Authors: Akwasi Asare, Ulas Bagci
Abstract: Colorectal cancer (CRC) remains one of the leading causes of cancer-related morbidity and mortality worldwide, with gastrointestinal (GI) polyps serving as critical precursors according to the World Health Organization (WHO). Early and accurate segmentation of polyps during colonoscopy is essential for reducing CRC progression, yet manual delineation is labor-intensive and prone to observer variability. Deep learning methods have demonstrated strong potential for automated polyp analysis, but their limited interpretability remains a barrier to clinical adoption. In this study, we present PolypSeg-GradCAM, an explainable deep learning framework that integrates the U-Net architecture with Gradient-weighted Class Activation Mapping (Grad-CAM) for transparent polyp segmentation. The model was trained and evaluated on the Kvasir-SEG dataset of 1000 annotated endoscopic images. Experimental results demonstrate robust segmentation performance, achieving a mean Intersection over Union (IoU) of 0.9257 on the test set and consistently high Dice coefficients (F-score > 0.96) on training and validation sets. Grad-CAM visualizations further confirmed that predictions were guided by clinically relevant regions, enhancing transparency and trust in the model's decisions. By coupling high segmentation accuracy with interpretability, PolypSeg-GradCAM represents a step toward reliable, trustworthy AI-assisted colonoscopy and improved early colorectal cancer prevention.
Authors: Akwasi Asare, Isaac Baffour Senkyire, Emmanuel Freeman, Simon Hilary Ayinedenaba Aluze-Ele, Kelvin Kwao
Abstract: Diabetic retinopathy is a leading cause of vision loss among adults and a major global health challenge, particularly in underserved regions. This study presents PerceptronCARE, a deep learning-based teleophthalmology application designed for automated diabetic retinopathy detection using retinal images. The system was developed and evaluated using multiple convolutional neural networks, including ResNet-18, EfficientNet-B0, and SqueezeNet, to determine the optimal balance between accuracy and computational efficiency. The final model classifies disease severity with an accuracy of 85.4%, enabling real-time screening in clinical and telemedicine settings. PerceptronCARE integrates cloud-based scalability, secure patient data management, and a multi-user framework, facilitating early diagnosis, improving doctor-patient interactions, and reducing healthcare costs. This study highlights the potential of AI-driven telemedicine solutions in expanding access to diabetic retinopathy screening, particularly in remote and resource-constrained environments.
Authors: Xiuding Cai, Yaoyao Zhu, Linjie Fu, Dong Miao, Yu Yao
Abstract: Regularization is essential in deep learning to enhance generalization and mitigate overfitting. However, conventional techniques often rely on heuristics, making them less reliable or effective across diverse settings. We propose Self Identity Mapping (SIM), a simple yet effective, data-intrinsic regularization framework that leverages an inverse mapping mechanism to enhance representation learning. By reconstructing the input from its transformed output, SIM reduces information loss during forward propagation and facilitates smoother gradient flow. To address computational inefficiencies, We instantiate SIM as $ \rho\text{SIM} $ by incorporating patch-level feature sampling and projection-based method to reconstruct latent features, effectively lowering complexity. As a model-agnostic, task-agnostic regularizer, SIM can be seamlessly integrated as a plug-and-play module, making it applicable to different network architectures and tasks. We extensively evaluate $\rho\text{SIM}$ across three tasks: image classification, few-shot prompt learning, and domain generalization. Experimental results show consistent improvements over baseline methods, highlighting $\rho\text{SIM}$'s ability to enhance representation learning across various tasks. We also demonstrate that $\rho\text{SIM}$ is orthogonal to existing regularization methods, boosting their effectiveness. Moreover, our results confirm that $\rho\text{SIM}$ effectively preserves semantic information and enhances performance in dense-to-dense tasks, such as semantic segmentation and image translation, as well as in non-visual domains including audio classification and time series anomaly detection. The code is publicly available at https://github.com/XiudingCai/SIM-pytorch.
Authors: Zhanting Zhou, Jinbo Wang, Zeqin Wu, Fengli Zhang
Abstract: We study gradient inversion in the challenging single round averaged gradient SAG regime where per sample cues are entangled within a single batch mean gradient. We introduce MAGIA a momentum based adaptive correction on gradient inversion attack a novel label inference free framework that senses latent per image signals by probing random data subsets. MAGIA objective integrates two core innovations 1 a closed form combinatorial rescaling that creates a provably tighter optimization bound and 2 a momentum based mixing of whole batch and subset losses to ensure reconstruction robustness. Extensive experiments demonstrate that MAGIA significantly outperforms advanced methods achieving high fidelity multi image reconstruction in large batch scenarios where prior works fail. This is all accomplished with a computational footprint comparable to standard solvers and without requiring any auxiliary information.
Authors: Khalil Hennara, Muhammad Hreden, Mohamed Motasim Hamed, Ahmad Bastati, Zeina Aldallal, Sara Chrouf, Safwan AlModhayan
Abstract: Arabic document OCR remains a challenging task due to the language's cursive script, diverse fonts, diacritics, and right-to-left orientation. While modern Multimodal Large Language Models (MLLMs) have advanced document understanding for high-resource languages, their performance on Arabic remains limited. In this work, we introduce Baseer, a vision-language model fine- tuned specifically for Arabic document OCR. Leveraging a large-scale dataset combining synthetic and real-world documents, Baseer is trained using a decoder-only fine-tuning strategy to adapt a pre-trained MLLM while preserving general visual features. We also present Misraj-DocOCR, a high-quality, expert-verified benchmark designed for rigorous evaluation of Arabic OCR systems. Our experiments show that Baseer significantly outperforms existing open-source and commercial solutions, achieving a WER of 0.25 and establishing a new state-of-the-art in the domain of Arabic document OCR. Our results highlight the benefits of domain-specific adaptation of general-purpose MLLMs and establish a strong baseline for high-accuracy OCR on morphologically rich languages like Arabic.
Authors: Wendong Yao, Saeed Azadnejad, Binhua Huang, Shane Donohue, Soumyabrata Dev
Abstract: Monitoring ground displacement is crucial for urban infrastructure stability and mitigating geological hazards. However, forecasting future deformation from sparse Interferometric Synthetic Aperture Radar (InSAR) time-series data remains a significant challenge. This paper introduces a novel deep learning framework that transforms these sparse point measurements into a dense spatio-temporal tensor. This methodological shift allows, for the first time, the direct application of advanced computer vision architectures to this forecasting problem. We design and implement a hybrid Convolutional Neural Network and Long-Short Term Memory (CNN-LSTM) model, specifically engineered to simultaneously learn spatial patterns and temporal dependencies from the generated data tensor. The model's performance is benchmarked against powerful machine learning baselines, Light Gradient Boosting Machine and LASSO regression, using Sentinel-1 data from eastern Ireland. Results demonstrate that the proposed architecture provides significantly more accurate and spatially coherent forecasts, establishing a new performance benchmark for this task. Furthermore, an interpretability analysis reveals that baseline models often default to simplistic persistence patterns, highlighting the necessity of our integrated spatio-temporal approach to capture the complex dynamics of ground deformation. Our findings confirm the efficacy and potential of spatio-temporal deep learning for high-resolution deformation forecasting.
Authors: George Corr\^ea de Ara\'ujo, Helena de Almeida Maia, Helio Pedrini
Abstract: In this paper, we present the Scrapbook framework, a novel methodology designed to generate extensive datasets for probing the learned concepts of artificial intelligence (AI) models. The framework focuses on fundamental concepts such as object recognition, absolute and relative positions, and attribute identification. By generating datasets with a large number of questions about individual concepts and a wide linguistic variation, the Scrapbook framework aims to validate the model's understanding of these basic elements before tackling more complex tasks. Our experimental findings reveal that, while contemporary models demonstrate proficiency in recognizing and enumerating objects, they encounter challenges in comprehending positional information and addressing inquiries with additional constraints. Specifically, the MobileVLM-V2 model showed significant answer disagreements and plausible wrong answers, while other models exhibited a bias toward affirmative answers and struggled with questions involving geometric shapes and positional information, indicating areas for improvement in understanding and consistency. The proposed framework offers a valuable instrument for generating diverse and comprehensive datasets, which can be utilized to systematically assess and enhance the performance of AI models.
Authors: Sai Varun Kodathala, Rakesh Vunnam
Abstract: With the increasing integration of multimodal AI systems in creative workflows, understanding information loss in vision-language-vision pipelines has become important for evaluating system limitations. However, the degradation that occurs when visual content passes through textual intermediation remains poorly quantified. In this work, we provide empirical analysis of the describe-then-generate bottleneck, where natural language serves as an intermediate representation for visual information. We generated 150 image pairs through the describe-then-generate pipeline and applied existing metrics (LPIPS, SSIM, and color distance) to measure information preservation across perceptual, structural, and chromatic dimensions. Our evaluation reveals that 99.3% of samples exhibit substantial perceptual degradation and 91.5% demonstrate significant structural information loss, providing empirical evidence that the describe-then-generate bottleneck represents a measurable and consistent limitation in contemporary multimodal systems.
Authors: Isabelle Tingzon, Yoji Toriumi, Caroline Gevaert
Abstract: Detailed structural building information is used to estimate potential damage from hazard events like cyclones, floods, and landslides, making them critical for urban resilience planning and disaster risk reduction. However, such information is often unavailable in many small island developing states (SIDS) in climate-vulnerable regions like the Caribbean. To address this data gap, we present an AI-driven workflow to automatically infer rooftop attributes from high-resolution satellite imagery, with Saint Vincent and the Grenadines as our case study. Here, we compare the utility of geospatial foundation models combined with shallow classifiers against fine-tuned deep learning models for rooftop classification. Furthermore, we assess the impact of incorporating additional training data from neighboring SIDS to improve model performance. Our best models achieve F1 scores of 0.88 and 0.83 for roof pitch and roof material classification, respectively. Combined with local capacity building, our work aims to provide SIDS with novel capabilities to harness AI and Earth Observation (EO) data to enable more efficient, evidence-based urban governance.
Authors: Jinyue Bian, Zhaoxing Zhang, Zhengyu Liang, Shiwei Zheng, Shengtao Zhang, Rong Shen, Chen Yang, Anzhou Hou
Abstract: The Visual-Language-Action (VLA) models can follow text instructions according to visual observations of the surrounding environment. This ability to map multimodal inputs to actions is derived from the training of the VLA model on extensive standard demonstrations. These visual observations captured by third-personal global and in-wrist local cameras are inevitably varied in number and perspective across different environments, resulting in significant differences in the visual features. This perspective heterogeneity constrains the generality of VLA models. In light of this, we first propose the lightweight module VLA-LPAF to foster the perspective adaptivity of VLA models using only 2D data. VLA-LPAF is finetuned using images from a single view and fuses other multiview observations in the latent space, which effectively and efficiently bridge the gap caused by perspective inconsistency. We instantiate our VLA-LPAF framework with the VLA model RoboFlamingo to construct RoboFlamingo-LPAF. Experiments show that RoboFlamingo-LPAF averagely achieves around 8% task success rate improvement on CALVIN, 15% on LIBERO, and 30% on a customized simulation benchmark. We also demonstrate the developed viewadaptive characteristics of the proposed RoboFlamingo-LPAF through real-world tasks.
Authors: Yifeng Cheng, Alois Knoll, Hu Cao
Abstract: Event cameras provide high temporal resolution, high dynamic range, and low latency, offering significant advantages over conventional frame-based cameras. In this work, we introduce an uncertainty-aware refinement network called URNet for event-based stereo depth estimation. Our approach features a local-global refinement module that effectively captures fine-grained local details and long-range global context. Additionally, we introduce a Kullback-Leibler (KL) divergence-based uncertainty modeling method to enhance prediction reliability. Extensive experiments on the DSEC dataset demonstrate that URNet consistently outperforms state-of-the-art (SOTA) methods in both qualitative and quantitative evaluations.
Authors: Giammarco La Barbera, Enzo Bonnot, Thomas Isla, Juan Pablo de la Plata, Joy-Rose Dunoyer de Segonzac, Jennifer Attali, C\'ecile Lozach, Alexandre Bellucci, Louis Marcellin, Laure Fournier, Sabine Sarnacki, Pietro Gori, Isabelle Bloch
Abstract: Endometriosis often leads to chronic pelvic pain and possible nerve involvement, yet imaging the peripheral nerves remains a challenge. We introduce Visionerves, a novel hybrid AI framework for peripheral nervous system recognition from multi-gradient DWI and morphological MRI data. Unlike conventional tractography, Visionerves encodes anatomical knowledge through fuzzy spatial relationships, removing the need for selection of manual ROIs. The pipeline comprises two phases: (A) automatic segmentation of anatomical structures using a deep learning model, and (B) tractography and nerve recognition by symbolic spatial reasoning. Applied to the lumbosacral plexus in 10 women with (confirmed or suspected) endometriosis, Visionerves demonstrated substantial improvements over standard tractography, with Dice score improvements of up to 25% and spatial errors reduced to less than 5 mm. This automatic and reproducible approach enables detailed nerve analysis and paves the way for non-invasive diagnosis of endometriosis-related neuropathy, as well as other conditions with nerve involvement.
Authors: Muhammad Naveed, Nazia Perwaiz, Sidra Sultana, Mohaira Ahmad, Muhammad Moazam Fraz
Abstract: Road traffic accidents remain a major public health challenge, particularly in countries with heterogeneous road conditions, mixed traffic flow, and variable driving discipline, such as Pakistan. Reliable detection of unsafe driving behaviours is a prerequisite for improving road safety, enabling advanced driver assistance systems (ADAS), and supporting data driven decisions in insurance and fleet management. Most of existing datasets originate from the developed countries with limited representation of the behavioural diversity observed in emerging economies and the driver's face recording voilates the privacy preservation. We present V-SenseDrive, the first privacy-preserving multimodal driver behaviour dataset collected entirely within the Pakistani driving environment. V-SenseDrive combines smartphone based inertial and GPS sensor data with synchronized road facing video to record three target driving behaviours (normal, aggressive, and risky) on multiple types of roads, including urban arterials, secondary roads, and motorways. Data was gathered using a custom Android application designed to capture high frequency accelerometer, gyroscope, and GPS streams alongside continuous video, with all sources precisely time aligned to enable multimodal analysis. The focus of this work is on the data acquisition process, covering participant selection, driving scenarios, environmental considerations, and sensor video synchronization techniques. The dataset is structured into raw, processed, and semantic layers, ensuring adaptability for future research in driver behaviour classification, traffic safety analysis, and ADAS development. By representing real world driving in Pakistan, V-SenseDrive fills a critical gap in the global landscape of driver behaviour datasets and lays the groundwork for context aware intelligent transportation solutions.
Authors: Daxiang Dong, Mingming Zheng, Dong Xu, Bairong Zhuang, Wenyu Zhang, Chunhua Luo, Haoran Wang, Zijian Zhao, Jie Li, Yuxuan Li, Hanjun Zhong, Mengyue Liu, Jieting Chen, Shupeng Li, Lun Tian, Yaping Feng, Xin Li, Donggang Jiang, Yong Chen, Yehua Xu, Duohao Qin, Chen Feng, Dan Wang, Henghua Zhang, Jingjing Ha, Jinhui He, Yanfeng Zhai, Chengxin Zheng, Jiayi Mao, Jiacheng Chen, Ruchang Yao, Ziye Yuan, Jianmin Wu, Guangjun Xie, Dou Shen
Abstract: We present Qianfan-VL, a series of multimodal large language models ranging from 3B to 70B parameters, achieving state-of-the-art performance through innovative domain enhancement techniques. Our approach employs multi-stage progressive training and high-precision data synthesis pipelines, which prove to be critical technologies for enhancing domain-specific capabilities while maintaining strong general performance. Qianfan-VL achieves comparable results to leading open-source models on general benchmarks, with state-of-the-art performance on benchmarks such as CCBench, SEEDBench IMG, ScienceQA, and MMStar. The domain enhancement strategy delivers significant advantages in OCR and document understanding, validated on both public benchmarks (OCRBench 873, DocVQA 94.75%) and in-house evaluations. Notably, Qianfan-VL-8B and 70B variants incorporate long chain-of-thought capabilities, demonstrating superior performance on mathematical reasoning (MathVista 78.6%) and logical inference tasks. All models are trained entirely on Baidu's Kunlun P800 chips, validating the capability of large-scale AI infrastructure to train SOTA-level multimodal models with over 90% scaling efficiency on 5000 chips for a single task. This work establishes an effective methodology for developing domain-enhanced multimodal models suitable for diverse enterprise deployment scenarios.
Authors: Junseong Shin, Seungwoo Chung, Yunjeong Yang, Tae Hyun Kim
Abstract: Dehazing involves removing haze or fog from images to restore clarity and improve visibility by estimating atmospheric scattering effects. While deep learning methods show promise, the lack of paired real-world training data and the resulting domain gap hinder generalization to real-world scenarios. In this context, physics-grounded learning becomes crucial; however, traditional methods based on the Atmospheric Scattering Model (ASM) often fall short in handling real-world complexities and diverse haze patterns. To solve this problem, we propose HazeFlow, a novel ODE-based framework that reformulates ASM as an ordinary differential equation (ODE). Inspired by Rectified Flow (RF), HazeFlow learns an optimal ODE trajectory to map hazy images to clean ones, enhancing real-world dehazing performance with only a single inference step. Additionally, we introduce a non-homogeneous haze generation method using Markov Chain Brownian Motion (MCBM) to address the scarcity of paired real-world data. By simulating realistic haze patterns through MCBM, we enhance the adaptability of HazeFlow to diverse real-world scenarios. Through extensive experiments, we demonstrate that HazeFlow achieves state-of-the-art performance across various real-world dehazing benchmark datasets.
Authors: Omar H. Khater, Abdul Jabbar Siddiqui, Aiman El-Maleh, M. Shamim Hossain
Abstract: Deploying deep learning models in agriculture is difficult because edge devices have limited resources, but this work presents a compressed version of EcoWeedNet using structured channel pruning, quantization-aware training (QAT), and acceleration with NVIDIA's TensorRT on the Jetson Orin Nano. Despite the challenges of pruning complex architectures with residual shortcuts, attention mechanisms, concatenations, and CSP blocks, the model size was reduced by up to 68.5% and computations by 3.2 GFLOPs, while inference speed reached 184 FPS at FP16, 28.7% faster than the baseline. On the CottonWeedDet12 dataset, the pruned EcoWeedNet with a 39.5% pruning ratio outperformed YOLO11n and YOLO12n (with only 20% pruning), achieving 83.7% precision, 77.5% recall, and 85.9% mAP50, proving it to be both efficient and effective for precision agriculture.
Authors: Yi Gu, Kuniaki Saito, Jiaxin Ma
Abstract: As medical diagnoses increasingly leverage multimodal data, machine learning models are expected to effectively fuse heterogeneous information while remaining robust to missing modalities. In this work, we propose a novel multimodal learning framework that integrates enhanced modalities dropout and contrastive learning to address real-world limitations such as modality imbalance and missingness. Our approach introduces learnable modality tokens for improving missingness-aware fusion of modalities and augments conventional unimodal contrastive objectives with fused multimodal representations. We validate our framework on large-scale clinical datasets for disease detection and prediction tasks, encompassing both visual and tabular modalities. Experimental results demonstrate that our method achieves state-of-the-art performance, particularly in challenging and practical scenarios where only a single modality is available. Furthermore, we show its adaptability through successful integration with a recent CT foundation model. Our findings highlight the effectiveness, efficiency, and generalizability of our approach for multimodal learning, offering a scalable, low-cost solution with significant potential for real-world clinical applications. The code is available at https://github.com/omron-sinicx/medical-modality-dropout.
URLs: https://github.com/omron-sinicx/medical-modality-dropout.
Authors: Yixin Zhang, Ryan Chamberlain, Lawrance Ngo, Kevin Kramer, Maciej A. Mazurowski
Abstract: In this study, we curated a densely annotated in-house dataset comprising 490 CTPA scans. Using this dataset, we systematically evaluated nine widely used segmentation architectures from both the CNN and Vision Transformer (ViT) families, initialized with either pretrained or random weights, under a unified testing framework as a performance audit. Our study leads to several important observations: (1) 3D U-Net with a ResNet encoder remains a highly effective architecture for PE segmentation; (2) 3D models are particularly well-suited to this task given the morphological characteristics of emboli; (3) CNN-based models generally yield superior performance compared to their ViT-based counterparts in PE segmentation; (4) classification-based pretraining, even on large PE datasets, can adversely impact segmentation performance compared to training from scratch, suggesting that PE classification and segmentation may rely on different sets of discriminative features; (5) different model architectures show a highly consistent pattern of segmentation performance when trained on the same data; and (6) while central and large emboli can be segmented with satisfactory accuracy, distal emboli remain challenging due to both task complexity and the scarcity of high-quality datasets. Besides these findings, our best-performing model achieves a mean Dice score of 0.7131 for segmentation. It detects 181 emboli with 49 false positives and 28 false negatives from 60 in-house testing scans. Its generalizability is further validated on public datasets.
Authors: Alessa Carbo, Eric Nalisnick
Abstract: Handshapes serve a fundamental phonological role in signed languages, with American Sign Language employing approximately 50 distinct shapes. However,computational approaches rarely model handshapes explicitly, limiting both recognition accuracy and linguistic analysis.We introduce a novel graph neural network that separates temporal dynamics from static handshape configurations. Our approach combines anatomically-informed graph structures with contrastive learning to address key challenges in handshape recognition, including subtle interclass distinctions and temporal variations. We establish the first benchmark for structured handshape recognition in signing sequences, achieving 46% accuracy across 37 handshape classes (with baseline methods achieving 25%).
Authors: Chun Kit Wong, Anders N. Christensen, Cosmin I. Bercea, Julia A. Schnabel, Martin G. Tolsgaard, Aasa Feragen
Abstract: Reliable out-of-distribution (OOD) detection is important for safe deployment of deep learning models in fetal ultrasound amidst heterogeneous image characteristics and clinical settings. OOD detection relies on estimating a classification model's uncertainty, which should increase for OOD samples. While existing research has largely focused on uncertainty quantification methods, this work investigates the impact of the classification task itself. Through experiments with eight uncertainty quantification methods across four classification tasks, we demonstrate that OOD detection performance significantly varies with the task, and that the best task depends on the defined ID-OOD criteria; specifically, whether the OOD sample is due to: i) an image characteristic shift or ii) an anatomical feature shift. Furthermore, we reveal that superior OOD detection does not guarantee optimal abstained prediction, underscoring the necessity to align task selection and uncertainty strategies with the specific downstream application in medical image analysis.
Authors: Oussema Dhaouadi, Riccardo Marin, Johannes Meier, Jacques Kaiser, Daniel Cremers
Abstract: Accurate visual localization from aerial views is a fundamental problem with applications in mapping, large-area inspection, and search-and-rescue operations. In many scenarios, these systems require high-precision localization while operating with limited resources (e.g., no internet connection or GNSS/GPS support), making large image databases or heavy 3D models impractical. Surprisingly, little attention has been given to leveraging orthographic geodata as an alternative paradigm, which is lightweight and increasingly available through free releases by governmental authorities (e.g., the European Union). To fill this gap, we propose OrthoLoC, the first large-scale dataset comprising 16,425 UAV images from Germany and the United States with multiple modalities. The dataset addresses domain shifts between UAV imagery and geospatial data. Its paired structure enables fair benchmarking of existing solutions by decoupling image retrieval from feature matching, allowing isolated evaluation of localization and calibration performance. Through comprehensive evaluation, we examine the impact of domain shifts, data resolutions, and covisibility on localization accuracy. Finally, we introduce a refinement technique called AdHoP, which can be integrated with any feature matcher, improving matching by up to 95% and reducing translation error by up to 63%. The dataset and code are available at: https://deepscenario.github.io/OrthoLoC.
Authors: Mehrdad Moradi, Shengzhe Chen, Hao Yan, Kamran Paynabar
Abstract: Anomaly detection in images is typically addressed by learning from collections of training data or relying on reference samples. In many real-world scenarios, however, such training data may be unavailable, and only the test image itself is provided. We address this zero-shot setting by proposing a single-image anomaly localization method that leverages the inductive bias of convolutional neural networks, inspired by Deep Image Prior (DIP). Our method is named Single Shot Decomposition Network (SSDnet). Our key assumption is that natural images often exhibit unified textures and patterns, and that anomalies manifest as localized deviations from these repetitive or stochastic patterns. To learn the deep image prior, we design a patch-based training framework where the input image is fed directly into the network for self-reconstruction, rather than mapping random noise to the image as done in DIP. To avoid the model simply learning an identity mapping, we apply masking, patch shuffling, and small Gaussian noise. In addition, we use a perceptual loss based on inner-product similarity to capture structure beyond pixel fidelity. Our approach needs no external training data, labels, or references, and remains robust in the presence of noise or missing pixels. SSDnet achieves 0.99 AUROC and 0.60 AUPRC on MVTec-AD and 0.98 AUROC and 0.67 AUPRC on the fabric dataset, outperforming state-of-the-art methods. The implementation code will be released at https://github.com/mehrdadmoradi124/SSDnet
Authors: Riad Ahmed Anonto, Sardar Md. Saffat Zabin, M. Saifur Rahman
Abstract: Grounding vision--language models in low-resource languages remains challenging, as they often produce fluent text about the wrong objects. This stems from scarce paired data, translation pivots that break alignment, and English-centric pretraining that ignores target-language semantics. We address this with a compute-aware Bengali captioning pipeline trained on LaBSE-verified EN--BN pairs and 110k bilingual-prompted synthetic images. A frozen MaxViT yields stable visual patches, a Bengali-native mBART-50 decodes, and a lightweight bridge links the modalities. Our core novelty is a tri-loss objective: Patch-Alignment Loss (PAL) aligns real and synthetic patch descriptors using decoder cross-attention, InfoNCE enforces global real--synthetic separation, and Sinkhorn-based OT ensures balanced fine-grained patch correspondence. This PAL+InfoNCE+OT synergy improves grounding, reduces spurious matches, and drives strong gains on Flickr30k-1k (BLEU-4 12.29, METEOR 27.98, BERTScore-F1 71.20) and MSCOCO-1k (BLEU-4 12.00, METEOR 28.14, BERTScore-F1 75.40), outperforming strong CE baselines and narrowing the real--synthetic centroid gap by 41%.
Authors: Reeshad Khan, John Gauch
Abstract: We present TinyBEV, a unified, camera only Bird's Eye View (BEV) framework that distills the full-stack capabilities of a large planning-oriented teacher (UniAD [19]) into a compact, real-time student model. Unlike prior efficient camera only baselines such as VAD[23] and VADv2[7], TinyBEV supports the complete autonomy stack 3D detection, HD-map segmentation, motion forecasting, occupancy prediction, and goal-directed planning within a streamlined 28M-parameter backbone, achieving a 78% reduction in parameters over UniAD [19]. Our model-agnostic, multi-stage distillation strategy combines feature-level, output-level, and adaptive region-aware supervision to effectively transfer high-capacity multi-modal knowledge to a lightweight BEV representation. On nuScenes[4], Tiny-BEV achieves 39.0 mAP for detection, 1.08 minADE for motion forecasting, and a 0.32 collision rate, while running 5x faster (11 FPS) and requiring only camera input. These results demonstrate that full-stack driving intelligence can be retained in resource-constrained settings, bridging the gap between large-scale, multi-modal perception-planning models and deployment-ready real-time autonomy.
Authors: Thomas Gossard, Filip Radovic, Andreas Ziegler, Andrea Zell
Abstract: Motion blur reduces the clarity of fast-moving objects, posing challenges for detection systems, especially in racket sports, where balls often appear as streaks rather than distinct points. Existing labeling conventions mark the ball at the leading edge of the blur, introducing asymmetry and ignoring valuable motion cues correlated with velocity. This paper introduces a new labeling strategy that places the ball at the center of the blur streak and explicitly annotates blur attributes. Using this convention, we release a new table tennis ball detection dataset. We demonstrate that this labeling approach consistently enhances detection performance across various models. Furthermore, we introduce BlurBall, a model that jointly estimates ball position and motion blur attributes. By incorporating attention mechanisms such as Squeeze-and-Excitation over multi-frame inputs, we achieve state-of-the-art results in ball detection. Leveraging blur not only improves detection accuracy but also enables more reliable trajectory prediction, benefiting real-time sports analytics.
Authors: Binhua Huang, Ni Wang, Wendong Yao, Soumyabrata Dev
Abstract: Running a large open-vocabulary (Open-vocab) detector on every video frame is accurate but expensive. We introduce a training-free pipeline that invokes OWLv2 only on fixed-interval keyframes and propagates detections to intermediate frames using compressed-domain motion vectors (MV). A simple 3x3 grid aggregation of motion vectors provides translation and uniform-scale updates, augmented with an area-growth check and an optional single-class switch. The method requires no labels, no fine-tuning, and uses the same prompt list for all open-vocabulary methods. On ILSVRC2015-VID (validation dataset), our approach (MVP) attains mAP@0.5=0.609 and mAP@[0.5:0.95]=0.316. At loose intersection-over-union (IoU) thresholds it remains close to framewise OWLv2-Large (0.747/0.721 at 0.2/0.3 versus 0.784/0.780), reflecting that coarse localization is largely preserved. Under the same keyframe schedule, MVP outperforms tracker-based propagation (MOSSE, KCF, CSRT) at mAP@0.5. A supervised reference (YOLOv12x) reaches 0.631 at mAP@0.5 but requires labeled training, whereas our method remains label-free and open-vocabulary. These results indicate that compressed-domain propagation is a practical way to reduce detector invocations while keeping strong zero-shot coverage in videos. Our code and models are available at https://github.com/microa/MVP.
Authors: Zitian Zhang, Joshua Urban Davis, Jeanne Phuong Anh Vu, Jiangtao Kuang, Jean-Fran\c{c}ois Lalonde
Abstract: Advances in high dynamic range (HDR) lighting estimation from a single image have opened new possibilities for augmented reality (AR) applications. Predicting complex lighting environments from a single input image allows for the realistic rendering and compositing of virtual objects. In this work, we investigate the color robustness of such methods -- an often overlooked yet critical factor for achieving visual realism. While most evaluations conflate color with other lighting attributes (e.g., intensity, direction), we isolate color as the primary variable of interest. Rather than introducing a new lighting estimation algorithm, we explore whether simple adaptation techniques can enhance the color accuracy of existing models. Using a novel HDR dataset featuring diverse lighting colors, we systematically evaluate several adaptation strategies. Our results show that preprocessing the input image with a pre-trained white balance network improves color robustness, outperforming other strategies across all tested scenarios. Notably, this approach requires no retraining of the lighting estimation model. We further validate the generality of this finding by applying the technique to three state-of-the-art lighting estimation methods from recent literature.
Authors: Sourav Halder, Jinjun Tong, Xinyu Wu
Abstract: Checks remain a foundational instrument in the financial ecosystem, facilitating substantial transaction volumes across institutions. However, their continued use also renders them a persistent target for fraud, underscoring the importance of robust check fraud detection mechanisms. At the core of such systems lies the accurate identification and localization of critical fields, such as the signature, magnetic ink character recognition (MICR) line, courtesy amount, legal amount, payee, and payer, which are essential for subsequent verification against reference checks belonging to the same customer. This field-level detection is traditionally dependent on object detection models trained on large, diverse, and meticulously labeled datasets, a resource that is scarce due to proprietary and privacy concerns. In this paper, we introduce a novel, training-free framework for automated check field detection, leveraging the power of a vision language model (VLM) in conjunction with a multimodal large language model (MLLM). Our approach enables zero-shot detection of check components, significantly lowering the barrier to deployment in real-world financial settings. Quantitative evaluation of our model on a hand-curated dataset of 110 checks spanning multiple formats and layouts demonstrates strong performance and generalization capability. Furthermore, this framework can serve as a bootstrap mechanism for generating high-quality labeled datasets, enabling the development of specialized real-time object detection models tailored to institutional needs.
Authors: Philip Wootaek Shin, Jack Sampson, Vijaykrishnan Narayanan, Andres Marquez, Mahantesh Halappanavar
Abstract: Vision language models (VLMs) show strong results on chart understanding, yet existing benchmarks assume clean figures and fact based queries. Real world charts often contain distortions and demand reasoning beyond simple matching. We evaluate ChatGPT 4o, Claude Sonnet 4, and Gemini 2.5 Pro, finding sharp performance drops under corruption or occlusion, with hallucinations such as value fabrication, trend misinterpretation, and entity confusion becoming more frequent. Models remain overconfident in degraded settings, generating plausible but unsupported explanations. To address this gap, we introduce CHART NOISe(Chart Hallucinations, Answers, and Reasoning Testing on Noisy and Occluded Input Selections), a dataset combining chart corruptions, occlusions, and exam style multiple choice questions inspired by Korea's CSAT English section. A key innovation is prompt reverse inconsistency, where models contradict themselves when asked to confirm versus deny the same statement. Our contributions are threefold: (1) benchmarking state of the art VLMs, exposing systematic vulnerabilities in chart reasoning; (2) releasing CHART NOISe, the first dataset unifying corruption, occlusion, and reverse inconsistency; and (3) proposing baseline mitigation strategies such as quality filtering and occlusion detection. Together, these efforts establish a rigorous testbed for advancing robustness and reliability in chart understanding.
Authors: Xinyang Wu, Muheng Li, Xia Li, Orso Pusterla, Sairos Safai, Philippe C. Cattin, Antony J. Lomax, Ye Zhang
Abstract: Four-dimensional MRI (4D-MRI) is an promising technique for capturing respiratory-induced motion in radiation therapy planning and delivery. Conventional 4D reconstruction methods, which typically rely on phase binning or separate template scans, struggle to capture temporal variability, complicate workflows, and impose heavy computational loads. We introduce a neural representation framework that considers respiratory motion as a smooth, continuous deformation steered by a 1D surrogate signal, completely replacing the conventional discrete sorting approach. The new method fuses motion modeling with image reconstruction through two synergistic networks: the Spatial Anatomy Network (SAN) encodes a continuous 3D anatomical representation, while a Temporal Motion Network (TMN), guided by Transformer-derived respiratory signals, produces temporally consistent deformation fields. Evaluation using a free-breathing dataset of 19 volunteers demonstrates that our template- and phase-free method accurately captures both regular and irregular respiratory patterns, while preserving vessel and bronchial continuity with high anatomical fidelity. The proposed method significantly improves efficiency, reducing the total processing time from approximately five hours required by conventional discrete sorting methods to just 15 minutes of training. Furthermore, it enables inference of each 3D volume in under one second. The framework accurately reconstructs 3D images at any respiratory state, achieves superior performance compared to conventional methods, and demonstrates strong potential for application in 4D radiation therapy planning and real-time adaptive treatment.
