloth achieves state-of-the-art performance with superior visual fidelity and realism, particularly in the most challenging cases.
Authors: Anton Voronov, Denis Kuznedelev, Mikhail Khoroshikh, Valentin Khrulkov, Dmitry Baranchuk
Abstract: This work presents Switti, a scale-wise transformer for text-to-image generation. We start by adapting an existing next-scale prediction autoregressive (AR) architecture to T2I generation, investigating and mitigating training stability issues in the process. Next, we argue that scale-wise transformers do not require causality and propose a non-causal counterpart facilitating ~21% faster sampling and lower memory usage while also achieving slightly better generation quality. Furthermore, we reveal that classifier-free guidance at high-resolution scales is often unnecessary and can even degrade performance. By disabling guidance at these scales, we achieve an additional sampling acceleration of ~32% and improve the generation of fine-grained details. Extensive human preference studies and automated evaluations show that Switti outperforms existing T2I AR models and competes with state-of-the-art T2I diffusion models while being up to 7x faster.
Authors: Mingzhe Zheng, Yongqi Xu, Haojian Huang, Xuran Ma, Yexin Liu, Wenjie Shu, Yatian Pang, Feilong Tang, Qifeng Chen, Harry Yang, Ser-Nam Lim
Abstract: Current video generation models excel at short clips but fail to produce cohesive multi-shot narratives due to disjointed visual dynamics and fractured storylines. Existing solutions either rely on extensive manual scripting/editing or prioritize single-shot fidelity over cross-scene continuity, limiting their practicality for movie-like content. We introduce VideoGen-of-Thought (VGoT), a step-by-step framework that automates multi-shot video synthesis from a single sentence by systematically addressing three core challenges: (1) Narrative Fragmentation: Existing methods lack structured storytelling. We propose dynamic storyline modeling, which first converts the user prompt into concise shot descriptions, then elaborates them into detailed, cinematic specifications across five domains (character dynamics, background continuity, relationship evolution, camera movements, HDR lighting), ensuring logical narrative progression with self-validation. (2) Visual Inconsistency: Existing approaches struggle with maintaining visual consistency across shots. Our identity-aware cross-shot propagation generates identity-preserving portrait (IPP) tokens that maintain character fidelity while allowing trait variations (expressions, aging) dictated by the storyline. (3) Transition Artifacts: Abrupt shot changes disrupt immersion. Our adjacent latent transition mechanisms implement boundary-aware reset strategies that process adjacent shots' features at transition points, enabling seamless visual flow while preserving narrative continuity. VGoT generates multi-shot videos that outperform state-of-the-art baselines by 20.4% in within-shot face consistency and 17.4% in style consistency, while achieving over 100% better cross-shot consistency and 10x fewer manual adjustments than alternatives.
Authors: Aniruddha Ganguly, Debolina Chatterjee, Wentao Huang, Jie Zhang, Alisa Yurovsky, Travis Steele Johnson, Chao Chen
Abstract: Recent advances in Spatial Transcriptomics (ST) pair histology images with spatially resolved gene expression profiles, enabling predictions of gene expression across different tissue locations based on image patches. This opens up new possibilities for enhancing whole slide image (WSI) prediction tasks with localized gene expression. However, existing methods fail to fully leverage the interactions between different tissue locations, which are crucial for accurate joint prediction. To address this, we introduce MERGE (Multi-faceted hiErarchical gRaph for Gene Expressions), which combines a multi-faceted hierarchical graph construction strategy with graph neural networks (GNN) to improve gene expression predictions from WSIs. By clustering tissue image patches based on both spatial and morphological features, and incorporating intra- and inter-cluster edges, our approach fosters interactions between distant tissue locations during GNN learning. As an additional contribution, we evaluate different data smoothing techniques that are necessary to mitigate artifacts in ST data, often caused by technical imperfections. We advocate for adopting gene-aware smoothing methods that are more biologically justified. Experimental results on gene expression prediction show that our GNN method outperforms state-of-the-art techniques across multiple metrics.
Authors: Chamuditha Jayanga Galappaththige, Jason Lai, Lloyd Windrim, Donald Dansereau, Niko Suenderhauf, Dimity Miller
Abstract: Autonomous agents often require accurate methods for detecting and localizing changes in their environment, particularly when observations are captured from unconstrained and inconsistent viewpoints. We propose a novel label-free, pose-agnostic change detection method that integrates information from multiple viewpoints to construct a change-aware 3D Gaussian Splatting (3DGS) representation of the scene. With as few as 5 images of the post-change scene, our approach can learn an additional change channel in a 3DGS and produce change masks that outperform single-view techniques. Our change-aware 3D scene representation additionally enables the generation of accurate change masks for unseen viewpoints. Experimental results demonstrate state-of-the-art performance in complex multi-object scenes, achieving a 1.7x and 1.5x improvement in Mean Intersection Over Union and F1 score respectively over other baselines. We also contribute a new real-world dataset to benchmark change detection in diverse challenging scenes in the presence of lighting variations.
Authors: Susung Hong, Johanna Karras, Ricardo Martin-Brualla, Ira Kemelmacher-Shlizerman
Abstract: Recent advancements in text-based diffusion models have accelerated progress in 3D reconstruction and text-based 3D editing. Although existing 3D editing methods excel at modifying color, texture, and style, they struggle with extensive geometric or appearance changes, thus limiting their applications. To this end, we propose Perturb-and-Revise, which makes possible a variety of NeRF editing. First, we perturb the NeRF parameters with random initializations to create a versatile initialization. The level of perturbation is determined automatically through analysis of the local loss landscape. Then, we revise the edited NeRF via generative trajectories. Combined with the generative process, we impose identity-preserving gradients to refine the edited NeRF. Extensive experiments demonstrate that Perturb-and-Revise facilitates flexible, effective, and consistent editing of color, appearance, and geometry in 3D. For 360{\deg} results, please visit our project page: https://susunghong.github.io/Perturb-and-Revise.
Authors: Xinyue Liu, Jianyuan Wang, Biao Leng, Shuo Zhang
Abstract: Knowledge distillation (KD) has been widely studied in unsupervised Industrial Image Anomaly Detection (AD), but its application to unsupervised multimodal AD remains underexplored. Existing KD-based methods for multimodal AD that use fused multimodal features to obtain teacher representations face challenges. Anomalies in one modality may not be effectively captured in the fused teacher features, leading to detection failures. Besides, these methods do not fully leverage the rich intra- and inter-modality information. In this paper, we propose Crossmodal Reverse Distillation (CRD) based on Multi-branch design to realize Multimodal Industrial AD. By assigning independent branches to each modality, our method enables finer detection of anomalies within each modality. Furthermore, we enhance the interaction between modalities during the distillation process by designing Crossmodal Filter and Amplifier. With the idea of crossmodal mapping, the student network is allowed to better learn normal features while anomalies in all modalities are ensured to be effectively detected. Experimental verifications on the MVTec 3D-AD dataset demonstrate that our method achieves state-of-the-art performance in multimodal anomaly detection and localization.
Authors: Messi H. J. Lee, Soyeon Jeon
Abstract: Vision-Language Models (VLMs) extend Large Language Models' capabilities by integrating image processing, but concerns persist about their potential to reproduce and amplify human biases. While research has documented how these models perpetuate stereotypes across demographic groups, most work has focused on between-group biases rather than within-group differences. This study investigates homogeneity bias-the tendency to portray groups as more uniform than they are-within Black Americans, examining how perceived racial phenotypicality influences VLMs' outputs. Using computer-generated images that systematically vary in phenotypicality, we prompted VLMs to generate stories about these individuals and measured text similarity to assess content homogeneity. Our findings reveal three key patterns: First, VLMs generate significantly more homogeneous stories about Black individuals with higher phenotypicality compared to those with lower phenotypicality. Second, stories about Black women consistently display greater homogeneity than those about Black men across all models tested. Third, in two of three VLMs, this homogeneity bias is primarily driven by a pronounced interaction where phenotypicality strongly influences content variation for Black women but has minimal impact for Black men. These results demonstrate how intersectionality shapes AI-generated representations and highlight the persistence of stereotyping that mirror documented biases in human perception, where increased racial phenotypicality leads to greater stereotyping and less individualized representation.
Authors: Zican Shi, Jing Hu, Jie Ren, Hengkang Ye, Xuyang Yuan, Yan Ouyang, Jia He, Bo Ji, Junyu Guo
Abstract: The introduction of Feature Pyramid Network (FPN) has significantly improved object detection performance. However, substantial challenges remain in detecting tiny objects, as their features occupy only a very small proportion of the feature maps. Although FPN integrates multi-scale features, it does not directly enhance or enrich the features of tiny objects. Furthermore, FPN lacks spatial perception ability. To address these issues, we propose a novel High Frequency and Spatial Perception Feature Pyramid Network (HS-FPN) with two innovative modules. First, we designed a high frequency perception module (HFP) that generates high frequency responses through high pass filters. These high frequency responses are used as mask weights from both spatial and channel perspectives to enrich and highlight the features of tiny objects in the original feature maps. Second, we developed a spatial dependency perception module (SDP) to capture the spatial dependencies that FPN lacks. Our experiments demonstrate that detectors based on HS-FPN exhibit competitive advantages over state-of-the-art models on the AI-TOD dataset for tiny object detection.
Authors: Andrea Dunn Beltran, Daniel Rho, Stephen Pizer, Marc Niethammer, Roni Sengupta
Abstract: Simultaneous Localization And Mapping (SLAM) from endoscopy videos can enable autonomous navigation, guidance to unsurveyed regions, blindspot detections, and 3D visualizations, which can significantly improve patient outcomes and endoscopy experience for both physicians and patients. Existing dense SLAM algorithms often assume distant and static lighting and optimize scene geometry and camera parameters by minimizing a photometric rendering loss, often called Photometric Bundle Adjustment. However, endoscopy videos exhibit dynamic near-field lighting due to the co-located light and camera moving extremely close to the surface. In addition, low texture surfaces in endoscopy videos cause photometric bundle adjustment of the existing SLAM frameworks to perform poorly compared to indoor/outdoor scenes. To mitigate this problem, we introduce Near-Field Lighting Bundle Adjustment Loss (NFL-BA) which explicitly models near-field lighting as a part of Bundle Adjustment loss and enables better performance for low texture surfaces. Our proposed NFL-BA can be applied to any neural-rendering based SLAM framework. We show that by replacing traditional photometric bundle adjustment loss with our proposed NFL-BA results in improvement, using neural implicit SLAM and 3DGS SLAMs. In addition to producing state-of-the-art tracking and mapping results on colonoscopy C3VD dataset we also show improvement on real colonoscopy videos. See results at https://asdunnbe.github.io/NFL-BA/
Authors: Zhifang Zhang, Shuo He, Haobo Wang, Bingquan Shen, Lei Feng
Abstract: Multimodal contrastive learning models (e.g., CLIP) can learn high-quality representations from large-scale image-text datasets, yet they exhibit significant vulnerabilities to backdoor attacks, raising serious safety concerns. In this paper, we disclose that CLIP's vulnerabilities primarily stem from its excessive encoding of class-irrelevant features, which can compromise the model's visual feature resistivity to input perturbations, making it more susceptible to capturing the trigger patterns inserted by backdoor attacks. Inspired by this finding, we propose Repulsive Visual Prompt Tuning (RVPT), a novel defense approach that employs specially designed deep visual prompt tuning and feature-repelling loss to eliminate excessive class-irrelevant features while simultaneously optimizing cross-entropy loss to maintain clean accuracy. Unlike existing multimodal backdoor defense methods that typically require the availability of poisoned data or involve fine-tuning the entire model, RVPT leverages few-shot downstream clean samples and only tunes a small number of parameters. Empirical results demonstrate that RVPT tunes only 0.27\% of the parameters relative to CLIP, yet it significantly outperforms state-of-the-art baselines, reducing the attack success rate from 67.53\% to 2.76\% against SoTA attacks and effectively generalizing its defensive capabilities across multiple datasets.