Authors: Prithvi Raj Singh, Raju Gottumukkala, Anthony Maida
Abstract: Unpredictable movement patterns and small visual mark make precise tracking of fast-moving tiny objects like a racquetball one of the challenging problems in computer vision. This challenge is particularly relevant for sport robotics applications, where lightweight and accurate tracking systems can improve robot perception and planning capabilities. While Kalman filter-based tracking methods have shown success in general object tracking scenarios, their performance degrades substantially when dealing with rapidly moving objects that exhibit irregular bouncing behavior. In this study, we evaluate the performance of five state-of-the-art Kalman filter-based tracking methods-OCSORT, DeepOCSORT, ByteTrack, BoTSORT, and StrongSORT-using a custom dataset containing 10,000 annotated racquetball frames captured at 720p-1280p resolution. We focus our analysis on two critical performance factors: inference speed and update frequency per image, examining how these parameters affect tracking accuracy and reliability for fast-moving tiny objects. Our experimental evaluation across four distinct scenarios reveals that DeepOCSORT achieves the lowest tracking error with an average ADE of 31.15 pixels compared to ByteTrack's 114.3 pixels, while ByteTrack demonstrates the fastest processing at 26.6ms average inference time versus DeepOCSORT's 26.8ms. However, our results show that all Kalman filter-based trackers exhibit significant tracking drift with spatial errors ranging from 3-11cm (ADE values: 31-114 pixels), indicating fundamental limitations in handling the unpredictable motion patterns of fast-moving tiny objects like racquetballs. Our analysis demonstrates that current tracking approaches require substantial improvements, with error rates 3-4x higher than standard object tracking benchmarks, highlighting the need for specialized methodologies for fast-moving tiny object tracking applications.
Authors: Binhua Huang, Wendong Yao, Shaowu Chen, Guoxin Wang, Qingyuan Wang, Soumyabrata Dev
Abstract: We introduce MoCrop, a motion-aware adaptive cropping module for efficient video action recognition in the compressed domain. MoCrop uses motion vectors that are available in H.264 video to locate motion-dense regions and produces a single clip-level crop that is applied to all I-frames at inference. The module is training free, adds no parameters, and can be plugged into diverse backbones. A lightweight pipeline that includes denoising & merge (DM), Monte Carlo sampling (MCS), and adaptive cropping (AC) via a motion-density submatrix search yields robust crops with negligible overhead. On UCF101, MoCrop improves accuracy or reduces compute. With ResNet-50, it delivers +3.5% Top-1 accuracy at equal FLOPs (attention setting), or +2.4% Top-1 accuracy with 26.5% fewer FLOPs (efficiency setting). Applied to CoViAR, it reaches 89.2% Top-1 accuracy at the original cost and 88.5% Top-1 accuracy while reducing compute from 11.6 to 8.5 GFLOPs. Consistent gains on MobileNet-V3, EfficientNet-B1, and Swin-B indicate strong generality and make MoCrop practical for real-time deployment in the compressed domain. Our code and models are available at https://github.com/microa/MoCrop.
Authors: Xinyu Wang, Zikun Zhou, Yingjian Li, Xin An, Hongpeng Wang
Abstract: Coding images for machines with minimal bitrate and strong analysis performance is key to effective edge-cloud systems. Several approaches deploy an image codec and perform analysis on the reconstructed image. Other methods compress intermediate features using entropy models and subsequently perform analysis on the decoded features. Nevertheless, these methods both perform poorly under low-bitrate conditions, as they retain many redundant details or learn over-concentrated symbol distributions. In this paper, we propose a Codebook-based Adaptive Feature Compression framework with Semantic Enhancement, named CAFC-SE. It maps continuous visual features to discrete indices with a codebook at the edge via Vector Quantization (VQ) and selectively transmits them to the cloud. The VQ operation that projects feature vectors onto the nearest visual primitives enables us to preserve more informative visual patterns under low-bitrate conditions. Hence, CAFC-SE is less vulnerable to low-bitrate conditions. Extensive experiments demonstrate the superiority of our method in terms of rate and accuracy.
Authors: Md Mostafijur Rahman, Radu Marculescu
Abstract: In this paper, we introduce MK-UNet, a paradigm shift towards ultra-lightweight, multi-kernel U-shaped CNNs tailored for medical image segmentation. Central to MK-UNet is the multi-kernel depth-wise convolution block (MKDC) we design to adeptly process images through multiple kernels, while capturing complex multi-resolution spatial relationships. MK-UNet also emphasizes the images salient features through sophisticated attention mechanisms, including channel, spatial, and grouped gated attention. Our MK-UNet network, with a modest computational footprint of only 0.316M parameters and 0.314G FLOPs, represents not only a remarkably lightweight, but also significantly improved segmentation solution that provides higher accuracy over state-of-the-art (SOTA) methods across six binary medical imaging benchmarks. Specifically, MK-UNet outperforms TransUNet in DICE score with nearly 333$\times$ and 123$\times$ fewer parameters and FLOPs, respectively. Similarly, when compared against UNeXt, MK-UNet exhibits superior segmentation performance, improving the DICE score up to 6.7% margins while operating with 4.7$\times$ fewer #Params. Our MK-UNet also outperforms other recent lightweight networks, such as MedT, CMUNeXt, EGE-UNet, and Rolling-UNet, with much lower computational resources. This leap in performance, coupled with drastic computational gains, positions MK-UNet as an unparalleled solution for real-time, high-fidelity medical diagnostics in resource-limited settings, such as point-of-care devices. Our implementation is available at https://github.com/SLDGroup/MK-UNet.
Authors: Maximilian Fehrentz, Alexander Winkler, Thomas Heiliger, Nazim Haouchine, Christian Heiliger, Nassir Navab
Abstract: We introduce BridgeSplat, a novel approach for deformable surgical navigation that couples intraoperative 3D reconstruction with preoperative CT data to bridge the gap between surgical video and volumetric patient data. Our method rigs 3D Gaussians to a CT mesh, enabling joint optimization of Gaussian parameters and mesh deformation through photometric supervision. By parametrizing each Gaussian relative to its parent mesh triangle, we enforce alignment between Gaussians and mesh and obtain deformations that can be propagated back to update the CT. We demonstrate BridgeSplat's effectiveness on visceral pig surgeries and synthetic data of a human liver under simulation, showing sensible deformations of the preoperative CT on monocular RGB data. Code, data, and additional resources can be found at https://maxfehrentz.github.io/ct-informed-splatting/ .
Authors: Wenjie Liu, Hongmin Liu, Lixin Zhang, Bin Fan
Abstract: Research on unsupervised domain adaptation (UDA) for semantic segmentation of remote sensing images has been extensively conducted. However, research on how to achieve domain adaptation in practical scenarios where source domain data is inaccessible namely, source-free domain adaptation (SFDA) remains limited. Self-training has been widely used in SFDA, which requires obtaining as many high-quality pseudo-labels as possible to train models on target domain data. Most existing methods optimize the entire pseudo-label set to obtain more supervisory information. However, as pseudo-label sets often contain substantial noise, simultaneously optimizing all labels is challenging. This limitation undermines the effectiveness of optimization approaches and thus restricts the performance of self-training. To address this, we propose a novel pseudo-label optimization framework called Diffusion-Guided Label Enrichment (DGLE), which starts from a few easily obtained high-quality pseudo-labels and propagates them to a complete set of pseudo-labels while ensuring the quality of newly generated labels. Firstly, a pseudo-label fusion method based on confidence filtering and super-resolution enhancement is proposed, which utilizes cross-validation of details and contextual information to obtain a small number of high-quality pseudo-labels as initial seeds. Then, we leverage the diffusion model to propagate incomplete seed pseudo-labels with irregular distributions due to its strong denoising capability for randomly distributed noise and powerful modeling capacity for complex distributions, thereby generating complete and high-quality pseudo-labels. This method effectively avoids the difficulty of directly optimizing the complete set of pseudo-labels, significantly improves the quality of pseudo-labels, and thus enhances the model's performance in the target domain.
Authors: Jiaxin Dai, Xiang Xiang
Abstract: In the field of machine learning, hyperbolic space demonstrates superior representation capabilities for hierarchical data compared to conventional Euclidean space. This work focuses on the Coarse-To-Fine Few-Shot Class-Incremental Learning (C2FSCIL) task. Our study follows the Knowe approach, which contrastively learns coarse class labels and subsequently normalizes and freezes the classifier weights of learned fine classes in the embedding space. To better interpret the "coarse-to-fine" paradigm, we propose embedding the feature extractor into hyperbolic space. Specifically, we employ the Poincar\'e ball model of hyperbolic space, enabling the feature extractor to transform input images into feature vectors within the Poincar\'e ball instead of Euclidean space. We further introduce hyperbolic contrastive loss and hyperbolic fully-connected layers to facilitate model optimization and classification in hyperbolic space. Additionally, to enhance performance under few-shot conditions, we implement maximum entropy distribution in hyperbolic space to estimate the probability distribution of fine-class feature vectors. This allows generation of augmented features from the distribution to mitigate overfitting during training with limited samples. Experiments on C2FSCIL benchmarks show that our method effectively improves both coarse and fine class accuracies.
Authors: Zixin Zhu, Haoxiang Li, Xuelu Feng, He Wu, Chunming Qiao, Junsong Yuan
Abstract: Towards intelligent image editing, object removal should eliminate both the target object and its causal visual artifacts, such as shadows and reflections. However, existing image appearance-based methods either follow strictly mask-aligned training and fail to remove these causal effects which are not explicitly masked, or adopt loosely mask-aligned strategies that lack controllability and may unintentionally over-erase other objects. We identify that these limitations stem from ignoring the causal relationship between an object's geometry presence and its visual effects. To address this limitation, we propose a geometry-aware two-stage framework that decouples object removal into (1) geometry removal and (2) appearance rendering. In the first stage, we remove the object directly from the geometry (e.g., depth) using strictly mask-aligned supervision, enabling structure-aware editing with strong geometric constraints. In the second stage, we render a photorealistic RGB image conditioned on the updated geometry, where causal visual effects are considered implicitly as a result of the modified 3D geometry. To guide learning in the geometry removal stage, we introduce a preference-driven objective based on positive and negative sample pairs, encouraging the model to remove objects as well as their causal visual artifacts while avoiding new structural insertions. Extensive experiments demonstrate that our method achieves state-of-the-art performance in removing both objects and their associated artifacts on two popular benchmarks. The code is available at https://github.com/buxiangzhiren/GeoRemover.
Authors: Yujia Liu, Dingquan Li, Tiejun Huang
Abstract: No-Reference Image Quality Assessment (NR-IQA) models play an important role in various real-world applications. Recently, adversarial attacks against NR-IQA models have attracted increasing attention, as they provide valuable insights for revealing model vulnerabilities and guiding robust system design. Some effective attacks have been proposed against NR-IQA models in white-box settings, where the attacker has full access to the target model. However, these attacks often suffer from poor transferability to unknown target models in more realistic black-box scenarios, where the target model is inaccessible. This work makes the first attempt to address the challenge of low transferability in attacking NR-IQA models by proposing a transferable Signed Ensemble Gaussian black-box Attack (SEGA). The main idea is to approximate the gradient of the target model by applying Gaussian smoothing to source models and ensembling their smoothed gradients. To ensure the imperceptibility of adversarial perturbations, SEGA further removes inappropriate perturbations using a specially designed perturbation filter mask. Experimental results on the CLIVE dataset demonstrate the superior transferability of SEGA, validating its effectiveness in enabling successful transfer-based black-box attacks against NR-IQA models.
Authors: Mohammad Junayed Hasan, Nabeel Mohammed, Shafin Rahman, Philipp Koehn
Abstract: The distinction between genuine and posed emotions represents a fundamental pattern recognition challenge with significant implications for data mining applications in social sciences, healthcare, and human-computer interaction. While recent multi-task learning frameworks have shown promise in combining deep learning architectures with handcrafted D-Marker features for smile facial emotion recognition, these approaches exhibit computational inefficiencies due to auxiliary task supervision and complex loss balancing requirements. This paper introduces HadaSmileNet, a novel feature fusion framework that directly integrates transformer-based representations with physiologically grounded D-Markers through parameter-free multiplicative interactions. Through systematic evaluation of 15 fusion strategies, we demonstrate that Hadamard multiplicative fusion achieves optimal performance by enabling direct feature interactions while maintaining computational efficiency. The proposed approach establishes new state-of-the-art results for deep learning methods across four benchmark datasets: UvA-NEMO (88.7 percent, +0.8), MMI (99.7 percent), SPOS (98.5 percent, +0.7), and BBC (100 percent, +5.0). Comprehensive computational analysis reveals 26 percent parameter reduction and simplified training compared to multi-task alternatives, while feature visualization demonstrates enhanced discriminative power through direct domain knowledge integration. The framework's efficiency and effectiveness make it particularly suitable for practical deployment in multimedia data mining applications that require real-time affective computing capabilities.
Authors: Xiaoting Yin, Hao Shi, Kailun Yang, Jiajun Zhai, Shangwei Guo, Lin Wang, Kaiwei Wang
Abstract: Reconstructing dynamic humans together with static scenes from monocular videos remains difficult, especially under fast motion, where RGB frames suffer from motion blur. Event cameras exhibit distinct advantages, e.g., microsecond temporal resolution, making them a superior sensing choice for dynamic human reconstruction. Accordingly, we present a novel event-guided human-scene reconstruction framework that jointly models human and scene from a single monocular event camera via 3D Gaussian Splatting. Specifically, a unified set of 3D Gaussians carries a learnable semantic attribute; only Gaussians classified as human undergo deformation for animation, while scene Gaussians stay static. To combat blur, we propose an event-guided loss that matches simulated brightness changes between consecutive renderings with the event stream, improving local fidelity in fast-moving regions. Our approach removes the need for external human masks and simplifies managing separate Gaussian sets. On two benchmark datasets, ZJU-MoCap-Blur and MMHPSD-Blur, it delivers state-of-the-art human-scene reconstruction, with notable gains over strong baselines in PSNR/SSIM and reduced LPIPS, especially for high-speed subjects.
Authors: Yuhan Wang, Cheng Liu, Zihan Zhao, Weichao Wu
Abstract: Real-time threat monitoring identifies threatening behaviors in video streams and provides reasoning and assessment of threat events through explanatory text. However, prevailing methodologies, whether based on supervised learning or generative models, struggle to concurrently satisfy the demanding requirements of real-time performance and decision explainability. To bridge this gap, we introduce Live-E2T, a novel framework that unifies these two objectives through three synergistic mechanisms. First, we deconstruct video frames into structured Human-Object-Interaction-Place semantic tuples. This approach creates a compact, semantically focused representation, circumventing the information degradation common in conventional feature compression. Second, an efficient online event deduplication and updating mechanism is proposed to filter spatio-temporal redundancies, ensuring the system's real time responsiveness. Finally, we fine-tune a Large Language Model using a Chain-of-Thought strategy, endow it with the capability for transparent and logical reasoning over event sequences to produce coherent threat assessment reports. Extensive experiments on benchmark datasets, including XD-Violence and UCF-Crime, demonstrate that Live-E2T significantly outperforms state-of-the-art methods in terms of threat detection accuracy, real-time efficiency, and the crucial dimension of explainability.
Authors: Daiqing Qi, Handong Zhao, Jing Shi, Simon Jenni, Yifei Fan, Franck Dernoncourt, Scott Cohen, Sheng Li
Abstract: While editing directly from life, photographers have found it too difficult to see simultaneously both the blue and the sky. Photographer and curator, Szarkowski insightfully revealed one of the notable gaps between general and aesthetic visual understanding: while the former focuses on identifying the factual element in an image (sky), the latter transcends such object identification, viewing it instead as an aesthetic component--a pure color block (blue). Such fundamental distinctions between general (detection, localization, etc.) and aesthetic (color, lighting, composition, etc.) visual understanding present a significant challenge for Multimodal Large Language Models (MLLMs). Although some recent works have made initial explorations, they are often limited to general and basic aesthetic commonsense. As a result, they frequently fall short in real-world scenarios (Fig. 1), which require extensive expertise--including photographic techniques, photo pre/post-processing knowledge, and more, to provide a detailed analysis and description. To fundamentally enhance the aesthetics understanding of MLLMs, we first introduce a novel dataset, PhotoCritique, derived from extensive discussions among professional photographers and enthusiasts, and characterized by the large scale, expertise, and diversity. Then, to better learn visual aesthetics from PhotoCritique, we furthur propose a novel model, PhotoEye, featuring a languageguided multi-view vision fusion mechanism to understand image aesthetics from multiple perspectives. Finally, we present a novel benchmark, PhotoBench, a comprehensive and professional benchmark for aesthetic visual understanding. On existing benchmarks and PhotoBench, our model demonstrates clear advantages over existing models.
Authors: Pengchao Deng, Shengqi Chen
Abstract: This paper presents an advanced tumor segmentation framework for real-time MRI-guided radiotherapy, designed for the TrackRAD2025 challenge. Our method leverages the XMem model, a memory-augmented architecture, to segment tumors across long cine-MRI sequences. The proposed system efficiently integrates memory mechanisms to track tumor motion in real-time, achieving high segmentation accuracy even under challenging conditions with limited annotated data. Unfortunately, the detailed experimental records have been lost, preventing us from reporting precise quantitative results at this stage. Nevertheless, From our preliminary impressions during development, the XMem-based framework demonstrated reasonable segmentation performance and satisfied the clinical real-time requirement. Our work contributes to improving the precision of tumor tracking during MRI-guided radiotherapy, which is crucial for enhancing the accuracy and safety of cancer treatments.
Authors: Xiaoman Wu, Lubin Gan, Siying Wu, Jing Zhang, Yunwei Ou, Xiaoyan Sun
Abstract: Multi-contrast Magnetic Resonance Imaging super-resolution (MC-MRI SR) aims to enhance low-resolution (LR) contrasts leveraging high-resolution (HR) references, shortening acquisition time and improving imaging efficiency while preserving anatomical details. The main challenge lies in maintaining spatial-semantic consistency, ensuring anatomical structures remain well-aligned and coherent despite structural discrepancies and motion between the target and reference images. Conventional methods insufficiently model spatial-semantic consistency and underuse frequency-domain information, which leads to poor fine-grained alignment and inadequate recovery of high-frequency details. In this paper, we propose the Spatial-Semantic Consistent Model (SSCM), which integrates a Dynamic Spatial Warping Module for inter-contrast spatial alignment, a Semantic-Aware Token Aggregation Block for long-range semantic consistency, and a Spatial-Frequency Fusion Block for fine structure restoration. Experiments on public and private datasets show that SSCM achieves state-of-the-art performance with fewer parameters while ensuring spatially and semantically consistent reconstructions.
Authors: Zhuoxiao Chen, Hongyang Yu, Ying Xu, Yadan Luo, Long Duong, Yuan-Fang Li
Abstract: Radiology report generation (RRG) aims to automatically produce clinically faithful reports from chest X-ray images. Prevailing work typically follows a scale-driven paradigm, by multi-stage training over large paired corpora and oversized backbones, making pipelines highly data- and compute-intensive. In this paper, we propose Oracle-educated GRPO {OraPO) with a FactScore-based reward (FactS) to tackle the RRG task under constrained budgets. OraPO enables single-stage, RL-only training by converting failed GRPO explorations on rare or difficult studies into direct preference supervision via a lightweight oracle step. FactS grounds learning in diagnostic evidence by extracting atomic clinical facts and checking entailment against ground-truth labels, yielding dense, interpretable sentence-level rewards. Together, OraPO and FactS create a compact and powerful framework that significantly improves learning efficiency on clinically challenging cases, setting the new SOTA performance on the CheXpert Plus dataset (0.341 in F1) with 2--3 orders of magnitude less training data using a small base VLM on modest hardware.
Authors: Xu Liu, Yibo Lu, Xinxian Wang, Xinyu Wu
Abstract: We propose Adaptive Multi-Style Fusion (AMSF), a reference-based training-free framework that enables controllable fusion of multiple reference styles in diffusion models. Most of the existing reference-based methods are limited by (a) acceptance of only one style image, thus prohibiting hybrid aesthetics and scalability to more styles, and (b) lack of a principled mechanism to balance several stylistic influences. AMSF mitigates these challenges by encoding all style images and textual hints with a semantic token decomposition module that is adaptively injected into every cross-attention layer of an frozen diffusion model. A similarity-aware re-weighting module then recalibrates, at each denoising step, the attention allocated to every style component, yielding balanced and user-controllable blends without any fine-tuning or external adapters. Both qualitative and quantitative evaluations show that AMSF produces multi-style fusion results that consistently outperform the state-of-the-art approaches, while its fusion design scales seamlessly to two or more styles. These capabilities position AMSF as a practical step toward expressive multi-style generation in diffusion models.
Authors: Yuzhi Wu, Li Xiao, Jun Liu, Guangfeng Jiang, XiangGen Xia
Abstract: The emerging 4D millimeter-wave radar, measuring the range, azimuth, elevation, and Doppler velocity of objects, is recognized for its cost-effectiveness and robustness in autonomous driving. Nevertheless, its point clouds exhibit significant sparsity and noise, restricting its standalone application in 3D object detection. Recent 4D radar-camera fusion methods have provided effective perception. Most existing approaches, however, adopt explicit Bird's-Eye-View fusion paradigms originally designed for LiDAR-camera fusion, neglecting radar's inherent drawbacks. Specifically, they overlook the sparse and incomplete geometry of radar point clouds and restrict fusion to coarse scene-level integration. To address these problems, we propose MLF-4DRCNet, a novel two-stage framework for 3D object detection via multi-level fusion of 4D radar and camera images. Our model incorporates the point-, scene-, and proposal-level multi-modal information, enabling comprehensive feature representation. It comprises three crucial components: the Enhanced Radar Point Encoder (ERPE) module, the Hierarchical Scene Fusion Pooling (HSFP) module, and the Proposal-Level Fusion Enhancement (PLFE) module. Operating at the point-level, ERPE densities radar point clouds with 2D image instances and encodes them into voxels via the proposed Triple-Attention Voxel Feature Encoder. HSFP dynamically integrates multi-scale voxel features with 2D image features using deformable attention to capture scene context and adopts pooling to the fused features. PLFE refines region proposals by fusing image features, and further integrates with the pooled features from HSFP. Experimental results on the View-of-Delft (VoD) and TJ4DRadSet datasets demonstrate that MLF-4DRCNet achieves the state-of-the-art performance. Notably, it attains performance comparable to LiDAR-based models on the VoD dataset.
Authors: Yichen Wu, Xu Liu, Chenxuan Zhao, Xinyu Wu
Abstract: Inverting corrupted images into the latent space of diffusion models is challenging. Current methods, which encode an image into a single latent vector, struggle to balance structural fidelity with semantic accuracy, leading to reconstructions with semantic drift, such as blurred details or incorrect attributes. To overcome this, we introduce Prompt-Guided Dual Latent Steering (PDLS), a novel, training-free framework built upon Rectified Flow models for their stable inversion paths. PDLS decomposes the inversion process into two complementary streams: a structural path to preserve source integrity and a semantic path guided by a prompt. We formulate this dual guidance as an optimal control problem and derive a closed-form solution via a Linear Quadratic Regulator (LQR). This controller dynamically steers the generative trajectory at each step, preventing semantic drift while ensuring the preservation of fine detail without costly, per-image optimization. Extensive experiments on FFHQ-1K and ImageNet-1K under various inversion tasks, including Gaussian deblurring, motion deblurring, super-resolution and freeform inpainting, demonstrate that PDLS produces reconstructions that are both more faithful to the original image and better aligned with the semantic information than single-latent baselines.
Authors: Yiwei Lyu, Samir Harake, Asadur Chowdury, Soumyanil Banerjee, Rachel Gologorsky, Shixuan Liu, Anna-Katharina Meissner, Akshay Rao, Chenhui Zhao, Akhil Kondepudi, Cheng Jiang, Xinhai Hou, Rushikesh S. Joshi, Volker Neuschmelting, Ashok Srinivasan, Dawn Kleindorfer, Brian Athey, Vikas Gulani, Aditya Pandey, Honglak Lee, Todd Hollon
Abstract: Neuroimaging is a ubiquitous tool for evaluating patients with neurological diseases. The global demand for magnetic resonance imaging (MRI) studies has risen steadily, placing significant strain on health systems, prolonging turnaround times, and intensifying physician burnout \cite{Chen2017-bt, Rula2024-qp-1}. These challenges disproportionately impact patients in low-resource and rural settings. Here, we utilized a large academic health system as a data engine to develop Prima, the first vision language model (VLM) serving as an AI foundation for neuroimaging that supports real-world, clinical MRI studies as input. Trained on over 220,000 MRI studies, Prima uses a hierarchical vision architecture that provides general and transferable MRI features. Prima was tested in a 1-year health system-wide study that included 30K MRI studies. Across 52 radiologic diagnoses from the major neurologic disorders, including neoplastic, inflammatory, infectious, and developmental lesions, Prima achieved a mean diagnostic area under the ROC curve of 92.0, outperforming other state-of-the-art general and medical AI models. Prima offers explainable differential diagnoses, worklist priority for radiologists, and clinical referral recommendations across diverse patient demographics and MRI systems. Prima demonstrates algorithmic fairness across sensitive groups and can help mitigate health system biases, such as prolonged turnaround times for low-resource populations. These findings highlight the transformative potential of health system-scale VLMs and Prima's role in advancing AI-driven healthcare.
Authors: Yuanhuiyi Lyu, Chi Kit Wong, Chenfei Liao, Lutao Jiang, Xu Zheng, Zexin Lu, Linfeng Zhang, Xuming Hu
Abstract: Recent works have made notable advancements in enhancing unified models for text-to-image generation through the Chain-of-Thought (CoT). However, these reasoning methods separate the processes of understanding and generation, which limits their ability to guide the reasoning of unified models in addressing the deficiencies of their generative capabilities. To this end, we propose a novel reasoning framework for unified models, Understanding-in-Generation (UiG), which harnesses the robust understanding capabilities of unified models to reinforce their performance in image generation. The core insight of our UiG is to integrate generative guidance by the strong understanding capabilities during the reasoning process, thereby mitigating the limitations of generative abilities. To achieve this, we introduce "Image Editing" as a bridge to infuse understanding into the generation process. Initially, we verify the generated image and incorporate the understanding of unified models into the editing instructions. Subsequently, we enhance the generated image step by step, gradually infusing the understanding into the generation process. Our UiG framework demonstrates a significant performance improvement in text-to-image generation over existing text-to-image reasoning methods, e.g., a 3.92% gain on the long prompt setting of the TIIF benchmark. The project code: https://github.com/QC-LY/UiG
Authors: Nicolas Toussaint, Emanuele Colleoni, Ricardo Sanchez-Matilla, Joshua Sutcliffe, Vanessa Thompson, Muhammad Asad, Imanol Luengo, Danail Stoyanov
Abstract: Monocular relative and metric depth estimation has seen a tremendous boost in the last few years due to the sharp advancements in foundation models and in particular transformer based networks. As we start to see applications to the domain of endoscopic images, there is still a lack of robust benchmarks and high-quality datasets in that area. This paper addresses these limitations by presenting a comprehensive benchmark of state-of-the-art (metric and relative) depth estimation models evaluated on real, unseen endoscopic images, providing critical insights into their generalisation and performance in clinical scenarios. Additionally, we introduce and publish a novel synthetic dataset (EndoSynth) of endoscopic surgical instruments paired with ground truth metric depth and segmentation masks, designed to bridge the gap between synthetic and real-world data. We demonstrate that fine-tuning depth foundation models using our synthetic dataset boosts accuracy on most unseen real data by a significant margin. By providing both a benchmark and a synthetic dataset, this work advances the field of depth estimation for endoscopic images and serves as an important resource for future research. Project page, EndoSynth dataset and trained weights are available at https://github.com/TouchSurgery/EndoSynth.
Authors: Lanhu Wu, Zilin Gao, Hao Fei, Mong-Li Lee, Wynne Hsu
Abstract: RGB-D salient object detection (SOD) aims to identify the most conspicuous objects in a scene with the incorporation of depth cues. Existing methods mainly rely on CNNs, limited by the local receptive fields, or Vision Transformers that suffer from the cost of quadratic complexity, posing a challenge in balancing performance and computational efficiency. Recently, state space models (SSM), Mamba, have shown great potential for modeling long-range dependency with linear complexity. However, directly applying SSM to RGB-D SOD may lead to deficient local semantics as well as the inadequate cross-modality fusion. To address these issues, we propose a Local Emphatic and Adaptive Fusion state space model (LEAF-Mamba) that contains two novel components: 1) a local emphatic state space module (LE-SSM) to capture multi-scale local dependencies for both modalities. 2) an SSM-based adaptive fusion module (AFM) for complementary cross-modality interaction and reliable cross-modality integration. Extensive experiments demonstrate that the LEAF-Mamba consistently outperforms 16 state-of-the-art RGB-D SOD methods in both efficacy and efficiency. Moreover, our method can achieve excellent performance on the RGB-T SOD task, proving a powerful generalization ability.
Authors: Xinle Gao, Linghui Ye, Zhiyong Xiao
Abstract: With the rapid development of society and continuous advances in science and technology, the food industry increasingly demands higher production quality and efficiency. Food image classification plays a vital role in enabling automated quality control on production lines, supporting food safety supervision, and promoting intelligent agricultural production. However, this task faces challenges due to the large number of parameters and high computational complexity of Vision Transformer models. To address these issues, we propose a lightweight food image classification algorithm that integrates a Window Multi-Head Attention Mechanism (WMHAM) and a Spatial Attention Mechanism (SAM). The WMHAM reduces computational cost by capturing local and global contextual features through efficient window partitioning, while the SAM adaptively emphasizes key spatial regions to improve discriminative feature representation. Experiments conducted on the Food-101 and Vireo Food-172 datasets demonstrate that our model achieves accuracies of 95.24% and 94.33%, respectively, while significantly reducing parameters and FLOPs compared with baseline methods. These results confirm that the proposed approach achieves an effective balance between computational efficiency and classification performance, making it well-suited for deployment in resource-constrained environments.
Authors: Siyi Chen, Kai Wang, Weicong Pang, Ruiming Yang, Ziru Chen, Renjun Gao, Alexis Kai Hon Lau, Dasa Gu, Chenchen Zhang, Cheng Li
Abstract: Open-set land-cover analysis in remote sensing requires the ability to achieve fine-grained spatial localization and semantically open categorization. This involves not only detecting and segmenting novel objects without categorical supervision but also assigning them interpretable semantic labels through multimodal reasoning. In this study, we introduce OSDA, an integrated three-stage framework for annotation-free open-set land-cover discovery, segmentation, and description. The pipeline consists of: (1) precise discovery and mask extraction with a promptable fine-tuned segmentation model (SAM), (2) semantic attribution and contextual description via a two-phase fine-tuned multimodal large language model (MLLM), and (3) LLM-as-judge and manual scoring of the MLLMs evaluation. By combining pixel-level accuracy with high-level semantic understanding, OSDA addresses key challenges in open-world remote sensing interpretation. Designed to be architecture-agnostic and label-free, the framework supports robust evaluation across diverse satellite imagery without requiring manual annotation. Our work provides a scalable and interpretable solution for dynamic land-cover monitoring, showing strong potential for automated cartographic updating and large-scale earth observation analysis.
Authors: Herve Goeau, Pierre Bonnet, Alexis Joly
Abstract: Automated plant identification has improved considerably thanks to recent advances in deep learning and the availability of training data with more and more field photos. However, this profusion of data concerns only a few tens of thousands of species, mainly located in North America and Western Europe, much less in the richest regions in terms of biodiversity such as tropical countries. On the other hand, for several centuries, botanists have systematically collected, catalogued and stored plant specimens in herbaria, especially in tropical regions, and recent efforts by the biodiversity informatics community have made it possible to put millions of digitised records online. The LifeCLEF 2021 plant identification challenge (or "PlantCLEF 2021") was designed to assess the extent to which automated identification of flora in data-poor regions can be improved by using herbarium collections. It is based on a dataset of about 1,000 species mainly focused on the Guiana Shield of South America, a region known to have one of the highest plant diversities in the world. The challenge was evaluated as a cross-domain classification task where the training set consisted of several hundred thousand herbarium sheets and a few thousand photos to allow learning a correspondence between the two domains. In addition to the usual metadata (location, date, author, taxonomy), the training data also includes the values of 5 morphological and functional traits for each species. The test set consisted exclusively of photos taken in the field. This article presents the resources and evaluations of the assessment carried out, summarises the approaches and systems used by the participating research groups and provides an analysis of the main results.
Authors: Zedong Zhang, Ying Tai, Jianjun Qian, Jian Yang, Jun Li
Abstract: Fusing cross-category objects to a single coherent object has gained increasing attention in text-to-image (T2I) generation due to its broad applications in virtual reality, digital media, film, and gaming. However, existing methods often produce biased, visually chaotic, or semantically inconsistent results due to overlapping artifacts and poor integration. Moreover, progress in this field has been limited by the absence of a comprehensive benchmark dataset. To address these problems, we propose \textbf{Adaptive Group Swapping (AGSwap)}, a simple yet highly effective approach comprising two key components: (1) Group-wise Embedding Swapping, which fuses semantic attributes from different concepts through feature manipulation, and (2) Adaptive Group Updating, a dynamic optimization mechanism guided by a balance evaluation score to ensure coherent synthesis. Additionally, we introduce \textbf{Cross-category Object Fusion (COF)}, a large-scale, hierarchically structured dataset built upon ImageNet-1K and WordNet. COF includes 95 superclasses, each with 10 subclasses, enabling 451,250 unique fusion pairs. Extensive experiments demonstrate that AGSwap outperforms state-of-the-art compositional T2I methods, including GPT-Image-1 using simple and complex prompts.
Authors: Herve Goeau, Pierre Bonnet, Alexis Joly
Abstract: Automated identification of plants has improved considerably thanks to the recent progress in deep learning and the availability of training data. However, this profusion of data only concerns a few tens of thousands of species, while the planet has nearly 369K. The LifeCLEF 2019 Plant Identification challenge (or "PlantCLEF 2019") was designed to evaluate automated identification on the flora of data deficient regions. It is based on a dataset of 10K species mainly focused on the Guiana shield and the Northern Amazon rainforest, an area known to have one of the greatest diversity of plants and animals in the world. As in the previous edition, a comparison of the performance of the systems evaluated with the best tropical flora experts was carried out. This paper presents the resources and assessments of the challenge, summarizes the approaches and systems employed by the participating research groups, and provides an analysis of the main outcomes.
Authors: Ke Li, Di Wang, Ting Wang, Fuyu Dong, Yiming Zhang, Luyao Zhang, Xiangyu Wang, Shaofeng Li, Quan Wang
Abstract: Remote sensing visual grounding (RSVG) aims to localize objects in remote sensing images based on free-form natural language expressions. Existing approaches are typically constrained to closed-set vocabularies, limiting their applicability in open-world scenarios. While recent attempts to leverage generic foundation models for open-vocabulary RSVG, they overly rely on expensive high-quality datasets and time-consuming fine-tuning. To address these limitations, we propose \textbf{RSVG-ZeroOV}, a training-free framework that aims to explore the potential of frozen generic foundation models for zero-shot open-vocabulary RSVG. Specifically, RSVG-ZeroOV comprises three key stages: (i) Overview: We utilize a vision-language model (VLM) to obtain cross-attention\footnote[1]{In this paper, although decoder-only VLMs use self-attention over all tokens, we refer to the image-text interaction part as cross-attention to distinguish it from pure visual self-attention.}maps that capture semantic correlations between text queries and visual regions. (ii) Focus: By leveraging the fine-grained modeling priors of a diffusion model (DM), we fill in gaps in structural and shape information of objects, which are often overlooked by VLM. (iii) Evolve: A simple yet effective attention evolution module is introduced to suppress irrelevant activations, yielding purified segmentation masks over the referred objects. Without cumbersome task-specific training, RSVG-ZeroOV offers an efficient and scalable solution. Extensive experiments demonstrate that the proposed framework consistently outperforms existing weakly-supervised and zero-shot methods.
Authors: Yingquan Wang, Pingping Zhang, Chong Sun, Dong Wang, Huchuan Lu
Abstract: Object Re-IDentification (ReID) aims to recognize individuals across non-overlapping camera views. While recent advances have achieved remarkable progress, most existing models are constrained to either single-domain or cross-domain scenarios, limiting their real-world applicability. Single-domain models tend to overfit to domain-specific features, whereas cross-domain models often rely on diverse normalization strategies that may inadvertently suppress identity-specific discriminative cues. To address these limitations, we propose an Attribute Prompt Composition (APC) framework, which exploits textual semantics to jointly enhance discrimination and generalization. Specifically, we design an Attribute Prompt Generator (APG) consisting of a Semantic Attribute Dictionary (SAD) and a Prompt Composition Module (PCM). SAD is an over-complete attribute dictionary to provide rich semantic descriptions, while PCM adaptively composes relevant attributes from SAD to generate discriminative attribute-aware features. In addition, motivated by the strong generalization ability of Vision-Language Models (VLM), we propose a Fast-Slow Training Strategy (FSTS) to balance ReID-specific discrimination and generalizable representation learning. Specifically, FSTS adopts a Fast Update Stream (FUS) to rapidly acquire ReID-specific discriminative knowledge and a Slow Update Stream (SUS) to retain the generalizable knowledge inherited from the pre-trained VLM. Through a mutual interaction, the framework effectively focuses on ReID-relevant features while mitigating overfitting. Extensive experiments on both conventional and Domain Generalized (DG) ReID datasets demonstrate that our framework surpasses state-of-the-art methods, exhibiting superior performances in terms of both discrimination and generalization. The source code is available at https://github.com/AWangYQ/APC.
Authors: Tong Zhang, Kuofeng Gao, Jiawang Bai, Leo Yu Zhang, Xin Yin, Zonghui Wang, Shouling Ji, Wenzhi Chen
Abstract: Recent studies have shown that Contrastive Language-Image Pre-training (CLIP) models are threatened by targeted data poisoning and backdoor attacks due to massive training image-caption pairs crawled from the Internet. Previous defense methods correct poisoned image-caption pairs by matching a new caption for each image. However, the matching process relies solely on the global representations of images and captions, overlooking fine-grained features of visual and textual features. It may introduce incorrect image-caption pairs and harm the CLIP pre-training. To address their limitations, we propose an Optimal Transport-based framework to reconstruct image-caption pairs, named OTCCLIP. We propose a new optimal transport-based distance measure between fine-grained visual and textual feature sets and re-assign new captions based on the proposed optimal transport distance. Additionally, to further reduce the negative impact of mismatched pairs, we encourage the inter- and intra-modality fine-grained alignment by employing optimal transport-based objective functions. Our experiments demonstrate that OTCCLIP can successfully decrease the attack success rates of poisoning attacks. Also, compared to previous methods, OTCCLIP significantly improves CLIP's zero-shot and linear probing performance trained on poisoned datasets.
Authors: Yilin Gao, Kangyi Chen, Zhongxing Peng, Hengjie Lu, Shugong Xu
Abstract: Current visual foundation models (VFMs) face a fundamental limitation in transferring knowledge from vision language models (VLMs), while VLMs excel at modeling cross-modal interactions through unified representation spaces, existing VFMs predominantly adopt result-oriented paradigms that neglect the underlying interaction processes. This representational discrepancy hinders effective knowledge transfer and limits generalization across diverse vision tasks. We propose Learning from Interactions (LFI), a cognitive-inspired framework that addresses this gap by explicitly modeling visual understanding as an interactive process. Our key insight is that capturing the dynamic interaction patterns encoded in pre-trained VLMs enables more faithful and efficient knowledge transfer to VFMs. The approach centers on two technical innovations, Interaction Queries, which maintain persistent relational structures across network layers, and interaction-based supervision, derived from the cross-modal attention mechanisms of VLMs. Comprehensive experiments demonstrate consistent improvements across multiple benchmarks, achieving 3.3 and 1.6mAP/2.4AP absolute gains on TinyImageNet classification and COCO detection/segmentation respectively, with minimal parameter overhead and faster convergence. The framework particularly excels in cross-domain settings, delivering 2.4 and 9.3 zero-shot improvements on PACS and VLCS. Human evaluations further confirm its cognitive alignment, outperforming result-oriented methods by 2.7 times in semantic consistency metrics.
Authors: Ruichao Hou, Xingyuan Li, Tongwei Ren, Dongming Zhou, Gangshan Wu, Jinde Cao
Abstract: RGB-thermal salient object detection (RGB-T SOD) aims to identify prominent objects by integrating complementary information from RGB and thermal modalities. However, learning the precise boundaries and complete objects remains challenging due to the intrinsic insufficient feature fusion and the extrinsic limitations of data scarcity. In this paper, we propose a novel hybrid prompt-driven segment anything model (HyPSAM), which leverages the zero-shot generalization capabilities of the segment anything model (SAM) for RGB-T SOD. Specifically, we first propose a dynamic fusion network (DFNet) that generates high-quality initial saliency maps as visual prompts. DFNet employs dynamic convolution and multi-branch decoding to facilitate adaptive cross-modality interaction, overcoming the limitations of fixed-parameter kernels and enhancing multi-modal feature representation. Moreover, we propose a plug-and-play refinement network (P2RNet), which serves as a general optimization strategy to guide SAM in refining saliency maps by using hybrid prompts. The text prompt ensures reliable modality input, while the mask and box prompts enable precise salient object localization. Extensive experiments on three public datasets demonstrate that our method achieves state-of-the-art performance. Notably, HyPSAM has remarkable versatility, seamlessly integrating with different RGB-T SOD methods to achieve significant performance gains, thereby highlighting the potential of prompt engineering in this field. The code and results of our method are available at: https://github.com/milotic233/HyPSAM.
Authors: Susmit Neogi
Abstract: LiDAR-based perception is central to autonomous driving and robotics, yet raw point clouds remain highly vulnerable to noise, occlusion, and adversarial corruptions. Autoencoders offer a natural framework for denoising and reconstruction, but their performance degrades under challenging real-world conditions. In this work, we propose TriFusion-AE, a multimodal cross-attention autoencoder that integrates textual priors, monocular depth maps from multi-view images, and LiDAR point clouds to improve robustness. By aligning semantic cues from text, geometric (depth) features from images, and spatial structure from LiDAR, TriFusion-AE learns representations that are resilient to stochastic noise and adversarial perturbations. Interestingly, while showing limited gains under mild perturbations, our model achieves significantly more robust reconstruction under strong adversarial attacks and heavy noise, where CNN-based autoencoders collapse. We evaluate on the nuScenes-mini dataset to reflect realistic low-data deployment scenarios. Our multimodal fusion framework is designed to be model-agnostic, enabling seamless integration with any CNN-based point cloud autoencoder for joint representation learning.
Authors: Yuyang Liu, Xinyuan Shi, Bang Yang, Peilin Zhou, Jiahua Dong, Long Chen, Ian Reid, Xiaondan Liang
Abstract: The success of Large Language Models (LLMs) has significantly propelled the research of video understanding. To harvest the benefits of well-trained expert models (i.e., tools), video LLMs prioritize the exploration of tool usage capabilities. Existing methods either prompt closed-source LLMs or employ the instruction tuning paradigm for tool-use fine-tuning. These methods, however, assume an established repository of fixed tools and struggle to generalize to real-world environments where tool data is perpetually evolving and streaming in. To this end, we propose to enhance open-source video LLMs with COntinuaL Tool usage (termed COLT), which automatically acquires tool-use ability in a successive tool stream without suffering 'catastrophic forgetting' of the past learned tools. Specifically, our COLT incorporates a learnable tool codebook as a tool-specific memory system. Then relevant tools are dynamically selected based on the similarity between user instruction and tool features within the codebook. To unleash the tool usage potential of video LLMs, we collect a video-centric tool-use instruction tuning dataset VideoToolBench. Extensive experiments on both previous video LLM benchmarks and the tool-use-specific VideoToolBench dataset demonstrate the state-of-the-art performance of our proposed COLT.
Authors: Zhaorui Wang, Yi Gu, Deming Zhou, Renjing Xu
Abstract: Recently, 3D Gaussian Splatting (3DGS) has demonstrated remarkable success in 3D reconstruction and novel view synthesis. However, reconstructing 3D scenes from sparse viewpoints remains highly challenging due to insufficient visual information, which results in noticeable artifacts persisting across the 3D representation. To address this limitation, recent methods have resorted to generative priors to remove artifacts and complete missing content in under-constrained areas. Despite their effectiveness, these approaches struggle to ensure multi-view consistency, resulting in blurred structures and implausible details. In this work, we propose FixingGS, a training-free method that fully exploits the capabilities of the existing diffusion model for sparse-view 3DGS reconstruction enhancement. At the core of FixingGS is our distillation approach, which delivers more accurate and cross-view coherent diffusion priors, thereby enabling effective artifact removal and inpainting. In addition, we propose an adaptive progressive enhancement scheme that further refines reconstructions in under-constrained regions. Extensive experiments demonstrate that FixingGS surpasses existing state-of-the-art methods with superior visual quality and reconstruction performance. Our code will be released publicly.
Authors: Xijun Wang, Junyun Huang, Rayyan Abdalla, Chengyuan Zhang, Ruiqi Xian, Dinesh Manocha
Abstract: We address the critical gap between the computational demands of vision-language models and the possible ultra-low-bit weight precision (bitwidth $\leq2$ bits) we can use for higher efficiency. Our work is motivated by the substantial computational cost and memory requirements of VLMs, which restrict their applicability in hardware-constrained environments. We propose Bi-VLM, which separates model weights non-uniformly based on the Gaussian quantiles. Our formulation groups the model weights into outlier (salient) and multiple inlier (unsalient) subsets, ensuring that each subset contains a proportion of weights corresponding to its quantile in the distribution. We propose a saliency-aware hybrid quantization algorithm and use it to quantize weights by imposing different constraints on the scaler and binary matrices based on the saliency metric and compression objective. We have evaluated our approach on different VLMs. For the language model part of the VLM, our Bi-VLM outperforms the SOTA by 3%-47% on the visual question answering task in terms of four different benchmarks and three different models. For the overall VLM, our Bi-VLM outperforms the SOTA by 4%-45%. We also perform token pruning on the quantized models and observe that there is redundancy of image tokens 90% - 99% in the quantized models. This helps us to further prune the visual tokens to improve efficiency.
Authors: Azad Singh, Deepak Mishra
Abstract: Self-supervised learning (SSL) has emerged as a powerful paradigm for medical image representation learning, particularly in settings with limited labeled data. However, existing SSL methods often rely on complex architectures, anatomy-specific priors, or heavily tuned augmentations, which limit their scalability and generalizability. More critically, these models are prone to shortcut learning, especially in modalities like chest X-rays, where anatomical similarity is high and pathology is subtle. In this work, we introduce DiSSECT -- Discrete Self-Supervision for Efficient Clinical Transferable Representations, a framework that integrates multi-scale vector quantization into the SSL pipeline to impose a discrete representational bottleneck. This constrains the model to learn repeatable, structure-aware features while suppressing view-specific or low-utility patterns, improving representation transfer across tasks and domains. DiSSECT achieves strong performance on both classification and segmentation tasks, requiring minimal or no fine-tuning, and shows particularly high label efficiency in low-label regimes. We validate DiSSECT across multiple public medical imaging datasets, demonstrating its robustness and generalizability compared to existing state-of-the-art approaches.
Authors: Hemanth Puppala, Wayne Sarasua, Srinivas Biyaguda, Farhad Farzinpour, Mashrur Chowdhury
Abstract: Deer-vehicle collisions represent a critical safety challenge in the United States, causing nearly 2.1 million incidents annually and resulting in approximately 440 fatalities, 59,000 injuries, and 10 billion USD in economic damages. These collisions also contribute significantly to declining deer populations. This paper presents a real-time detection and driver warning system that integrates thermal imaging, deep learning, and vehicle-to-everything communication to help mitigate deer-vehicle collisions. Our system was trained and validated on a custom dataset of over 12,000 thermal deer images collected in Mars Hill, North Carolina. Experimental evaluation demonstrates exceptional performance with 98.84 percent mean average precision, 95.44 percent precision, and 95.96 percent recall. The system was field tested during a follow-up visit to Mars Hill and readily sensed deer providing the driver with advanced warning. Field testing validates robust operation across diverse weather conditions, with thermal imaging maintaining between 88 and 92 percent detection accuracy in challenging scenarios where conventional visible light based cameras achieve less than 60 percent effectiveness. When a high probability threshold is reached sensor data sharing messages are broadcast to surrounding vehicles and roadside units via cellular vehicle to everything (CV2X) communication devices. Overall, our system achieves end to end latency consistently under 100 milliseconds from detection to driver alert. This research establishes a viable technological pathway for reducing deer-vehicle collisions through thermal imaging and connected vehicles.
Authors: Danush Kumar Venkatesh, Stefanie Speidel
Abstract: The scarcity of annotated surgical data poses a significant challenge for developing deep learning systems in computer-assisted interventions. While diffusion models can synthesize realistic images, they often suffer from data memorization, resulting in inconsistent or non-diverse samples that may fail to improve, or even harm, downstream performance. We introduce \emph{Surgical Application-Aligned Diffusion} (SAADi), a new framework that aligns diffusion models with samples preferred by downstream models. Our method constructs pairs of \emph{preferred} and \emph{non-preferred} synthetic images and employs lightweight fine-tuning of diffusion models to align the image generation process with downstream objectives explicitly. Experiments on three surgical datasets demonstrate consistent gains of $7$--$9\%$ in classification and $2$--$10\%$ in segmentation tasks, with the considerable improvements observed for underrepresented classes. Iterative refinement of synthetic samples further boosts performance by $4$--$10\%$. Unlike baseline approaches, our method overcomes sample degradation and establishes task-aware alignment as a key principle for mitigating data scarcity and advancing surgical vision applications.
Authors: Kuang Xiaodong, Li Bingxuan, Li Yuan, Rao Fan, Ma Gege, Xie Qingguo, Mok Greta S P, Liu Huafeng, Zhu Wentao
Abstract: Achieving high image quality for temporal frames in dynamic positron emission tomography (PET) is challenging due to the limited statistic especially for the short frames. Recent studies have shown that deep learning (DL) is useful in a wide range of medical image denoising tasks. In this paper, we propose a model-based neural network for dynamic PET image denoising. The inter-frame spatial correlation and intra-frame structural consistency in dynamic PET are used to establish the kernel space-based multidimensional sparse (KMDS) model. We then substitute the inherent forms of the parameter estimation with neural networks to enable adaptive parameters optimization, forming the end-to-end neural KMDS-Net. Extensive experimental results from simulated and real data demonstrate that the neural KMDS-Net exhibits strong denoising performance for dynamic PET, outperforming previous baseline methods. The proposed method may be used to effectively achieve high temporal and spatial resolution for dynamic PET. Our source code is available at https://github.com/Kuangxd/Neural-KMDS-Net/tree/main.
Authors: Garam Kim, Tae Kyeong Jeong, Juyoun Park
Abstract: Robot-assisted surgery (RAS) has become a critical paradigm in modern surgery, promoting patient recovery and reducing the burden on surgeons through minimally invasive approaches. To fully realize its potential, however, a precise understanding of the visual data generated during surgical procedures is essential. Previous studies have predominantly focused on single-task approaches, but real surgical scenes involve complex temporal dynamics and diverse instrument interactions that limit comprehensive understanding. Moreover, the effective application of multi-task learning (MTL) requires sufficient pixel-level segmentation data, which are difficult to obtain due to the high cost and expertise required for annotation. In particular, long-term annotations such as phases and steps are available for every frame, whereas short-term annotations such as surgical instrument segmentation and action detection are provided only for key frames, resulting in a significant temporal-spatial imbalance. To address these challenges, we propose a novel framework that combines optical flow-based segmentation label interpolation with multi-task learning. optical flow estimated from annotated key frames is used to propagate labels to adjacent unlabeled frames, thereby enriching sparse spatial supervision and balancing temporal and spatial information for training. This integration improves both the accuracy and efficiency of surgical scene understanding and, in turn, enhances the utility of RAS.
Authors: Yanzuo Lu, Xin Xia, Manlin Zhang, Huafeng Kuang, Jianbin Zheng, Yuxi Ren, Xuefeng Xiao
Abstract: Unified multimodal models have recently attracted considerable attention for their remarkable abilities in jointly understanding and generating diverse content. However, as contexts integrate increasingly numerous interleaved multimodal tokens, the iterative processes of diffusion denoising and autoregressive decoding impose significant computational overhead. To address this, we propose Hyper-Bagel, a unified acceleration framework designed to simultaneously speed up both multimodal understanding and generation tasks. Our approach uses a divide-and-conquer strategy, employing speculative decoding for next-token prediction and a multi-stage distillation process for diffusion denoising. The framework delivers substantial performance gains, achieving over a 2x speedup in multimodal understanding. For generative tasks, our resulting lossless 6-NFE model yields a 16.67x speedup in text-to-image generation and a 22x speedup in image editing, all while preserving the high-quality output of the original model. We further develop a highly efficient 1-NFE model that enables near real-time interactive editing and generation. By combining advanced adversarial distillation with human feedback learning, this model achieves ultimate cost-effectiveness and responsiveness, making complex multimodal interactions seamless and instantaneous.
Authors: Gianmarco Spinaci (Department of Classical Philology and Italian Studies, University of Bologna, Italy, Villa i Tatti, The Harvard University Center for Italian Renaissance Studies, Florence, Italy), Lukas Klic (Villa i Tatti, The Harvard University Center for Italian Renaissance Studies, Florence, Italy), Giovanni Colavizza (Department of Classical Philology and Italian Studies, University of Bologna, Italy, Department of Communication, University of Copenhagen, Denmark)
Abstract: This study evaluates the capabilities of Multimodal Large Language Models (LLMs) and Vision Language Models (VLMs) in the task of single-label classification of Christian Iconography. The goal was to assess whether general-purpose VLMs (CLIP and SigLIP) and LLMs, such as GPT-4o and Gemini 2.5, can interpret the Iconography, typically addressed by supervised classifiers, and evaluate their performance. Two research questions guided the analysis: (RQ1) How do multimodal LLMs perform on image classification of Christian saints? And (RQ2), how does performance vary when enriching input with contextual information or few-shot exemplars? We conducted a benchmarking study using three datasets supporting Iconclass natively: ArtDL, ICONCLASS, and Wikidata, filtered to include the top 10 most frequent classes. Models were tested under three conditions: (1) classification using class labels, (2) classification with Iconclass descriptions, and (3) few-shot learning with five exemplars. Results were compared against ResNet50 baselines fine-tuned on the same datasets. The findings show that Gemini-2.5 Pro and GPT-4o outperformed the ResNet50 baselines. Accuracy dropped significantly on the Wikidata dataset, where Siglip reached the highest accuracy score, suggesting model sensitivity to image size and metadata alignment. Enriching prompts with class descriptions generally improved zero-shot performance, while few-shot learning produced lower results, with only occasional and minimal increments in accuracy. We conclude that general-purpose multimodal LLMs are capable of classification in visually complex cultural heritage domains. These results support the application of LLMs as metadata curation tools in digital humanities workflows, suggesting future research on prompt optimization and the expansion of the study to other classification strategies and models.
Authors: Ismael Elsharkawi, Hossam Sharara, Ahmed Rafea
Abstract: Image Representation Learning is an important problem in Computer Vision. Traditionally, images were processed as grids, using Convolutional Neural Networks or as a sequence of visual tokens, using Vision Transformers. Recently, Vision Graph Neural Networks (ViG) have proposed the treatment of images as a graph of nodes; which provides a more intuitive image representation. The challenge is to construct a graph of nodes in each layer that best represents the relations between nodes and does not need a hyper-parameter search. ViG models in the literature depend on non-parameterized and non-learnable statistical methods that operate on the latent features of nodes to create a graph. This might not select the best neighborhood for each node. Starting from k-NN graph construction to HyperGraph Construction and Similarity-Thresholded graph construction, these methods lack the ability to provide a learnable hyper-parameter-free graph construction method. To overcome those challenges, we present the Learnable Reparameterized Graph Construction (LRGC) for Vision Graph Neural Networks. LRGC applies key-query attention between every pair of nodes; then uses soft-threshold reparameterization for edge selection, which allows the use of a differentiable mathematical model for training. Using learnable parameters to select the neighborhood removes the bias that is induced by any clustering or thresholding methods previously introduced in the literature. In addition, LRGC allows tuning the threshold in each layer to the training data since the thresholds are learnable through training and are not provided as hyper-parameters to the model. We demonstrate that the proposed ViG-LRGC approach outperforms state-of-the-art ViG models of similar sizes on the ImageNet-1k benchmark dataset.
Authors: Junhao Su, Yuanliang Wan, Junwei Yang, Hengyu Shi, Tianyang Han, Junfeng Luo, Yurui Qiu
Abstract: Tool-augmented large language models (LLMs) are usually trained with supervised imitation or coarse-grained reinforcement learning that optimizes single tool calls. Current self-reflection practices rely on heuristic prompts or one-way reasoning: the model is urged to 'think more' instead of learning error diagnosis and repair. This is fragile in multi-turn interactions; after a failure the model often repeats the same mistake. We propose structured reflection, which turns the path from error to repair into an explicit, controllable, and trainable action. The agent produces a short yet precise reflection: it diagnoses the failure using evidence from the previous step and then proposes a correct, executable follow-up call. For training we combine DAPO and GSPO objectives with a reward scheme tailored to tool use, optimizing the stepwise strategy Reflect, then Call, then Final. To evaluate, we introduce Tool-Reflection-Bench, a lightweight benchmark that programmatically checks structural validity, executability, parameter correctness, and result consistency. Tasks are built as mini trajectories of erroneous call, reflection, and corrected call, with disjoint train and test splits. Experiments on BFCL v3 and Tool-Reflection-Bench show large gains in multi-turn tool-call success and error recovery, and a reduction of redundant calls. These results indicate that making reflection explicit and optimizing it directly improves the reliability of tool interaction and offers a reproducible path for agents to learn from failure.
Authors: Xueyu Liu, Xiaoyi Zhang, Guangze Shi, Meilin Liu, Yexin Lai, Yongfei Wu, Mingqiang Wei
Abstract: Prompt quality plays a critical role in the performance of the Segment Anything Model (SAM), yet existing approaches often rely on heuristic or manually crafted prompts, limiting scalability and generalization. In this paper, we propose Point Prompt Defender, an adversarial reinforcement learning framework that adopts an attack-for-defense paradigm to automatically optimize point prompts. We construct a task-agnostic point prompt environment by representing image patches as nodes in a dual-space graph, where edges encode both physical and semantic distances. Within this environment, an attacker agent learns to activate a subset of prompts that maximally degrade SAM's segmentation performance, while a defender agent learns to suppress these disruptive prompts and restore accuracy. Both agents are trained using Deep Q-Networks with a reward signal based on segmentation quality variation. During inference, only the defender is deployed to refine arbitrary coarse prompt sets, enabling enhanced SAM segmentation performance across diverse tasks without retraining. Extensive experiments show that Point Prompt Defender effectively improves SAM's robustness and generalization, establishing a flexible, interpretable, and plug-and-play framework for prompt-based segmentation.
Authors: Jenna Kline, Anirudh Potlapally, Bharath Pillai, Tanishka Wani, Rugved Katole, Vedant Patil, Penelope Covey, Hari Subramoni, Tanya Berger-Wolf, Christopher Stewart
Abstract: We present the first release of SmartWilds, a multimodal wildlife monitoring dataset. SmartWilds is a synchronized collection of drone imagery, camera trap photographs and videos, and bioacoustic recordings collected during summer 2025 at The Wilds safari park in Ohio. This dataset supports multimodal AI research for comprehensive environmental monitoring, addressing critical needs in endangered species research, conservation ecology, and habitat management. Our pilot deployment captured four days of synchronized monitoring across three modalities in a 220-acre pasture containing Pere David's deer, Sichuan takin, Przewalski's horses, as well as species native to Ohio, including bald eagles, white-tailed deer, and coyotes. We provide a comparative analysis of sensor modality performance, demonstrating complementary strengths for landuse patterns, species detection, behavioral analysis, and habitat monitoring. This work establishes reproducible protocols for multimodal wildlife monitoring while contributing open datasets to advance conservation computer vision research. Future releases will include synchronized GPS tracking data from tagged individuals, citizen science data, and expanded temporal coverage across multiple seasons.
Authors: Jiayu Wang, Ruizhi Wang, Jie Song, Haofei Zhang, Mingli Song, Zunlei Feng, Li Sun
Abstract: In this paper, we introduce a novel benchmark designed to propel the advancement of general-purpose, large-scale 3D vision models for remote sensing imagery. While several datasets have been proposed within the realm of remote sensing, many existing collections either lack comprehensive depth information or fail to establish precise alignment between depth data and remote sensing images. To address this deficiency, we present a visual Benchmark for 3D understanding of Remotely Sensed images, dubbed RS3DBench. This dataset encompasses 54,951 pairs of remote sensing images and pixel-level aligned depth maps, accompanied by corresponding textual descriptions, spanning a broad array of geographical contexts. It serves as a tool for training and assessing 3D visual perception models within remote sensing image spatial understanding tasks. Furthermore, we introduce a remotely sensed depth estimation model derived from stable diffusion, harnessing its multimodal fusion capabilities, thereby delivering state-of-the-art performance on our dataset. Our endeavor seeks to make a profound contribution to the evolution of 3D visual perception models and the advancement of geographic artificial intelligence within the remote sensing domain. The dataset, models and code will be accessed on the https://rs3dbench.github.io.
Authors: Pengteng Li, Yunfan Lu, Pinhao Song, Weiyu Guo, Huizai Yao, F. Richard Yu, Hui Xiong
Abstract: In this paper, we propose the first Structure-from-Motion (SfM)-free deblurring 3D Gaussian Splatting method via event camera, dubbed DeblurSplat. We address the motion-deblurring problem in two ways. First, we leverage the pretrained capability of the dense stereo module (DUSt3R) to directly obtain accurate initial point clouds from blurred images. Without calculating camera poses as an intermediate result, we avoid the cumulative errors transfer from inaccurate camera poses to the initial point clouds' positions. Second, we introduce the event stream into the deblur pipeline for its high sensitivity to dynamic change. By decoding the latent sharp images from the event stream and blurred images, we can provide a fine-grained supervision signal for scene reconstruction optimization. Extensive experiments across a range of scenes demonstrate that DeblurSplat not only excels in generating high-fidelity novel views but also achieves significant rendering efficiency compared to the SOTAs in deblur 3D-GS.
Authors: Shuwei Guo, Simin Luan, Yan Ke, Zeyd Boukhers, John See, Cong Yang
Abstract: Moir\'e patterns arise from spectral aliasing between display pixel lattices and camera sensor grids, manifesting as anisotropic, multi-scale artifacts that pose significant challenges for digital image demoir\'eing. We propose Moir\'eNet, a convolutional neural U-Net-based framework that synergistically integrates frequency and spatial domain features for effective artifact removal. Moir\'eNet introduces two key components: a Directional Frequency-Spatial Encoder (DFSE) that discerns moir\'e orientation via directional difference convolution, and a Frequency-Spatial Adaptive Selector (FSAS) that enables precise, feature-adaptive suppression. Extensive experiments demonstrate that Moir\'eNet achieves state-of-the-art performance on public and actively used datasets while being highly parameter-efficient. With only 5.513M parameters, representing a 48% reduction compared to ESDNet-L, Moir\'eNet combines superior restoration quality with parameter efficiency, making it well-suited for resource-constrained applications including smartphone photography, industrial imaging, and augmented reality.
Authors: Yunzhe Shen, Kai Peng, Leiye Liu, Wei Ji, Jingjing Li, Miao Zhang, Yongri Piao, Huchuan Lu
Abstract: Audio-visual segmentation (AVS) plays a critical role in multimodal machine learning by effectively integrating audio and visual cues to precisely segment objects or regions within visual scenes. Recent AVS methods have demonstrated significant improvements. However, they overlook the inherent frequency-domain contradictions between audio and visual modalities--the pervasively interfering noise in audio high-frequency signals vs. the structurally rich details in visual high-frequency signals. Ignoring these differences can result in suboptimal performance. In this paper, we rethink the AVS task from a deeper perspective by reformulating AVS task as a frequency-domain decomposition and recomposition problem. To this end, we introduce a novel Frequency-Aware Audio-Visual Segmentation (FAVS) framework consisting of two key modules: Frequency-Domain Enhanced Decomposer (FDED) module and Synergistic Cross-Modal Consistency (SCMC) module. FDED module employs a residual-based iterative frequency decomposition to discriminate modality-specific semantics and structural features, and SCMC module leverages a mixture-of-experts architecture to reinforce semantic consistency and modality-specific feature preservation through dynamic expert routing. Extensive experiments demonstrate that our FAVS framework achieves state-of-the-art performance on three benchmark datasets, and abundant qualitative visualizations further verify the effectiveness of the proposed FDED and SCMC modules. The code will be released as open source upon acceptance of the paper.
Authors: Nguyen Van Tu, Pham Nguyen Hai Long, Vo Hoai Viet
Abstract: Deep learning has become the de facto standard and dominant paradigm in image analysis tasks, achieving state-of-the-art performance. However, this approach often results in "black-box" models, whose decision-making processes are difficult to interpret, raising concerns about reliability in critical applications. To address this challenge and provide human a method to understand how AI model process and make decision, the field of xAI has emerged. This paper surveys four representative approaches in xAI for visual perception tasks: (i) Saliency Maps, (ii) Concept Bottleneck Models (CBM), (iii) Prototype-based methods, and (iv) Hybrid approaches. We analyze their underlying mechanisms, strengths and limitations, as well as evaluation metrics, thereby providing a comprehensive overview to guide future research and applications.
Authors: Amirhesam Aghanouri, Cristina Olaverri-Monreal
Abstract: Autonomous vehicles (AVs) are expected to revolutionize transportation by improving efficiency and safety. Their success relies on 3D vision systems that effectively sense the environment and detect traffic agents. Among sensors AVs use to create a comprehensive view of surroundings, LiDAR provides high-resolution depth data enabling accurate object detection, safe navigation, and collision avoidance. However, collecting real-world LiDAR data is time-consuming and often affected by noise and sparsity due to adverse weather or sensor limitations. This work applies a denoising diffusion probabilistic model (DDPM), enhanced with novel noise scheduling and time-step embedding techniques to generate high-quality synthetic data for augmentation, thereby improving performance across a range of computer vision tasks, particularly in AV perception. These modifications impact the denoising process and the model's temporal awareness, allowing it to produce more realistic point clouds based on the projection. The proposed method was extensively evaluated under various configurations using the IAMCV and KITTI-360 datasets, with four performance metrics compared against state-of-the-art (SOTA) methods. The results demonstrate the model's superior performance over most existing baselines and its effectiveness in mitigating the effects of noisy and sparse LiDAR data, producing diverse point clouds with rich spatial relationships and structural detail.
Authors: Chuni Liu, Hongjie Li, Jiaqi Du, Yangyang Hou, Qian Sun, Lei Jin, Ke Xu
Abstract: The pretraining-finetuning paradigm is a crucial strategy in metallic surface defect detection for mitigating the challenges posed by data scarcity. However, its implementation presents a critical dilemma. Pretraining on natural image datasets such as ImageNet, faces a significant domain gap. Meanwhile, naive self-supervised pretraining on in-domain industrial data is often ineffective due to the inability of existing learning objectives to distinguish subtle defect patterns from complex background noise and textures. To resolve this, we introduce Anomaly-Guided Self-Supervised Pretraining (AGSSP), a novel paradigm that explicitly guides representation learning through anomaly priors. AGSSP employs a two-stage framework: (1) it first pretrains the model's backbone by distilling knowledge from anomaly maps, encouraging the network to capture defect-salient features; (2) it then pretrains the detector using pseudo-defect boxes derived from these maps, aligning it with localization tasks. To enable this, we develop a knowledge-enhanced method to generate high-quality anomaly maps and collect a large-scale industrial dataset of 120,000 images. Additionally, we present two small-scale, pixel-level labeled metallic surface defect datasets for validation. Extensive experiments demonstrate that AGSSP consistently enhances performance across various settings, achieving up to a 10\% improvement in mAP@0.5 and 11.4\% in mAP@0.5:0.95 compared to ImageNet-based models. All code, pretrained models, and datasets are publicly available at https://clovermini.github.io/AGSSP-Dev/.