Authors: Chenyang Liu, Keyan Chen, Rui Zhao, Zhengxia Zou, Zhenwei Shi
Abstract: Generative foundation models have advanced large-scale text-driven natural image generation, becoming a prominent research trend across various vertical domains. However, in the remote sensing field, there is still a lack of research on large-scale text-to-image (text2image) generation technology. Existing remote sensing image-text datasets are small in scale and confined to specific geographic areas and scene types. Besides, existing text2image methods have struggled to achieve global-scale, multi-resolution controllable, and unbounded image generation. To address these challenges, this paper presents two key contributions: the Git-10M dataset and the Text2Earth foundation model. Git-10M is a global-scale image-text dataset comprising 10.5 million image-text pairs, 5 times larger than the previous largest one. The dataset covers a wide range of geographic scenes and contains resolution information, significantly surpassing existing datasets in both size and diversity. Building on Git-10M, we propose Text2Earth, a 1.3 billion parameter generative foundation model based on the diffusion framework to model global-scale remote sensing scenes. Text2Earth integrates a resolution guidance mechanism, enabling users to specify image resolutions. A dynamic condition adaptation strategy is proposed for training and inference to improve image quality. Text2Earth excels in zero-shot text2image generation and demonstrates robust generalization and flexibility across multiple tasks, including unbounded scene construction, image editing, and cross-modal image generation. This robust capability surpasses previous models restricted to the basic fixed size and limited scene types. On the previous benchmark dataset, Text2Earth outperforms previous models with an improvement of +26.23 FID and +20.95% Zero-shot Cls-OA metric.Our project page is https://chen-yang-liu.github.io/Text2Earth
Authors: Louis Mahon
Abstract: Data complexity is an important concept in the natural sciences and related areas, but lacks a rigorous and computable definition. In this paper, we focus on a particular sense of complexity that is high if the data is structured in a way that could serve to communicate a message. In this sense, human speech, written language, drawings, diagrams and photographs are high complexity, whereas data that is close to uniform throughout or populated by random values is low complexity. We describe a general framework for measuring data complexity based on dividing the shortest description of the data into a structured and an unstructured portion, and taking the size of the former as the complexity score. We outline an application of this framework in statistical mechanics that may allow a more objective characterisation of the macrostate and entropy of a physical system. Then, we derive a more precise and computable definition geared towards human communication, by proposing local compositionality as an appropriate specific structure. We demonstrate experimentally that this method can distinguish meaningful signals from noise or repetitive signals in auditory, visual and text domains, and could potentially help determine whether an extra-terrestrial signal contained a message.
Authors: Xiang Xu, Lingdong Kong, Hui Shuai, Liang Pan, Ziwei Liu, Qingshan Liu
Abstract: LiDAR data pretraining offers a promising approach to leveraging large-scale, readily available datasets for enhanced data utilization. However, existing methods predominantly focus on sparse voxel representation, overlooking the complementary attributes provided by other LiDAR representations. In this work, we propose LiMoE, a framework that integrates the Mixture of Experts (MoE) paradigm into LiDAR data representation learning to synergistically combine multiple representations, such as range images, sparse voxels, and raw points. Our approach consists of three stages: i) Image-to-LiDAR Pretraining, which transfers prior knowledge from images to point clouds across different representations; ii) Contrastive Mixture Learning (CML), which uses MoE to adaptively activate relevant attributes from each representation and distills these mixed features into a unified 3D network; iii) Semantic Mixture Supervision (SMS), which combines semantic logits from multiple representations to boost downstream segmentation performance. Extensive experiments across eleven large-scale LiDAR datasets demonstrate our effectiveness and superiority. The code has been made publicly accessible.
Authors: Tsai-Shien Chen, Aliaksandr Siarohin, Willi Menapace, Yuwei Fang, Kwot Sin Lee, Ivan Skorokhodov, Kfir Aberman, Jun-Yan Zhu, Ming-Hsuan Yang, Sergey Tulyakov
Abstract: Video personalization methods allow us to synthesize videos with specific concepts such as people, pets, and places. However, existing methods often focus on limited domains, require time-consuming optimization per subject, or support only a single subject. We present Video Alchemist $-$ a video model with built-in multi-subject, open-set personalization capabilities for both foreground objects and background, eliminating the need for time-consuming test-time optimization. Our model is built on a new Diffusion Transformer module that fuses each conditional reference image and its corresponding subject-level text prompt with cross-attention layers. Developing such a large model presents two main challenges: dataset and evaluation. First, as paired datasets of reference images and videos are extremely hard to collect, we sample selected video frames as reference images and synthesize a clip of the target video. However, while models can easily denoise training videos given reference frames, they fail to generalize to new contexts. To mitigate this issue, we design a new automatic data construction pipeline with extensive image augmentations. Second, evaluating open-set video personalization is a challenge in itself. To address this, we introduce a personalization benchmark that focuses on accurate subject fidelity and supports diverse personalization scenarios. Finally, our extensive experiments show that our method significantly outperforms existing personalization methods in both quantitative and qualitative evaluations.
Authors: Wojciech Zielonka, Stephan J. Garbin, Alexandros Lattas, George Kopanas, Paulo Gotardo, Thabo Beeler, Justus Thies, Timo Bolkart
Abstract: We present SynShot, a novel method for the few-shot inversion of a drivable head avatar based on a synthetic prior. We tackle three major challenges. First, training a controllable 3D generative network requires a large number of diverse sequences, for which pairs of images and high-quality tracked meshes are not always available. Second, the use of real data is strictly regulated (e.g., under the General Data Protection Regulation, which mandates frequent deletion of models and data to accommodate a situation when a participant's consent is withdrawn). Synthetic data, free from these constraints, is an appealing alternative. Third, state-of-the-art monocular avatar models struggle to generalize to new views and expressions, lacking a strong prior and often overfitting to a specific viewpoint distribution. Inspired by machine learning models trained solely on synthetic data, we propose a method that learns a prior model from a large dataset of synthetic heads with diverse identities, expressions, and viewpoints. With few input images, SynShot fine-tunes the pretrained synthetic prior to bridge the domain gap, modeling a photorealistic head avatar that generalizes to novel expressions and viewpoints. We model the head avatar using 3D Gaussian splatting and a convolutional encoder-decoder that outputs Gaussian parameters in UV texture space. To account for the different modeling complexities over parts of the head (e.g., skin vs hair), we embed the prior with explicit control for upsampling the number of per-part primitives. Compared to SOTA monocular and GAN-based methods, SynShot significantly improves novel view and expression synthesis.
Authors: Xinyang Zhou, Fanyue Wei, Lixin Duan, Angela Yao, Wen Li
Abstract: Given a textual query along with a corresponding video, the objective of moment retrieval aims to localize the moments relevant to the query within the video. While commendable results have been demonstrated by existing transformer-based approaches, predicting the accurate temporal span of the target moment is still a major challenge. This paper reveals that a crucial reason stems from the spurious correlation between the text query and the moment context. Namely, the model makes predictions by overly associating queries with background frames rather than distinguishing target moments. To address this issue, we propose a dynamic learning approach for moment retrieval, where two strategies are designed to mitigate the spurious correlation. First, we introduce a novel video synthesis approach to construct a dynamic context for the queried moment, enabling the model to attend to the target moment of the corresponding query across dynamic backgrounds. Second, to alleviate the over-association with backgrounds, we enhance representations temporally by incorporating text-dynamics interaction, which encourages the model to align text with target moments through complementary dynamic representations. With the proposed method, our model significantly alleviates the spurious correlation issue in moment retrieval and establishes new state-of-the-art performance on two popular benchmarks, \ie, QVHighlights and Charades-STA. In addition, detailed ablation studies and evaluations across different architectures demonstrate the generalization and effectiveness of the proposed strategies. Our code will be publicly available.
Authors: Jianing Yang, Alexander Sax, Kevin J. Liang, Mikael Henaff, Hao Tang, Ang Cao, Joyce Chai, Franziska Meier, Matt Feiszli
Abstract: Multi-view 3D reconstruction remains a core challenge in computer vision, particularly in applications requiring accurate and scalable representations across diverse perspectives. Current leading methods such as DUSt3R employ a fundamentally pairwise approach, processing images in pairs and necessitating costly global alignment procedures to reconstruct from multiple views. In this work, we propose Fast 3D Reconstruction (Fast3R), a novel multi-view generalization to DUSt3R that achieves efficient and scalable 3D reconstruction by processing many views in parallel. Fast3R's Transformer-based architecture forwards N images in a single forward pass, bypassing the need for iterative alignment. Through extensive experiments on camera pose estimation and 3D reconstruction, Fast3R demonstrates state-of-the-art performance, with significant improvements in inference speed and reduced error accumulation. These results establish Fast3R as a robust alternative for multi-view applications, offering enhanced scalability without compromising reconstruction accuracy.
Authors: Karahan Sar{\i}ta\c{s}, Peter Dayan, Tingke Shen, Surabhi S Nath
Abstract: Understanding how humans perceive visual complexity is a key area of study in visual cognition. Previous approaches to modeling visual complexity assessments have often resulted in intricate, difficult-to-interpret algorithms that employ numerous features or sophisticated deep learning architectures. While these complex models achieve high performance on specific datasets, they often sacrifice interpretability, making it challenging to understand the factors driving human perception of complexity. Recently (Shen, et al. 2024) proposed an interpretable segmentation-based model that accurately predicted complexity across various datasets, supporting the idea that complexity can be explained simply. In this work, we investigate the failure of their model to capture structural, color and surprisal contributions to complexity. To this end, we propose Multi-Scale Sobel Gradient (MSG) which measures spatial intensity variations, Multi-Scale Unique Color (MUC) which quantifies colorfulness across multiple scales, and surprise scores generated using a Large Language Model. We test our features on existing benchmarks and a novel dataset (Surprising Visual Genome) containing surprising images from Visual Genome. Our experiments demonstrate that modeling complexity accurately is not as simple as previously thought, requiring additional perceptual and semantic factors to address dataset biases. Our model improves predictive performance while maintaining interpretability, offering deeper insights into how visual complexity is perceived and assessed. Our code, analysis and data are available at https://github.com/Complexity-Project/Complexity-in-Complexity.
URLs: https://github.com/Complexity-Project/Complexity-in-Complexity.
Authors: Agnieszka Anna Tomaka, Dariusz Pojda, Micha{\l} Tarnawski, Leszek Luchowski
Abstract: Multimodal image registration plays a key role in creating digital patient models by combining data from different imaging techniques into a single coordinate system. This process often involves multiple sequential and interconnected transformations, which must be well-documented to ensure transparency and reproducibility. In this paper, we propose the use of transformation trees as a method for structured recording and management of these transformations. This approach has been implemented in the dpVision software and uses a dedicated .dpw file format to store hierarchical relationships between images, transformations, and motion data. Transformation trees allow precise tracking of all image processing steps, reduce the need to store multiple copies of the same data, and enable the indirect registration of images that do not share common reference points. This improves the reproducibility of the analyses and facilitates later processing and integration of images from different sources. The practical application of this method is demonstrated with examples from orthodontics, including the integration of 3D face scans, intraoral scans, and CBCT images, as well as the documentation of mandibular motion. Beyond orthodontics, this method can be applied in other fields that require systematic management of image registration processes, such as maxillofacial surgery, oncology, and biomechanical analysis. Maintaining long-term data consistency is essential for both scientific research and clinical practice. It enables easier comparison of results in longitudinal studies, improves retrospective analysis, and supports the development of artificial intelligence algorithms by providing standardized and well-documented datasets. The proposed approach enhances data organization, allows for efficient analysis, and facilitates the reuse of information in future studies and diagnostic procedures.
Authors: Alexander Nikulin, Ilya Zisman, Denis Tarasov, Nikita Lyubaykin, Andrei Polubarov, Igor Kiselev, Vladislav Kurenkov
Abstract: Recently, latent action learning, pioneered by Latent Action Policies (LAPO), have shown remarkable pre-training efficiency on observation-only data, offering potential for leveraging vast amounts of video available on the web for embodied AI. However, prior work has focused on distractor-free data, where changes between observations are primarily explained by ground-truth actions. Unfortunately, real-world videos contain action-correlated distractors that may hinder latent action learning. Using Distracting Control Suite (DCS) we empirically investigate the effect of distractors on latent action learning and demonstrate that LAPO struggle in such scenario. We propose LAOM, a simple LAPO modification that improves the quality of latent actions by 8x, as measured by linear probing. Importantly, we show that providing supervision with ground-truth actions, as few as 2.5% of the full dataset, during latent action learning improves downstream performance by 4.2x on average. Our findings suggest that integrating supervision during Latent Action Models (LAM) training is critical in the presence of distractors, challenging the conventional pipeline of first learning LAM and only then decoding from latent to ground-truth actions.
Authors: Tao Zhang, Cheng Da, Kun Ding, Kun Jin, Yan Li, Tingting Gao, Di Zhang, Shiming Xiang, Chunhong Pan
Abstract: Preference optimization for diffusion models aims to align them with human preferences for images. Previous methods typically leverage Vision-Language Models (VLMs) as pixel-level reward models to approximate human preferences. However, when used for step-level preference optimization, these models face challenges in handling noisy images of different timesteps and require complex transformations into pixel space. In this work, we demonstrate that diffusion models are inherently well-suited for step-level reward modeling in the latent space, as they can naturally extract features from noisy latent images. Accordingly, we propose the Latent Reward Model (LRM), which repurposes components of diffusion models to predict preferences of latent images at various timesteps. Building on LRM, we introduce Latent Preference Optimization (LPO), a method designed for step-level preference optimization directly in the latent space. Experimental results indicate that LPO not only significantly enhances performance in aligning diffusion models with general, aesthetic, and text-image alignment preferences, but also achieves 2.5-28$\times$ training speedup compared to existing preference optimization methods. Our code and models are available at https://github.com/Kwai-Kolors/LPO.
Authors: Tao Zhang, Jinyong Wen, Zhen Chen, Kun Ding, Shiming Xiang, Chunhong Pan
Abstract: Pre-training techniques significantly enhance the performance of semantic segmentation tasks with limited training data. However, the efficacy under a large domain gap between pre-training (e.g. RGB) and fine-tuning (e.g. infrared) remains underexplored. In this study, we first benchmark the infrared semantic segmentation performance of various pre-training methods and reveal several phenomena distinct from the RGB domain. Next, our layerwise analysis of pre-trained attention maps uncovers that: (1) There are three typical attention patterns (local, hybrid, and global); (2) Pre-training tasks notably influence the pattern distribution across layers; (3) The hybrid pattern is crucial for semantic segmentation as it attends to both nearby and foreground elements; (4) The texture bias impedes model generalization in infrared tasks. Building on these insights, we propose UNIP, a UNified Infrared Pre-training framework, to enhance the pre-trained model performance. This framework uses the hybrid-attention distillation NMI-HAD as the pre-training target, a large-scale mixed dataset InfMix for pre-training, and a last-layer feature pyramid network LL-FPN for fine-tuning. Experimental results show that UNIP outperforms various pre-training methods by up to 13.5\% in average mIoU on three infrared segmentation tasks, evaluated using fine-tuning and linear probing metrics. UNIP-S achieves performance on par with MAE-L while requiring only 1/10 of the computational cost. Furthermore, UNIP significantly surpasses state-of-the-art (SOTA) infrared or RGB segmentation methods and demonstrates broad potential for application in other modalities, such as RGB and depth. Our code is available at https://github.com/casiatao/UNIP.
Authors: Chaoyin She, Ruifang Lu, Danni He, Jiayi Lv, Yadan Lin, Meiqing Cheng, Hui Huang, Fengyu Ye, Lida Chen, Wei Wang, Qinghua Huang
Abstract: Hepatocellular carcinoma (HCC), ranking as the third leading cause of cancer-related mortality worldwide, demands urgent improvements in early detection to enhance patient survival. While ultrasound remains the preferred screening modality due to its cost-effectiveness and real-time capabilities, its sensitivity (59%-78%) heavily relies on radiologists' expertise, leading to inconsistent diagnostic outcomes and operational inefficiencies. Recent advancements in AI technology offer promising solutions to bridge this gap. This study introduces the Hierarchical Sparse Query Transformer (HSQformer), a novel hybrid architecture that synergizes CNNs' local feature extraction with Vision Transformers' global contextual awareness through latent space representation and sparse learning. By dynamically activating task-specific experts via a Mixture-of-Experts (MoE) framework, HSQformer achieves hierarchical feature integration without structural redundancy. Evaluated across three clinical scenarios: single-center, multi-center, and high-risk patient cohorts, HSQformer outperforms state-of-the-art models (e.g., 95.38% AUC in multi-center testing) and matches senior radiologists' diagnostic accuracy while significantly surpassing junior counterparts. These results highlight the potential of AI-assisted tools to standardize HCC screening, reduce dependency on human expertise, and improve early diagnosis rates. The full code is available at https://github.com/Asunatan/HSQformer.
Authors: Pedro V\'elez, Luisa F. Polan\'ia, Yi Yang, Chuhan Zhang, Rishabh Kabra, Anurag Arnab, Mehdi S. M. Sajjadi
Abstract: Diffusion models have revolutionized generative modeling, enabling unprecedented realism in image and video synthesis. This success has sparked interest in leveraging their representations for visual understanding tasks. While recent works have explored this potential for image generation, the visual understanding capabilities of video diffusion models remain largely uncharted. To address this gap, we systematically compare the same model architecture trained for video versus image generation, analyzing the performance of their latent representations on various downstream tasks including image classification, action recognition, depth estimation, and tracking. Results show that video diffusion models consistently outperform their image counterparts, though we find a striking range in the extent of this superiority. We further analyze features extracted from different layers and with varying noise levels, as well as the effect of model size and training budget on representation and generation quality. This work marks the first direct comparison of video and image diffusion objectives for visual understanding, offering insights into the role of temporal information in representation learning.
Authors: Saad Lahrichi, Jake Bova, Jesse Johnson, Jordan Malof
Abstract: Recent research has demonstrated the potential of deep neural networks (DNNs) to accurately predict wildfire spread on a given day based upon high-dimensional explanatory data from a single preceding day, or from a time series of T preceding days. Here, we introduce a variety of modeling improvements that achieve state-of-the-art (SOTA) accuracy for both single-day and multi-day input scenarios, as evaluated on a large public benchmark for next-day wildfire spread, termed the WildfireSpreadTS (WSTS) benchmark. Consistent with prior work, we found that models using time-series input obtained the best overall accuracy. Furthermore, we create a new benchmark, WSTS+, by incorporating four additional years of historical wildfire data into the WSTS benchmark. Our benchmark doubles the number of unique years of historical data, expands its geographic scope, and, to our knowledge, represents the largest public benchmark for time-series-based wildfire spread prediction.
Authors: Sang-Eun Lee, Ko Nishino, Shohei Nobuhara
Abstract: We propose a novel method for spatiotemporal multi-camera calibration using freely moving people in multiview videos. Since calibrating multiple cameras and finding matches across their views are inherently interdependent, performing both in a unified framework poses a significant challenge. We address these issues as a single registration problem of matching two sets of 3D points, leveraging human motion in dynamic multi-person scenes. To this end, we utilize 3D human poses obtained from an off-the-shelf monocular 3D human pose estimator and transform them into 3D points on a unit sphere, to solve the rotation, time offset, and the association alternatingly. We employ a probabilistic approach that can jointly solve both problems of aligning spatiotemporal data and establishing correspondences through soft assignment between two views. The translation is determined by applying coplanarity constraints. The pairwise registration results are integrated into a multiview setup, and then a nonlinear optimization method is used to improve the accuracy of the camera poses, temporal offsets, and multi-person associations. Extensive experiments on synthetic and real data demonstrate the effectiveness and flexibility of the proposed method as a practical marker-free calibration tool.