Authors: Kartik Teotia, Helge Rhodin, Mohit Mendiratta, Hyeongwoo Kim, Marc Habermann, Christian Theobalt
Abstract: We introduce the first method for audio-driven universal photorealistic avatar synthesis, combining a person-agnostic speech model with our novel Universal Head Avatar Prior (UHAP). UHAP is trained on cross-identity multi-view videos. In particular, our UHAP is supervised with neutral scan data, enabling it to capture the identity-specific details at high fidelity. In contrast to previous approaches, which predominantly map audio features to geometric deformations only while ignoring audio-dependent appearance variations, our universal speech model directly maps raw audio inputs into the UHAP latent expression space. This expression space inherently encodes, both, geometric and appearance variations. For efficient personalization to new subjects, we employ a monocular encoder, which enables lightweight regression of dynamic expression variations across video frames. By accounting for these expression-dependent changes, it enables the subsequent model fine-tuning stage to focus exclusively on capturing the subject's global appearance and geometry. Decoding these audio-driven expression codes via UHAP generates highly realistic avatars with precise lip synchronization and nuanced expressive details, such as eyebrow movement, gaze shifts, and realistic mouth interior appearance as well as motion. Extensive evaluations demonstrate that our method is not only the first generalizable audio-driven avatar model that can account for detailed appearance modeling and rendering, but it also outperforms competing (geometry-only) methods across metrics measuring lip-sync accuracy, quantitative image quality, and perceptual realism.
Authors: Pamela Osuna-Vargas, Altug Kamacioglu, Dominik F. Aschauer, Petros E. Vlachos, Sercan Alipek, Jochen Triesch, Simon Rumpel, Matthias Kaschube
Abstract: Dendritic spines are key structural components of excitatory synapses in the brain. Given the size of dendritic spines provides a proxy for synaptic efficacy, their detection and tracking across time is important for studies of the neural basis of learning and memory. Despite their relevance, large-scale analyses of the structural dynamics of dendritic spines in 3D+time microscopy data remain challenging and labor-intense. Here, we present a modular machine learning-based pipeline designed to automate the detection, time-tracking, and feature extraction of dendritic spines in volumes chronically recorded with two-photon microscopy. Our approach tackles the challenges posed by biological data by combining a transformer-based detection module, a depth-tracking component that integrates spatial features, a time-tracking module to associate 3D spines across time by leveraging spatial consistency, and a feature extraction unit that quantifies biologically relevant spine properties. We validate our method on open-source labeled spine data, and on two complementary annotated datasets that we publish alongside this work: one for detection and depth-tracking, and one for time-tracking, which, to the best of our knowledge, is the first data of this kind. To encourage future research, we release our data, code, and pre-trained weights at https://github.com/pamelaosuna/SynapFlow, establishing a baseline for scalable, end-to-end analysis of dendritic spine dynamics.
Authors: Matheus Vin\'icius Todescato, Joel Lu\'is Carbonera
Abstract: While deep learning, including Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs), has significantly advanced classification performance, its typical reliance on extensive annotated datasets presents a major obstacle in many practical scenarios where such data is scarce. Vision-language models (VLMs) and transfer learning with pre-trained visual models appear as promising techniques to deal with this problem. This paper proposes a novel zero-shot image classification framework that combines a VLM and a pre-trained visual model within a self-learning cycle. Requiring only the set of class names and no labeled training data, our method utilizes a confidence-based pseudo-labeling strategy to train a lightweight classifier directly on the test data, enabling dynamic adaptation. The VLM identifies high-confidence samples, and the pre-trained visual model enhances their visual representations. These enhanced features then iteratively train the classifier, allowing the system to capture complementary semantic and visual cues without supervision. Notably, our approach avoids VLM fine-tuning and the use of large language models, relying on the visual-only model to reduce the dependence on semantic representation. Experimental evaluations on ten diverse datasets demonstrate that our approach outperforms the baseline zero-shot method.
Authors: Zijing Guo, Yunyang Zhao, Lin Wang
Abstract: Mirror-containing environments pose unique challenges for 3D reconstruction and novel view synthesis (NVS), as reflective surfaces introduce view-dependent distortions and inconsistencies. While cutting-edge methods such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) excel in typical scenes, their performance deteriorates in the presence of mirrors. Existing solutions mainly focus on handling mirror surfaces through symmetry mapping but often overlook the rich information carried by mirror reflections. These reflections offer complementary perspectives that can fill in absent details and significantly enhance reconstruction quality. To advance 3D reconstruction in mirror-rich environments, we present MirrorScene3D, a comprehensive dataset featuring diverse indoor scenes, 1256 high-quality images, and annotated mirror masks, providing a benchmark for evaluating reconstruction methods in reflective settings. Building on this, we propose ReflectiveGS, an extension of 3D Gaussian Splatting that utilizes mirror reflections as complementary viewpoints rather than simple symmetry artifacts, enhancing scene geometry and recovering absent details. Experiments on MirrorScene3D show that ReflectiveGaussian outperforms existing methods in SSIM, PSNR, LPIPS, and training speed, setting a new benchmark for 3D reconstruction in mirror-rich environments.
Authors: Cristina Iacono, Mariarosaria Meola, Federica Conte, Laura Mecozzi, Umberto Bracale, Pietro Falco, Fanny Ficuciello
Abstract: Cholecystectomy is one of the most frequently performed procedures in gastrointestinal surgery, and the laparoscopic approach is the gold standard for symptomatic cholecystolithiasis and acute cholecystitis. In addition to the advantages of a significantly faster recovery and better cosmetic results, the laparoscopic approach bears a higher risk of bile duct injury, which has a significant impact on quality of life and survival. To avoid bile duct injury, it is essential to improve the intraoperative visualization of the bile duct. This work aims to address this problem by leveraging a deep-learning approach for the localization of the biliary tract from white-light images acquired during the surgical procedures. To this end, the construction and annotation of an image database to train the Yolo detection algorithm has been employed. Besides classical data augmentation techniques, the paper proposes Generative Adversarial Network (GAN) for the generation of a synthetic portion of the training dataset. Experimental results have been discussed along with ethical considerations.
Authors: Jiabao Chen, Shan Xiong, Jialin Peng
Abstract: Domain adaptive segmentation (DAS) of numerous organelle instances from large-scale electron microscopy (EM) is a promising way to enable annotation-efficient learning. Inspired by SAM, we propose a promptable multitask framework, namely Prompt-DAS, which is flexible enough to utilize any number of point prompts during the adaptation training stage and testing stage. Thus, with varying prompt configurations, Prompt-DAS can perform unsupervised domain adaptation (UDA) and weakly supervised domain adaptation (WDA), as well as interactive segmentation during testing. Unlike the foundation model SAM, which necessitates a prompt for each individual object instance, Prompt-DAS is only trained on a small dataset and can utilize full points on all instances, sparse points on partial instances, or even no points at all, facilitated by the incorporation of an auxiliary center-point detection task. Moreover, a novel prompt-guided contrastive learning is proposed to enhance discriminative feature learning. Comprehensive experiments conducted on challenging benchmarks demonstrate the effectiveness of the proposed approach over existing UDA, WDA, and SAM-based approaches.
Authors: Hao Wang, Eiki Murata, Lingfang Zhang, Ayako Sato, So Fukuda, Ziqi Yin, Wentao Hu, Keisuke Nakao, Yusuke Nakamura, Sebastian Zwirner, Yi-Chia Chen, Hiroyuki Otomo, Hiroki Ouchi, Daisuke Kawahara
Abstract: Recent advances in multimodal large language models (MLLMs) have significantly enhanced video understanding capabilities, opening new possibilities for practical applications. Yet current video benchmarks focus largely on indoor scenes or short-range outdoor activities, leaving the challenges associated with long-distance travel largely unexplored. Mastering extended geospatial-temporal trajectories is critical for next-generation MLLMs, underpinning real-world tasks such as embodied-AI planning and navigation. To bridge this gap, we present VIR-Bench, a novel benchmark consisting of 200 travel videos that frames itinerary reconstruction as a challenging task designed to evaluate and push forward MLLMs' geospatial-temporal intelligence. Experimental results reveal that state-of-the-art MLLMs, including proprietary ones, struggle to achieve high scores, underscoring the difficulty of handling videos that span extended spatial and temporal scales. Moreover, we conduct an in-depth case study in which we develop a prototype travel-planning agent that leverages the insights gained from VIR-Bench. The agent's markedly improved itinerary recommendations verify that our evaluation protocol not only benchmarks models effectively but also translates into concrete performance gains in user-facing applications.
Authors: Honghao Chen, Xingzhou Lou, Xiaokun Feng, Kaiqi Huang, Xinlong Wang
Abstract: Chain of thought reasoning has demonstrated remarkable success in large language models, yet its adaptation to vision-language reasoning remains an open challenge with unclear best practices. Existing attempts typically employ reasoning chains at a coarse-grained level, which struggles to perform fine-grained structured reasoning and, more importantly, are difficult to evaluate the reward and quality of intermediate reasoning. In this work, we delve into chain of step reasoning for vision-language models, enabling assessing reasoning step quality accurately and leading to effective reinforcement learning and inference-time scaling with fine-grained rewards. We present a simple, effective, and fully transparent framework, including the step-level reasoning data, process reward model (PRM), and reinforcement learning training. With the proposed approaches, our models set strong baselines with consistent improvements on challenging vision-language benchmarks. More importantly, we conduct a thorough empirical analysis and ablation study, unveiling the impact of each component and several intriguing properties of inference-time scaling. We believe this paper serves as a baseline for vision-language models and offers insights into more complex multimodal reasoning. Our dataset, PRM, and code will be available at https://github.com/baaivision/CoS.
Authors: Ioannis Sarafis, Alexandros Papadopoulos, Anastasios Delopoulos
Abstract: In this paper, we propose a weakly supervised semantic segmentation approach for food images which takes advantage of the zero-shot capabilities and promptability of the Segment Anything Model (SAM) along with the attention mechanisms of Vision Transformers (ViTs). Specifically, we use class activation maps (CAMs) from ViTs to generate prompts for SAM, resulting in masks suitable for food image segmentation. The ViT model, a Swin Transformer, is trained exclusively using image-level annotations, eliminating the need for pixel-level annotations during training. Additionally, to enhance the quality of the SAM-generated masks, we examine the use of image preprocessing techniques in combination with single-mask and multi-mask SAM generation strategies. The methodology is evaluated on the FoodSeg103 dataset, generating an average of 2.4 masks per image (excluding background), and achieving an mIoU of 0.54 for the multi-mask scenario. We envision the proposed approach as a tool to accelerate food image annotation tasks or as an integrated component in food and nutrition tracking applications.
Authors: Jierui Qu, Jianchun Zhao
Abstract: Accurate segmentation of cardiac anatomy in echocardiography is essential for cardiovascular diagnosis and treatment. Yet echocardiography is prone to deformation and speckle noise, causing frame-to-frame segmentation jitter. Even with high accuracy in single-frame segmentation, temporal instability can weaken functional estimates and impair clinical interpretability. To address these issues, we propose DyL-UNet, a dynamic learning-based temporal consistency U-Net segmentation architecture designed to achieve temporally stable and precise echocardiographic segmentation. The framework constructs an Echo-Dynamics Graph (EDG) through dynamic learning to extract dynamic information from videos. DyL-UNet incorporates multiple Swin-Transformer-based encoder-decoder branches for processing single-frame images. It further introduces Cardiac Phase-Dynamics Attention (CPDA) at the skip connections, which uses EDG-encoded dynamic features and cardiac-phase cues to enforce temporal consistency during segmentation. Extensive experiments on the CAMUS and EchoNet-Dynamic datasets demonstrate that DyL-UNet maintains segmentation accuracy comparable to existing methods while achieving superior temporal consistency, providing a reliable solution for automated clinical echocardiography.
Authors: Zijian Ling, Han Zhang, Yazhuo Zhou, Jiahao Cui
Abstract: This paper presents ColorBlindnessEval, a novel benchmark designed to evaluate the robustness of Vision-Language Models (VLMs) in visually adversarial scenarios inspired by the Ishihara color blindness test. Our dataset comprises 500 Ishihara-like images featuring numbers from 0 to 99 with varying color combinations, challenging VLMs to accurately recognize numerical information embedded in complex visual patterns. We assess 9 VLMs using Yes/No and open-ended prompts and compare their performance with human participants. Our experiments reveal limitations in the models' ability to interpret numbers in adversarial contexts, highlighting prevalent hallucination issues. These findings underscore the need to improve the robustness of VLMs in complex visual environments. ColorBlindnessEval serves as a valuable tool for benchmarking and improving the reliability of VLMs in real-world applications where accuracy is critical.
Authors: Hung Nguyen, Runfa Li, An Le, Truong Nguyen
Abstract: 3D Gaussian Splatting (3DGS) has become a powerful representation for image-based object reconstruction, yet its performance drops sharply in sparse-view settings. Prior works address this limitation by employing diffusion models to repair corrupted renders, subsequently using them as pseudo ground truths for later optimization. While effective, such approaches incur heavy computation from the diffusion fine-tuning and repair steps. We present WaveletGaussian, a framework for more efficient sparse-view 3D Gaussian object reconstruction. Our key idea is to shift diffusion into the wavelet domain: diffusion is applied only to the low-resolution LL subband, while high-frequency subbands are refined with a lightweight network. We further propose an efficient online random masking strategy to curate training pairs for diffusion fine-tuning, replacing the commonly used, but inefficient, leave-one-out strategy. Experiments across two benchmark datasets, Mip-NeRF 360 and OmniObject3D, show WaveletGaussian achieves competitive rendering quality while substantially reducing training time.
Authors: Alexey Nekrasov, Ali Athar, Daan de Geus, Alexander Hermans, Bastian Leibe
Abstract: Sa2VA is a recent model for language-guided dense grounding in images and video that achieves state-of-the-art results on multiple segmentation benchmarks and that has become widely popular. However, we found that Sa2VA does not perform according to its full potential for referring video object segmentation tasks. We identify inconsistencies between training and inference procedures as the key factor holding it back. To mitigate this issue, we propose an improved version of Sa2VA, Sa2VA-i, that rectifies these issues and improves the results. In fact, Sa2VA-i sets a new state of the art for multiple video benchmarks and achieves improvements of up to +11.6 J&F on MeViS, +1.4 on Ref-YT-VOS, +3.3 on Ref-DAVIS and +4.1 on ReVOS using the same Sa2VA checkpoints. With our fixes, the Sa2VA-i-1B model even performs on par with the original Sa2VA-26B model on the MeViS benchmark. We hope that this work will show the importance of seemingly trivial implementation details and that it will provide valuable insights for the referring video segmentation field. We provide the code and updated models at https://github.com/kumuji/sa2va-i
Authors: Ganesh Mallya, Yotam Gigi, Dahun Kim, Maxim Neumann, Genady Beryozkin, Tomer Shekel, Anelia Angelova
Abstract: Multi-spectral imagery plays a crucial role in diverse Remote Sensing applications including land-use classification, environmental monitoring and urban planning. These images are widely adopted because their additional spectral bands correlate strongly with physical materials on the ground, such as ice, water, and vegetation. This allows for more accurate identification, and their public availability from missions, such as Sentinel-2 and Landsat, only adds to their value. Currently, the automatic analysis of such data is predominantly managed through machine learning models specifically trained for multi-spectral input, which are costly to train and support. Furthermore, although providing a lot of utility for Remote Sensing, such additional inputs cannot be used with powerful generalist large multimodal models, which are capable of solving many visual problems, but are not able to understand specialized multi-spectral signals. To address this, we propose a training-free approach which introduces new multi-spectral data in a Zero-Shot-only mode, as inputs to generalist multimodal models, trained on RGB-only inputs. Our approach leverages the multimodal models' understanding of the visual space, and proposes to adapt to inputs to that space, and to inject domain-specific information as instructions into the model. We exemplify this idea with the Gemini2.5 model and observe strong Zero-Shot performance gains of the approach on popular Remote Sensing benchmarks for land cover and land use classification and demonstrate the easy adaptability of Gemini2.5 to new inputs. These results highlight the potential for geospatial professionals, working with non-standard specialized inputs, to easily leverage powerful multimodal models, such as Gemini2.5, to accelerate their work, benefiting from their rich reasoning and contextual capabilities, grounded in the specialized sensor data.
Authors: Guoxin Wang, Jun Zhao, Xinyi Liu, Yanbo Liu, Xuyang Cao, Chao Li, Zhuoyun Liu, Qintian Sun, Fangru Zhou, Haoqiang Xing, Zhenhong Yang
Abstract: Medical imaging provides critical evidence for clinical diagnosis, treatment planning, and surgical decisions, yet most existing imaging models are narrowly focused and require multiple specialized networks, limiting their generalization. Although large-scale language and multimodal models exhibit strong reasoning and multi-task capabilities, real-world clinical applications demand precise visual grounding, multimodal integration, and chain-of-thought reasoning. We introduce Citrus-V, a multimodal medical foundation model that combines image analysis with textual reasoning. The model integrates detection, segmentation, and multimodal chain-of-thought reasoning, enabling pixel-level lesion localization, structured report generation, and physician-like diagnostic inference in a single framework. We propose a novel multimodal training approach and release a curated open-source data suite covering reasoning, detection, segmentation, and document understanding tasks. Evaluations demonstrate that Citrus-V outperforms existing open-source medical models and expert-level imaging systems across multiple benchmarks, delivering a unified pipeline from visual grounding to clinical reasoning and supporting precise lesion quantification, automated reporting, and reliable second opinions.
Authors: Ilhan Skender, Kailin Tong, Selim Solmaz, Daniel Watzenig
Abstract: Traffic safety remains a critical global concern, with timely and accurate accident detection essential for hazard reduction and rapid emergency response. Infrastructure-based vision sensors offer scalable and efficient solutions for continuous real-time monitoring, facilitating automated detection of acci- dents directly from captured images. This research investigates the zero-shot capabilities of multimodal large language models (MLLMs) for detecting and describing traffic accidents using images from infrastructure cameras, thus minimizing reliance on extensive labeled datasets. Main contributions include: (1) Evaluation of MLLMs using the simulated DeepAccident dataset from CARLA, explicitly addressing the scarcity of diverse, realistic, infrastructure-based accident data through controlled simulations; (2) Comparative performance analysis between Gemini 1.5 and 2.0, Gemma 3 and Pixtral models in acci- dent identification and descriptive capabilities without prior fine-tuning; and (3) Integration of advanced visual analytics, specifically YOLO for object detection, Deep SORT for multi- object tracking, and Segment Anything (SAM) for instance segmentation, into enhanced prompts to improve model accuracy and explainability. Key numerical results show Pixtral as the top performer with an F1-score of 0.71 and 83% recall, while Gemini models gained precision with enhanced prompts (e.g., Gemini 1.5 rose to 90%) but suffered notable F1 and recall losses. Gemma 3 offered the most balanced performance with minimal metric fluctuation. These findings demonstrate the substantial potential of integrating MLLMs with advanced visual analytics techniques, enhancing their applicability in real-world automated traffic monitoring systems.
Authors: G\"orkay Aydemir, Weidi Xie, Fatma G\"uney
Abstract: In this paper, we consider the problem of long-term point tracking, which requires consistent identification of points across video frames under significant appearance changes, motion, and occlusion. We target the online setting, i.e. tracking points frame-by-frame, making it suitable for real-time and streaming applications. We extend our prior model Track-On into Track-On2, a simple and efficient transformer-based model for online long-term tracking. Track-On2 improves both performance and efficiency through architectural refinements, more effective use of memory, and improved synthetic training strategies. Unlike prior approaches that rely on full-sequence access or iterative updates, our model processes frames causally and maintains temporal coherence via a memory mechanism, which is key to handling drift and occlusions without requiring future frames. At inference, we perform coarse patch-level classification followed by refinement. Beyond architecture, we systematically study synthetic training setups and their impact on memory behavior, showing how they shape temporal robustness over long sequences. Through comprehensive experiments, Track-On2 achieves state-of-the-art results across five synthetic and real-world benchmarks, surpassing prior online trackers and even strong offline methods that exploit bidirectional context. These results highlight the effectiveness of causal, memory-based architectures trained purely on synthetic data as scalable solutions for real-world point tracking. Project page: https://kuis-ai.github.io/track_on2
Authors: Adam Romlein, Benjamin X. Hou, Yuval Boss, Cynthia L. Christman, Stacie Koslovsky, Erin E. Moreland, Jason Parham, Anthony Hoogs
Abstract: We introduce KAMERA: a comprehensive system for multi-camera, multi-spectral synchronization and real-time detection of seals and polar bears. Utilized in aerial surveys for ice-associated seals in the Bering, Chukchi, and Beaufort seas around Alaska, KAMERA provides up to an 80% reduction in dataset processing time over previous methods. Our rigorous calibration and hardware synchronization enable using multiple spectra for object detection. All collected data are annotated with metadata so they can be easily referenced later. All imagery and animal detections from a survey are mapped onto a world plane for accurate surveyed area estimates and quick assessment of survey results. We hope KAMERA will inspire other mapping and detection efforts in the scientific community, with all software, models, and schematics fully open-sourced.
Authors: Maurf Hassan, Steven Davy, Muhammad Zawish, Owais Bin Zuber, Nouman Ashraf
Abstract: Spiking Neural Networks (SNNs) offer significant potential for enabling energy-efficient intelligence at the edge. However, performing full SNN inference at the edge can be challenging due to the latency and energy constraints arising from fixed and high timestep overheads. Edge-cloud co-inference systems present a promising solution, but their deployment is often hindered by high latency and feature transmission costs. To address these issues, we introduce NeuCODEX, a neuromorphic co-inference architecture that jointly optimizes both spatial and temporal redundancy. NeuCODEX incorporates a learned spike-driven compression module to reduce data transmission and employs a dynamic early-exit mechanism to adaptively terminate inference based on output confidence. We evaluated NeuCODEX on both static images (CIFAR10 and Caltech) and neuromorphic event streams (CIFAR10-DVS and N-Caltech). To demonstrate practicality, we prototyped NeuCODEX on ResNet-18 and VGG-16 backbones in a real edge-to-cloud testbed. Our proposed system reduces data transfer by up to 2048x and edge energy consumption by over 90%, while reducing end-to-end latency by up to 3x compared to edge-only inference, all with a negligible accuracy drop of less than 2%. In doing so, NeuCODEX enables practical, high-performance SNN deployment in resource-constrained environments.
Authors: Yun Wang, Junjie Hu, Junhui Hou, Chenghao Zhang, Renwei Yang, Dapeng Oliver Wu
Abstract: Recent self-supervised stereo matching methods have made significant progress, but their performance significantly degrades under adverse weather conditions such as night, rain, and fog. We identify two primary weaknesses contributing to this performance degradation. First, adverse weather introduces noise and reduces visibility, making CNN-based feature extractors struggle with degraded regions like reflective and textureless areas. Second, these degraded regions can disrupt accurate pixel correspondences, leading to ineffective supervision based on the photometric consistency assumption. To address these challenges, we propose injecting robust priors derived from the visual foundation model into the CNN-based feature extractor to improve feature representation under adverse weather conditions. We then introduce scene correspondence priors to construct robust supervisory signals rather than relying solely on the photometric consistency assumption. Specifically, we create synthetic stereo datasets with realistic weather degradations. These datasets feature clear and adverse image pairs that maintain the same semantic context and disparity, preserving the scene correspondence property. With this knowledge, we propose a robust self-supervised training paradigm, consisting of two key steps: robust self-supervised scene correspondence learning and adverse weather distillation. Both steps aim to align underlying scene results from clean and adverse image pairs, thus improving model disparity estimation under adverse weather effects. Extensive experiments demonstrate the effectiveness and versatility of our proposed solution, which outperforms existing state-of-the-art self-supervised methods. Codes are available at \textcolor{blue}{https://github.com/cocowy1/RoSe-Robust-Self-supervised-Stereo-Matching-under-Adverse-Weather-Conditions}.
Authors: Siddharth Gupta, Jitin Singla
Abstract: Colorectal cancer (CRC), a lethal disease, begins with the growth of abnormal mucosal cell proliferation called polyps in the inner wall of the colon. When left undetected, polyps can become malignant tumors. Colonoscopy is the standard procedure for detecting polyps, as it enables direct visualization and removal of suspicious lesions. Manual detection by colonoscopy can be inconsistent and is subject to oversight. Therefore, object detection based on deep learning offers a better solution for a more accurate and real-time diagnosis during colonoscopy. In this work, we propose YOLO-LAN, a YOLO-based polyp detection pipeline, trained using M2IoU loss, versatile data augmentations and negative data to replicate real clinical situations. Our pipeline outperformed existing methods for the Kvasir-seg and BKAI-IGH NeoPolyp datasets, achieving mAP$_{50}$ of 0.9619, mAP$_{50:95}$ of 0.8599 with YOLOv12 and mAP$_{50}$ of 0.9540, mAP$_{50:95}$ of 0.8487 with YOLOv8 on the Kvasir-seg dataset. The significant increase is achieved in mAP$_{50:95}$ score, showing the precision of polyp detection. We show robustness based on polyp size and precise location detection, making it clinically relevant in AI-assisted colorectal screening.
Authors: Mingqi Gao, Jingkun Chen, Yunqi Miao, Gengshen Wu, Zhijin Qin, Jungong Han
Abstract: This technical report explores the MOSEv2 track of the LSVOS Challenge, which targets complex semi-supervised video object segmentation. By analysing and adapting SeC, an enhanced SAM-2 framework, we conduct a detailed study of its long-term memory and concept-aware memory, showing that long-term memory preserves temporal continuity under occlusion and reappearance, while concept-aware memory supplies semantic priors that suppress distractors; together, these traits directly benefit several MOSEv2's core challenges. Our solution achieves a JF score of 39.89% on the test set, ranking 1st in the MOSEv2 track of the LSVOS Challenge.
Authors: Yueyan Li, Chenggong Zhao, Zeyuan Zang, Caixia Yuan, Xiaojie Wang
Abstract: Vision-Language Models (VLMs) have demonstrated remarkable performance across a variety of real-world tasks. However, existing VLMs typically process visual information by serializing images, a method that diverges significantly from the parallel nature of human vision. Moreover, their opaque internal mechanisms hinder both deeper understanding and architectural innovation. Inspired by the dual-stream hypothesis of human vision, which distinguishes the "what" and "where" pathways, we deconstruct the visual processing in VLMs into object recognition and spatial perception for separate study. For object recognition, we convert images into text token maps and find that the model's perception of image content unfolds as a two-stage process from shallow to deep layers, beginning with attribute recognition and culminating in semantic disambiguation. For spatial perception, we theoretically derive and empirically verify the geometric structure underlying the positional representation in VLMs. Based on these findings, we introduce an instruction-agnostic token compression algorithm based on a plug-and-play visual decoder to improve decoding efficiency, and a RoPE scaling technique to enhance spatial reasoning. Through rigorous experiments, our work validates these analyses, offering a deeper understanding of VLM internals and providing clear principles for designing more capable future architectures.
Authors: Ioanna Ntinou, Alexandros Xenos, Yassine Ouali, Adrian Bulat, Georgios Tzimiropoulos
Abstract: Contrastively-trained Vision-Language Models (VLMs), such as CLIP, have become the standard approach for learning discriminative vision-language representations. However, these models often exhibit shallow language understanding, manifesting bag-of-words behaviour. These limitations are reinforced by their dual-encoder design, which induces a modality gap. Additionally, the reliance on vast web-collected data corpora for training makes the process computationally expensive and introduces significant privacy concerns. To address these limitations, in this work, we challenge the necessity of vision encoders for retrieval tasks by introducing a vision-free, single-encoder retrieval pipeline. Departing from the traditional text-to-image retrieval paradigm, we migrate to a text-to-text paradigm with the assistance of VLLM-generated structured image descriptions. We demonstrate that this paradigm shift has significant advantages, including a substantial reduction of the modality gap, improved compositionality, and better performance on short and long caption queries, all attainable with only a few hours of calibration on two GPUs. Additionally, substituting raw images with textual descriptions introduces a more privacy-friendly alternative for retrieval. To further assess generalisation and address some of the shortcomings of prior compositionality benchmarks, we release two benchmarks derived from Flickr30k and COCO, containing diverse compositional queries made of short captions, which we coin subFlickr and subCOCO. Our vision-free retriever matches and often surpasses traditional multimodal models. Importantly, our approach achieves state-of-the-art zero-shot performance on multiple retrieval and compositionality benchmarks, with models as small as 0.3B parameters. Code is available at: https://github.com/IoannaNti/LexiCLIP
Authors: Israfel Salazar, Desmond Elliott, Yova Kementchedjhieva
Abstract: Contrastive vision-language models (VLMs) have made significant progress in binding visual and textual information, but understanding long, dense captions remains an open challenge. We hypothesize that compositionality, the capacity to reason about object-attribute bindings and inter-object relationships, is key to understanding longer captions. In this paper, we investigate the interaction between compositionality and long-caption understanding, asking whether training for one property enhances the other. We train and evaluate a range of models that target each of these capabilities. Our results reveal a bidirectional relationship: compositional training improves performance on long-caption retrieval, and training on long captions promotes compositionality. However, these gains are sensitive to data quality and model design. We find that training on poorly structured captions, or with limited parameter updates, fails to support generalization. Likewise, strategies that aim at retaining general alignment, such as freezing positional embeddings, do not improve compositional understanding. Overall, we find that compositional understanding and long-caption understanding are intertwined capabilities that can be jointly learned through training on dense, grounded descriptions. Despite these challenges, we show that models trained on high-quality, long-caption data can achieve strong performance in both tasks, offering practical guidance for improving VLM generalization.
Authors: Earl Ranario, Ismael Mayanja, Heesup Yun, Brian N. Bailey, J. Mason Earles
Abstract: Accurate plant segmentation in thermal imagery remains a significant challenge for high throughput field phenotyping, particularly in outdoor environments where low contrast between plants and weeds and frequent occlusions hinder performance. To address this, we present a framework that leverages synthetic RGB imagery, a limited set of real annotations, and GAN-based cross-modality alignment to enhance semantic segmentation in thermal images. We trained models on 1,128 synthetic images containing complex mixtures of crop and weed plants in order to generate image segmentation masks for crop and weed plants. We additionally evaluated the benefit of integrating as few as five real, manually segmented field images within the training process using various sampling strategies. When combining all the synthetic images with a few labeled real images, we observed a maximum relative improvement of 22% for the weed class and 17% for the plant class compared to the full real-data baseline. Cross-modal alignment was enabled by translating RGB to thermal using CycleGAN-turbo, allowing robust template matching without calibration. Results demonstrated that combining synthetic data with limited manual annotations and cross-domain translation via generative models can significantly boost segmentation performance in complex field environments for multi-model imagery.
Authors: Yunzhi Xu, Yushuang Ding, Hu Sun, Hongxi Zhang, Li Zhao
Abstract: Evaluation of hydrocephalus in children is challenging, and the related research is limited by a lack of publicly available, expert-annotated datasets, particularly those with segmentation of the choroid plexus. To address this, we present HyKid, an open-source dataset from 48 pediatric patients with hydrocephalus. 3D MRIs were provided with 1mm isotropic resolution, which was reconstructed from routine low-resolution images using a slice-to-volume algorithm. Manually corrected segmentations of brain tissues, including white matter, grey matter, lateral ventricle, external CSF, and the choroid plexus, were provided by an experienced neurologist. Additionally, structured data was extracted from clinical radiology reports using a Retrieval-Augmented Generation framework. The strong correlation between choroid plexus volume and total CSF volume provided a potential biomarker for hydrocephalus evaluation, achieving excellent performance in a predictive model (AUC = 0.87). The proposed HyKid dataset provided a high-quality benchmark for neuroimaging algorithms development, and it revealed the choroid plexus-related features in hydrocephalus assessments. Our datasets are publicly available at https://www.synapse.org/Synapse:syn68544889.
Authors: Tongshuai Wu, Chao Lu, Ze Song, Yunlong Lin, Sizhe Fan, Xuemei Chen
Abstract: With the widespread deployment of dashcams and advancements in computer vision, developing accident prediction models from the dashcam perspective has become critical for proactive safety interventions. However, two key challenges persist: modeling feature-level interactions among traffic participants (often occluded in dashcam views) and capturing complex, asynchronous multi-temporal behavioral cues preceding accidents. To deal with these two challenges, a Multi-scale Feature Interaction Network (MsFIN) is proposed for early-stage accident anticipation from dashcam videos. MsFIN has three layers for multi-scale feature aggregation, temporal feature processing and multi-scale feature post fusion, respectively. For multi-scale feature aggregation, a Multi-scale Module is designed to extract scene representations at short-term, mid-term and long-term temporal scales. Meanwhile, the Transformer architecture is leveraged to facilitate comprehensive feature interactions. Temporal feature processing captures the sequential evolution of scene and object features under causal constraints. In the multi-scale feature post fusion stage, the network fuses scene and object features across multiple temporal scales to generate a comprehensive risk representation. Experiments on DAD and DADA datasets show that MsFIN significantly outperforms state-of-the-art models with single-scale feature extraction in both prediction correctness and earliness. Ablation studies validate the effectiveness of each module in MsFIN, highlighting how the network achieves superior performance through multi-scale feature fusion and contextual interaction modeling.
Authors: Tianshuo Zhang, Li Gao, Siran Peng, Xiangyu Zhu, Zhen Lei
Abstract: The rise of realistic digital face generation and manipulation poses significant social risks. The primary challenge lies in the rapid and diverse evolution of generation techniques, which often outstrip the detection capabilities of existing models. To defend against the ever-evolving new types of forgery, we need to enable our model to quickly adapt to new domains with limited computation and data while avoiding forgetting previously learned forgery types. In this work, we posit that genuine facial samples are abundant and relatively stable in acquisition methods, while forgery faces continuously evolve with the iteration of manipulation techniques. Given the practical infeasibility of exhaustively collecting all forgery variants, we frame face forgery detection as a continual learning problem and allow the model to develop as new forgery types emerge. Specifically, we employ a Developmental Mixture of Experts (MoE) architecture that uses LoRA models as its individual experts. These experts are organized into two groups: a Real-LoRA to learn and refine knowledge of real faces, and multiple Fake-LoRAs to capture incremental information from different forgery types. To prevent catastrophic forgetting, we ensure that the learning direction of Fake-LoRAs is orthogonal to the established subspace. Moreover, we integrate orthogonal gradients into the orthogonal loss of Fake-LoRAs, preventing gradient interference throughout the training process of each task. Experimental results under both the datasets and manipulation types incremental protocols demonstrate the effectiveness of our method.
Authors: Shufan Li, Jiuxiang Gu, Kangning Liu, Zhe Lin, Zijun Wei, Aditya Grover, Jason Kuen
Abstract: We proposed Lavida-O, a unified multi-modal Masked Diffusion Model (MDM) capable of image understanding and generation tasks. Unlike existing multimodal diffsion language models such as MMaDa and Muddit which only support simple image-level understanding tasks and low-resolution image generation, Lavida-O exhibits many new capabilities such as object grounding, image-editing, and high-resolution (1024px) image synthesis. It is also the first unified MDM that uses its understanding capabilities to improve image generation and editing results through planning and iterative self-reflection. To allow effective and efficient training and sampling, Lavida-O ntroduces many novel techniques such as Elastic Mixture-of-Transformer architecture, universal text conditioning, and stratified sampling. \ours~achieves state-of-the-art performance on a wide range of benchmarks such as RefCOCO object grounding, GenEval text-to-image generation, and ImgEdit image editing, outperforming existing autoregressive and continuous diffusion models such as Qwen2.5-VL and FluxKontext-dev, while offering considerable speedup at inference.