Authors: Woo Kyoung Han, Byeonghun Lee, Hyunmin Cho, Sunghoon Im, Kyong Hwan Jin
Abstract: We quantify the upper bound on the size of the implicit neural representation (INR) model from a digital perspective. The upper bound of the model size increases exponentially as the required bit-precision increases. To this end, we present a bit-plane decomposition method that makes INR predict bit-planes, producing the same effect as reducing the upper bound of the model size. We validate our hypothesis that reducing the upper bound leads to faster convergence with constant model size. Our method achieves lossless representation in 2D image and audio fitting, even for high bit-depth signals, such as 16-bit, which was previously unachievable. We pioneered the presence of bit bias, which INR prioritizes as the most significant bit (MSB). We expand the application of the INR task to bit depth expansion, lossless image compression, and extreme network quantization. Our source code is available at https://github.com/WooKyoungHan/LosslessINR
Authors: Ranjan Sapkota, Manoj Karkee
Abstract: This study evaluated the performance of the YOLOv12 object detection model, and compared against the performances YOLOv11 and YOLOv10 for apple detection in commercial orchards based on the model training completed entirely on synthetic images generated by Large Language Models (LLMs). The YOLOv12n configuration achieved the highest precision at 0.916, the highest recall at 0.969, and the highest mean Average Precision (mAP@50) at 0.978. In comparison, the YOLOv11 series was led by YOLO11x, which achieved the highest precision at 0.857, recall at 0.85, and mAP@50 at 0.91. For the YOLOv10 series, YOLOv10b and YOLOv10l both achieved the highest precision at 0.85, with YOLOv10n achieving the highest recall at 0.8 and mAP@50 at 0.89. These findings demonstrated that YOLOv12, when trained on realistic LLM-generated datasets surpassed its predecessors in key performance metrics. The technique also offered a cost-effective solution by reducing the need for extensive manual data collection in the agricultural field. In addition, this study compared the computational efficiency of all versions of YOLOv12, v11 and v10, where YOLOv11n reported the lowest inference time at 4.7 ms, compared to YOLOv12n's 5.6 ms and YOLOv10n's 5.9 ms. Although YOLOv12 is new and more accurate than YOLOv11, and YOLOv10, YOLO11n still stays the fastest YOLO model among YOLOv10, YOLOv11 and YOLOv12 series of models. (Index: YOLOv12, YOLOv11, YOLOv10, YOLOv13, YOLOv14, YOLOv15, YOLOE, YOLO Object detection)
Authors: Xuehao Gao, Yang Yang, Shaoyi Du, Guo-Jun Qi, Junwei Han
Abstract: As two intimate reciprocal tasks, scene-aware human motion synthesis and analysis require a joint understanding between multiple modalities, including 3D body motions, 3D scenes, and textual descriptions. In this paper, we integrate these two paired processes into a Co-Evolving Synthesis-Analysis (CESA) pipeline and mutually benefit their learning. Specifically, scene-aware text-to-human synthesis generates diverse indoor motion samples from the same textual description to enrich human-scene interaction intra-class diversity, thus significantly benefiting training a robust human motion analysis system. Reciprocally, human motion analysis would enforce semantic scrutiny on each synthesized motion sample to ensure its semantic consistency with the given textual description, thus improving realistic motion synthesis. Considering that real-world indoor human motions are goal-oriented and path-guided, we propose a cascaded generation strategy that factorizes text-driven scene-specific human motion generation into three stages: goal inferring, path planning, and pose synthesizing. Coupling CESA with this powerful cascaded motion synthesis model, we jointly improve realistic human motion synthesis and robust human motion analysis in 3D scenes.
Authors: Sohyun Jeong, Taewoong Kang, Hyojin Jang, Jaegul Choo
Abstract: With growing demand in media and social networks for personalized images, the need for advanced head-swapping techniques, integrating an entire head from the head image with the body from the body image, has increased. However, traditional head swapping methods heavily rely on face-centered cropped data with primarily frontal facing views, which limits their effectiveness in real world applications. Additionally, their masking methods, designed to indicate regions requiring editing, are optimized for these types of dataset but struggle to achieve seamless blending in complex situations, such as when the original data includes features like long hair extending beyond the masked area. To overcome these limitations and enhance adaptability in diverse and complex scenarios, we propose a novel head swapping method, HID, that is robust to images including the full head and the upper body, and handles from frontal to side views, while automatically generating context aware masks. For automatic mask generation, we introduce the IOMask, which enables seamless blending of the head and body, effectively addressing integration challenges. We further introduce the hair injection module to capture hair details with greater precision. Our experiments demonstrate that the proposed approach achieves state-of-the-art performance in head swapping, providing visually consistent and realistic results across a wide range of challenging conditions.
Authors: Runmao Yao, Yi Du, Zhuoqun Chen, Haoze Zheng, Chen Wang
Abstract: Room reidentification (ReID) is a challenging yet essential task with numerous applications in fields such as augmented reality (AR) and homecare robotics. Existing visual place recognition (VPR) methods, which typically rely on global descriptors or aggregate local features, often struggle in cluttered indoor environments densely populated with man-made objects. These methods tend to overlook the crucial role of object-oriented information. To address this, we propose AirRoom, an object-aware pipeline that integrates multi-level object-oriented information-from global context to object patches, object segmentation, and keypoints-utilizing a coarse-to-fine retrieval approach. Extensive experiments on four newly constructed datasets-MPReID, HMReID, GibsonReID, and ReplicaReID-demonstrate that AirRoom outperforms state-of-the-art (SOTA) models across nearly all evaluation metrics, with improvements ranging from 6% to 80%. Moreover, AirRoom exhibits significant flexibility, allowing various modules within the pipeline to be substituted with different alternatives without compromising overall performance. It also shows robust and consistent performance under diverse viewpoint variations.
Authors: Xiangjun Tang, Biao Zhang, Peter Wonka
Abstract: Realistic human geometry generation is an important yet challenging task, requiring both the preservation of fine clothing details and the accurate modeling of clothing-pose interactions. Geometry distributions, which can model the geometry of a single human as a distribution, provide a promising representation for high-fidelity synthesis. However, applying geometry distributions for human generation requires learning a dataset-level distribution over numerous individual geometry distributions. To address the resulting challenges, we propose a novel 3D human generative framework that, for the first time, models the distribution of human geometry distributions. Our framework operates in two stages: first, generating the human geometry distribution, and second, synthesizing high-fidelity humans by sampling from this distribution. We validate our method on two tasks: pose-conditioned 3D human generation and single-view-based novel pose generation. Experimental results demonstrate that our approach achieves the best quantitative results in terms of realism and geometric fidelity, outperforming state-of-the-art generative methods.
Authors: Guande Wu, Huan Song, Yawei Wang, Qiaojing Yan, Yijun Tian, Lin Lee Cheong, Panpan Xu
Abstract: Reasoning is increasingly crucial for various tasks. While chain-of-thought prompting enables large language models to leverage reasoning effectively, harnessing the reasoning capabilities of Vision-Language Models (VLMs) remains challenging. To solve this problem, we propose a novel self-distillation framework that enhances the reasoning capabilities of the model. The proposed framework introduces several key innovations. We start by employing a prompt library tailored to visual reasoning tasks to generate diverse in-context questions and utilize a two-step reasoning procedure to derive reasoning-guided responses. These responses are then used for self-distillation, enabling the model to internalize the reasoning process. Additionally, we improve the model architecture with several innovative components, including an intervention adapter for efficient parameter updates, a cross-modal skip connection to facilitate information exchange between modalities, and an ensemble learning algorithm to integrate diverse reasoning from multiple in-context questions. Extensive experiments show that our method significantly improves the baseline performance across five VQA datasets.
Authors: Yanlong Xu, Haoxuan Qu, Jun Liu, Wenxiao Zhang, Xun Yang
Abstract: The goal of point cloud localization based on linguistic description is to identify a 3D position using textual description in large urban environments, which has potential applications in various fields, such as determining the location for vehicle pickup or goods delivery. Ideally, for a textual description and its corresponding 3D location, the objects around the 3D location should be fully described in the text description. However, in practical scenarios, e.g., vehicle pickup, passengers usually describe only the part of the most significant and nearby surroundings instead of the entire environment. In response to this $\textbf{partially relevant}$ challenge, we propose $\textbf{CMMLoc}$, an uncertainty-aware $\textbf{C}$auchy-$\textbf{M}$ixture-$\textbf{M}$odel ($\textbf{CMM}$) based framework for text-to-point-cloud $\textbf{Loc}$alization. To model the uncertain semantic relations between text and point cloud, we integrate CMM constraints as a prior during the interaction between the two modalities. We further design a spatial consolidation scheme to enable adaptive aggregation of different 3D objects with varying receptive fields. To achieve precise localization, we propose a cardinal direction integration module alongside a modality pre-alignment strategy, helping capture the spatial relationships among objects and bringing the 3D objects closer to the text modality. Comprehensive experiments validate that CMMLoc outperforms existing methods, achieving state-of-the-art results on the KITTI360Pose dataset. Codes are available in this GitHub repository https://github.com/kevin301342/CMMLoc.
Authors: Xiaojun Bi, Shuo Li, Ziyue Wang, Fuwen Luo, Weizheng Qiao, Lu Han, Ziwei Sun, Peng Li, Yang Liu
Abstract: Dongba pictographs are the only pictographs still in use in the world. They have pictorial ideographic features, and their symbols carry rich cultural and contextual information. Due to the lack of relevant datasets, existing research has difficulty in advancing the study of semantic understanding of Dongba pictographs. To this end, we propose \textbf{DongbaMIE}, the first multimodal dataset for semantic understanding and extraction of Dongba pictographs, consisting of Dongba pictograph images and corresponding Chinese semantic annotations. DongbaMIE contains 23,530 sentence-level and 2,539 paragraph-level images, covering four semantic dimensions: objects, actions, relations, and attributes. We systematically evaluate multimodal large language models (MLLMs), such as GPT-4o, Gemini-2.0, and Qwen2-VL. Experimental results show that best F1 scores of proprietary models, GPT-4o and Gemini, for object extraction task are only 3.16 and 3.11 respectively. For the open-source model Qwen2-VL, it achieves only 11.49 after supervised fine-tuning. These suggest that current MLLMs still face significant challenges in accurately recognizing diverse semantic information in Dongba pictographs.
Authors: Paul J. Krassnig, Dieter P. Gruber
Abstract: Automatic visual inspection using machine learning-based methods plays a key role in achieving zero-defect policies in industry. Research on anomaly detection approaches is constrained by the availability of datasets that represent complex defect appearances and imperfect imaging conditions, which are typical to industrial processes. Recent benchmarks indicate that most publicly available datasets are biased towards optimal imaging conditions, leading to an overestimation of the methods' applicability to real-world industrial scenarios. To address this gap, we introduce the Industrial Screen Printing Anomaly Detection dataset (ISP-AD). It presents challenging small and weakly contrasted surface defects embedded within structured patterns exhibiting high permitted design variability. To the best of our knowledge, it is the largest publicly available industrial dataset to date, including both synthetic and real defects collected directly from the factory floor. In addition to the evaluation of defect detection performance of recent unsupervised anomaly detection methods, experiments on a mixed supervised training approach, incorporating both synthesized and real defects, were conducted. Even small amounts of injected real defects prove beneficial for model generalization. Furthermore, starting from training on purely synthetic defects, emerging real defective samples can be efficiently integrated into subsequent scalable training. Research findings indicate that supervision by means of both synthetic and accumulated real defects can complement each other, meeting demanded industrial inspection requirements such as low false positive rates and high recall. The presented unsupervised and supervised dataset splits are designed to emphasize research on unsupervised, self-supervised, and supervised approaches, enhancing their applicability to industrial settings.
Authors: Zisheng Chen, Chunwei Wang, Xiuwei Chen, Hang Xu, Jianhua Han, Xiaodan Liang
Abstract: We present SemHiTok, a unified image Tokenizer via Semantic-Guided Hierarchical codebook that provides consistent discrete feature representations for multimodal understanding and generation tasks. Recently, unified multimodal large models (MLLMs) for understanding and generation have sparked exploration within research community. Previous works attempt to train a unified image tokenizer by combining loss functions for semantic feature reconstruction and pixel reconstruction. However, due to the differing levels of features prioritized by multimodal understanding and generation tasks, joint training methods face significant challenges in achieving a good trade-off. SemHiTok addresses this challenge through Semantic-Guided Hierarchical codebook which builds texture sub-codebooks on pre-trained semantic codebook. This design decouples the training of semantic reconstruction and pixel reconstruction and equips the tokenizer with low-level texture feature extraction capability without degradation of high-level semantic feature extraction ability. Our experiments demonstrate that SemHiTok achieves excellent rFID score at 256X256resolution compared to other unified tokenizers, and exhibits competitive performance on multimodal understanding and generation tasks.
Authors: Nadarasar Bahavan, Sachith Seneviratne, Saman Halgamuge
Abstract: The widespread use of deep learning classifiers necessitates Open-set recognition (OSR), which enables the identification of input data not only from classes known during training but also from unknown classes that might be present in test data. Many existing OSR methods are computationally expensive due to the reliance on complex generative models or suffer from high training costs. We investigate OSR from a representation-learning perspective, specifically through spherical embeddings. We introduce SphOR, a computationally efficient representation learning method that models the feature space as a mixture of von Mises-Fisher distributions. This approach enables the use of semantically ambiguous samples during training, to improve the detection of samples from unknown classes. We further explore the relationship between OSR performance and key representation learning properties which influence how well features are structured in high-dimensional space. Extensive experiments on multiple OSR benchmarks demonstrate the effectiveness of our method, producing state-of-the-art results, with improvements up-to 6% that validate its performance. Code at https://github.com/nadarasarbahavan/SpHOR
Authors: Fei Wang, Chengcheng Chen, Hongyu Chen, Yugang Chang, Weiming Zeng
Abstract: Recently, large language models (LLMs) and vision-language models (VLMs) have achieved significant success, demonstrating remarkable capabilities in understanding various images and videos, particularly in classification and detection tasks. However, due to the substantial differences between remote sensing images and conventional optical images, these models face considerable challenges in comprehension, especially in detection tasks. Directly prompting VLMs with detection instructions often leads to unsatisfactory results. To address this issue, this letter explores the application of VLMs for object detection in remote sensing images. Specifically, we constructed supervised fine-tuning (SFT) datasets using publicly available remote sensing object detection datasets, including SSDD, HRSID, and NWPU-VHR-10. In these new datasets, we converted annotation information into JSON-compliant natural language descriptions, facilitating more effective understanding and training for the VLM. We then evaluate the detection performance of various fine-tuning strategies for VLMs and derive optimized model weights for object detection in remote sensing images. Finally, we evaluate the model's prior knowledge capabilities using natural language queries. Experimental results demonstrate that, without modifying the model architecture, remote sensing object detection can be effectively achieved using natural language alone. Additionally, the model exhibits the ability to perform certain vision question answering (VQA) tasks. Our datasets and related code will be released soon.
Authors: Jaa-Yeon Lee, Byunghee Cha, Jeongsol Kim, Jong Chul Ye
Abstract: While recent advancements in generative modeling have significantly improved text-image alignment, some residual misalignment between text and image representations still remains. Although many approaches have attempted to address this issue by fine-tuning models using various reward models, etc., we revisit the challenge from the perspective of representation alignment-an approach that has gained popularity with the success of REPresentation Alignment (REPA). We first argue that conventional text-to-image (T2I) diffusion models, typically trained on paired image and text data (i.e., positive pairs) by minimizing score matching or flow matching losses, is suboptimal from the standpoint of representation alignment. Instead, a better alignment can be achieved through contrastive learning that leverages both positive and negative pairs. To achieve this efficiently even with pretrained models, we introduce a lightweight contrastive fine tuning strategy called SoftREPA that uses soft text tokens. This approach improves alignment with minimal computational overhead by adding fewer than 1M trainable parameters to the pretrained model. Our theoretical analysis demonstrates that our method explicitly increases the mutual information between text and image representations, leading to enhanced semantic consistency. Experimental results across text-to-image generation and text-guided image editing tasks validate the effectiveness of our approach in improving the semantic consistency of T2I generative models.
Authors: Dong Li, Guihong Wan, Xintao Wu, Xinyu Wu, Xiaohui Chen, Yi He, Christine G. Lian, Peter K. Sorger, Yevgeniy R. Semenov, Chen Zhao
Abstract: Foundation models have emerged as a powerful paradigm in computational pathology (CPath), enabling scalable and generalizable analysis of histopathological images. While early developments centered on uni-modal models trained solely on visual data, recent advances have highlighted the promise of multi-modal foundation models that integrate heterogeneous data sources such as textual reports, structured domain knowledge, and molecular profiles. In this survey, we provide a comprehensive and up-to-date review of multi-modal foundation models in CPath, with a particular focus on models built upon hematoxylin and eosin (H&E) stained whole slide images (WSIs) and tile-level representations. We categorize 32 state-of-the-art multi-modal foundation models into three major paradigms: vision-language, vision-knowledge graph, and vision-gene expression. We further divide vision-language models into non-LLM-based and LLM-based approaches. Additionally, we analyze 28 available multi-modal datasets tailored for pathology, grouped into image-text pairs, instruction datasets, and image-other modality pairs. Our survey also presents a taxonomy of downstream tasks, highlights training and evaluation strategies, and identifies key challenges and future directions. We aim for this survey to serve as a valuable resource for researchers and practitioners working at the intersection of pathology and AI.
Authors: Beier Zhu, Jiequan Cui, Hanwang Zhang, Chi Zhang
Abstract: While image-text foundation models have succeeded across diverse downstream tasks, they still face challenges in the presence of spurious correlations between the input and label. To address this issue, we propose a simple three-step approach,Project-Probe-Aggregate (PPA), that enables parameter-efficient fine-tuning for foundation models without relying on group annotations. Building upon the failure-based debiasing scheme, our method, PPA, improves its two key components: minority samples identification and the robust training algorithm. Specifically, we first train biased classifiers by projecting image features onto the nullspace of class proxies from text encoders. Next, we infer group labels using the biased classifier and probe group targets with prior correction. Finally, we aggregate group weights of each class to produce the debiased classifier. Our theoretical analysis shows that our PPA enhances minority group identification and is Bayes optimal for minimizing the balanced group error, mitigating spurious correlations. Extensive experimental results confirm the effectiveness of our PPA: it outperforms the state-of-the-art by an average worst-group accuracy while requiring less than 0.01% tunable parameters without training group labels.
Authors: Ayush Jain, Alexander Swerdlow, Yuzhou Wang, Sergio Arnaud, Ada Martin, Alexander Sax, Franziska Meier, Katerina Fragkiadaki
Abstract: Progress in 3D vision-language learning has been hindered by the scarcity of large-scale 3D datasets. We introduce UniVLG, a unified architecture for 2D and 3D vision-language understanding that bridges the gap between existing 2D-centric models and the rich 3D sensory data available in embodied systems. Our approach initializes most model weights from pre-trained 2D models and trains on both 2D and 3D vision-language data. We propose a novel language-conditioned mask decoder shared across 2D and 3D modalities to ground objects effectively in both RGB and RGB-D images, outperforming box-based approaches. To further reduce the domain gap between 2D and 3D, we incorporate 2D-to-3D lifting strategies, enabling UniVLG to utilize 2D data to enhance 3D performance. With these innovations, our model achieves state-of-the-art performance across multiple 3D vision-language grounding tasks, demonstrating the potential of transferring advances from 2D vision-language learning to the data-constrained 3D domain. Furthermore, co-training on both 2D and 3D data enhances performance across modalities without sacrificing 2D capabilities. By removing the reliance on 3D mesh reconstruction and ground-truth object proposals, UniVLG sets a new standard for realistic, embodied-aligned evaluation. Code and additional visualizations are available at https://univlg.github.io .
URLs: https://univlg.github.io
Authors: Lilin Zhang, Chengpei Wu, Ning Yang
Abstract: Existing adversarial training (AT) methods often suffer from incomplete perturbation, meaning that not all non-robust features are perturbed when generating adversarial examples (AEs). This results in residual correlations between non-robust features and labels, leading to suboptimal learning of robust features. However, achieving complete perturbation, i.e., perturbing as many non-robust features as possible, is challenging due to the difficulty in distinguishing robust and non-robust features and the sparsity of labeled data. To address these challenges, we propose a novel approach called Weakly Supervised Contrastive Adversarial Training (WSCAT). WSCAT ensures complete perturbation for improved learning of robust features by disrupting correlations between non-robust features and labels through complete AE generation over partially labeled data, grounded in information theory. Extensive theoretical analysis and comprehensive experiments on widely adopted benchmarks validate the superiority of WSCAT. Our code is available at https://github.com/zhang-lilin/WSCAT.
Authors: Fanhu Zeng, Hao Tang, Yihua Shao, Siyu Chen, Ling Shao, Yan Wang
Abstract: A high-performance image compression algorithm is crucial for real-time information transmission across numerous fields. Despite rapid progress in image compression, computational inefficiency and poor redundancy modeling still pose significant bottlenecks, limiting practical applications. Inspired by the effectiveness of state space models (SSMs) in capturing long-range dependencies, we leverage SSMs to address computational inefficiency in existing methods and improve image compression from multiple perspectives. In this paper, we integrate the advantages of SSMs for better efficiency-performance trade-off and propose an enhanced image compression approach through refined context modeling, which we term MambaIC. Specifically, we explore context modeling to adaptively refine the representation of hidden states. Additionally, we introduce window-based local attention into channel-spatial entropy modeling to reduce potential spatial redundancy during compression, thereby increasing efficiency. Comprehensive qualitative and quantitative results validate the effectiveness and efficiency of our approach, particularly for high-resolution image compression. Code is released at https://github.com/AuroraZengfh/MambaIC.
Authors: Tianyu Li, Yihang Qiu, Zhenhua Wu, Carl Lindstr\"om, Peng Su, Matthias Nie{\ss}ner, Hongyang Li
Abstract: Multi-traversal data, commonly collected through daily commutes or by self-driving fleets, provides multiple viewpoints for scene reconstruction within a road block. This data offers significant potential for high-quality novel view synthesis, which is crucial for applications such as autonomous vehicle simulators. However, inherent challenges in multi-traversal data often result in suboptimal reconstruction quality, including variations in appearance and the presence of dynamic objects. To address these issues, we propose Multi-Traversal Gaussian Splatting (MTGS), a novel approach that reconstructs high-quality driving scenes from arbitrarily collected multi-traversal data by modeling a shared static geometry while separately handling dynamic elements and appearance variations. Our method employs a multi-traversal dynamic scene graph with a shared static node and traversal-specific dynamic nodes, complemented by color correction nodes with learnable spherical harmonics coefficient residuals. This approach enables high-fidelity novel view synthesis and provides flexibility to navigate any viewpoint. We conduct extensive experiments on a large-scale driving dataset, nuPlan, with multi-traversal data. Our results demonstrate that MTGS improves LPIPS by 23.5% and geometry accuracy by 46.3% compared to single-traversal baselines. The code and data would be available to the public.