Authors: Benedetta Liberatori, Alessandro Conti, Lorenzo Vaquero, Yiming Wang, Elisa Ricci, Paolo Rota
Abstract: What does it mean for two videos to be similar? Videos may appear similar when judged by the actions they depict, yet entirely different if evaluated based on the locations where they were filmed. While humans naturally compare videos by taking different aspects into account, this ability has not been thoroughly studied and presents a challenge for models that often depend on broad global similarity scores. Large Multimodal Models (LMMs) with video understanding capabilities open new opportunities for leveraging natural language in comparative video tasks. We introduce Concept-based Video Similarity estimation (ConViS), a novel task that compares pairs of videos by computing interpretable similarity scores across a predefined set of key semantic concepts. ConViS allows for human-like reasoning about video similarity and enables new applications such as concept-conditioned video retrieval. To support this task, we also introduce ConViS-Bench, a new benchmark comprising carefully annotated video pairs spanning multiple domains. Each pair comes with concept-level similarity scores and textual descriptions of both differences and similarities. Additionally, we benchmark several state-of-the-art models on ConViS, providing insights into their alignment with human judgments. Our results reveal significant performance differences on ConViS, indicating that some concepts present greater challenges for estimating video similarity. We believe that ConViS-Bench will serve as a valuable resource for advancing research in language-driven video understanding.
Authors: Gabriel Maldonado, Narges Rashvand, Armin Danesh Pazho, Ghazal Alinezhad Noghre, Vinit Katariya, Hamed Tabkhi
Abstract: Continuous human motion understanding remains a core challenge in computer vision due to its high dimensionality and inherent redundancy. Efficient compression and representation are crucial for analyzing complex motion dynamics. In this work, we introduce an adversarially-refined VQ-GAN framework with dense motion tokenization for compressing spatio-temporal heatmaps while preserving the fine-grained traces of human motion. Our approach combines dense motion tokenization with adversarial refinement, which eliminates reconstruction artifacts like motion smearing and temporal misalignment observed in non-adversarial baselines. Our experiments on the CMU Panoptic dataset provide conclusive evidence of our method's superiority, outperforming the dVAE baseline by 9.31% SSIM and reducing temporal instability by 37.1%. Furthermore, our dense tokenization strategy enables a novel analysis of motion complexity, revealing that 2D motion can be optimally represented with a compact 128-token vocabulary, while 3D motion's complexity demands a much larger 1024-token codebook for faithful reconstruction. These results establish practical deployment feasibility across diverse motion analysis applications. The code base for this work is available at https://github.com/TeCSAR-UNCC/Pose-Quantization.
Authors: Dheerendranath Battalapalli, Apoorva Safai, Maria Jaramillo, Hyemin Um, Gustavo Adalfo Pineda Ortiz, Ulas Bagci, Manmeet Singh Ahluwalia, Marwa Ismail, Pallavi Tiwari
Abstract: A significant challenge in solid tumors is reliably distinguishing confounding pathologies from malignant neoplasms on routine imaging. While radiomics methods seek surrogate markers of lesion heterogeneity on CT/MRI, many aggregate features across the region of interest (ROI) and miss complex spatial relationships among varying intensity compositions. We present a new Graph-Radiomic Learning (GrRAiL) descriptor for characterizing intralesional heterogeneity (ILH) on clinical MRI scans. GrRAiL (1) identifies clusters of sub-regions using per-voxel radiomic measurements, then (2) computes graph-theoretic metrics to quantify spatial associations among clusters. The resulting weighted graphs encode higher-order spatial relationships within the ROI, aiming to reliably capture ILH and disambiguate confounding pathologies from malignancy. To assess efficacy and clinical feasibility, GrRAiL was evaluated in n=947 subjects spanning three use cases: differentiating tumor recurrence from radiation effects in glioblastoma (GBM; n=106) and brain metastasis (n=233), and stratifying pancreatic intraductal papillary mucinous neoplasms (IPMNs) into no+low vs high risk (n=608). In a multi-institutional setting, GrRAiL consistently outperformed state-of-the-art baselines - Graph Neural Networks (GNNs), textural radiomics, and intensity-graph analysis. In GBM, cross-validation (CV) and test accuracies for recurrence vs pseudo-progression were 89% and 78% with >10% test-accuracy gains over comparators. In brain metastasis, CV and test accuracies for recurrence vs radiation necrosis were 84% and 74% (>13% improvement). For IPMN risk stratification, CV and test accuracies were 84% and 75%, showing >10% improvement.
Authors: Markos Diomataris, Berat Mert Albaba, Giorgio Becherini, Partha Ghosh, Omid Taheri, Michael J. Black
Abstract: The way we perceive the world fundamentally shapes how we move, whether it is how we navigate in a room or how we interact with other humans. Current human motion generation methods, neglect this interdependency and use task-specific ``perception'' that differs radically from that of humans. We argue that the generation of human-like avatar behavior requires human-like perception. Consequently, in this work we present CLOPS, the first human avatar that solely uses egocentric vision to perceive its surroundings and navigate. Using vision as the primary driver of motion however, gives rise to a significant challenge for training avatars: existing datasets have either isolated human motion, without the context of a scene, or lack scale. We overcome this challenge by decoupling the learning of low-level motion skills from learning of high-level control that maps visual input to motion. First, we train a motion prior model on a large motion capture dataset. Then, a policy is trained using Q-learning to map egocentric visual inputs to high-level control commands for the motion prior. Our experiments empirically demonstrate that egocentric vision can give rise to human-like motion characteristics in our avatars. For example, the avatars walk such that they avoid obstacles present in their visual field. These findings suggest that equipping avatars with human-like sensors, particularly egocentric vision, holds promise for training avatars that behave like humans.
Authors: Bingnan Li, Chen-Yu Wang, Haiyang Xu, Xiang Zhang, Ethan Armand, Divyansh Srivastava, Xiaojun Shan, Zeyuan Chen, Jianwen Xie, Zhuowen Tu
Abstract: Despite steady progress in layout-to-image generation, current methods still struggle with layouts containing significant overlap between bounding boxes. We identify two primary challenges: (1) large overlapping regions and (2) overlapping instances with minimal semantic distinction. Through both qualitative examples and quantitative analysis, we demonstrate how these factors degrade generation quality. To systematically assess this issue, we introduce OverLayScore, a novel metric that quantifies the complexity of overlapping bounding boxes. Our analysis reveals that existing benchmarks are biased toward simpler cases with low OverLayScore values, limiting their effectiveness in evaluating model performance under more challenging conditions. To bridge this gap, we present OverLayBench, a new benchmark featuring high-quality annotations and a balanced distribution across different levels of OverLayScore. As an initial step toward improving performance on complex overlaps, we also propose CreatiLayout-AM, a model fine-tuned on a curated amodal mask dataset. Together, our contributions lay the groundwork for more robust layout-to-image generation under realistic and challenging scenarios. Project link: https://mlpc-ucsd.github.io/OverLayBench.
Authors: Sherwin Bahmani, Tianchang Shen, Jiawei Ren, Jiahui Huang, Yifeng Jiang, Haithem Turki, Andrea Tagliasacchi, David B. Lindell, Zan Gojcic, Sanja Fidler, Huan Ling, Jun Gao, Xuanchi Ren
Abstract: The ability to generate virtual environments is crucial for applications ranging from gaming to physical AI domains such as robotics, autonomous driving, and industrial AI. Current learning-based 3D reconstruction methods rely on the availability of captured real-world multi-view data, which is not always readily available. Recent advancements in video diffusion models have shown remarkable imagination capabilities, yet their 2D nature limits the applications to simulation where a robot needs to navigate and interact with the environment. In this paper, we propose a self-distillation framework that aims to distill the implicit 3D knowledge in the video diffusion models into an explicit 3D Gaussian Splatting (3DGS) representation, eliminating the need for multi-view training data. Specifically, we augment the typical RGB decoder with a 3DGS decoder, which is supervised by the output of the RGB decoder. In this approach, the 3DGS decoder can be purely trained with synthetic data generated by video diffusion models. At inference time, our model can synthesize 3D scenes from either a text prompt or a single image for real-time rendering. Our framework further extends to dynamic 3D scene generation from a monocular input video. Experimental results show that our framework achieves state-of-the-art performance in static and dynamic 3D scene generation.
Authors: Weijie Wang, Yeqing Chen, Zeyu Zhang, Hengyu Liu, Haoxiao Wang, Zhiyuan Feng, Wenkang Qin, Zheng Zhu, Donny Y. Chen, Bohan Zhuang
Abstract: Feed-forward 3D Gaussian Splatting (3DGS) has emerged as a highly effective solution for novel view synthesis. Existing methods predominantly rely on a pixel-aligned Gaussian prediction paradigm, where each 2D pixel is mapped to a 3D Gaussian. We rethink this widely adopted formulation and identify several inherent limitations: it renders the reconstructed 3D models heavily dependent on the number of input views, leads to view-biased density distributions, and introduces alignment errors, particularly when source views contain occlusions or low texture. To address these challenges, we introduce VolSplat, a new multi-view feed-forward paradigm that replaces pixel alignment with voxel-aligned Gaussians. By directly predicting Gaussians from a predicted 3D voxel grid, it overcomes pixel alignment's reliance on error-prone 2D feature matching, ensuring robust multi-view consistency. Furthermore, it enables adaptive control over Gaussian density based on 3D scene complexity, yielding more faithful Gaussian point clouds, improved geometric consistency, and enhanced novel-view rendering quality. Experiments on widely used benchmarks including RealEstate10K and ScanNet demonstrate that VolSplat achieves state-of-the-art performance while producing more plausible and view-consistent Gaussian reconstructions. In addition to superior results, our approach establishes a more scalable framework for feed-forward 3D reconstruction with denser and more robust representations, paving the way for further research in wider communities. The video results, code and trained models are available on our project page: https://lhmd.top/volsplat.
Authors: Chen Chen, Pengsheng Guo, Liangchen Song, Jiasen Lu, Rui Qian, Xinze Wang, Tsu-Jui Fu, Wei Liu, Yinfei Yang, Alex Schwing
Abstract: Conditional generative modeling aims to learn a conditional data distribution from samples containing data-condition pairs. For this, diffusion and flow-based methods have attained compelling results. These methods use a learned (flow) model to transport an initial standard Gaussian noise that ignores the condition to the conditional data distribution. The model is hence required to learn both mass transport and conditional injection. To ease the demand on the model, we propose Condition-Aware Reparameterization for Flow Matching (CAR-Flow) -- a lightweight, learned shift that conditions the source, the target, or both distributions. By relocating these distributions, CAR-Flow shortens the probability path the model must learn, leading to faster training in practice. On low-dimensional synthetic data, we visualize and quantify the effects of CAR. On higher-dimensional natural image data (ImageNet-256), equipping SiT-XL/2 with CAR-Flow reduces FID from 2.07 to 1.68, while introducing less than 0.6% additional parameters.
Authors: Mrigank Dhingra, Romit Maulik, Adil Rasheed, Omer San
Abstract: Neural operator learning has emerged as a powerful approach for solving partial differential equations (PDEs) in a data-driven manner. However, applying principal component analysis (PCA) to high-dimensional solution fields incurs significant computational overhead. To address this, we propose a patch-based PCA-Net framework that decomposes the solution fields into smaller patches, applies PCA within each patch, and trains a neural operator in the reduced PCA space. We investigate two different patch-based approaches that balance computational efficiency and reconstruction accuracy: (1) local-to-global patch PCA, and (2) local-to-local patch PCA. The trade-off between computational cost and accuracy is analyzed, highlighting the advantages and limitations of each approach. Furthermore, within each approach, we explore two refinements for the most computationally efficient method: (i) introducing overlapping patches with a smoothing filter and (ii) employing a two-step process with a convolutional neural network (CNN) for refinement. Our results demonstrate that patch-based PCA significantly reduces computational complexity while maintaining high accuracy, reducing end-to-end pipeline processing time by a factor of 3.7 to 4 times compared to global PCA, thefore making it a promising technique for efficient operator learning in PDE-based systems.
Authors: Faizul Rakib Sayem, Shahana Ibrahim
Abstract: The reliability of artificial intelligence (AI) systems in open-world settings depends heavily on their ability to flag out-of-distribution (OOD) inputs unseen during training. Recent advances in large-scale vision-language models (VLMs) have enabled promising few-shot OOD detection frameworks using only a handful of in-distribution (ID) samples. However, existing prompt learning-based OOD methods rely solely on softmax probabilities, overlooking the rich discriminative potential of the feature embeddings learned by VLMs trained on millions of samples. To address this limitation, we propose a novel context optimization (CoOp)-based framework that integrates subspace representation learning with prompt tuning. Our approach improves ID-OOD separability by projecting the ID features into a subspace spanned by prompt vectors, while projecting ID-irrelevant features into an orthogonal null space. To train such OOD detection framework, we design an easy-to-handle end-to-end learning criterion that ensures strong OOD detection performance as well as high ID classification accuracy. Experiments on real-world datasets showcase the effectiveness of our approach.
Authors: Yao Zhao, Haoyue Sun, Yantian Ding, Yanxun Xu
Abstract: Reconstructing individual patient data (IPD) from Kaplan-Meier (KM) plots provides valuable insights for evidence synthesis in clinical research. However, existing approaches often rely on manual digitization, which is error-prone and lacks scalability. To address these limitations, we develop KM-GPT, the first fully automated, AI-powered pipeline for reconstructing IPD directly from KM plots with high accuracy, robustness, and reproducibility. KM-GPT integrates advanced image preprocessing, multi-modal reasoning powered by GPT-5, and iterative reconstruction algorithms to generate high-quality IPD without manual input or intervention. Its hybrid reasoning architecture automates the conversion of unstructured information into structured data flows and validates data extraction from complex KM plots. To improve accessibility, KM-GPT is equipped with a user-friendly web interface and an integrated AI assistant, enabling researchers to reconstruct IPD without requiring programming expertise. KM-GPT was rigorously evaluated on synthetic and real-world datasets, consistently demonstrating superior accuracy. To illustrate its utility, we applied KM-GPT to a meta-analysis of gastric cancer immunotherapy trials, reconstructing IPD to facilitate evidence synthesis and biomarker-based subgroup analyses. By automating traditionally manual processes and providing a scalable, web-based solution, KM-GPT transforms clinical research by leveraging reconstructed IPD to enable more informed downstream analyses, supporting evidence-based decision-making.
Authors: Tianyu Yu, Zefan Wang, Chongyi Wang, Fuwei Huang, Wenshuo Ma, Zhihui He, Tianchi Cai, Weize Chen, Yuxiang Huang, Yuanqian Zhao, Bokai Xu, Junbo Cui, Yingjing Xu, Liqing Ruan, Luoyuan Zhang, Hanyu Liu, Jingkun Tang, Hongyuan Liu, Qining Guo, Wenhao Hu, Bingxiang He, Jie Zhou, Jie Cai, Ji Qi, Zonghao Guo, Chi Chen, Guoyang Zeng, Yuxuan Li, Ganqu Cui, Ning Ding, Xu Han, Yuan Yao, Zhiyuan Liu, Maosong Sun
Abstract: Multimodal Large Language Models (MLLMs) are undergoing rapid progress and represent the frontier of AI development. However, their training and inference efficiency have emerged as a core bottleneck in making MLLMs more accessible and scalable. To address the challenges, we present MiniCPM-V 4.5, an 8B parameter model designed for high efficiency and strong performance. We introduce three core improvements in model architecture, data strategy and training method: a unified 3D-Resampler model architecture for highly compact encoding over images and videos, a unified learning paradigm for document knowledge and text recognition without heavy data engineering, and a hybrid reinforcement learning strategy for proficiency in both short and long reasoning modes. Comprehensive experimental results in OpenCompass evaluation show that MiniCPM-V 4.5 surpasses widely used proprietary models such as GPT-4o-latest, and significantly larger open-source models such as Qwen2.5-VL 72B. Notably, the strong performance is achieved with remarkable efficiency. For example, on the widely adopted VideoMME benchmark, MiniCPM-V 4.5 achieves state-of-the-art performance among models under 30B size, using just 46.7\% GPU memory cost and 8.7\% inference time of Qwen2.5-VL 7B.
Authors: Rajitha de Silva, Jonathan Cox, James R. Heselden, Marija Popovic, Cesar Cadena, Riccardo Polvara
Abstract: Accurate localisation is critical for mobile robots in structured outdoor environments, yet LiDAR-based methods often fail in vineyards due to repetitive row geometry and perceptual aliasing. We propose a semantic particle filter that incorporates stable object-level detections, specifically vine trunks and support poles into the likelihood estimation process. Detected landmarks are projected into a birds eye view and fused with LiDAR scans to generate semantic observations. A key innovation is the use of semantic walls, which connect adjacent landmarks into pseudo-structural constraints that mitigate row aliasing. To maintain global consistency in headland regions where semantics are sparse, we introduce a noisy GPS prior that adaptively supports the filter. Experiments in a real vineyard demonstrate that our approach maintains localisation within the correct row, recovers from deviations where AMCL fails, and outperforms vision-based SLAM methods such as RTAB-Map.
Authors: Muheng Li, Evangelia Choulilitsa, Lisa Fankhauser, Francesca Albertini, Antony Lomax, Ye Zhang
Abstract: Accurate dose calculation on cone beam computed tomography (CBCT) images is essential for modern proton treatment planning workflows, particularly when accounting for inter-fractional anatomical changes in adaptive treatment scenarios. Traditional CBCT-based dose calculation suffers from image quality limitations, requiring complex correction workflows. This study develops and validates a deep learning approach for direct proton dose calculation from CBCT images using extended Long Short-Term Memory (xLSTM) neural networks. A retrospective dataset of 40 head-and-neck cancer patients with paired planning CT and treatment CBCT images was used to train an xLSTM-based neural network (CBCT-NN). The architecture incorporates energy token encoding and beam's-eye-view sequence modelling to capture spatial dependencies in proton dose deposition patterns. Training utilized 82,500 paired beam configurations with Monte Carlo-generated ground truth doses. Validation was performed on 5 independent patients using gamma analysis, mean percentage dose error assessment, and dose-volume histogram comparison. The CBCT-NN achieved gamma pass rates of 95.1 $\pm$ 2.7% using 2mm/2% criteria. Mean percentage dose errors were 2.6 $\pm$ 1.4% in high-dose regions ($>$90% of max dose) and 5.9 $\pm$ 1.9% globally. Dose-volume histogram analysis showed excellent preservation of target coverage metrics (Clinical Target Volume V95% difference: -0.6 $\pm$ 1.1%) and organ-at-risk constraints (parotid mean dose difference: -0.5 $\pm$ 1.5%). Computation time is under 3 minutes without sacrificing Monte Carlo-level accuracy. This study demonstrates the proof-of-principle of direct CBCT-based proton dose calculation using xLSTM neural networks. The approach eliminates traditional correction workflows while achieving comparable accuracy and computational efficiency suitable for adaptive protocols.
Authors: Gaurav Seth, Hoa Pham, Giles Hamilton-Fletcher, Charles Leclercq, John-Ross Rizzo
Abstract: Visual assistive technologies, such as Microsoft Seeing AI, can improve access to environmental information for persons with blindness or low vision (pBLV). Yet, the physical and functional implications of different device embodiments remain unclear. In this study, 11 pBLV participants used Seeing AI on a hand-held smartphone and on a head-mounted ARx Vision system to perform six activities of daily living, while their movements were captured with Xsens motion capture. Functional outcomes included task time, success rate, and number of attempts, and biomechanical measures included joint range of motion, angular path length, working volume, and movement smoothness. The head-mounted system generally reduced upper-body movement and task time, especially for document-scanning style tasks, whereas the hand-held system yielded higher success rates for tasks involving small or curved text. These findings indicate that both embodiments are viable, but they differ in terms of physical demands and ease of use. Incorporating biomechanical measures into assistive technology evaluations can inform designs that optimise user experience by balancing functional efficiency, physical sustainability, and intuitive interaction.
Authors: Bahey Tharwat, Yara Nasser, Ali Abouzeid, Ian Reid
Abstract: Vision-Language-Action (VLA) models have gained popularity for learning robotic manipulation tasks that follow language instructions. State-of-the-art VLAs, such as OpenVLA and $\pi_{0}$, were trained on large-scale, manually labeled action datasets collected through teleoperation. More recent approaches, including LAPA and villa-X, introduce latent action representations that enable unsupervised pretraining on unlabeled datasets by modeling abstract visual changes between frames. Although these methods have shown strong results, their large model sizes make deployment in real-world settings challenging. In this work, we propose LAWM, a model-agnostic framework to pretrain imitation learning models in a self-supervised way, by learning latent action representations from unlabeled video data through world modeling. These videos can be sourced from robot recordings or videos of humans performing actions with everyday objects. Our framework is designed to be effective for transferring across tasks, environments, and embodiments. It outperforms models trained with ground-truth robotics actions and similar pretraining methods on the LIBERO benchmark and real-world setup, while being significantly more efficient and practical for real-world settings.
Authors: Ayan Sar, Sampurna Roy, Tanupriya Choudhury, Ajith Abraham
Abstract: Generative adversarial networks (GANs) and diffusion models have dramatically advanced deepfake technology, and its threats to digital security, media integrity, and public trust have increased rapidly. This research explored zero-shot deepfake detection, an emerging method even when the models have never seen a particular deepfake variation. In this work, we studied self-supervised learning, transformer-based zero-shot classifier, generative model fingerprinting, and meta-learning techniques that better adapt to the ever-evolving deepfake threat. In addition, we suggested AI-driven prevention strategies that mitigated the underlying generation pipeline of the deepfakes before they occurred. They consisted of adversarial perturbations for creating deepfake generators, digital watermarking for content authenticity verification, real-time AI monitoring for content creation pipelines, and blockchain-based content verification frameworks. Despite these advancements, zero-shot detection and prevention faced critical challenges such as adversarial attacks, scalability constraints, ethical dilemmas, and the absence of standardized evaluation benchmarks. These limitations were addressed by discussing future research directions on explainable AI for deepfake detection, multimodal fusion based on image, audio, and text analysis, quantum AI for enhanced security, and federated learning for privacy-preserving deepfake detection. This further highlighted the need for an integrated defense framework for digital authenticity that utilized zero-shot learning in combination with preventive deepfake mechanisms. Finally, we highlighted the important role of interdisciplinary collaboration between AI researchers, cybersecurity experts, and policymakers to create resilient defenses against the rising tide of deepfake attacks.
Authors: Louis Rossignol, Tangui Aladjidi, Myrann Baker-Rasooli, Quentin Glorieux
Abstract: The nonlinear Schr\"odinger equation (NLSE) is a fundamental model for wave dynamics in nonlinear media ranging from optical fibers to Bose-Einstein condensates. Accurately estimating its parameters, which are often strongly correlated, from a single measurement remains a significant challenge. We address this problem by treating parameter estimation as an inverse problem and training a neural network to invert the NLSE mapping. We combine a fast numerical solver with a machine learning approach based on the ConvNeXt architecture and a multivariate Gaussian negative log-likelihood loss function. From single-shot field (density and phase) images, our model estimates three key parameters: the nonlinear coefficient $n_2$, the saturation intensity $I_{sat}$, and the linear absorption coefficient $\alpha$. Trained on 100,000 simulated images, the model achieves a mean absolute error of $3.22\%$ on 12,500 unseen test samples, demonstrating strong generalization and close agreement with ground-truth values. This approach provides an efficient route for characterizing nonlinear systems and has the potential to bridge theoretical modeling and experimental data when realistic noise is incorporated.
Authors: Kaiwen Jiang, Jia-Mu Sun, Zilu Li, Dan Wang, Tzu-Mao Li, Ravi Ramamoorthi
Abstract: Radiance fields have gained tremendous success with applications ranging from novel view synthesis to geometry reconstruction, especially with the advent of Gaussian splatting. However, they sacrifice modeling of material reflective properties and lighting conditions, leading to significant geometric ambiguities and the inability to easily perform relighting. One way to address these limitations is to incorporate physically-based rendering, but it has been prohibitively expensive to include full global illumination within the inner loop of the optimization. Therefore, previous works adopt simplifications that make the whole optimization with global illumination effects efficient but less accurate. In this work, we adopt Gaussian surfels as the primitives and build an efficient framework for differentiable light transport, inspired from the classic radiosity theory. The whole framework operates in the coefficient space of spherical harmonics, enabling both diffuse and specular materials. We extend the classic radiosity into non-binary visibility and semi-opaque primitives, propose novel solvers to efficiently solve the light transport, and derive the backward pass for gradient optimizations, which is more efficient than auto-differentiation. During inference, we achieve view-independent rendering where light transport need not be recomputed under viewpoint changes, enabling hundreds of FPS for global illumination effects, including view-dependent reflections using a spherical harmonics representation. Through extensive qualitative and quantitative experiments, we demonstrate superior geometry reconstruction, view synthesis and relighting than previous inverse rendering baselines, or data-driven baselines given relatively sparse datasets with known or unknown lighting conditions.
Authors: Mohammad Hosseini, Maryam M. Shanechi
Abstract: High-dimensional imaging of neural activity, such as widefield calcium and functional ultrasound imaging, provide a rich source of information for understanding the relationship between brain activity and behavior. Accurately modeling neural dynamics in these modalities is crucial for understanding this relationship but is hindered by the high-dimensionality, complex spatiotemporal dependencies, and prevalent behaviorally irrelevant dynamics in these modalities. Existing dynamical models often employ preprocessing steps to obtain low-dimensional representations from neural image modalities. However, this process can discard behaviorally relevant information and miss spatiotemporal structure. We propose SBIND, a novel data-driven deep learning framework to model spatiotemporal dependencies in neural images and disentangle their behaviorally relevant dynamics from other neural dynamics. We validate SBIND on widefield imaging datasets, and show its extension to functional ultrasound imaging, a recent modality whose dynamical modeling has largely remained unexplored. We find that our model effectively identifies both local and long-range spatial dependencies across the brain while also dissociating behaviorally relevant neural dynamics. Doing so, SBIND outperforms existing models in neural-behavioral prediction. Overall, SBIND provides a versatile tool for investigating the neural mechanisms underlying behavior using imaging modalities.
Authors: Richa Rawat, Faisal Ahmed
Abstract: Cancer is one of the leading health challenges for women, specifically breast and ovarian cancer. Early detection can help improve the survival rate through timely intervention and treatment. Traditional methods of detecting cancer involve manually examining mammograms, CT scans, ultrasounds, and other imaging types. However, this makes the process labor-intensive and requires the expertise of trained pathologists. Hence, making it both time-consuming and resource-intensive. In this paper, we introduce a novel vision transformer (ViT)-based method for detecting and classifying breast and ovarian cancer. We use a pre-trained ViT-Base-Patch16-224 model, which is fine-tuned for both binary and multi-class classification tasks using publicly available histopathological image datasets. Further, we use a preprocessing pipeline that converts raw histophological images into standardized PyTorch tensors, which are compatible with the ViT architecture and also help improve the model performance. We evaluated the performance of our model on two benchmark datasets: the BreakHis dataset for binary classification and the UBC-OCEAN dataset for five-class classification without any data augmentation. Our model surpasses existing CNN, ViT, and topological data analysis-based approaches in binary classification. For multi-class classification, it is evaluated against recent topological methods and demonstrates superior performance. Our study highlights the effectiveness of Vision Transformer-based transfer learning combined with efficient preprocessing in oncological diagnostics.
Authors: Neel P. Bhatt, Yunhao Yang, Rohan Siva, Pranay Samineni, Daniel Milan, Zhangyang Wang, Ufuk Topcu
Abstract: Rapid adaptation in unseen environments is essential for scalable real-world autonomy, yet existing approaches rely on exhaustive exploration or rigid navigation policies that fail to generalize. We present VLN-Zero, a two-phase vision-language navigation framework that leverages vision-language models to efficiently construct symbolic scene graphs and enable zero-shot neurosymbolic navigation. In the exploration phase, structured prompts guide VLM-based search toward informative and diverse trajectories, yielding compact scene graph representations. In the deployment phase, a neurosymbolic planner reasons over the scene graph and environmental observations to generate executable plans, while a cache-enabled execution module accelerates adaptation by reusing previously computed task-location trajectories. By combining rapid exploration, symbolic reasoning, and cache-enabled execution, the proposed framework overcomes the computational inefficiency and poor generalization of prior vision-language navigation methods, enabling robust and scalable decision-making in unseen environments. VLN-Zero achieves 2x higher success rate compared to state-of-the-art zero-shot models, outperforms most fine-tuned baselines, and reaches goal locations in half the time with 55% fewer VLM calls on average compared to state-of-the-art models across diverse environments. Codebase, datasets, and videos for VLN-Zero are available at: https://vln-zero.github.io/.
Authors: Maryam Viqar, Erdem Sahin, Elena Stoykova, Violeta Madjarova
Abstract: Conventional Fourier-domain Optical Coherence Tomography (FD-OCT) systems depend on resampling into wavenumber (k) domain to extract the depth profile. This either necessitates additional hardware resources or amplifies the existing computational complexity. Moreover, the OCT images also suffer from speckle noise, due to systemic reliance on low coherence interferometry. We propose a streamlined and computationally efficient approach based on Deep-Learning (DL) which enables reconstructing speckle-reduced OCT images directly from the wavelength domain. For reconstruction, two encoder-decoder styled networks namely Spatial Domain Convolution Neural Network (SD-CNN) and Fourier Domain CNN (FD-CNN) are used sequentially. The SD-CNN exploits the highly degraded images obtained by Fourier transforming the domain fringes to reconstruct the deteriorated morphological structures along with suppression of unwanted noise. The FD-CNN leverages this output to enhance the image quality further by optimization in Fourier domain (FD). We quantitatively and visually demonstrate the efficacy of the method in obtaining high-quality OCT images. Furthermore, we illustrate the computational complexity reduction by harnessing the power of DL models. We believe that this work lays the framework for further innovations in the realm of OCT image reconstruction.
Authors: Johannes A. Gaus, Loris Schneider, Yitian Shi, Jongseok Lee, Rania Rayyes, Rudolph Triebel
Abstract: In this paper, we address the point cloud registration problem, where well-known methods like ICP fail under uncertainty arising from sensor noise, pose-estimation errors, and partial overlap due to occlusion. We develop a novel approach, Gaussian Process Concept Attribution (GP-CA), which not only quantifies registration uncertainty but also explains it by attributing uncertainty to well-known sources of errors in registration problems. Our approach leverages active learning to discover new uncertainty sources in the wild by querying informative instances. We validate GP-CA on three publicly available datasets and in our real-world robot experiment. Extensive ablations substantiate our design choices. Our approach outperforms other state-of-the-art methods in terms of runtime, high sample-efficiency with active learning, and high accuracy. Our real-world experiment clearly demonstrates its applicability. Our video also demonstrates that GP-CA enables effective failure-recovery behaviors, yielding more robust robotic perception.
Authors: Suzannah Wistreich, Baiyu Shi, Stephen Tian, Samuel Clarke, Michael Nath, Chengyi Xu, Zhenan Bao, Jiajun Wu
Abstract: Human skin provides a rich tactile sensing stream, localizing intentional and unintentional contact events over a large and contoured region. Replicating these tactile sensing capabilities for dexterous robotic manipulation systems remains a longstanding challenge. In this work, we take a step towards this goal by introducing DexSkin. DexSkin is a soft, conformable capacitive electronic skin that enables sensitive, localized, and calibratable tactile sensing, and can be tailored to varying geometries. We demonstrate its efficacy for learning downstream robotic manipulation by sensorizing a pair of parallel jaw gripper fingers, providing tactile coverage across almost the entire finger surfaces. We empirically evaluate DexSkin's capabilities in learning challenging manipulation tasks that require sensing coverage across the entire surface of the fingers, such as reorienting objects in hand and wrapping elastic bands around boxes, in a learning-from-demonstration framework. We then show that, critically for data-driven approaches, DexSkin can be calibrated to enable model transfer across sensor instances, and demonstrate its applicability to online reinforcement learning on real robots. Our results highlight DexSkin's suitability and practicality for learning real-world, contact-rich manipulation. Please see our project webpage for videos and visualizations: https://dex-skin.github.io/.
Authors: Pin-Yen Chiu, I-Sheng Fang, Jun-Cheng Chen
Abstract: Recent advances in diffusion models have significantly improved image and video synthesis. In addition, several concept control methods have been proposed to enable fine-grained, continuous, and flexible control over free-form text prompts. However, these methods not only require intensive training time and GPU memory usage to learn the sliders or embeddings but also need to be retrained for different diffusion backbones, limiting their scalability and adaptability. To address these limitations, we introduce Text Slider, a lightweight, efficient and plug-and-play framework that identifies low-rank directions within a pre-trained text encoder, enabling continuous control of visual concepts while significantly reducing training time, GPU memory consumption, and the number of trainable parameters. Furthermore, Text Slider supports multi-concept composition and continuous control, enabling fine-grained and flexible manipulation in both image and video synthesis. We show that Text Slider enables smooth and continuous modulation of specific attributes while preserving the original spatial layout and structure of the input. Text Slider achieves significantly better efficiency: 5$\times$ faster training than Concept Slider and 47$\times$ faster than Attribute Control, while reducing GPU memory usage by nearly 2$\times$ and 4$\times$, respectively.
Authors: Hillol Biswas
Abstract: An integer winding, i.e., topological charge, is a characteristic of skyrmions, which are topologically nontrivial spin patterns in magnets. They emerge when smooth two-dimensional spin configurations are stabilized by conflicting interactions such as exchange, anisotropy, the Dzyaloshinskii-Moriya interaction, or geometric frustration. These nanoscale textures, which are typically a few to tens of nanometers in size, are strong 'particle-like' excitations because they are shielded by energy barriers connected to their topology. By exploiting their helicity, i.e., spin rotation angle or associated internal modes, as a two-level system, skyrmions can function as quantum bits or qubits. Two quantized helicity states of a nanometer-scale skyrmion encode the logical value states in a 'skyrmion qubit.' Interestingly, skyrmion qubits are topologically protected and macroscopic, i.e., they involve a large number of spins; however, external influences can still affect them. When the texture is tiny and disconnected, the helicity angle of the skyrmion becomes quantized. A qubit basis is made up of the lowest two energy eigenstates, i.e., symmetric or antisymmetric superpositions of opposite helicity, for example. Therefore, Skyrmion textures can provide valuable insights for different purposes. However, is it possible to synthetically generate skyrmion textures using quantum computing? This paper investigates the possibility and generates a few hundred different textures, producing sample comparisons from various types, which indicate a novel direction for skyrmion-based research based on quantum randomness and other criteria.