Authors: Tony Danjun Wang, Lennart Bastian, Tobias Czempiel, Christian Heiliger, Nassir Navab
Abstract: Surgical domain models improve workflow optimization through automated predictions of each staff member's surgical role. However, mounting evidence indicates that team familiarity and individuality impact surgical outcomes. We present a novel staff-centric modeling approach that characterizes individual team members through their distinctive movement patterns and physical characteristics, enabling long-term tracking and analysis of surgical personnel across multiple procedures. To address the challenge of inter-clinic variability, we develop a generalizable re-identification framework that encodes sequences of 3D point clouds to capture shape and articulated motion patterns unique to each individual. Our method achieves 86.19% accuracy on realistic clinical data while maintaining 75.27% accuracy when transferring between different environments - a 12% improvement over existing methods. When used to augment markerless personnel tracking, our approach improves accuracy by over 50%. Through extensive validation across three datasets and the introduction of a novel workflow visualization technique, we demonstrate how our framework can reveal novel insights into surgical team dynamics and space utilization patterns, advancing methods to analyze surgical workflows and team coordination.
Authors: Jie Huang, Haorui Chen, Jiaxuan Ren, Siran Peng, Liangjian Deng
Abstract: Currently, deep learning-based methods for remote sensing pansharpening have advanced rapidly. However, many existing methods struggle to fully leverage feature heterogeneity and redundancy, thereby limiting their effectiveness. We use the covariance matrix to model the feature heterogeneity and redundancy and propose Correlation-Aware Covariance Weighting (CACW) to adjust them. CACW captures these correlations through the covariance matrix, which is then processed by a nonlinear function to generate weights for adjustment. Building upon CACW, we introduce a general adaptive dual-level weighting mechanism (ADWM) to address these challenges from two key perspectives, enhancing a wide range of existing deep-learning methods. First, Intra-Feature Weighting (IFW) evaluates correlations among channels within each feature to reduce redundancy and enhance unique information. Second, Cross-Feature Weighting (CFW) adjusts contributions across layers based on inter-layer correlations, refining the final output. Extensive experiments demonstrate the superior performance of ADWM compared to recent state-of-the-art (SOTA) methods. Furthermore, we validate the effectiveness of our approach through generality experiments, redundancy visualization, comparison experiments, key variables and complexity analysis, and ablation studies. Our code is available at https://github.com/Jie-1203/ADWM.
Authors: Shashikant Verma, Harish Katti, Soumyaratna Debnath, Yamuna Swamy, Shanmuganathan Raman
Abstract: We introduce STEP, a novel framework utilizing Transformer-based discriminative model prediction for simultaneous tracking and estimation of pose across diverse animal species and humans. We are inspired by the fact that the human brain exploits spatiotemporal continuity and performs concurrent localization and pose estimation despite the specialization of brain areas for form and motion processing. Traditional discriminative models typically require predefined target states for determining model weights, a challenge we address through Gaussian Map Soft Prediction (GMSP) and Offset Map Regression Adapter (OMRA) Modules. These modules remove the necessity of keypoint target states as input, streamlining the process. Our method starts with a known target state in the initial frame of a given video sequence. It then seamlessly tracks the target and estimates keypoints of anatomical importance as output for subsequent frames. Unlike prevalent top-down pose estimation methods, our approach doesn't rely on per-frame target detections due to its tracking capability. This facilitates a significant advancement in inference efficiency and potential applications. We train and validate our approach on datasets encompassing diverse species. Our experiments demonstrate superior results compared to existing methods, opening doors to various applications, including but not limited to action recognition and behavioral analysis.
Authors: Yuxiang Lai, Jike Zhong, Ming Li, Shitian Zhao, Xiaofeng Yang
Abstract: Vision-language models (VLMs) have advanced reasoning in natural scenes, but their role in medical imaging remains underexplored. Medical reasoning tasks demand robust image analysis and well-justified answers, posing challenges due to the complexity of medical images. Transparency and trustworthiness are essential for clinical adoption and regulatory compliance. We introduce Med-R1, a framework exploring reinforcement learning (RL) to enhance VLMs' generalizability and trustworthiness in medical reasoning. Leveraging the DeepSeek strategy, we employ Group Relative Policy Optimization (GRPO) to guide reasoning paths via reward signals. Unlike supervised fine-tuning (SFT), which often overfits and lacks generalization, RL fosters robust and diverse reasoning. Med-R1 is evaluated across eight medical imaging modalities: CT, MRI, Ultrasound, Dermoscopy, Fundus Photography, Optical Coherence Tomography (OCT), Microscopy, and X-ray Imaging. Compared to its base model, Qwen2-VL-2B, Med-R1 achieves a 29.94% accuracy improvement and outperforms Qwen2-VL-72B, which has 36 times more parameters. Testing across five question types-modality recognition, anatomy identification, disease diagnosis, lesion grading, and biological attribute analysis Med-R1 demonstrates superior generalization, exceeding Qwen2-VL-2B by 32.06% and surpassing Qwen2-VL-72B in question-type generalization. These findings show that RL improves medical reasoning and enables parameter-efficient models to outperform significantly larger ones. With interpretable reasoning outputs, Med-R1 represents a promising step toward generalizable, trustworthy, and clinically viable medical VLMs.
Authors: Baiqin Wang, Xiangyu Zhu, Fan Shen, Hao Xu, Zhen Lei
Abstract: Recent advancements in audio-driven talking face generation have made great progress in lip synchronization. However, current methods often lack sufficient control over facial animation such as speaking style and emotional expression, resulting in uniform outputs. In this paper, we focus on improving two key factors: lip-audio alignment and emotion control, to enhance the diversity and user-friendliness of talking videos. Lip-audio alignment control focuses on elements like speaking style and the scale of lip movements, whereas emotion control is centered on generating realistic emotional expressions, allowing for modifications in multiple attributes such as intensity. To achieve precise control of facial animation, we propose a novel framework, PC-Talk, which enables lip-audio alignment and emotion control through implicit keypoint deformations. First, our lip-audio alignment control module facilitates precise editing of speaking styles at the word level and adjusts lip movement scales to simulate varying vocal loudness levels, maintaining lip synchronization with the audio. Second, our emotion control module generates vivid emotional facial features with pure emotional deformation. This module also enables the fine modification of intensity and the combination of multiple emotions across different facial regions. Our method demonstrates outstanding control capabilities and achieves state-of-the-art performance on both HDTF and MEAD datasets in extensive experiments.
Authors: Shoubin Yu, Difan Liu, Ziqiao Ma, Yicong Hong, Yang Zhou, Hao Tan, Joyce Chai, Mohit Bansal
Abstract: Recent video diffusion models have enhanced video editing, but it remains challenging to handle instructional editing and diverse tasks (e.g., adding, removing, changing) within a unified framework. In this paper, we introduce VEGGIE, a Video Editor with Grounded Generation from Instructions, a simple end-to-end framework that unifies video concept editing, grounding, and reasoning based on diverse user instructions. Specifically, given a video and text query, VEGGIE first utilizes an MLLM to interpret user intentions in instructions and ground them to the video contexts, generating frame-specific grounded task queries for pixel-space responses. A diffusion model then renders these plans and generates edited videos that align with user intent. To support diverse tasks and complex instructions, we employ a curriculum learning strategy: first aligning the MLLM and video diffusion model with large-scale instructional image editing data, followed by end-to-end fine-tuning on high-quality multitask video data. Additionally, we introduce a novel data synthesis pipeline to generate paired instructional video editing data for model training. It transforms static image data into diverse, high-quality video editing samples by leveraging Image-to-Video models to inject dynamics. VEGGIE shows strong performance in instructional video editing with different editing skills, outperforming the best instructional baseline as a versatile model, while other models struggle with multi-tasking. VEGGIE also excels in video object grounding and reasoning segmentation, where other baselines fail. We further reveal how the multiple tasks help each other and highlight promising applications like zero-shot multimodal instructional and in-context video editing.
Authors: Qing Li, Jiahui Geng, Derui Zhu, Fengyu Cai, Chenyang Lyu, Fakhri Karray
Abstract: Unlearning methods for vision-language models (VLMs) have primarily adapted techniques from large language models (LLMs), relying on weight updates that demand extensive annotated forget sets. Moreover, these methods perform unlearning at a coarse granularity, often leading to excessive forgetting and reduced model utility. To address this issue, we introduce SAUCE, a novel method that leverages sparse autoencoders (SAEs) for fine-grained and selective concept unlearning in VLMs. Briefly, SAUCE first trains SAEs to capture high-dimensional, semantically rich sparse features. It then identifies the features most relevant to the target concept for unlearning. During inference, it selectively modifies these features to suppress specific concepts while preserving unrelated information. We evaluate SAUCE on two distinct VLMs, LLaVA-v1.5-7B and LLaMA-3.2-11B-Vision-Instruct, across two types of tasks: concrete concept unlearning (objects and sports scenes) and abstract concept unlearning (emotions, colors, and materials), encompassing a total of 60 concepts. Extensive experiments demonstrate that SAUCE outperforms state-of-the-art methods by 18.04% in unlearning quality while maintaining comparable model utility. Furthermore, we investigate SAUCE's robustness against widely used adversarial attacks, its transferability across models, and its scalability in handling multiple simultaneous unlearning requests. Our findings establish SAUCE as an effective and scalable solution for selective concept unlearning in VLMs.
Authors: Ying Liu, Yijing Hua, Haojiang Chai, Yanbo Wang, TengQi Ye
Abstract: Open-vocabulary detectors are proposed to locate and recognize objects in novel classes. However, variations in vision-aware language vocabulary data used for open-vocabulary learning can lead to unfair and unreliable evaluations. Recent evaluation methods have attempted to address this issue by incorporating object properties or adding locations and characteristics to the captions. Nevertheless, since these properties and locations depend on the specific details of the images instead of classes, detectors can not make accurate predictions without precise descriptions provided through human annotation. This paper introduces 3F-OVD, a novel task that extends supervised fine-grained object detection to the open-vocabulary setting. Our task is intuitive and challenging, requiring a deep understanding of Fine-grained captions and careful attention to Fine-grained details in images in order to accurately detect Fine-grained objects. Additionally, due to the scarcity of qualified fine-grained object detection datasets, we have created a new dataset, NEU-171K, tailored for both supervised and open-vocabulary settings. We benchmark state-of-the-art object detectors on our dataset for both settings. Furthermore, we propose a simple yet effective post-processing technique.
Authors: Imanol G. Estepa, Jes\'us M. Rodr\'iguez-de-Vera, Ignacio Saras\'ua, Bhalaji Nagarajan, Petia Radeva
Abstract: While representation learning and generative modeling seek to understand visual data, unifying both domains remains unexplored. Recent Unified Self-Supervised Learning (SSL) methods have started to bridge the gap between both paradigms. However, they rely solely on semantic token reconstruction, which requires an external tokenizer during training -- introducing a significant overhead. In this work, we introduce Sorcen, a novel unified SSL framework, incorporating a synergic Contrastive-Reconstruction objective. Our Contrastive objective, "Echo Contrast", leverages the generative capabilities of Sorcen, eliminating the need for additional image crops or augmentations during training. Sorcen "generates" an echo sample in the semantic token space, forming the contrastive positive pair. Sorcen operates exclusively on precomputed tokens, eliminating the need for an online token transformation during training, thereby significantly reducing computational overhead. Extensive experiments on ImageNet-1k demonstrate that Sorcen outperforms the previous Unified SSL SoTA by 0.4%, 1.48 FID, 1.76%, and 1.53% on linear probing, unconditional image generation, few-shot learning, and transfer learning, respectively, while being 60.8% more efficient. Additionally, Sorcen surpasses previous single-crop MIM SoTA in linear probing and achieves SoTA performance in unconditional image generation, highlighting significant improvements and breakthroughs in Unified SSL models.
Authors: Amir Hamza, Andrea Caraffa, Davide Boscaini, Fabio Poiesi
Abstract: Three-dimensional local descriptors are crucial for encoding geometric surface properties, making them essential for various point cloud understanding tasks. Among these descriptors, GeDi has demonstrated strong zero-shot 6D pose estimation capabilities but remains computationally impractical for real-world applications due to its expensive inference process. Can we retain GeDi's effectiveness while significantly improving its efficiency? In this paper, we explore this question by introducing a knowledge distillation framework that trains an efficient student model to regress local descriptors from a GeDi teacher. Our key contributions include: an efficient large-scale training procedure that ensures robustness to occlusions and partial observations while operating under compute and storage constraints, and a novel loss formulation that handles weak supervision from non-distinctive teacher descriptors. We validate our approach on five BOP Benchmark datasets and demonstrate a significant reduction in inference time while maintaining competitive performance with existing methods, bringing zero-shot 6D pose estimation closer to real-time feasibility. Project Website: https://tev-fbk.github.io/dGeDi/
Authors: Zinqin Huang, Gu Wang, Chenyangguang Zhang, Ruida Zhang, Xiu Li, Xiangyang Ji
Abstract: Recent advances in RGBD-based category-level object pose estimation have been limited by their reliance on precise depth information, restricting their broader applicability. In response, RGB-based methods have been developed. Among these methods, geometry-guided pose regression that originated from instance-level tasks has demonstrated strong performance. However, we argue that the NOCS map is an inadequate intermediate representation for geometry-guided pose regression method, as its many-to-one correspondence with category-level pose introduces redundant instance-specific information, resulting in suboptimal results. This paper identifies the intra-class variation problem inherent in pose regression based solely on the NOCS map and proposes the Intra-class Variation-Free Consensus (IVFC) map, a novel coordinate representation generated from the category-level consensus model. By leveraging the complementary strengths of the NOCS map and the IVFC map, we introduce GIVEPose, a framework that implements Gradual Intra-class Variation Elimination for category-level object pose estimation. Extensive evaluations on both synthetic and real-world datasets demonstrate that GIVEPose significantly outperforms existing state-of-the-art RGB-based approaches, achieving substantial improvements in category-level object pose estimation. Our code is available at https://github.com/ziqin-h/GIVEPose.
Authors: Mingzhe Zheng, Yongqi Xu, Haojian Huang, Xuran Ma, Yexin Liu, Wenjie Shu, Yatian Pang, Feilong Tang, Qifeng Chen, Harry Yang, Ser-Nam Lim
Abstract: Current video generation models excel at short clips but fail to produce cohesive multi-shot narratives due to disjointed visual dynamics and fractured storylines. Existing solutions either rely on extensive manual scripting/editing or prioritize single-shot fidelity over cross-scene continuity, limiting their practicality for movie-like content. We introduce VideoGen-of-Thought (VGoT), a step-by-step framework that automates multi-shot video synthesis from a single sentence by systematically addressing three core challenges: (1) Narrative Fragmentation: Existing methods lack structured storytelling. We propose dynamic storyline modeling, which first converts the user prompt into concise shot descriptions, then elaborates them into detailed, cinematic specifications across five domains (character dynamics, background continuity, relationship evolution, camera movements, HDR lighting), ensuring logical narrative progression with self-validation. (2) Visual Inconsistency: Existing approaches struggle with maintaining visual consistency across shots. Our identity-aware cross-shot propagation generates identity-preserving portrait (IPP) tokens that maintain character fidelity while allowing trait variations (expressions, aging) dictated by the storyline. (3) Transition Artifacts: Abrupt shot changes disrupt immersion. Our adjacent latent transition mechanisms implement boundary-aware reset strategies that process adjacent shots' features at transition points, enabling seamless visual flow while preserving narrative continuity. VGoT generates multi-shot videos that outperform state-of-the-art baselines by 20.4% in within-shot face consistency and 17.4% in style consistency, while achieving over 100% better cross-shot consistency and 10x fewer manual adjustments than alternatives.
Authors: Giorgia Crosilla, Lukas Klic, Giovanni Colavizza
Abstract: Traditional machine learning models for Handwritten Text Recognition (HTR) rely on supervised training, requiring extensive manual annotations, and often produce errors due to the separation between layout and text processing. In contrast, Multimodal Large Language Models (MLLMs) offer a general approach to recognizing diverse handwriting styles without the need for model-specific training. The study benchmarks various proprietary and open-source LLMs against Transkribus models, evaluating their performance on both modern and historical datasets written in English, French, German, and Italian. In addition, emphasis is placed on testing the models' ability to autonomously correct previously generated outputs. Findings indicate that proprietary models, especially Claude 3.5 Sonnet, outperform open-source alternatives in zero-shot settings. MLLMs achieve excellent results in recognizing modern handwriting and exhibit a preference for the English language due to their pre-training dataset composition. Comparisons with Transkribus show no consistent advantage for either approach. Moreover, LLMs demonstrate limited ability to autonomously correct errors in zero-shot transcriptions.
Authors: Jiamian Wang, Kunpeng Li, Yulun Zhang, Xin Yuan, Zhiqiang Tao
Abstract: Snapshot compressive imaging (SCI) surges as a novel way of capturing hyperspectral images. It operates an optical encoder to compress the 3D data into a 2D measurement and adopts a software decoder for the signal reconstruction. Recently, a representative SCI set-up of coded aperture snapshot compressive imager (CASSI) with Transformer reconstruction backend remarks high-fidelity sensing performance. However, dominant spatial and spectral attention designs show limitations in hyperspectral modeling. The spatial attention values describe the inter-pixel correlation but overlook the across-spectra variation within each pixel. The spectral attention size is unscalable to the token spatial size and thus bottlenecks information allocation. Besides, CASSI entangles the spatial and spectral information into a 2D measurement, placing a barrier for information disentanglement and modeling. In addition, CASSI blocks the light with a physical binary mask, yielding the masked data loss. To tackle above challenges, we propose a spatial-spectral (S2-) Transformer implemented by a paralleled attention design and a mask-aware learning strategy. Firstly, we systematically explore pros and cons of different spatial (-spectral) attention designs, based on which we find performing both attentions in parallel well disentangles and models the blended information. Secondly, the masked pixels induce higher prediction difficulty and should be treated differently from unmasked ones. We adaptively prioritize the loss penalty attributing to the mask structure by referring to the mask-encoded prediction as an uncertainty estimator. We theoretically discuss the distinct convergence tendencies between masked/unmasked regions of the proposed learning strategy. Extensive experiments demonstrate that on average, the results of the proposed method are superior over the state-of-the-art method.
Authors: Zahra Shamsi, Drew Bryant, Jacob Wilson, Xiaoyu Qu, Avinava Dubey, Konik Kothari, Mostafa Dehghani, Mariya Chavarha, Valerii Likhosherstov, Brian Williams, Michael Frumkin, Fred Appelbaum, Krzysztof Choromanski, Ali Bashir, Min Fang
Abstract: We present a machine learning method capable of accurately detecting chromosome abnormalities that cause blood cancers directly from microscope images of the metaphase stage of cell division. The pipeline is built on a series of fine-tuned Vision Transformers. Current state of the art (and standard clinical practice) requires expensive, manual expert analysis, whereas our pipeline takes only 15 seconds per metaphase image. Using a novel pretraining-finetuning strategy to mitigate the challenge of data scarcity, we achieve a high precision-recall score of 94% AUC for the clinically significant del(5q) and t(9;22) anomalies. Our method also unlocks zero-shot detection of rare aberrations based on model latent embeddings. The ability to quickly, accurately, and scalably diagnose genetic abnormalities directly from metaphase images could transform karyotyping practice and improve patient outcomes. We will make code publicly available.
Authors: Jingyun Xue, Tao Wang, Pengwen Dai, Kaihao Zhang
Abstract: Under-Display Camera (UDC) is an emerging technology that achieves full-screen display via hiding the camera under the display panel. However, the current implementation of UDC causes serious degradation. The incident light required for camera imaging undergoes attenuation and diffraction when passing through the display panel, leading to various artifacts in UDC imaging. Presently, the prevailing UDC image restoration methods predominantly utilize convolutional neural network architectures, whereas Transformer-based methods have exhibited superior performance in the majority of image restoration tasks. This is attributed to the Transformer's capability to sample global features for the local reconstruction of images, thereby achieving high-quality image restoration. In this paper, we observe that when using the Vision Transformer for UDC degraded image restoration, the global attention samples a large amount of redundant information and noise. Furthermore, compared to the ordinary Transformer employing dense attention, the Transformer utilizing sparse attention can alleviate the adverse impact of redundant information and noise. Building upon this discovery, we propose a Segmentation Guided Sparse Transformer method (SGSFormer) for the task of restoring high-quality images from UDC degraded images. Specifically, we utilize sparse self-attention to filter out redundant information and noise, directing the model's attention to focus on the features more relevant to the degraded regions in need of reconstruction. Moreover, we integrate the instance segmentation map as prior information to guide the sparse self-attention in filtering and focusing on the correct regions.
Authors: Shuhong Liu, Tianchen Deng, Heng Zhou, Liuzhuozheng Li, Hongyu Wang, Danwei Wang, Mingrui Li
Abstract: Gaussian Splatting SLAMs have made significant advancements in improving the efficiency and fidelity of real-time reconstructions. However, these systems often encounter incomplete reconstructions in complex indoor environments, characterized by substantial holes due to unobserved geometry caused by obstacles or limited view angles. To address this challenge, we present Manhattan Gaussian SLAM, an RGB-D system that leverages the Manhattan World hypothesis to enhance geometric accuracy and completeness. By seamlessly integrating fused line segments derived from structured scenes, our method ensures robust tracking in textureless indoor areas. Moreover, The extracted lines and planar surface assumption allow strategic interpolation of new Gaussians in regions of missing geometry, enabling efficient scene completion. Extensive experiments conducted on both synthetic and real-world scenes demonstrate that these advancements enable our method to achieve state-of-the-art performance, marking a substantial improvement in the capabilities of Gaussian SLAM systems.
Authors: Omer Sahin Tas, Royden Wagner
Abstract: Transformer-based models generate hidden states that are difficult to interpret. In this work, we analyze hidden states and modify them at inference, with a focus on motion forecasting. We use linear probing to analyze whether interpretable features are embedded in hidden states. Our experiments reveal high probing accuracy, indicating latent space regularities with functionally important directions. Building on this, we use the directions between hidden states with opposing features to fit control vectors. At inference, we add our control vectors to hidden states and evaluate their impact on predictions. Remarkably, such modifications preserve the feasibility of predictions. We further refine our control vectors using sparse autoencoders (SAEs). This leads to more linear changes in predictions when scaling control vectors. Our approach enables mechanistic interpretation as well as zero-shot generalization to unseen dataset characteristics with negligible computational overhead.