Authors: Jun Ma, Qian He, Gaofeng He, Huang Chen, Chen Liu, Xiaogang Jin, Huamin Wang
Abstract: Diffusion models have significantly advanced image manipulation techniques, and their ability to generate photorealistic images is beginning to transform retail workflows, particularly in presale visualization. Beyond artistic style transfer, the capability to perform fine-grained visual feature transfer is becoming increasingly important. Embroidery is a textile art form characterized by intricate interplay of diverse stitch patterns and material properties, which poses unique challenges for existing style transfer methods. To explore the customization for such fine-grained features, we propose a novel contrastive learning framework that disentangles fine-grained style and content features with a single reference image, building on the classic concept of image analogy. We first construct an image pair to define the target style, and then adopt a similarity metric based on the decoupled representations of pretrained diffusion models for style-content separation. Subsequently, we propose a two-stage contrastive LoRA modulation technique to capture fine-grained style features. In the first stage, we iteratively update the whole LoRA and the selected style blocks to initially separate style from content. In the second stage, we design a contrastive learning strategy to further decouple style and content through self-knowledge distillation. Finally, we build an inference pipeline to handle image or text inputs with only the style blocks. To evaluate our method on fine-grained style transfer, we build a benchmark for embroidery customization. Our approach surpasses prior methods on this task and further demonstrates strong generalization to three additional domains: artistic style transfer, sketch colorization, and appearance transfer.
Authors: Minoo Dolatabadi, Fardin Ayar, Ehsan Javanmardi, Manabu Tsukada, Mahdi Javanmardi
Abstract: LiDAR-based localization and SLAM often rely on iterative matching algorithms, particularly the Iterative Closest Point (ICP) algorithm, to align sensor data with pre-existing maps or previous scans. However, ICP is prone to errors in featureless environments and dynamic scenes, leading to inaccurate pose estimation. Accurately predicting the uncertainty associated with ICP is crucial for robust state estimation but remains challenging, as existing approaches often rely on handcrafted models or simplified assumptions. Moreover, a few deep learning-based methods for localizability estimation either depend on a pre-built map, which may not always be available, or provide a binary classification of localizable versus non-localizable, which fails to properly model uncertainty. In this work, we propose a data-driven framework that leverages deep learning to estimate the registration error covariance of ICP before matching, even in the absence of a reference map. By associating each LiDAR scan with a reliable 6-DoF error covariance estimate, our method enables seamless integration of ICP within Kalman filtering, enhancing localization accuracy and robustness. Extensive experiments on the KITTI dataset demonstrate the effectiveness of our approach, showing that it accurately predicts covariance and, when applied to localization using a pre-built map or SLAM, reduces localization errors and improves robustness.
Authors: Lorenzo Shaikewitz, Tim Nguyen, Luca Carlone
Abstract: Object shape and pose estimation is a foundational robotics problem, supporting tasks from manipulation to scene understanding and navigation. We present a fast local solver for shape and pose estimation which requires only category-level object priors and admits an efficient certificate of global optimality. Given an RGB-D image of an object, we use a learned front-end to detect sparse, category-level semantic keypoints on the target object. We represent the target object's unknown shape using a linear active shape model and pose a maximum a posteriori optimization problem to solve for position, orientation, and shape simultaneously. Expressed in unit quaternions, this problem admits first-order optimality conditions in the form of an eigenvalue problem with eigenvector nonlinearities. Our primary contribution is to solve this problem efficiently with self-consistent field iteration, which only requires computing a 4-by-4 matrix and finding its minimum eigenvalue-vector pair at each iterate. Solving a linear system for the corresponding Lagrange multipliers gives a simple global optimality certificate. One iteration of our solver runs in about 100 microseconds, enabling fast outlier rejection. We test our method on synthetic data and a variety of real-world settings, including two public datasets and a drone tracking scenario. Code is released at https://github.com/MIT-SPARK/Fast-ShapeAndPose.
Authors: Yang Li, Chenyu Wang, Tingrui Wang, Yongwei Wang, Haonan Li, Zhunga Liu, Quan Pan
Abstract: Black-box adversarial attacks remain challenging due to limited access to model internals. Existing methods often depend on specific network architectures or require numerous queries, resulting in limited cross-architecture transferability and high query costs. To address these limitations, we propose JAD, a latent diffusion model framework for black-box adversarial attacks. JAD generates adversarial examples by leveraging a latent diffusion model guided by attention maps distilled from both a convolutional neural network (CNN) and a Vision Transformer (ViT) models. By focusing on image regions that are commonly sensitive across architectures, this approach crafts adversarial perturbations that transfer effectively between different model types. This joint attention distillation strategy enables JAD to be architecture-agnostic, achieving superior attack generalization across diverse models. Moreover, the generative nature of the diffusion framework yields high adversarial sample generation efficiency by reducing reliance on iterative queries. Experiments demonstrate that JAD offers improved attack generalization, generation efficiency, and cross-architecture transferability compared to existing methods, providing a promising and effective paradigm for black-box adversarial attacks.
Authors: Hongli Xu, Lei Zhang, Xiaoyue Hu, Boyang Zhong, Kaixin Bai, Zolt\'an-Csaba M\'arton, Zhenshan Bing, Zhaopeng Chen, Alois Christian Knoll, Jianwei Zhang
Abstract: General-purpose robotic skills from end-to-end demonstrations often leads to task-specific policies that fail to generalize beyond the training distribution. Therefore, we introduce FunCanon, a framework that converts long-horizon manipulation tasks into sequences of action chunks, each defined by an actor, verb, and object. These chunks focus policy learning on the actions themselves, rather than isolated tasks, enabling compositionality and reuse. To make policies pose-aware and category-general, we perform functional object canonicalization for functional alignment and automatic manipulation trajectory transfer, mapping objects into shared functional frames using affordance cues from large vision language models. An object centric and action centric diffusion policy FuncDiffuser trained on this aligned data naturally respects object affordances and poses, simplifying learning and improving generalization ability. Experiments on simulated and real-world benchmarks demonstrate category-level generalization, cross-task behavior reuse, and robust sim2real deployment, showing that functional canonicalization provides a strong inductive bias for scalable imitation learning in complex manipulation domains. Details of the demo and supplemental material are available on our project website https://sites.google.com/view/funcanon.
Authors: Georgii Kolokolnikov, Marie-Lena Schmalhofer, Sophie G\"otz, Lennart Well, Said Farschtschi, Victor-Felix Mautner, Inka Ristow, Rene Werner
Abstract: Background and Objectives: Neurofibromatosis type 1 is a genetic disorder characterized by the development of numerous neurofibromas (NFs) throughout the body. Whole-body MRI (WB-MRI) is the clinical standard for detection and longitudinal surveillance of NF tumor growth. Existing interactive segmentation methods fail to combine high lesion-wise precision with scalability to hundreds of lesions. This study proposes a novel interactive segmentation model tailored to this challenge. Methods: We introduce MOIS-SAM2, a multi-object interactive segmentation model that extends the state-of-the-art, transformer-based, promptable Segment Anything Model 2 (SAM2) with exemplar-based semantic propagation. MOIS-SAM2 was trained and evaluated on 119 WB-MRI scans from 84 NF1 patients acquired using T2-weighted fat-suppressed sequences. The dataset was split at the patient level into a training set and four test sets (one in-domain and three reflecting different domain shift scenarios, e.g., MRI field strength variation, low tumor burden, differences in clinical site and scanner vendor). Results: On the in-domain test set, MOIS-SAM2 achieved a scan-wise DSC of 0.60 against expert manual annotations, outperforming baseline 3D nnU-Net (DSC: 0.54) and SAM2 (DSC: 0.35). Performance of the proposed model was maintained under MRI field strength shift (DSC: 0.53) and scanner vendor variation (DSC: 0.50), and improved in low tumor burden cases (DSC: 0.61). Lesion detection F1 scores ranged from 0.62 to 0.78 across test sets. Preliminary inter-reader variability analysis showed model-to-expert agreement (DSC: 0.62-0.68), comparable to inter-expert agreement (DSC: 0.57-0.69). Conclusions: The proposed MOIS-SAM2 enables efficient and scalable interactive segmentation of NFs in WB-MRI with minimal user input and strong generalization, supporting integration into clinical workflows.
Authors: R\'emi Giraud, Vinh-Thong Ta, Nicolas Papadakis
Abstract: In the superpixel literature, the comparison of state-of-the-art methods can be biased by the non-robustness of some metrics to decomposition aspects, such as the superpixel scale. Moreover, most recent decomposition methods allow to set a shape regularity parameter, which can have a substantial impact on the measured performances. In this paper, we introduce an evaluation framework, that aims to unify the comparison process of superpixel methods. We investigate the limitations of existing metrics, and propose to evaluate each of the three core decomposition aspects: color homogeneity, respect of image objects and shape regularity. To measure the regularity aspect, we propose a new global regularity measure (GR), which addresses the non-robustness of state-of-the-art metrics. We evaluate recent superpixel methods with these criteria, at several superpixel scales and regularity levels. The proposed framework reduces the bias in the comparison process of state-of-the-art superpixel methods. Finally, we demonstrate that the proposed GR measure is correlated with the performances of various applications.
Authors: Zanyar Zohourianshahzadi, Terrance E. Boult, Jugal K. Kalita
Abstract: In the past few years the transformer model has been utilized for a variety of tasks such as image captioning, image classification natural language generation, and natural language understanding. As a key component of the transformer model, self-attention calculates the attention values by mapping the relationships among the head elements of the source and target sequence, yet there is no explicit mechanism to refine and intensify the attention values with respect to the context of the input and target sequences. Based on this intuition, we introduce a novel refine and intensify attention mechanism that is called Zoneup Dropout Injection Attention Calculation (ZoDIAC), in which the intensities of attention values in the elements of the input source and target sequences are first refined using GELU and dropout and then intensified using a proposed zoneup process which includes the injection of a learned scalar factor. Our extensive experiments show that ZoDIAC achieves statistically significant higher scores under all image captioning metrics using various feature extractors in comparison to the conventional self-attention module in the transformer model on the MS-COCO dataset. Our proposed ZoDIAC attention modules can be used as a drop-in replacement for the attention components in all transformer models. The code for our experiments is publicly available at: https://github.com/zanyarz/zodiac
Authors: Luigi Celona, Simone Bianco, Marco Donzella, Paolo Napoletano
Abstract: State-of-The-Art (SoTA) image captioning models are often trained on the MicroSoft Common Objects in Context (MS-COCO) dataset, which contains human-annotated captions with an average length of approximately ten tokens. Although effective for general scene understanding, these short captions often fail to capture complex scenes and convey detailed information. Moreover, captioning models tend to exhibit bias towards the ``average'' caption, which captures only the more general aspects, thus overlooking finer details. In this paper, we present a novel approach to generate richer and more informative image captions by combining the captions generated from different SoTA captioning models. Our proposed method requires no additional model training: given an image, it leverages pre-trained models from the literature to generate the initial captions, and then ranks them using a newly introduced image-text-based metric, which we name BLIPScore. Subsequently, the top two captions are fused using a Large Language Model (LLM) to produce the final, more detailed description. Experimental results on the MS-COCO and Flickr30k test sets demonstrate the effectiveness of our approach in terms of caption-image alignment and hallucination reduction according to the ALOHa, CAPTURE, and Polos metrics. A subjective study lends additional support to these results, suggesting that the captions produced by our model are generally perceived as more consistent with human judgment. By combining the strengths of diverse SoTA models, our method enhances the quality and appeal of image captions, bridging the gap between automated systems and the rich and informative nature of human-generated descriptions. This advance enables the generation of more suitable captions for the training of both vision-language and captioning models.
Authors: Julia Grabinski, Steffen Jung, Janis Keuper, Margret Keuper
Abstract: Convolutional Neural Networks (CNNs) are successful in various computer vision tasks. From an image and signal processing point of view, this success is counter-intuitive, as the inherent spatial pyramid design of most CNNs is apparently violating basic signal processing laws, i.e. the Sampling Theorem in their downsampling operations. This issue has been broadly neglected until recent work in the context of adversarial attacks and distribution shifts showed that there is a strong correlation between the vulnerability of CNNs and aliasing artifacts induced by bandlimit-violating downsampling. As a remedy, we propose an alias-free downsampling operation in the frequency domain, denoted Frequency Low Cut Pooling (FLC Pooling) which we further extend to Aliasing and Sinc Artifact-free Pooling (ASAP). ASAP is alias-free and removes further artifacts from sinc-interpolation. Our experimental evaluation on ImageNet-1k, ImageNet-C and CIFAR datasets on various CNN architectures demonstrates that networks using FLC Pooling and ASAP as downsampling methods learn more stable features as measured by their robustness against common corruptions and adversarial attacks, while maintaining a clean accuracy similar to the respective baseline models.
Authors: Christian Wachinger, Dennis Hedderich, Melissa Thalhammer, Fabian Bongratz
Abstract: Understanding individual differences in cortical structure is key to advancing diagnostics in neurology and psychiatry. Reference models aid in detecting aberrant cortical thickness, yet site-specific biases limit their direct application to unseen data, and region-wise averages prevent the detection of localized cortical changes. To address these limitations, we developed the Stochastic Cortical Self-Reconstruction (SCSR), a novel method that leverages deep learning to reconstruct cortical thickness maps at the vertex level without needing additional subject information. Trained on over 25,000 healthy individuals, SCSR generates highly individualized cortical reconstructions that can detect subtle thickness deviations. Our evaluations on independent test sets demonstrated that SCSR achieved significantly lower reconstruction errors and identified atrophy patterns that enabled better disease discrimination than established methods. It also hints at cortical thinning in preterm infants that went undetected by existing models, showcasing its versatility. Finally, SCSR excelled in mapping highly resolved cortical deviations of dementia patients from clinical data, highlighting its potential for supporting diagnosis in clinical practice.
Authors: Joseph Cho, Mrudang Mathur, Cyril Zakka, Dhamanpreet Kaur, Matthew Leipzig, Alex Dalal, Aravind Krishnan, Eubee Koo, Karen Wai, Cindy S. Zhao, Akshay Chaudhari, Matthew Duda, Ashley Choi, Ehsan Rahimy, Lyna Azzouz, Robyn Fong, Rohan Shad, William Hiesinger
Abstract: Deep learning algorithms require extensive data to achieve robust performance. However, data availability is often restricted in the medical domain due to patient privacy concerns. Synthetic data presents a possible solution to these challenges. Recently, image generative models have found increasing use for medical applications but are often designed for singular medical specialties and imaging modalities, thus limiting their broader utility. To address this, we introduce MediSyn: a text-guided, latent diffusion model capable of generating synthetic images from 6 medical specialties and 10 image types. Through extensive experimentation, we first demonstrate that MediSyn quantitatively matches or surpasses the performance of specialist models. Second, we show that our synthetic images are realistic and exhibit strong alignment with their corresponding text prompts, as validated by a team of expert physicians. Third, we provide empirical evidence that our synthetic images are visually distinct from their corresponding real patient images. Finally, we demonstrate that in data-limited settings, classifiers trained solely on synthetic data or real data supplemented with synthetic data can outperform those trained solely on real data. Our findings highlight the immense potential of generalist image generative models to accelerate algorithmic research and development in medicine.
Authors: Ma\"elic Neau, Paulo E. Santos, Anne-Gwenn Bosser, C\'edric Buche, Akihiro Sugimoto
Abstract: Scene Graph Generation (SGG) is a task that encodes visual relationships between objects in images as graph structures. SGG shows significant promise as a foundational component for downstream tasks, such as reasoning for embodied agents. To enable real-time applications, SGG must address the trade-off between performance and inference speed. However, current methods tend to focus on one of the following: (1) improving relation prediction accuracy, (2) enhancing object detection accuracy, or (3) reducing latency, without aiming to balance all three objectives simultaneously. To address this limitation, we propose the Real-time Efficiency and Accuracy Compromise for Tradeoffs in Scene Graph Generation (REACT) architecture, which achieves the highest inference speed among existing SGG models, improving object detection accuracy without sacrificing relation prediction performance. Compared to state-of-the-art approaches, REACT is 2.7 times faster and improves object detection accuracy by 58\%. Furthermore, our proposal significantly reduces model size, with an average of 5.5x fewer parameters. The code is available at https://github.com/Maelic/SGG-Benchmark
Authors: R\'emi Giraud, Micha\"el Cl\'ement
Abstract: Over the years, the use of superpixel segmentation has become very popular in various applications, serving as a preprocessing step to reduce data size by adapting to the content of the image, regardless of its semantic content. While the superpixel segmentation of standard planar images, captured with a 90{\deg} field of view, has been extensively studied, there has been limited focus on dedicated methods to omnidirectional or spherical images, captured with a 360{\deg} field of view. In this study, we introduce the first deep learning-based superpixel segmentation approach tailored for omnidirectional images called DSS (for Deep Spherical Superpixels). Our methodology leverages on spherical CNN architectures and the differentiable K-means clustering paradigm for superpixels, to generate superpixels that follow the spherical geometry. Additionally, we propose to use data augmentation techniques specifically designed for 360{\deg} images, enabling our model to efficiently learn from a limited set of annotated omnidirectional data. Our extensive validation across two datasets demonstrates that taking into account the inherent circular geometry of such images into our framework improves the segmentation performance over traditional and deep learning-based superpixel methods. Our code is available online.
Authors: Qiushuo Cheng, Catherine Morgan, Arindam Sikdar, Alessandro Masullo, Alan Whone, Majid Mirmehdi
Abstract: People with Parkinson's Disease (PD) often experience progressively worsening gait, including changes in how they turn around, as the disease progresses. Existing clinical rating tools are not capable of capturing hour-by-hour variations of PD symptoms, as they are confined to brief assessments within clinic settings. Measuring gait turning angles continuously and passively is a component step towards using gait characteristics as sensitive indicators of disease progression in PD. This paper presents a deep learning-based approach to automatically quantify turning angles by extracting 3D skeletons from videos and calculating the rotation of hip and knee joints. We utilise state-of-the-art human pose estimation models, Fastpose and Strided Transformer, on a total of 1386 turning video clips from 24 subjects (12 people with PD and 12 healthy control volunteers), trimmed from a PD dataset of unscripted free-living videos in a home-like setting (Turn-REMAP). We also curate a turning video dataset, Turn-H3.6M, from the public Human3.6M human pose benchmark with 3D ground truth, to further validate our method. Previous gait research has primarily taken place in clinics or laboratories evaluating scripted gait outcomes, but this work focuses on free-living home settings where complexities exist, such as baggy clothing and poor lighting. Due to difficulties in obtaining accurate ground truth data in a free-living setting, we quantise the angle into the nearest bin $45^\circ$ based on the manual labelling of expert clinicians. Our method achieves a turning calculation accuracy of 41.6%, a Mean Absolute Error (MAE) of 34.7{\deg}, and a weighted precision WPrec of 68.3% for Turn-REMAP. This is the first work to explore the use of single monocular camera data to quantify turns by PD patients in a home setting.
Authors: Toon Van de Maele, Ozan Catal, Alexander Tschantz, Christopher L. Buckley, Tim Verbelen
Abstract: Recently, 3D Gaussian Splatting has emerged as a promising approach for modeling 3D scenes using mixtures of Gaussians. The predominant optimization method for these models relies on backpropagating gradients through a differentiable rendering pipeline, which struggles with catastrophic forgetting when dealing with continuous streams of data. To address this limitation, we propose Variational Bayes Gaussian Splatting (VBGS), a novel approach that frames training a Gaussian splat as variational inference over model parameters. By leveraging the conjugacy properties of multivariate Gaussians, we derive a closed-form variational update rule, allowing efficient updates from partial, sequential observations without the need for replay buffers. Our experiments show that VBGS not only matches state-of-the-art performance on static datasets, but also enables continual learning from sequentially streamed 2D and 3D data, drastically improving performance in this setting.
Authors: Ziyang Gong, Zhixiang Wei, Di Wang, Xiaoxing Hu, Xianzheng Ma, Hongruixuan Chen, Yuru Jia, Yupeng Deng, Zhenming Ji, Xiangwei Zhu, Xue Yang, Naoto Yokoya, Jing Zhang, Bo Du, Junchi Yan, Liangpei Zhang
Abstract: The field of Remote Sensing Domain Generalization (RSDG) has emerged as a critical and valuable research frontier, focusing on developing models that generalize effectively across diverse scenarios. Despite the substantial domain gaps in RS images that are characterized by variabilities such as location, wavelength, and sensor type, research in this area remains underexplored: (1) Current cross-domain methods primarily focus on Domain Adaptation (DA), which adapts models to predefined domains rather than to unseen ones; (2) Few studies targeting the RSDG issue, especially for semantic segmentation tasks, where existing models are developed for specific unknown domains, struggling with issues of underfitting on other unknown scenarios; (3) Existing RS foundation models tend to prioritize in-domain performance over cross-domain generalization. To this end, we introduce the first vision foundation model for RSDG semantic segmentation, CrossEarth. CrossEarth demonstrates strong cross-domain generalization through a specially designed data-level Earth-Style Injection pipeline and a model-level Multi-Task Training pipeline. In addition, for the semantic segmentation task, we have curated an RSDG benchmark comprising 32 cross-domain settings across various regions, spectral bands, platforms, and climates, providing a comprehensive framework for testing the generalizability of future RSDG models. Extensive experiments on this benchmark demonstrate the superiority of CrossEarth over existing state-of-the-art methods.
Authors: Jyh-Jing Hwang, Runsheng Xu, Hubert Lin, Wei-Chih Hung, Jingwei Ji, Kristy Choi, Di Huang, Tong He, Paul Covington, Benjamin Sapp, Yin Zhou, James Guo, Dragomir Anguelov, Mingxing Tan
Abstract: We introduce EMMA, an End-to-end Multimodal Model for Autonomous driving. Built upon a multi-modal large language model foundation like Gemini, EMMA directly maps raw camera sensor data into various driving-specific outputs, including planner trajectories, perception objects, and road graph elements. EMMA maximizes the utility of world knowledge from the pre-trained large language models, by representing all non-sensor inputs (e.g. navigation instructions and ego vehicle status) and outputs (e.g. trajectories and 3D locations) as natural language text. This approach allows EMMA to jointly process various driving tasks in a unified language space, and generate the outputs for each task using task-specific prompts. Empirically, we demonstrate EMMA's effectiveness by achieving state-of-the-art performance in motion planning on nuScenes as well as competitive results on the Waymo Open Motion Dataset (WOMD). EMMA also yields competitive results for camera-primary 3D object detection on the Waymo Open Dataset (WOD). We show that co-training EMMA with planner trajectories, object detection, and road graph tasks yields improvements across all three domains, highlighting EMMA's potential as a generalist model for autonomous driving applications. We hope that our results will inspire research to further evolve the state of the art in autonomous driving model architectures.
Authors: R\'emi Giraud, Micha\"el Cl\'ement
Abstract: For many years, image over-segmentation into superpixels has been essential to computer vision pipelines, by creating homogeneous and identifiable regions of similar sizes. Such constrained segmentation problem would require a clear definition and specific evaluation criteria. However, the validation framework for superpixel methods, typically viewed as standard object segmentation, has rarely been thoroughly studied. In this work, we first take a step back to show that superpixel segmentation is fundamentally an ill-posed problem, due to the implicit regularity constraint on the shape and size of superpixels. We also demonstrate through a novel comprehensive study that the literature suffers from only evaluating certain aspects, sometimes incorrectly and with inappropriate metrics. Concurrently, recent deep learning-based superpixel methods mainly focus on the object segmentation task at the expense of regularity. In this ill-posed context, we show that we can achieve competitive results using a recent architecture like the Segment Anything Model (SAM), without dedicated training for the superpixel segmentation task. This leads to rethinking superpixel segmentation and the necessary properties depending on the targeted downstream task.
Authors: Shuning Chang, Pichao Wang, Jiasheng Tang, Fan Wang, Yi Yang
Abstract: Diffusion Transformers (DiT) are renowned for their impressive generative performance; however, they are significantly constrained by considerable computational costs due to the quadratic complexity in self-attention and the extensive sampling steps required. While advancements have been made in expediting the sampling process, the underlying architectural inefficiencies within DiT remain underexplored. We introduce SparseDiT, a novel framework that implements token sparsification across spatial and temporal dimensions to enhance computational efficiency while preserving generative quality. Spatially, SparseDiT employs a tri-segment architecture that allocates token density based on feature requirements at each layer: Poolingformer in the bottom layers for efficient global feature extraction, Sparse-Dense Token Modules (SDTM) in the middle layers to balance global context with local detail, and dense tokens in the top layers to refine high-frequency details. Temporally, SparseDiT dynamically modulates token density across denoising stages, progressively increasing token count as finer details emerge in later timesteps. This synergy between SparseDiT spatially adaptive architecture and its temporal pruning strategy enables a unified framework that balances efficiency and fidelity throughout the generation process. Our experiments demonstrate SparseDiT effectiveness, achieving a 55% reduction in FLOPs and a 175% improvement in inference speed on DiT-XL with similar FID score on 512x512 ImageNet, a 56% reduction in FLOPs across video generation datasets, and a 69% improvement in inference speed on PixArt-$\alpha$ on text-to-image generation task with a 0.24 FID score decrease. SparseDiT provides a scalable solution for high-quality diffusion-based generation compatible with sampling optimization techniques.
Authors: Jihao Gu, Yingyao Wang, Meng Cao, Pi Bu, Jun Song, Yancheng He, Shilong Li, Bo Zheng
Abstract: Direct Preference Optimization (DPO) has been demonstrated to be highly effective in mitigating hallucinations in Large Vision Language Models (LVLMs) by aligning their outputs more closely with human preferences. Despite the recent progress, existing methods suffer from two drawbacks: 1) Lack of scalable token-level rewards; and 2) Neglect of visual-anchored tokens. To this end, we propose a novel Token Preference Optimization model with self-calibrated rewards (dubbed as TPO), which adaptively attends to visual-correlated tokens without fine-grained annotations. Specifically, we introduce a token-level \emph{visual-anchored} \emph{reward} as the difference of the logistic distributions of generated tokens conditioned on the raw image and the corrupted one. In addition, to highlight the informative visual-anchored tokens, a visual-aware training objective is proposed to enhance more accurate token-level optimization. Extensive experimental results have manifested the state-of-the-art performance of the proposed TPO. For example, by building on top of LLAVA-1.5-7B, our TPO boosts the performance absolute improvement for hallucination benchmarks.
Authors: Wen-Dong Jiang, Chih-Yung Chang, Hsiang-Chuan Chang, Ji-Yuan Chen, Diptendu Sinha Roy
Abstract: Weakly Supervised Monitoring Anomaly Detection (WSMAD) utilizes weak supervision learning to identify anomalies, a critical task for smart city monitoring. However, existing multimodal approaches often fail to meet the real-time and interpretability requirements of edge devices due to their complexity. This paper presents TCVADS (Two-stage Cross-modal Video Anomaly Detection System), which leverages knowledge distillation and cross-modal contrastive learning to enable efficient, accurate, and interpretable anomaly detection on edge devices.TCVADS operates in two stages: coarse-grained rapid classification and fine-grained detailed analysis. In the first stage, TCVADS extracts features from video frames and inputs them into a time series analysis module, which acts as the teacher model. Insights are then transferred via knowledge distillation to a simplified convolutional network (student model) for binary classification. Upon detecting an anomaly, the second stage is triggered, employing a fine-grained multi-class classification model. This stage uses CLIP for cross-modal contrastive learning with text and images, enhancing interpretability and achieving refined classification through specially designed triplet textual relationships. Experimental results demonstrate that TCVADS significantly outperforms existing methods in model performance, detection efficiency, and interpretability, offering valuable contributions to smart city monitoring applications.
Authors: Dong Shu, Haiyan Zhao, Jingyu Hu, Weiru Liu, Ali Payani, Lu Cheng, Mengnan Du
Abstract: Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities in processing both visual and textual information. However, the critical challenge of alignment between visual and textual representations is not fully understood. This survey presents a comprehensive examination of alignment and misalignment in LVLMs through an explainability lens. We first examine the fundamentals of alignment, exploring its representational and behavioral aspects, training methodologies, and theoretical foundations. We then analyze misalignment phenomena across three semantic levels: object, attribute, and relational misalignment. Our investigation reveals that misalignment emerges from challenges at multiple levels: the data level, the model level, and the inference level. We provide a comprehensive review of existing mitigation strategies, categorizing them into parameter-frozen and parameter-tuning approaches. Finally, we outline promising future research directions, emphasizing the need for standardized evaluation protocols and in-depth explainability studies.
Authors: Pengteng Li, Yunfan Lu, Pinghao Song, Wuyang Li, Huizai Yao, Hui Xiong
Abstract: The event-based Vision-Language Model (VLM) recently has made good progress for practical vision tasks. However, most of these works just utilize CLIP for focusing on traditional perception tasks, which obstruct model understanding explicitly the sufficient semantics and context from event streams. To address the deficiency, we propose EventVL, the first generative event-based MLLM (Multimodal Large Language Model) framework for explicit semantic understanding. Specifically, to bridge the data gap for connecting different modalities semantics, we first annotate a large event-image/video-text dataset, containing almost 1.4 million high-quality pairs of data, which enables effective learning across various scenes, e.g., drive scene or human motion. After that, we design Event Spatiotemporal Representation to fully explore the comprehensive information by diversely aggregating and segmenting the event stream. To further promote a compact semantic space, Dynamic Semantic Alignment is introduced to improve and complete sparse semantic spaces of events. Extensive experiments show that our EventVL can significantly surpass existing MLLM baselines in event captioning and scene description generation tasks. We hope our research could contribute to the development of the event vision community.
Authors: Zhongwei Chen, Zhao-Xu Yang, Hai-Jun Rong, Guoqi Li
Abstract: Drone-view Geo-Localization (DVGL) aims to achieve accurate localization of drones by retrieving the most relevant GPS-tagged satellite images. However, most existing methods heavily rely on strictly pre-paired drone-satellite images for supervised learning. When the target region shifts, new paired samples are typically required to adapt to the distribution changes. The high cost of annotation and the limited transferability of these methods significantly hinder the practical deployment of DVGL in open-world scenarios. To address these limitations, we propose a novel end-to-end self-supervised learning method with a shallow backbone network, called the dynamic memory-driven and neighborhood information learning (DMNIL) method. It employs a clustering algorithm to generate pseudo-labels and adopts a dual-path contrastive learning framework to learn discriminative intra-view representations. Furthermore, DMNIL incorporates two core modules, including the dynamic hierarchical memory learning (DHML) module and the information consistency evolution learning (ICEL) module. The DHML module combines short-term and long-term memory to enhance intra-view feature consistency and discriminability. Meanwhile, the ICEL module utilizes a neighborhood-driven dynamic constraint mechanism to systematically capture implicit cross-view semantic correlations, consequently improving cross-view feature alignment. To further stabilize and strengthen the self-supervised training process, a pseudo-label enhancement strategy is introduced to enhance the quality of pseudo supervision. Extensive experiments on three public benchmark datasets demonstrate that the proposed method consistently outperforms existing self-supervised methods and even surpasses several state-of-the-art supervised methods. Our code is available at https://github.com/ISChenawei/DMNIL.
Authors: Ziyuan Liu, Ruifei Zhu, Long Gao, Yuanxiu Zhou, Jingyu Ma, Yuantao Gu
Abstract: Change detection (CD) in remote sensing images plays a vital role in Earth observation. However, the scarcity of high-resolution, comprehensive open-source datasets and the difficulty in achieving robust performance across varying change types remain major challenges. To address these issues, we introduce JL1-CD, a large-scale, sub-meter CD dataset consisting of 5,000 image pairs. We further propose a novel Origin-Partition (O-P) strategy and integrate it into a Multi-Teacher Knowledge Distillation (MTKD) framework to enhance CD performance. The O-P strategy partitions the training set by Change Area Ratio (CAR) and trains specialized teacher models on each subset. The MTKD framework then distills complementary knowledge from these teachers into a single student model, enabling improved detection results across diverse CAR scenarios without additional inference cost. Our MTKD approach demonstrated strong performance in the 2024 ``Jilin-1'' Cup challenge, ranking first in the preliminary and second in the final rounds. Extensive experiments on the JL1-CD and SYSU-CD datasets show that the MTKD framework consistently improves the performance of CD models with various network architectures and parameter sizes, establishing new state-of-the-art results. Code and dataset are available at https://github.com/circleLZY/MTKD-CD.
Authors: Zhangkai Wu, Xuhui Fan, Hongyu Wu, Longbing Cao
Abstract: Pre-trained diffusion models are commonly used to generate clean data (e.g., images) from random noises, effectively forming pairs of noises and corresponding clean images. Distillation on these pre-trained models can be viewed as the process of constructing advanced trajectories within the pair to accelerate sampling. For instance, consistency model distillation develops consistent projection functions to regulate trajectories, although sampling efficiency remains a concern. Rectified flow method enforces straight trajectories to enable faster sampling, yet relies on numerical ODE solvers, which may introduce approximation errors. In this work, we bridge the gap between the consistency model and the rectified flow method by proposing a Straight Consistent Trajectory~(SCoT) model. SCoT enjoys the benefits of both approaches for fast sampling, producing trajectories with consistent and straight properties simultaneously. These dual properties are strategically balanced by targeting two critical objectives: (1) regulating the gradient of SCoT's mapping to a constant, (2) ensuring trajectory consistency. Extensive experimental results demonstrate the effectiveness and efficiency of SCoT.
Authors: Zekang Weng, Jinjin Shi, Jinwei Wang, Zeming Han
Abstract: Image anomaly detection plays a vital role in applications such as industrial quality inspection and medical imaging, where it directly contributes to improving product quality and system reliability. However, existing methods often struggle with complex and diverse anomaly patterns. In particular, the separation between generation and discrimination tasks limits the effective coordination between anomaly sample generation and anomaly region detection. To address these challenges, we propose a novel hybrid diffusion model (HDM) that integrates generation and discrimination into a unified framework. The model consists of three key modules: the Diffusion Anomaly Generation Module (DAGM), the Diffusion Discriminative Module (DDM), and the Probability Optimization Module (POM). DAGM generates realistic and diverse anomaly samples, improving their representativeness. DDM then applies a reverse diffusion process to capture the differences between generated and normal samples, enabling precise anomaly region detection and localization based on probability distributions. POM refines the probability distributions during both the generation and discrimination phases, ensuring high-quality samples are used for training. Extensive experiments on multiple industrial image datasets demonstrate that our method outperforms state-of-the-art approaches, significantly improving both image-level and pixel-level anomaly detection performance, as measured by AUROC.
Authors: Zhaoyi Joey Hou, Adriana Kovashka, Xiang Lorraine Li
Abstract: Evaluating creativity is challenging, even for humans, not only because of its subjectivity but also because it involves complex cognitive processes. Inspired by work in marketing, we attempt to break down visual advertisement creativity into atypicality and originality. With fine-grained human annotations on these dimensions, we propose a suite of tasks specifically for such a subjective problem. We also evaluate the alignment between state-of-the-art (SoTA) vision language models (VLMs) and humans on our proposed benchmark, demonstrating both the promises and challenges of using VLMs for automatic creativity assessment.
Authors: Jindong Jiang, Xiuyu Li, Zhijian Liu, Muyang Li, Guo Chen, Zhiqi Li, De-An Huang, Guilin Liu, Zhiding Yu, Kurt Keutzer, Sungjin Ahn, Jan Kautz, Hongxu Yin, Yao Lu, Song Han, Wonmin Byeon
Abstract: Recent advances in video-based multimodal large language models (Video-LLMs) have significantly improved video understanding by processing videos as sequences of image frames. However, many existing methods treat frames independently in the vision backbone, lacking explicit temporal modeling, which limits their ability to capture dynamic patterns and efficiently handle long videos. To address these limitations, we introduce STORM (Spatiotemporal TOken Reduction for Multimodal LLMs), a novel architecture incorporating a dedicated temporal encoder between the image encoder and the LLM. Our temporal encoder leverages the Mamba State Space Model to integrate temporal information into image tokens, generating enriched representations that preserve inter-frame dynamics across the entire video sequence. This enriched encoding not only enhances video reasoning capabilities but also enables effective token reduction strategies, including test-time sampling and training-based temporal and spatial pooling, substantially reducing computational demands on the LLM without sacrificing key temporal information. By integrating these techniques, our approach simultaneously reduces training and inference latency while improving performance, enabling efficient and robust video understanding over extended temporal contexts. Extensive evaluations show that STORM achieves state-of-the-art results across various long video understanding benchmarks (more than 5% improvement on MLVU and LongVideoBench) while reducing the computation costs by up to $8\times$ and the decoding latency by 2.4-2.9$\times$ for the fixed numbers of input frames. Project page is available at https://research.nvidia.com/labs/lpr/storm
Authors: Guilherme Fernandes, Vasco Ramos, Regev Cohen, Idan Szpektor, Jo\~ao Magalh\~aes
Abstract: While diffusion models excel at generating high-quality images from text prompts, they struggle with visual consistency when generating image sequences. Existing methods generate each image independently, leading to disjointed narratives - a challenge further exacerbated in non-linear storytelling, where scenes must connect beyond adjacent images. We introduce a novel beam search strategy for latent space exploration, enabling conditional generation of full image sequences with beam search decoding. In contrast to earlier methods that rely on fixed latent priors, our method dynamically samples past latents to search for an optimal sequence of latent representations, ensuring coherent visual transitions. As the latent denoising space is explored, the beam search graph is pruned with a cross-attention mechanism that efficiently scores search paths, prioritizing alignment with both textual prompts and visual context. Human and automatic evaluations confirm that BeamDiffusion outperforms other baseline methods, producing full sequences with superior coherence, visual continuity, and textual alignment.
Authors: Boyang Deng, Songyou Peng, Kyle Genova, Gordon Wetzstein, Noah Snavely, Leonidas Guibas, Thomas Funkhouser
Abstract: We present a system using Multimodal LLMs (MLLMs) to analyze a large database with tens of millions of images captured at different times, with the aim of discovering patterns in temporal changes. Specifically, we aim to capture frequent co-occurring changes ("trends") across a city over a certain period. Unlike previous visual analyses, our analysis answers open-ended queries (e.g., "what are the frequent types of changes in the city?") without any predetermined target subjects or training labels. These properties cast prior learning-based or unsupervised visual analysis tools unsuitable. We identify MLLMs as a novel tool for their open-ended semantic understanding capabilities. Yet, our datasets are four orders of magnitude too large for an MLLM to ingest as context. So we introduce a bottom-up procedure that decomposes the massive visual analysis problem into more tractable sub-problems. We carefully design MLLM-based solutions to each sub-problem. During experiments and ablation studies with our system, we find it significantly outperforms baselines and is able to discover interesting trends from images captured in large cities (e.g., "addition of outdoor dining,", "overpass was painted blue," etc.). See more results and interactive demos at https://boyangdeng.com/visual-chronicles.
Authors: Stathis Galanakis, Alexandros Lattas, Stylianos Moschoglou, Bernhard Kainz, Stefanos Zafeiriou
Abstract: Despite recent progress in diffusion models, generating realistic head portraits from novel viewpoints remains a significant challenge. Most current approaches are constrained to limited angular ranges, predominantly focusing on frontal or near-frontal views. Moreover, although the recent emerging large-scale diffusion models have been proven robust in handling 3D scenes, they underperform on facial data, given their complex structure and the uncanny valley pitfalls. In this paper, we propose SpinMeRound, a diffusion-based approach designed to generate consistent and accurate head portraits from novel viewpoints. By leveraging a number of input views alongside an identity embedding, our method effectively synthesizes diverse viewpoints of a subject whilst robustly maintaining its unique identity features. Through experimentation, we showcase our model's generation capabilities in 360 head synthesis, while beating current state-of-the-art multiview diffusion models.
Authors: Hasan Wehbi, Hasan Nasrallah, Mohamad Hasan Zahweh, Zeinab Takach, Veera Ganesh Yalla, Ali J. Ghandour
Abstract: Wheat accounts for approximately 20% of the world's caloric intake, making it a vital component of global food security. Given this importance, mapping wheat fields plays a crucial role in enabling various stakeholders, including policy makers, researchers, and agricultural organizations, to make informed decisions regarding food security, supply chain management, and resource allocation. In this paper, we tackle the problem of accurately mapping wheat fields out of satellite images by introducing an improved pipeline for winter wheat segmentation, as well as presenting a case study on a decade-long analysis of wheat mapping in Lebanon. We integrate a Temporal Spatial Vision Transformer (TSViT) with Parameter-Efficient Fine Tuning (PEFT) and a novel post-processing pipeline based on the Fields of The World (FTW) framework. Our proposed pipeline addresses key challenges encountered in existing approaches, such as the clustering of small agricultural parcels in a single large field. By merging wheat segmentation with precise field boundary extraction, our method produces geometrically coherent and semantically rich maps that enable us to perform in-depth analysis such as tracking crop rotation pattern over years. Extensive evaluations demonstrate improved boundary delineation and field-level precision, establishing the potential of the proposed framework in operational agricultural monitoring and historical trend analysis. By allowing for accurate mapping of wheat fields, this work lays the foundation for a range of critical studies and future advances, including crop monitoring and yield estimation.
Authors: Zechuan Zhang, Ji Xie, Yu Lu, Zongxin Yang, Yi Yang
Abstract: Instruction-based image editing enables precise modifications via natural language prompts, but existing methods face a precision-efficiency tradeoff: fine-tuning demands massive datasets (>10M) and computational resources, while training-free approaches suffer from weak instruction comprehension. We address this by proposing ICEdit, which leverages the inherent comprehension and generation abilities of large-scale Diffusion Transformers (DiTs) through three key innovations: (1) An in-context editing paradigm without architectural modifications; (2) Minimal parameter-efficient fine-tuning for quality improvement; (3) Early Filter Inference-Time Scaling, which uses VLMs to select high-quality noise samples for efficiency. Experiments show that ICEdit achieves state-of-the-art editing performance with only 0.1\% of the training data and 1\% trainable parameters compared to previous methods. Our approach establishes a new paradigm for balancing precision and efficiency in instructional image editing. Codes and demos can be found in https://river-zhang.github.io/ICEdit-gh-pages/.
Authors: Stefanos Gkikas, Raul Fernandez Rojas, Manolis Tsiknakis
Abstract: Pain is a manifold condition that impacts a significant percentage of the population. Accurate and reliable pain evaluation for the people suffering is crucial to developing effective and advanced pain management protocols. Automatic pain assessment systems provide continuous monitoring and support decision-making processes, ultimately aiming to alleviate distress and prevent functionality decline. This study introduces PainFormer, a vision foundation model based on multi-task learning principles trained simultaneously on 14 tasks/datasets with a total of 10.9 million samples. Functioning as an embedding extractor for various input modalities, the foundation model provides feature representations to the Embedding-Mixer, a transformer-based module that performs the final pain assessment. Extensive experiments employing behavioral modalities - including RGB, synthetic thermal, and estimated depth videos - and physiological modalities such as ECG, EMG, GSR, and fNIRS revealed that PainFormer effectively extracts high-quality embeddings from diverse input modalities. The proposed framework is evaluated on two pain datasets, BioVid and AI4Pain, and directly compared to 75 different methodologies documented in the literature. Experiments conducted in unimodal and multimodal settings demonstrate state-of-the-art performances across modalities and pave the way toward general-purpose models for automatic pain assessment. The foundation model's architecture (code) and weights are available at: https://github.com/GkikasStefanos/PainFormer.
Authors: Jialei Chen, Xu Zheng, Dongyue Li, Chong Yi, Seigo Ito, Danda Pani Paudel, Luc Van Gool, Hiroshi Murase, Daisuke Deguchi
Abstract: Zero-shot Semantic Segmentation (ZSS) aims to segment categories that are not annotated during training. While fine-tuning vision-language models has achieved promising results, these models often overfit to seen categories due to the lack of supervision for unseen classes. As an alternative to fully supervised approaches, query-based segmentation has shown great latent in ZSS, as it enables object localization without relying on explicit labels. However, conventional Hungarian matching, a core component in query-based frameworks, needs full supervision and often misclassifies unseen categories as background in the setting of ZSS. To address this issue, we propose Split Matching (SM), a novel assignment strategy that decouples Hungarian matching into two components: one for seen classes in annotated regions and another for latent classes in unannotated regions (referred to as unseen candidates). Specifically, we partition the queries into seen and candidate groups, enabling each to be optimized independently according to its available supervision. To discover unseen candidates, we cluster CLIP dense features to generate pseudo masks and extract region-level embeddings using CLS tokens. Matching is then conducted separately for the two groups based on both class-level similarity and mask-level consistency. Additionally, we introduce a Multi-scale Feature Enhancement (MFE) module that refines decoder features through residual multi-scale aggregation, improving the model's ability to capture spatial details across resolutions. SM is the first to introduce decoupled Hungarian matching under the inductive ZSS setting, and achieves state-of-the-art performance on two standard benchmarks.
Authors: Feng Li, Zhaoyue Wang, Enyuan Zhang, Mohammad Masum Billah, Yunduan Cui, Kun Xu
Abstract: BEV-based 3D perception has emerged as a focal point of research in end-to-end autonomous driving. However, existing BEV approaches encounter significant challenges due to the large feature space, complicating efficient modeling and hindering effective integration of global attention mechanisms. We propose a novel modeling strategy, called InstanceBEV, that synergistically combines the strengths of both map-centric approaches and object-centric approaches. Our method effectively extracts instance-level features within the BEV features, facilitating the implementation of global attention modeling in a highly compressed feature space, thereby addressing the efficiency challenges inherent in map-centric global modeling. Furthermore, our approach enables effective multi-task learning without introducing additional module. We validate the efficiency and accuracy of the proposed model through predicting occupancy, achieving 3D occupancy panoptic segmentation by combining instance information. Experimental results on the OCC3D-nuScenes dataset demonstrate that InstanceBEV, utilizing only 8 frames, achieves a RayPQ of 15.3 and a RayIoU of 38.2. This surpasses SparseOcc's RayPQ by 9.3% and RayIoU by 10.7%, showcasing the effectiveness of multi-task synergy.
Authors: Jianfei Zhao, Feng Zhang, Xin Sun, Chong Feng
Abstract: Due to the unidirectional masking mechanism, Decoder-Only models propagate information from left to right. LVLMs (Large Vision-Language Models) follow the same architecture, with visual information gradually integrated into semantic representations during forward propagation. Through systematic analysis, we observe that the majority of the visual information is absorbed into the semantic representations. However, the model's attention distribution does not exhibit sufficient emphasis on semantic representations. This misalignment between the attention distribution and the actual information flow undermines the model's visual understanding ability and contributes to hallucinations. To address this issue, we enhance the model's visual understanding by leveraging the core information embedded in semantic representations. Specifically, we identify attention heads that focus on core semantic representations based on their attention distributions. Then, through a two-stage optimization paradigm, we propagate the advantages of these attention heads across the entire model, aligning the attention distribution with the actual information flow. We evaluate our method on three image captioning benchmarks using five different LVLMs, demonstrating its effectiveness in significantly reducing hallucinations. Further experiments reveal a trade-off between reduced hallucinations and richer details. Notably, our method allows for manual adjustment of the model's conservativeness, enabling flexible control to meet diverse real-world requirements.
Authors: Ruoxin Chen, Junwei Xi, Zhiyuan Yan, Ke-Yue Zhang, Shuang Wu, Jingyi Xie, Xu Chen, Lei Xu, Isabel Guan, Taiping Yao, Shouhong Ding
Abstract: Existing detectors are often trained on biased datasets, leading to the possibility of overfitting on non-causal image attributes that are spuriously correlated with real/synthetic labels. While these biased features enhance performance on the training data, they result in substantial performance degradation when applied to unbiased datasets. One common solution is to perform dataset alignment through generative reconstruction, matching the semantic content between real and synthetic images. However, we revisit this approach and show that pixel-level alignment alone is insufficient. The reconstructed images still suffer from frequency-level misalignment, which can perpetuate spurious correlations. To illustrate, we observe that reconstruction models tend to restore the high-frequency details lost in real images (possibly due to JPEG compression), inadvertently creating a frequency-level misalignment, where synthetic images appear to have richer high-frequency content than real ones. This misalignment leads to models associating high-frequency features with synthetic labels, further reinforcing biased cues. To resolve this, we propose Dual Data Alignment (DDA), which aligns both the pixel and frequency domains. Moreover, we introduce two new test sets: DDA-COCO, containing DDA-aligned synthetic images for testing detector performance on the most aligned dataset, and EvalGEN, featuring the latest generative models for assessing detectors under new generative architectures such as visual auto-regressive generators. Finally, our extensive evaluations demonstrate that a detector trained exclusively on DDA-aligned MSCOCO could improve across 8 diverse benchmarks by a non-trivial margin, showing a +7.2% on in-the-wild benchmarks, highlighting the improved generalizability of unbiased detectors. Our code is available at: https://github.com/roy-ch/Dual-Data-Alignment.
Authors: Zhipei Xu, Xuanyu Zhang, Qing Huang, Xing Zhou, Jian Zhang
Abstract: Recent advances in Artificial Intelligence Generated Content have led to highly realistic synthetic videos, particularly in human-centric scenarios involving speech, gestures, and full-body motion, posing serious threats to information authenticity and public trust. Unlike DeepFake techniques that focus on localized facial manipulation, human-centric video generation methods can synthesize entire human bodies with controllable movements, enabling complex interactions with environments, objects, and even other people. However, existing detection methods largely overlook the growing risks posed by such full-body synthetic content. Meanwhile, a growing body of research has explored leveraging LLMs for interpretable fake detection, aiming to explain decisions in natural language. Yet these approaches heavily depend on supervised fine-tuning, which introduces limitations such as annotation bias, hallucinated supervision, and weakened generalization. To address these challenges, we propose AvatarShield, a novel multimodal human-centric synthetic video detection framework that eliminates the need for dense textual supervision by adopting Group Relative Policy Optimization, enabling LLMs to develop reasoning capabilities from simple binary labels. Our architecture combines a discrete vision tower for high-level semantic inconsistencies and a residual extractor for fine-grained artifact analysis. We further introduce FakeHumanVid, a large-scale benchmark containing 15K real and synthetic videos across nine state-of-the-art human generation methods driven by text, pose, or audio. Extensive experiments demonstrate that AvatarShield outperforms existing methods in both in-domain and cross-domain settings.
Authors: Shenghui Chen, Po-han Li, Sandeep Chinchali, Ufuk Topcu
Abstract: Many decision-making tasks, where both accuracy and efficiency matter, still require human supervision. For example, tasks like traffic officers reviewing hour-long dashcam footage or researchers screening conference videos can benefit from concise summaries that reduce cognitive load and save time. Yet current vision-language models (VLMs) often produce verbose, redundant outputs that hinder task performance. Existing video caption evaluation depends on costly human annotations and overlooks the summaries' utility in downstream tasks. We address these gaps with Video-to-text Information Bottleneck Evaluation (VIBE), an annotation-free method that scores VLM outputs using two metrics: grounding (how well the summary aligns with visual content) and utility (how informative it is for the task). VIBE selects from randomly sampled VLM outputs by ranking them according to the two scores to support effective human decision-making. Human studies on LearningPaper24, SUTD-TrafficQA, and LongVideoBench show that summaries selected by VIBE consistently improve performance-boosting task accuracy by up to 61.23% and reducing response time by 75.77% compared to naive VLM summaries or raw video.
Authors: Olaf D\"unkel, Thomas Wimmer, Christian Theobalt, Christian Rupprecht, Adam Kortylewski
Abstract: Finding correspondences between semantically similar points across images and object instances is one of the everlasting challenges in computer vision. While large pre-trained vision models have recently been demonstrated as effective priors for semantic matching, they still suffer from ambiguities for symmetric objects or repeated object parts. We propose improving semantic correspondence estimation through 3D-aware pseudo-labeling. Specifically, we train an adapter to refine off-the-shelf features using pseudo-labels obtained via 3D-aware chaining, filtering wrong labels through relaxed cyclic consistency, and 3D spherical prototype mapping constraints. While reducing the need for dataset-specific annotations compared to prior work, we establish a new state-of-the-art on SPair-71k, achieving an absolute gain of over 4% and of over 7% compared to methods with similar supervision requirements. The generality of our proposed approach simplifies the extension of training to other data sources, which we demonstrate in our experiments.
Authors: Prakriti Tripathi
Abstract: Industry partners provided a problem statement that involves classifying electronic waste using machine learning models that will be used by pick-and-place robots for waste segregation. This was achieved by taking common electronic waste items, such as a mouse and charger, unsoldering them, and taking pictures to create a custom dataset. Then state-of-the art YOLOv11 model was trained and run to achieve 70 mAP in real-time. Mask-RCNN model was also trained and achieved 41 mAP. The model can be integrated with pick-and-place robots to perform segregation of e-waste.
Authors: Tao Wang, Mengyu Li, Geduo Zeng, Cheng Meng, Qiong Zhang
Abstract: 3D Gaussian Splatting (3DGS) has emerged as a powerful technique for radiance field rendering, but it typically requires millions of redundant Gaussian primitives, overwhelming memory and rendering budgets. Existing compaction approaches address this by pruning Gaussians based on heuristic importance scores, without global fidelity guarantee. To bridge this gap, we propose a novel optimal transport perspective that casts 3DGS compaction as global Gaussian mixture reduction. Specifically, we first minimize the composite transport divergence over a KD- tree partition to produce a compact geometric representation, and then decouple appearance from geometry by fine-tuning color and opacity attributes with far fewer Gaussian primitives. Experiments on benchmark datasets show that our method (i) yields negligible loss in rendering quality (PSNR, SSIM, LPIPS) compared to vanilla 3DGS with only 10% Gaussians; and (ii) consistently outperforms state- of-the-art 3DGS compaction techniques. Notably, our method is applicable to any stage of vanilla or accelerated 3DGS pipelines, providing an efficient and agnostic pathway to lightweight neural rendering. The code is publicly available at https://github.com/DrunkenPoet/GHAP
Authors: Yanlong Chen, Mattia Orlandi, Pierangelo Maria Rapa, Simone Benatti, Luca Benini, Yawei Li
Abstract: Human-machine interaction, particularly in prosthetic and robotic control, has seen progress with gesture recognition via surface electromyographic (sEMG) signals.However, classifying similar gestures that produce nearly identical muscle signals remains a challenge, often reducing classification accuracy. Traditional deep learning models for sEMG gesture recognition are large and computationally expensive, limiting their deployment on resource-constrained embedded systems. In this work, we propose WaveFormer, a lightweight transformer-based architecture tailored for sEMG gesture recognition. Our model integrates time-domain and frequency-domain features through a novel learnable wavelet transform, enhancing feature extraction. In particular, the WaveletConv module, a multi-level wavelet decomposition layer with depthwise separable convolution, ensures both efficiency and compactness. With just 3.1 million parameters, WaveFormer achieves 95% classification accuracy on the EPN612 dataset, outperforming larger models. Furthermore, when profiled on a laptop equipped with an Intel CPU, INT8 quantization achieves real-time deployment with a 6.75 ms inference latency.
Authors: Nikolaos Dionelis, Riccardo Musto, Jente Bosmans, Simone Sarti, Giancarlo Paoletti, S\'ebastien Lef\`evre, Bertrand Le Saux, Nicolas Long\'ep\'e
Abstract: Today, Earth Observation (EO) satellites generate massive volumes of data. To fully exploit this, it is essential to pretrain EO Foundation Models (FMs) on large unlabeled datasets, enabling efficient fine-tuning for downstream tasks with minimal labeled data. In this paper, we study scaling-up FMs: we train our models on the pretraining dataset MajorTOM 23TB which includes all regions, and the performance on average is competitive versus models pretrained on more specialized datasets which are substantially smaller and include only land. The additional data of oceans and ice do not decrease the performance on land-focused downstream tasks. These results indicate that large FMs trained on global datasets for a wider variety of downstream tasks can be useful for downstream applications that only require a subset of the information included in their training. The second contribution is the exploration of U-Net Convolutional Neural Network (CNN), Vision Transformers (ViT), and Mamba State-Space Models (SSM) as FMs. U-Net captures local correlations amongst pixels, while ViT and Mamba capture local and distant correlations. We develop various models using different architectures, including U-Net, ViT, and Mamba, and different number of parameters. We evaluate the FLoating-point OPerations (FLOPs) needed by the models. We fine-tune on the PhilEO Bench for different downstream tasks: roads, buildings, and land cover. For most n-shots for roads and buildings, U-Net 200M-2T outperforms the other models. Using Mamba, we achieve comparable results on the downstream tasks, with less computational expenses. We also compare with the recent FM TerraMind which we evaluate on PhilEO Bench.
Authors: Md Nahiduzzaman, Steven Korevaar, Zongyuan Ge, Feng Xia, Alireza Bab-Hadiashar, Ruwan Tennakoon
Abstract: To be adopted in safety-critical domains like medical image analysis, AI systems must provide human-interpretable decisions. Variational Information Pursuit (V-IP) offers an interpretable-by-design framework by sequentially querying input images for human-understandable concepts, using their presence or absence to make predictions. However, existing V-IP methods overlook sample-specific uncertainty in concept predictions, which can arise from ambiguous features or model limitations, leading to suboptimal query selection and reduced robustness. In this paper, we propose an interpretable and uncertainty-aware framework for medical imaging that addresses these limitations by accounting for upstream uncertainties in concept-based, interpretable-by-design models. Specifically, we introduce two uncertainty-aware models, EUAV-IP and IUAV-IP, that integrate uncertainty estimates into the V-IP querying process to prioritize more reliable concepts per sample. EUAV-IP skips uncertain concepts via masking, while IUAV-IP incorporates uncertainty into query selection implicitly for more informed and clinically aligned decisions. Our approach allows models to make reliable decisions based on a subset of concepts tailored to each individual sample, without human intervention, while maintaining overall interpretability. We evaluate our methods on five medical imaging datasets across four modalities: dermoscopy, X-ray, ultrasound, and blood cell imaging. The proposed IUAV-IP model achieves state-of-the-art accuracy among interpretable-by-design approaches on four of the five datasets, and generates more concise explanations by selecting fewer yet more informative concepts. These advances enable more reliable and clinically meaningful outcomes, enhancing model trustworthiness and supporting safer AI deployment in healthcare.
Authors: Chenqiu Zhao, Anup Basu
Abstract: A common assumption in probabilistic generative models for image generation is that learning the global data distribution suffices to generate novel images via sampling. We investigate the limitation of this core assumption, namely that learning global distributions leads to memorization rather than generative behavior. We propose two theoretical frameworks, the Mutually Exclusive Probability Space (MEPS) and the Local Dependence Hypothesis (LDH), for investigation. MEPS arises from the observation that deterministic mappings (e.g. neural networks) involving random variables tend to reduce overlap coefficients among involved random variables, thereby inducing exclusivity. We further propose a lower bound in terms of the overlap coefficient, and introduce a Binary Latent Autoencoder (BL-AE) that encodes images into signed binary latent representations. LDH formalizes dependence within a finite observation radius, which motivates our $\gamma$-Autoregressive Random Variable Model ($\gamma$-ARVM). $\gamma$-ARVM is an autoregressive model, with a variable observation range $\gamma$, that predicts a histogram for the next token. Using $\gamma$-ARVM, we observe that as the observation range increases, autoregressive models progressively shift toward memorization. In the limit of global dependence, the model behaves as a pure memorizer when operating on the binary latents produced by our BL-AE. Comprehensive experiments and discussions support our investigation.
Authors: Paul McHard, Florent P. Audonnet, Oliver Summerell, Sebastian Andraos, Paul Henderson, Gerardo Aragon-Camarasa
Abstract: Surface defects are a primary source of yield loss in manufacturing, yet existing anomaly detection methods often fail in real-world deployment due to limited and unrepresentative datasets. To overcome this, we introduce 3D-ADAM, a 3D Anomaly Detection in Additive Manufacturing dataset, that is the first large-scale, industry-relevant dataset for RGB+3D surface defect detection in additive manufacturing. 3D-ADAM comprises 14,120 high-resolution scans of 217 unique parts, captured with four industrial depth sensors, and includes 27,346 annotated defects across 12 categories along with 27,346 annotations of machine element features in 16 classes. 3D-ADAM is captured in a real industrial environment and as such reflects real production conditions, including variations in part placement, sensor positioning, lighting, and partial occlusion. Benchmarking state-of-the-art models demonstrates that 3D-ADAM presents substantial challenges beyond existing datasets. Validation through expert labelling surveys with industry partners further confirms its industrial relevance. By providing this benchmark, 3D-ADAM establishes a foundation for advancing robust 3D anomaly detection capable of meeting manufacturing demands.
Authors: Hung Nguyen, Runfa Li, An Le, Truong Nguyen
Abstract: Sparse-view 3D Gaussian Splatting (3DGS) presents significant challenges in reconstructing high-quality novel views, as it often overfits to the widely-varying high-frequency (HF) details of the sparse training views. While frequency regularization can be a promising approach, its typical reliance on Fourier transforms causes difficult parameter tuning and biases towards detrimental HF learning. We propose DWTGS, a framework that rethinks frequency regularization by leveraging wavelet-space losses that provide additional spatial supervision. Specifically, we supervise only the low-frequency (LF) LL subbands at multiple DWT levels, while enforcing sparsity on the HF HH subband in a self-supervised manner. Experiments across benchmarks show that DWTGS consistently outperforms Fourier-based counterparts, as this LF-centric strategy improves generalization and reduces HF hallucinations.
Authors: Chengchang Tian, Jianwei Ma, Yan Huang, Zhanye Chen, Honghao Wei, Hui Zhang, Wei Hong
Abstract: Feature-level fusion shows promise in collaborative perception (CP) through balanced performance and communication bandwidth trade-off. However, its effectiveness critically relies on input feature quality. The acquisition of high-quality features faces domain gaps from hardware diversity and deployment conditions, alongside temporal misalignment from transmission delays. These challenges degrade feature quality with cumulative effects throughout the collaborative network. In this paper, we present the Domain-And-Time Alignment (DATA) network, designed to systematically align features while maximizing their semantic representations for fusion. Specifically, we propose a Consistency-preserving Domain Alignment Module (CDAM) that reduces domain gaps through proximal-region hierarchical downsampling and observability-constrained discriminator. We further propose a Progressive Temporal Alignment Module (PTAM) to handle transmission delays via multi-scale motion modeling and two-stage compensation. Building upon the aligned features, an Instance-focused Feature Aggregation Module (IFAM) is developed to enhance semantic representations. Extensive experiments demonstrate that DATA achieves state-of-the-art performance on three typical datasets, maintaining robustness with severe communication delays and pose errors. The code will be released at https://github.com/ChengchangTian/DATA.
Authors: Lianwei Yang, Haokun Lin, Tianchen Zhao, Yichen Wu, Hongyu Zhu, Ruiqi Xie, Zhenan Sun, Yu Wang, Qingyi Gu
Abstract: Diffusion Transformers (DiTs) have achieved impressive performance in text-to-image and text-to-video generation. However, their high computational cost and large parameter sizes pose significant challenges for usage in resource-constrained scenarios. Effective compression of models has become a crucial issue that urgently needs to be addressed. Post-training quantization (PTQ) is a promising solution to reduce memory usage and accelerate inference, but existing PTQ methods suffer from severe performance degradation under extreme low-bit settings. After experiments and analysis, we identify two key obstacles to low-bit PTQ for DiTs: (1) the weights of DiT models follow a Gaussian-like distribution with long tails, causing uniform quantization to poorly allocate intervals and leading to significant quantization errors. This issue has been observed in the linear layer weights of different DiT models, which deeply limits the performance. (2) two types of activation outliers in DiT models: (i) Mild Outliers with slightly elevated values, and (ii) Salient Outliers with large magnitudes concentrated in specific channels, which disrupt activation quantization. To address these issues, we propose LRQ-DiT, an efficient and accurate post-training quantization framework for image and video generation. First, we introduce Twin-Log Quantization (TLQ), a log-based method that allocates more quantization intervals to the intermediate dense regions, effectively achieving alignment with the weight distribution and reducing quantization errors. Second, we propose an Adaptive Rotation Scheme (ARS) that dynamically applies Hadamard or outlier-aware rotations based on activation fluctuation, effectively mitigating the impact of both types of outliers. Extensive experiments on various text-to-image and text-to-video DiT models demonstrate that LRQ-DiT preserves high generation quality.
Authors: Qihang Zhou, Shibo He, Jiangtao Yan, Wenchao Meng, Jiming Chen
Abstract: In this paper, we aim to transfer CLIP's robust 2D generalization capabilities to identify 3D anomalies across unseen objects of highly diverse class semantics. To this end, we propose a unified framework to comprehensively detect and segment 3D anomalies by leveraging both point- and pixel-level information. We first design PointAD, which leverages point-pixel correspondence to represent 3D anomalies through their associated rendering pixel representations. This approach is referred to as implicit 3D representation, as it focuses solely on rendering pixel anomalies but neglects the inherent spatial relationships within point clouds. Then, we propose PointAD+ to further broaden the interpretation of 3D anomalies by introducing explicit 3D representation, emphasizing spatial abnormality to uncover abnormal spatial relationships. Hence, we propose G-aggregation to involve geometry information to enable the aggregated point representations spatially aware. To simultaneously capture rendering and spatial abnormality, PointAD+ proposes hierarchical representation learning, incorporating implicit and explicit anomaly semantics into hierarchical text prompts: rendering prompts for the rendering layer and geometry prompts for the geometry layer. A cross-hierarchy contrastive alignment is further introduced to promote the interaction between the rendering and geometry layers, facilitating mutual anomaly learning. Finally, PointAD+ integrates anomaly semantics from both layers to capture the generalized anomaly semantics. During the test, PointAD+ can integrate RGB information in a plug-and-play manner and further improve its detection performance. Extensive experiments demonstrate the superiority of PointAD+ in ZS 3D anomaly detection across unseen objects with highly diverse class semantics, achieving a holistic understanding of abnormality.
Authors: Linqing Wang, Ximing Xing, Yiji Cheng, Zhiyuan Zhao, Donghao Li, Tiankai Hang, Jiale Tao, Qixun Wang, Ruihuang Li, Comi Chen, Xin Li, Mingrui Wu, Xinchi Deng, Shuyang Gu, Chunyu Wang, Qinglin Lu
Abstract: Recent advancements in text-to-image (T2I) diffusion models have demonstrated remarkable capabilities in generating high-fidelity images. However, these models often struggle to faithfully render complex user prompts, particularly in aspects like attribute binding, negation, and compositional relationships. This leads to a significant mismatch between user intent and the generated output. To address this challenge, we introduce PromptEnhancer, a novel and universal prompt rewriting framework that enhances any pretrained T2I model without requiring modifications to its weights. Unlike prior methods that rely on model-specific fine-tuning or implicit reward signals like image-reward scores, our framework decouples the rewriter from the generator. We achieve this by training a Chain-of-Thought (CoT) rewriter through reinforcement learning, guided by a dedicated reward model we term the AlignEvaluator. The AlignEvaluator is trained to provide explicit and fine-grained feedback based on a systematic taxonomy of 24 key points, which are derived from a comprehensive analysis of common T2I failure modes. By optimizing the CoT rewriter to maximize the reward from our AlignEvaluator, our framework learns to generate prompts that are more precisely interpreted by T2I models. Extensive experiments on the HunyuanImage 2.1 model demonstrate that PromptEnhancer significantly improves image-text alignment across a wide range of semantic and compositional challenges. Furthermore, we introduce a new, high-quality human preference benchmark to facilitate future research in this direction.
Authors: Feng Shen, Jiaming Cui, Shuai Zhou, Wenqiang Li, Ruifeng Qin
Abstract: Automated defect detection from UAV imagery of transmission lines is a challenging task due to the small size, ambiguity, and complex backgrounds of defects. This paper proposes TinyDef-DETR, a DETR-based framework designed to achieve accurate and efficient detection of transmission line defects from UAV-acquired images. The model integrates four major components: an edge-enhanced ResNet backbone to strengthen boundary-sensitive representations, a stride-free space-to-depth module to enable detail-preserving downsampling, a cross-stage dual-domain multi-scale attention mechanism to jointly model global context and local cues, and a Focaler-Wise-SIoU regression loss to improve the localization of small and difficult targets. Together, these designs effectively mitigate the limitations of conventional detectors. Extensive experiments on both public and real-world datasets demonstrate that TinyDef-DETR achieves superior detection performance and strong generalization capability, while maintaining modest computational overhead. The accuracy and efficiency of TinyDef-DETR make it a suitable method for UAV-based transmission line defect detection, particularly in scenarios involving small and ambiguous targets.