Authors: Boyi Li, Ligeng Zhu, Ran Tian, Shuhan Tan, Yuxiao Chen, Yao Lu, Yin Cui, Sushant Veer, Max Ehrlich, Jonah Philion, Xinshuo Weng, Fuzhao Xue, Linxi Fan, Yuke Zhu, Jan Kautz, Andrew Tao, Ming-Yu Liu, Sanja Fidler, Boris Ivanovic, Trevor Darrell, Jitendra Malik, Song Han, Marco Pavone
Abstract: We propose Wolf, a WOrLd summarization Framework for accurate video captioning. Wolf is an automated captioning framework that adopts a mixture-of-experts approach, leveraging complementary strengths of Vision Language Models (VLMs). By utilizing both image and video models, our framework captures different levels of information and summarizes them efficiently. Our approach can be applied to enhance video understanding, auto-labeling, and captioning. To evaluate caption quality, we introduce CapScore, an LLM-based metric to assess the similarity and quality of generated captions compared to the ground truth captions. We further build four human-annotated datasets in three domains: autonomous driving, general scenes, and robotics, to facilitate comprehensive comparisons. We show that Wolf achieves superior captioning performance compared to state-of-the-art approaches from the research community (VILA1.5, CogAgent) and commercial solutions (Gemini-Pro-1.5, GPT-4V). For instance, in comparison with GPT-4V, Wolf improves CapScore both quality-wise by 55.6% and similarity-wise by 77.4% on challenging driving videos. Finally, we establish a benchmark for video captioning and introduce a leaderboard, aiming to accelerate advancements in video understanding, captioning, and data alignment. Webpage: https://wolfv0.github.io/.
Authors: Ruining Yang, Yi Xu, Yun Fu, Lili Su
Abstract: Trajectory prediction is a core task in autonomous driving. However, training advanced trajectory prediction models on large-scale datasets is both time-consuming and computationally expensive. In addition, the imbalanced distribution of driving scenarios often biases models toward data-rich cases, limiting performance in safety-critical, data-scarce conditions. To address these challenges, we propose the Sample Selection for Trajectory Prediction (SSTP) framework, which constructs a compact yet balanced dataset for trajectory prediction. SSTP consists of two main stages (1) Extraction, in which a pretrained trajectory prediction model computes gradient vectors for each sample to capture their influence on parameter updates; and (2) Selection, where a submodular function is applied to greedily choose a representative subset that covers diverse driving scenarios. This approach significantly reduces the dataset size and mitigates scenario imbalance, without sacrificing prediction accuracy and even improving in high-density cases. We evaluate our proposed SSTP on the Argoverse 1 and Argoverse 2 benchmarks using a wide range of recent state-of-the-art models. Our experiments demonstrate that SSTP achieves comparable performance to full-dataset training using only half the data while delivering substantial improvements in high-density traffic scenes and significantly reducing training time. Importantly, SSTP exhibits strong generalization and robustness, and the selected subset is model-agnostic, offering a broadly applicable solution.
Authors: Daixuan Cheng, Shaohan Huang, Ziyu Zhu, Xintong Zhang, Wayne Xin Zhao, Zhongzhi Luan, Bo Dai, Zhenliang Zhang
Abstract: Adapting general multimodal large language models (MLLMs) to specific domains, such as scientific and industrial fields, is highly significant in promoting their practical applications. This paper systematically investigates domain adaptation of MLLMs through post-training, focusing on data synthesis, training pipelines, and task evaluation. (1) Data Synthesis: Using only open-source models, we develop a generate-then-filter pipeline that curates diverse visual instruction tasks based on domain-specific image-caption pairs. The resulting data surpass the data synthesized by manual rules or strong closed-source models (e.g., GPT-4V) in enhancing domain-specific performance. (2) Training Pipeline: While the two-stage training--initially on image-caption pairs followed by visual instruction tasks--is commonly adopted for developing general MLLMs, we apply a single-stage training pipeline to enhance task diversity for domain-specific post-training. (3) Task Evaluation: We conduct extensive experiments in high-impact domains such as biomedicine, food, and remote sensing, by post-training a variety of MLLMs and then evaluating MLLM performance on various domain-specific tasks. Furthermore, we fully open-source our models, code, and data to encourage future research in this area.
Authors: Yi Chen, Yuying Ge, Weiliang Tang, Yizhuo Li, Yixiao Ge, Mingyu Ding, Ying Shan, Xihui Liu
Abstract: Recent developments in Large Language Models pre-trained on extensive corpora have shown significant success in various natural language processing tasks with minimal fine-tuning. This success offers new promise for robotics, which has long been constrained by the high cost of action-labeled data. We ask: given the abundant video data containing interaction-related knowledge available as a rich "corpus", can a similar generative pre-training approach be effectively applied to enhance robot learning? The key challenge is to identify an effective representation for autoregressive pre-training that benefits robot manipulation tasks. Inspired by the way humans learn new skills through observing dynamic environments, we propose that effective robotic learning should emphasize motion-related knowledge, which is closely tied to low-level actions and is hardware-agnostic, facilitating the transfer of learned motions to actual robot actions. To this end, we introduce Moto, which converts video content into latent Motion Token sequences by a Latent Motion Tokenizer, learning a bridging "language" of motion from videos in an unsupervised manner. We pre-train Moto-GPT through motion token autoregression, enabling it to capture diverse visual motion knowledge. After pre-training, Moto-GPT demonstrates the promising ability to produce semantically interpretable motion tokens, predict plausible motion trajectories, and assess trajectory rationality through output likelihood. To transfer learned motion priors to real robot actions, we implement a co-fine-tuning strategy that seamlessly bridges latent motion token prediction and real robot control. Extensive experiments show that the fine-tuned Moto-GPT exhibits superior robustness and efficiency on robot manipulation benchmarks, underscoring its effectiveness in transferring knowledge from video data to downstream visual manipulation tasks.
Authors: Aleix Boquet-Pujadas
Abstract: Mechanobiology is gaining more and more traction as the fundamental role of physical forces in biological function becomes clearer. Forces at the microscale are often measured indirectly using inverse problems such as Traction Force Microscopy because biological experiments are hard to access with physical probes. In contrast with the experimental nature of biology and physics, these measurements do not come with error bars, confidence regions, or p-values. The aim of this manuscript is to publicize this issue and to propose a first step towards a remedy therefor in the form of a general reconstruction framework. We also show that this opens the door to hypothesis testing of seemingly abstract experimental questions.
Authors: Haoming Xu, Ningyuan Zhao, Liming Yang, Sendong Zhao, Shumin Deng, Mengru Wang, Bryan Hooi, Nay Oo, Huajun Chen, Ningyu Zhang
Abstract: Current unlearning methods for large language models usually rely on reverse optimization to reduce target token probabilities. However, this paradigm disrupts the subsequent tokens prediction, degrading model performance and linguistic coherence. Moreover, existing evaluation metrics overemphasize contextual forgetting while inadequately assessing response fluency and relevance. To address these challenges, we propose ReLearn, a data augmentation and fine-tuning pipeline for effective unlearning, along with a comprehensive evaluation framework. This framework introduces Knowledge Forgetting Rate (KFR) and Knowledge Retention Rate (KRR) to measure knowledge-level preservation, and Linguistic Score (LS) to evaluate generation quality. Our experiments show that ReLearn successfully achieves targeted forgetting while preserving high-quality output. Through mechanistic analysis, we further demonstrate how reverse optimization disrupts coherent text generation, while ReLearn preserves this essential capability. Code is available at https://github.com/zjunlp/unlearn.
Authors: Wei Dai, Peilin Chen, Malinda Lu, Daniel Li, Haowen Wei, Hejie Cui, Paul Pu Liang
Abstract: Recent advances in clinical AI have enabled remarkable progress across many clinical domains. However, existing benchmarks and models are primarily limited to a small set of modalities and tasks, which hinders the development of large-scale multimodal methods that can make holistic assessments of patient health and well-being. To bridge this gap, we introduce Clinical Large-Scale Integrative Multimodal Benchmark (CLIMB), a comprehensive clinical benchmark unifying diverse clinical data across imaging, language, temporal, and graph modalities. CLIMB comprises 4.51 million patient samples totaling 19.01 terabytes distributed across 2D imaging, 3D video, time series, graphs, and multimodal data. Through extensive empirical evaluation, we demonstrate that multitask pretraining significantly improves performance on understudied domains, achieving up to 29% improvement in ultrasound and 23% in ECG analysis over single-task learning. Pretraining on CLIMB also effectively improves models' generalization capability to new tasks, and strong unimodal encoder performance translates well to multimodal performance when paired with task-appropriate fusion strategies. Our findings provide a foundation for new architecture designs and pretraining strategies to advance clinical AI research. Code is released at https://github.com/DDVD233/climb.
Authors: Siyi Wu, Leyi Zhao, Haotian Ma, Xinyuan Song
Abstract: Accurate segmentation of tubular and curvilinear structures, such as blood vessels, neurons, and road networks, is crucial in various applications. A key challenge is ensuring topological correctness while maintaining computational efficiency. Existing approaches often employ topological loss functions based on persistent homology, such as Betti error, to enforce structural consistency. However, these methods suffer from high computational costs and are insensitive to pixel-level accuracy, often requiring additional loss terms like Dice or MSE to compensate. To address these limitations, we propose \textbf{SDF-TopoNet}, an improved topology-aware segmentation framework that enhances both segmentation accuracy and training efficiency. Our approach introduces a novel two-stage training strategy. In the pre-training phase, we utilize the signed distance function (SDF) as an auxiliary learning target, allowing the model to encode topological information without directly relying on computationally expensive topological loss functions. In the fine-tuning phase, we incorporate a dynamic adapter alongside a refined topological loss to ensure topological correctness while mitigating overfitting and computational overhead. We evaluate our method on five benchmark datasets. Experimental results demonstrate that SDF-TopoNet outperforms existing methods in both topological accuracy and quantitative segmentation metrics, while significantly reducing training complexity.
Authors: Ananya Ganapthy, Praveen Shastry, Naveen Kumarasami, Anandakumar D, Keerthana R, Mounigasri M, Varshinipriya M, Kishore Prasath Venkatesh, Bargava Subramanian, Kalyan Sivasailam
Abstract: Background: This study introduces a Vision-Language Model (VLM) leveraging SIGLIP and Gemma-3b architectures for automated acute tuberculosis (TB) screening. By integrating chest X-ray images and clinical notes, the model aims to enhance diagnostic accuracy and efficiency, particularly in resource-limited settings. Methods: The VLM combines visual data from chest X-rays with clinical context to generate detailed, context-aware diagnostic reports. The architecture employs SIGLIP for visual encoding and Gemma-3b for decoding, ensuring effective representation of acute TB-specific pathologies and clinical insights. Results: Key acute TB pathologies, including consolidation, cavities, and nodules, were detected with high precision (97percent) and recall (96percent). The model demonstrated strong spatial localization capabilities and robustness in distinguishing TB-positive cases, making it a reliable tool for acute TB diagnosis. Conclusion: The multimodal capability of the VLM reduces reliance on radiologists, providing a scalable solution for acute TB screening. Future work will focus on improving the detection of subtle pathologies and addressing dataset biases to enhance its generalizability and application in diverse global healthcare settings.