Authors: Jiarui Chen, Yikeng Chen, Yingshuang Zou, Ye Huang, Peng Wang, Yuan Liu, Yujing Sun, Wenping Wang
Abstract: 3D Gaussian Splatting (3DGS) has emerged as a dominant novel-view synthesis technique, but its high memory consumption severely limits its applicability on edge devices. A growing number of 3DGS compression methods have been proposed to make 3DGS more efficient, yet most only focus on storage compression and fail to address the critical bottleneck of rendering memory. To address this problem, we introduce MEGS$^{2}$, a novel memory-efficient framework that tackles this challenge by jointly optimizing two key factors: the total primitive number and the parameters per primitive, achieving unprecedented memory compression. Specifically, we replace the memory-intensive spherical harmonics with lightweight, arbitrarily oriented spherical Gaussian lobes as our color representations. More importantly, we propose a unified soft pruning framework that models primitive-number and lobe-number pruning as a single constrained optimization problem. Experiments show that MEGS$^{2}$ achieves a 50% static VRAM reduction and a 40% rendering VRAM reduction compared to existing methods, while maintaining comparable rendering quality. Project page: https://megs-2.github.io/
Authors: Payal Varshney, Adriano Lucieri, Christoph Balada, Sheraz Ahmed, Andreas Dengel
Abstract: Video-based AI systems are increasingly adopted in safety-critical domains such as autonomous driving and healthcare. However, interpreting their decisions remains challenging due to the inherent spatiotemporal complexity of video data and the opacity of deep learning models. Existing explanation techniques often suffer from limited temporal coherence, insufficient robustness, and a lack of actionable causal insights. Current counterfactual explanation methods typically do not incorporate guidance from the target model, reducing semantic fidelity and practical utility. We introduce Latent Diffusion for Video Counterfactual Explanations (LD-ViCE), a novel framework designed to explain the behavior of video-based AI models. Compared to previous approaches, LD-ViCE reduces the computational costs of generating explanations by operating in latent space using a state-of-the-art diffusion model, while producing realistic and interpretable counterfactuals through an additional refinement step. Our experiments demonstrate the effectiveness of LD-ViCE across three diverse video datasets, including EchoNet-Dynamic (cardiac ultrasound), FERV39k (facial expression), and Something-Something V2 (action recognition). LD-ViCE outperforms a recent state-of-the-art method, achieving an increase in R2 score of up to 68% while reducing inference time by half. Qualitative analysis confirms that LD-ViCE generates semantically meaningful and temporally coherent explanations, offering valuable insights into the target model behavior. LD-ViCE represents a valuable step toward the trustworthy deployment of AI in safety-critical domains.
Authors: Matthieu Vilain, R\'emi Giraud, Yannick Berthoumieu, Guillaume Bourmaud
Abstract: Dense image matching aims to find a correspondent for every pixel of a source image in a partially overlapping target image. State-of-the-art methods typically rely on a coarse-to-fine mechanism where a single correspondent hypothesis is produced per source location at each scale. In challenging cases -- such as at depth discontinuities or when the target image is a strong zoom-in of the source image -- the correspondents of neighboring source locations are often widely spread and predicting a single correspondent hypothesis per source location at each scale may lead to erroneous matches. In this paper, we investigate the idea of predicting multiple correspondent hypotheses per source location at each scale instead. We consider a beam search strategy to propagat multiple hypotheses at each scale and propose integrating these multiple hypotheses into cross-attention layers, resulting in a novel dense matching architecture called BEAMER. BEAMER learns to preserve and propagate multiple hypotheses across scales, making it significantly more robust than state-of-the-art methods, especially at depth discontinuities or when the target image is a strong zoom-in of the source image.
Authors: Rogerio Guimaraes, Frank Xiao, Pietro Perona, Markus Marks
Abstract: Humans can recognize the same actions despite large context and viewpoint variations, such as differences between species (walking in spiders vs. horses), viewpoints (egocentric vs. third-person), and contexts (real life vs movies). Current deep learning models struggle with such generalization. We propose using features generated by a Vision Diffusion Model (VDM), aggregated via a transformer, to achieve human-like action recognition across these challenging conditions. We find that generalization is enhanced by the use of a model conditioned on earlier timesteps of the diffusion process to highlight semantic information over pixel level details in the extracted features. We experimentally explore the generalization properties of our approach in classifying actions across animal species, across different viewing angles, and different recording contexts. Our model sets a new state-of-the-art across all three generalization benchmarks, bringing machine action recognition closer to human-like robustness. Project page: https://www.vision.caltech.edu/actiondiff. Code: https://github.com/frankyaoxiao/ActionDiff
URLs: https://www.vision.caltech.edu/actiondiff., https://github.com/frankyaoxiao/ActionDiff
Authors: Jing Huang, Zhiya Tan, Shutao Gong, Fanwei Zeng, Joey Tianyi Zhou, Jianshu Li
Abstract: As large vision language models (VLMs) advance, their capabilities in multilingual visual question answering (mVQA) have significantly improved. Chain-of-thought (CoT) reasoning has been proven to enhance interpretability and complex reasoning. However, most existing approaches rely primarily on textual CoT and provide limited support for multilingual multimodal reasoning, constraining their deployment in real-world applications. To address this gap, we introduce \textbf{LaV-CoT}, the first Language-aware Visual CoT framework with Multi-Aspect Reward Optimization. LaV-CoT incorporates an interpretable multi-stage reasoning pipeline consisting of Text Summary with Bounding Box (BBox), Language Identification, Spatial Object-level Captioning, and Step-by-step Logical Reasoning. Following this reasoning pipeline, we design an automated data curation method that generates multilingual CoT annotations through iterative generation, correction, and refinement, enabling scalable and high-quality training data. To improve reasoning and generalization, LaV-CoT adopts a two-stage training paradigm combining Supervised Fine-Tuning (SFT) with Language-aware Group Relative Policy Optimization (GRPO), guided by verifiable multi-aspect rewards including language consistency, structural accuracy, and semantic alignment. Extensive evaluations on public datasets including MMMB, Multilingual MMBench, and MTVQA show that LaV-CoT achieves up to ~9.5% accuracy improvements over open-source baselines of similar size and even surpasses models with 2$\times$ larger scales by ~2.6%. Moreover, LaV-CoT outperforms advanced proprietary models such as GPT-4o-0513 and Gemini-2.5-flash. We further conducted an online A/B test to validate our method on real-world data, highlighting its effectiveness for industrial deployment. Our code is available at this link: \href{https://github.com/HJNVR/LaV-CoT}
Authors: Salma Galaaoui, Eduardo Valle, David Picard, Nermin Samet
Abstract: In this paper, we present a comprehensive review of 3D human pose estimation and human mesh recovery from in-the-wild LiDAR point clouds. We compare existing approaches across several key dimensions, and propose a structured taxonomy to classify these methods. Following this taxonomy, we analyze each method's strengths, limitations, and design choices. In addition, (i) we perform a quantitative comparison of the three most widely used datasets, detailing their characteristics; (ii) we compile unified definitions of all evaluation metrics; and (iii) we establish benchmark tables for both tasks on these datasets to enable fair comparisons and promote progress in the field. We also outline open challenges and research directions critical for advancing LiDAR-based 3D human understanding. Moreover, we maintain an accompanying webpage that organizes papers according to our taxonomy and continuously update it with new studies: https://github.com/valeoai/3D-Human-Pose-Shape-Estimation-from-LiDAR
URLs: https://github.com/valeoai/3D-Human-Pose-Shape-Estimation-from-LiDAR
Authors: Elena Camuffo, Francesco Barbato, Mete Ozay, Simone Milani, Umberto Michieli
Abstract: We introduce MOCHA (Multi-modal Objects-aware Cross-arcHitecture Alignment), a knowledge distillation approach that transfers region-level multimodal semantics from a large vision-language teacher (e.g., LLaVa) into a lightweight vision-only object detector student (e.g., YOLO). A translation module maps student features into a joint space, where the training of the student and translator is guided by a dual-objective loss that enforces both local alignment and global relational consistency. Unlike prior approaches focused on dense or global alignment, MOCHA operates at the object level, enabling efficient transfer of semantics without modifying the teacher or requiring textual input at inference. We validate our method across four personalized detection benchmarks under few-shot regimes. Results show consistent gains over baselines, with a +10.1 average score improvement. Despite its compact architecture, MOCHA reaches performance on par with larger multimodal models, proving its suitability for real-world deployment.
Authors: Jialiang Kang, Han Shu, Wenshuo Li, Yingjie Zhai, Xinghao Chen
Abstract: Speculative decoding is a widely adopted technique for accelerating inference in large language models (LLMs), yet its application to vision-language models (VLMs) remains underexplored, with existing methods achieving only modest speedups (<1.5x). This gap is increasingly significant as multimodal capabilities become central to large-scale models. We hypothesize that large VLMs can effectively filter redundant image information layer by layer without compromising textual comprehension, whereas smaller draft models struggle to do so. To address this, we introduce Vision-Aware Speculative Decoding (ViSpec), a novel framework tailored for VLMs. ViSpec employs a lightweight vision adaptor module to compress image tokens into a compact representation, which is seamlessly integrated into the draft model's attention mechanism while preserving original image positional information. Additionally, we extract a global feature vector for each input image and augment all subsequent text tokens with this feature to enhance multimodal coherence. To overcome the scarcity of multimodal datasets with long assistant responses, we curate a specialized training dataset by repurposing existing datasets and generating extended outputs using the target VLM with modified prompts. Our training strategy mitigates the risk of the draft model exploiting direct access to the target model's hidden states, which could otherwise lead to shortcut learning when training solely on target model outputs. Extensive experiments validate ViSpec, achieving, to our knowledge, the first substantial speedup in VLM speculative decoding. Code is available at https://github.com/KangJialiang/ViSpec.
Authors: Aiden Chang, Celso De Melo, Stephanie M. Lukin
Abstract: Real-time understanding of continuous video streams is essential for intelligent agents operating in high-stakes environments, including autonomous vehicles, surveillance drones, and disaster response robots. Yet, most existing video understanding and highlight detection methods assume access to the entire video during inference, making them unsuitable for online or streaming scenarios. In particular, current models optimize for offline summarization, failing to support step-by-step reasoning needed for real-time decision-making. We introduce Aha, an autoregressive highlight detection framework that predicts the relevance of each video frame against a task described in natural language. Without accessing future video frames, Aha utilizes a multimodal vision-language model and lightweight, decoupled heads trained on a large, curated dataset of human-centric video labels. To enable scalability, we introduce the Dynamic SinkCache mechanism that achieves constant memory usage across infinite-length streams without degrading performance on standard benchmarks. This encourages the hidden representation to capture high-level task objectives, enabling effective frame-level rankings for informativeness, relevance, and uncertainty with respect to the natural language task. Aha achieves state-of-the-art (SOTA) performance on highlight detection benchmarks, surpassing even prior offline, full-context approaches and video-language models by +5.9% on TVSum and +8.3% on Mr. Hisum in mAP (mean Average Precision). We explore Aha's potential for real-world robotics applications given a task-oriented natural language input and a continuous, robot-centric video. Both experiments demonstrate Aha's potential effectiveness as a real-time reasoning module for downstream planning and long-horizon understanding.
Authors: Maria Taktasheva, Lily Goli, Alessandro Fiorini, Zhen Li, Daniel Rebain, Andrea Tagliasacchi
Abstract: Recent advances in radiance fields and novel view synthesis enable creation of realistic digital twins from photographs. However, current methods struggle with flat, texture-less surfaces, creating uneven and semi-transparent reconstructions, due to an ill-conditioned photometric reconstruction objective. Surface reconstruction methods solve this issue but sacrifice visual quality. We propose a novel hybrid 2D/3D representation that jointly optimizes constrained planar (2D) Gaussians for modeling flat surfaces and freeform (3D) Gaussians for the rest of the scene. Our end-to-end approach dynamically detects and refines planar regions, improving both visual fidelity and geometric accuracy. It achieves state-of-the-art depth estimation on ScanNet++ and ScanNetv2, and excels at mesh extraction without overfitting to a specific camera model, showing its effectiveness in producing high-quality reconstruction of indoor scenes.
Authors: Xin Lei Lin, Soroush Mehraban, Abhishek Moturu, Babak Taati
Abstract: Automated pain assessment from facial expressions is crucial for non-communicative patients, such as those with dementia. Progress has been limited by two challenges: (i) existing datasets exhibit severe demographic and label imbalance due to ethical constraints, and (ii) current generative models cannot precisely control facial action units (AUs), facial structure, or clinically validated pain levels. We present 3DPain, a large-scale synthetic dataset specifically designed for automated pain assessment, featuring unprecedented annotation richness and demographic diversity. Our three-stage framework generates diverse 3D meshes, textures them with diffusion models, and applies AU-driven face rigging to synthesize multi-view faces with paired neutral and pain images, AU configurations, PSPI scores, and the first dataset-level annotations of pain-region heatmaps. The dataset comprises 82,500 samples across 25,000 pain expression heatmaps and 2,500 synthetic identities balanced by age, gender, and ethnicity. We further introduce ViTPain, a Vision Transformer based cross-modal distillation framework in which a heatmap-trained teacher guides a student trained on RGB images, enhancing accuracy, interpretability, and clinical reliability. Together, 3DPain and ViTPain establish a controllable, diverse, and clinically grounded foundation for generalizable automated pain assessment.
Authors: Kai Jiang, Zhengyan Shi, Dell Zhang, Hongyuan Zhang, Xuelong Li
Abstract: Class Incremental Learning (CIL) aims to continuously learn new categories while retaining the knowledge of old ones. Pre-trained models (PTMs) show promising capabilities in CIL. However, existing approaches that apply lightweight fine-tuning to backbones still induce parameter drift, thereby compromising the generalization capability of pre-trained models. Parameter drift can be conceptualized as a form of noise that obscures critical patterns learned for previous tasks. However, recent researches have shown that noise is not always harmful. For example, the large number of visual patterns learned from pre-training can be easily abused by a single task, and introducing appropriate noise can suppress some low-correlation features, thus leaving a margin for future tasks. To this end, we propose learning beneficial noise for CIL guided by information theory and propose Mixture of Noise (Min), aiming to mitigate the degradation of backbone generalization from adapting new tasks. Specifically, task-specific noise is learned from high-dimension features of new tasks. Then, a set of weights is adjusted dynamically for optimal mixture of different task noise. Finally, Min embeds the beneficial noise into the intermediate features to mask the response of inefficient patterns. Extensive experiments on six benchmark datasets demonstrate that Min achieves state-of-the-art performance in most incremental settings, with particularly outstanding results in 50-steps incremental settings. This shows the significant potential for beneficial noise in continual learning. Code is available at https://github.com/ASCIIJK/MiN-NeurIPS2025.
Authors: Ziyang Xu, Benedikt Schwab, Yihui Yang, Thomas H. Kolbe, Christoph Holst
Abstract: Accurate registration between LiDAR (Light Detection and Ranging) point clouds and semantic 3D city models is a fundamental topic in urban digital twinning and a prerequisite for downstream tasks, such as digital construction, change detection and model refinement. However, achieving accurate LiDAR-to-Model registration at individual building level remains challenging, particularly due to the generalization uncertainty in semantic 3D city models at the Level of Detail 2 (LoD2). This paper addresses this gap by proposing L2M-Reg, a plane-based fine registration method that explicitly accounts for model uncertainty. L2M-Reg consists of three key steps: establishing reliable plane correspondence, building a pseudo-plane-constrained Gauss-Helmert model, and adaptively estimating vertical translation. Experiments on three real-world datasets demonstrate that L2M-Reg is both more accurate and computationally efficient than existing ICP-based and plane-based methods. Overall, L2M-Reg provides a novel building-level solution regarding LiDAR-to-Model registration when model uncertainty is present.
Authors: Yingzhen Hu, Yiheng Zhong, Ruobing Li, Yingxue Su, Jiabao An, Feilong Tang, Jionglong Su, Imran Razzak
Abstract: The Segment Anything Model (SAM) demonstrates impressive zero-shot segmentation ability on natural images but encounters difficulties in medical imaging due to domain shifts, anatomical variability, and its reliance on user-provided prompts. Recent prompt-free adaptations alleviate the need for expert intervention, yet still suffer from limited robustness and adaptability, often overlooking the issues of semantic over-smoothing and token uniformity. We propose SAM-DCE, which balances local discrimination and global semantics while mitigating token uniformity, enhancing inter-class separability, and enriching mask decoding with fine-grained, consistent representations. Extensive experiments on diverse medical benchmarks validate its effectiveness.
Authors: Youwei Pang, Xiaoqi Zhao, Lihe Zhang, Huchuan Lu, Georges El Fakhri, Xiaofeng Liu, Shijian Lu
Abstract: As an essential vision task, infrared small target detection (IRSTD) has seen significant advancements through deep learning. However, critical limitations in current evaluation protocols impede further progress. First, existing methods rely on fragmented pixel- and target-level specific metrics, which fails to provide a comprehensive view of model capabilities. Second, an excessive emphasis on overall performance scores obscures crucial error analysis, which is vital for identifying failure modes and improving real-world system performance. Third, the field predominantly adopts dataset-specific training-testing paradigms, hindering the understanding of model robustness and generalization across diverse infrared scenarios. This paper addresses these issues by introducing a hybrid-level metric incorporating pixel- and target-level performance, proposing a systematic error analysis method, and emphasizing the importance of cross-dataset evaluation. These aim to offer a more thorough and rational hierarchical analysis framework, ultimately fostering the development of more effective and robust IRSTD models. An open-source toolkit has be released to facilitate standardized benchmarking.
Authors: Haoyang Xu, Tianhao Zhao, Sibei Yang, Yutian Lin
Abstract: Diffusion models have emerged as a powerful technique for text-to-image (T2I) generation, creating high-quality, diverse images across various domains. However, a common limitation in these models is the incomplete display of objects, where fragments or missing parts undermine the model's performance in downstream applications. In this study, we conduct an in-depth analysis of the incompleteness issue and reveal that the primary factor behind incomplete object generation is the usage of RandomCrop during model training. This widely used data augmentation method, though enhances model generalization ability, disrupts object continuity during training. To address this, we propose a training-free solution that penalizes activation values at image boundaries during the early denoising steps. Our method is easily applicable to pre-trained Stable Diffusion models with minimal modifications and negligible computational overhead. Extensive experiments demonstrate the effectiveness of our method, showing substantial improvements in object integrity and image quality.
Authors: Zipeng Wang, Dan Xu
Abstract: Recently, 3D Gaussian Splatting (3DGS) has emerged as a powerful alternative to NeRF-based approaches, enabling real-time, high-quality novel view synthesis through explicit, optimizable 3D Gaussians. However, 3DGS suffers from significant memory overhead due to its reliance on per-Gaussian parameters to model view-dependent effects and anisotropic shapes. While recent works propose compressing 3DGS with neural fields, these methods struggle to capture high-frequency spatial variations in Gaussian properties, leading to degraded reconstruction of fine details. We present Hybrid Radiance Fields (HyRF), a novel scene representation that combines the strengths of explicit Gaussians and neural fields. HyRF decomposes the scene into (1) a compact set of explicit Gaussians storing only critical high-frequency parameters and (2) grid-based neural fields that predict remaining properties. To enhance representational capacity, we introduce a decoupled neural field architecture, separately modeling geometry (scale, opacity, rotation) and view-dependent color. Additionally, we propose a hybrid rendering scheme that composites Gaussian splatting with a neural field-predicted background, addressing limitations in distant scene representation. Experiments demonstrate that HyRF achieves state-of-the-art rendering quality while reducing model size by over 20 times compared to 3DGS and maintaining real-time performance. Our project page is available at https://wzpscott.github.io/hyrf/.
Authors: Zhitao Zeng, Guojian Yuan, Junyuan Mao, Yuxuan Wang, Xiaoshuang Jia, Yueming Jin
Abstract: Accurate temporal prediction is the bridge between comprehensive scene understanding and embodied artificial intelligence. However, predicting multiple fine-grained states of a scene at multiple temporal scales is difficult for vision-language models. We formalize the Multi-Scale Temporal Prediction (MSTP) task in general and surgical scenes by decomposing multi-scale into two orthogonal dimensions: the temporal scale, forecasting states of humans and surgery at varying look-ahead intervals, and the state scale, modeling a hierarchy of states in general and surgical scenes. For example, in general scenes, states of contact relationships are finer-grained than states of spatial relationships. In surgical scenes, medium-level steps are finer-grained than high-level phases yet remain constrained by their encompassing phase. To support this unified task, we introduce the first MSTP Benchmark, featuring synchronized annotations across multiple state scales and temporal scales. We further propose a method, Incremental Generation and Multi-agent Collaboration (IG-MC), which integrates two key innovations. First, we present a plug-and-play incremental generation module that continuously synthesizes up-to-date visual previews at expanding temporal scales to inform multiple decision-making agents, keeping decisions and generated visuals synchronized and preventing performance degradation as look-ahead intervals lengthen. Second, we present a decision-driven multi-agent collaboration framework for multi-state prediction, comprising generation, initiation, and multi-state assessment agents that dynamically trigger and evaluate prediction cycles to balance global coherence and local fidelity.
Authors: Gunjan Chhablani, Xiaomeng Ye, Muhammad Zubair Irshad, Zsolt Kira
Abstract: The field of Embodied AI predominantly relies on simulation for training and evaluation, often using either fully synthetic environments that lack photorealism or high-fidelity real-world reconstructions captured with expensive hardware. As a result, sim-to-real transfer remains a major challenge. In this paper, we introduce EmbodiedSplat, a novel approach that personalizes policy training by efficiently capturing the deployment environment and fine-tuning policies within the reconstructed scenes. Our method leverages 3D Gaussian Splatting (GS) and the Habitat-Sim simulator to bridge the gap between realistic scene capture and effective training environments. Using iPhone-captured deployment scenes, we reconstruct meshes via GS, enabling training in settings that closely approximate real-world conditions. We conduct a comprehensive analysis of training strategies, pre-training datasets, and mesh reconstruction techniques, evaluating their impact on sim-to-real predictivity in real-world scenarios. Experimental results demonstrate that agents fine-tuned with EmbodiedSplat outperform both zero-shot baselines pre-trained on large-scale real-world datasets (HM3D) and synthetically generated datasets (HSSD), achieving absolute success rate improvements of 20% and 40% on real-world Image Navigation task. Moreover, our approach yields a high sim-vs-real correlation (0.87-0.97) for the reconstructed meshes, underscoring its effectiveness in adapting policies to diverse environments with minimal effort. Project page: https://gchhablani.github.io/embodied-splat.
Authors: Keyu Du, Jingyu Hu, Haipeng Li, Hao Xu, Haibing Huang, Chi-Wing Fu, Shuaicheng Liu
Abstract: This paper presents a new approach to estimate accurate and robust 3D semantic correspondence with the hierarchical neural semantic representation. Our work has three key contributions. First, we design the hierarchical neural semantic representation (HNSR), which consists of a global semantic feature to capture high-level structure and multi-resolution local geometric features to preserve fine details, by carefully harnessing 3D priors from pre-trained 3D generative models. Second, we design a progressive global-to-local matching strategy, which establishes coarse semantic correspondence using the global semantic feature, then iteratively refines it with local geometric features, yielding accurate and semantically-consistent mappings. Third, our framework is training-free and broadly compatible with various pre-trained 3D generative backbones, demonstrating strong generalization across diverse shape categories. Our method also supports various applications, such as shape co-segmentation, keypoint matching, and texture transfer, and generalizes well to structurally diverse shapes, with promising results even in cross-category scenarios. Both qualitative and quantitative evaluations show that our method outperforms previous state-of-the-art techniques.
Authors: Qinghua Lin, Guang-Hai Liu, Zuoyong Li, Yang Li, Yuting Jiang, Xiang Wu
Abstract: Multimodal pathological images are usually in clinical diagnosis, but computer vision-based multimodal image-assisted diagnosis faces challenges with modality fusion, especially in the absence of expert-annotated data. To achieve the modality fusion in multimodal images with label scarcity, we propose a novel ``pretraining + fine-tuning" framework for multimodal semi-supervised medical image classification. Specifically, we propose a synergistic learning pretraining framework of consistency, reconstructive, and aligned learning. By treating one modality as an augmented sample of another modality, we implement a self-supervised learning pre-train, enhancing the baseline model's feature representation capability. Then, we design a fine-tuning method for multimodal fusion. During the fine-tuning stage, we set different encoders to extract features from the original modalities and provide a multimodal fusion encoder for fusion modality. In addition, we propose a distribution shift method for multimodal fusion features, which alleviates the prediction uncertainty and overfitting risks caused by the lack of labeled samples. We conduct extensive experiments on the publicly available gastroscopy image datasets Kvasir and Kvasirv2. Quantitative and qualitative results demonstrate that the proposed method outperforms the current state-of-the-art classification methods. The code will be released at: https://github.com/LQH89757/MICS.
Authors: Dian Jin, Yanghao Zhou, Jinxing Zhou, Jiaqi Ma, Ruohao Guo, Dan Guo
Abstract: Referring Audio-Visual Segmentation (Ref-AVS) aims to segment specific objects in videos based on natural language expressions involving audio, vision, and text information. This task poses significant challenges in cross-modal reasoning and fine-grained object localization. In this paper, we propose a simple framework, SimToken, that integrates a multimodal large language model (MLLM) with the Segment Anything Model (SAM). The MLLM is guided to generate a special semantic token representing the referred object. This compact token, enriched with contextual information from all modalities, acts as a prompt to guide SAM to segment objectsacross video frames. To further improve semantic learning, we introduce a novel target-consistent semantic alignment loss that aligns token embeddings from different expressions but referring to the same object. Experiments on the Ref-AVS benchmark demonstrate that our approach achieves superior performance compared to existing methods.
Authors: Yuxuan Li, Yicheng Zhang, Wenhao Tang, Yimian Dai, Ming-Ming Cheng, Xiang Li, Jian Yang
Abstract: Modern computer vision is converging on a closed loop in which perception, reasoning and generation mutually reinforce each other. However, this loop remains incomplete: the top-down influence of high-level reasoning on the foundational learning of low-level perceptual features is not yet underexplored. This paper addresses this gap by proposing a new paradigm for pretraining foundation models in downstream domains. We introduce Visual insTruction Pretraining (ViTP), a novel approach that directly leverages reasoning to enhance perception. ViTP embeds a Vision Transformer (ViT) backbone within a Vision-Language Model and pretrains it end-to-end using a rich corpus of visual instruction data curated from target downstream domains. ViTP is powered by our proposed Visual Robustness Learning (VRL), which compels the ViT to learn robust and domain-relevant features from a sparse set of visual tokens. Extensive experiments on 16 challenging remote sensing and medical imaging benchmarks demonstrate that ViTP establishes new state-of-the-art performance across a diverse range of downstream tasks. The code is available at https://github.com/zcablii/ViTP.
Authors: Sehyun Kim, Hye Jun Lee, Jiwoo Lee, Taemin Lee
Abstract: With the development of deep learning technology, virtual try-on technology has become an important application value in the fields of e-commerce, fashion, and entertainment. The recently proposed Leffa has improved the texture distortion problem of diffu-sion-based models, but there are limitations in that the bottom detection inaccuracy and the existing clothing silhouette remain in the synthesis results. To solve this problem, this study proposes CaP-VTON (Clothing agnostic Pre-inpainting Virtual Try-ON). CaP-VTON has improved the naturalness and consistency of whole-body clothing syn-thesis by integrating multi-category masking based on Dress Code and skin inpainting based on Stable Diffusion. In particular, a generate skin module was introduced to solve the skin restoration problem that occurs when long-sleeved images are converted into short-sleeved or sleeveless ones, and high-quality restoration was implemented consider-ing the human body posture and color. As a result, CaP-VTON recorded 92.5%, which is 15.4% better than Leffa in short-sleeved synthesis accuracy, and showed the performance of consistently reproducing the style and shape of reference clothing in visual evaluation. These structures maintain model-agnostic properties and are applicable to various diffu-sion-based virtual inspection systems, and can contribute to applications that require high-precision virtual wearing, such as e-commerce, custom styling, and avatar creation.
Authors: Yikun Ma, Bo Li, Ying Chen, Zijie Yue, Shuchang Xu, Jingyao Li, Lei Ma, Liang Zhong, Duowu Zou, Leiming Xu, Yunshi Zhong, Xiaobo Li, Weiqun Ding, Minmin Zhang, Dongli He, Zhenghong Li, Ye Chen, Ye Zhao, Jialong Zhuo, Xiaofen Wu, Lisha Yi, Miaojing Shi, Huihui Sun
Abstract: The early detection of esophagogastric junction adenocarcinoma (EGJA) is crucial for improving patient prognosis, yet its current diagnosis is highly operator-dependent. This paper aims to make the first attempt to develop an artificial intelligence (AI) foundation model-based method for both screening and staging diagnosis of EGJA using endoscopic images. In this cohort and learning study, we conducted a multicentre study across seven Chinese hospitals between December 28, 2016 and December 30, 2024. It comprises 12,302 images from 1,546 patients; 8,249 of them were employed for model training, while the remaining were divided into the held-out (112 patients, 914 images), external (230 patients, 1,539 images), and prospective (198 patients, 1,600 images) test sets for evaluation. The proposed model employs DINOv2 (a vision foundation model) and ResNet50 (a convolutional neural network) to extract features of global appearance and local details of endoscopic images for EGJA staging diagnosis. Our model demonstrates satisfactory performance for EGJA staging diagnosis across three test sets, achieving an accuracy of 0.9256, 0.8895, and 0.8956, respectively. In contrast, among representative AI models, the best one (ResNet50) achieves an accuracy of 0.9125, 0.8382, and 0.8519 on the three test sets, respectively; the expert endoscopists achieve an accuracy of 0.8147 on the held-out test set. Moreover, with the assistance of our model, the overall accuracy for the trainee, competent, and expert endoscopists improves from 0.7035, 0.7350, and 0.8147 to 0.8497, 0.8521, and 0.8696, respectively. To our knowledge, our model is the first application of foundation models for EGJA staging diagnosis and demonstrates great potential in both diagnostic accuracy and efficiency.
Authors: Chenglin Li, Feng Han, Feng Tao, Ruilin Li, Qianglong Chen, Jingqi Tong, Yin Zhang, Jiaqi Wang
Abstract: Large language models (LLMs) have shown promise in generating program workflows for visual tasks. However, previous approaches often rely on closed-source models, lack systematic reasoning, and struggle with long-form video question answering (videoQA). To address these challenges, we introduce the FS-VisPR framework, an adaptive visual program reasoning approach that balances fast reasoning for simple queries with slow reasoning for difficult ones. First, we design efficient visual modules (e.g., key clip retrieval and subtitle retrieval) to support long-form video tasks. Then, we construct a diverse and high-quality fast-slow reasoning dataset with a strong LLM to align open-source language models' ability to generate visual program workflows as FS-LLM. Next, we design a fast-slow reasoning framework with FS-LLM: Simple queries are directly solved by VideoLLMs, while difficult ones invoke visual program reasoning, motivated by human-like reasoning processes. During this process, low-confidence fast-thinking answers will trigger a second-stage slow-reasoning process, and a fallback mechanism to fast reasoning is activated if the program execution fails. Moreover, we improve visual programs through parameter search during both training and inference. By adjusting the parameters of the visual modules within the program, multiple variants are generated: during training, programs that yield correct answers are selected, while during inference, the program with the highest confidence result is applied. Experiments show that FS-VisPR improves both efficiency and reliability in visual program workflows. It achieves 50.4% accuracy on LVBench, surpassing GPT-4o, matching the performance of Qwen2.5VL-72B on VideoMME.
Authors: Haoxin Yang, Bangzhen Liu, Xuemiao Xu, Cheng Xu, Yuyang Yu, Zikai Huang, Yi Wang, Shengfeng He
Abstract: The advancement of diffusion models has enhanced the realism of AI-generated content but also raised concerns about misuse, necessitating robust copyright protection and tampering localization. Although recent methods have made progress toward unified solutions, their reliance on post hoc processing introduces considerable application inconvenience and compromises forensic reliability. We propose StableGuard, a novel framework that seamlessly integrates a binary watermark into the diffusion generation process, ensuring copyright protection and tampering localization in Latent Diffusion Models through an end-to-end design. We develop a Multiplexing Watermark VAE (MPW-VAE) by equipping a pretrained Variational Autoencoder (VAE) with a lightweight latent residual-based adapter, enabling the generation of paired watermarked and watermark-free images. These pairs, fused via random masks, create a diverse dataset for training a tampering-agnostic forensic network. To further enhance forensic synergy, we introduce a Mixture-of-Experts Guided Forensic Network (MoE-GFN) that dynamically integrates holistic watermark patterns, local tampering traces, and frequency-domain cues for precise watermark verification and tampered region detection. The MPW-VAE and MoE-GFN are jointly optimized in a self-supervised, end-to-end manner, fostering a reciprocal training between watermark embedding and forensic accuracy. Extensive experiments demonstrate that StableGuard consistently outperforms state-of-the-art methods in image fidelity, watermark verification, and tampering localization.
Authors: Brando Miranda, Patrick Yu, Saumya Goyal, Yu-Xiong Wang, Sanmi Koyejo
Abstract: In the context of few-shot learning, it is currently believed that a fixed pre-trained (PT) model, along with fine-tuning the final layer during evaluation, outperforms standard meta-learning algorithms. We re-evaluate these claims under an in-depth empirical examination of an extensive set of formally diverse datasets and compare PT to Model Agnostic Meta-Learning (MAML). Unlike previous work, we emphasize a fair comparison by using: the same architecture, the same optimizer, and all models trained to convergence. Crucially, we use a more rigorous statistical tool -- the effect size (Cohen's d) -- to determine the practical significance of the difference between a model trained with PT vs. a MAML. We then use a previously proposed metric -- the diversity coefficient -- to compute the average formal diversity of a dataset. Using this analysis, we demonstrate the following: 1. when the formal diversity of a data set is low, PT beats MAML on average and 2. when the formal diversity is high, MAML beats PT on average. The caveat is that the magnitude of the average difference between a PT vs. MAML using the effect size is low (according to classical statistical thresholds) -- less than 0.2. Nevertheless, this observation is contrary to the currently held belief that a pre-trained model is always better than a meta-learning model. Our extensive experiments consider 21 few-shot learning benchmarks, including the large-scale few-shot learning dataset Meta-Data set. We also show no significant difference between a MAML model vs. a PT model with GPT-2 on Openwebtext. We, therefore, conclude that a pre-trained model does not always beat a meta-learned model and that the formal diversity of a dataset is a driving factor.
Authors: Cheng Huang, Weizheng Xie, Tsengdar Lee, Karanjit Kooner, Ning Zhang, Jia Zhang
Abstract: Glaucoma is a complex group of eye diseases marked by optic nerve damage, commonly linked to elevated intraocular pressure and biomarkers like retinal nerve fiber layer thickness. Understanding how these biomarkers interact is crucial for unraveling glaucoma's underlying mechanisms. In this paper, we propose GlaLSTM, a novel concurrent LSTM stream framework for glaucoma detection, leveraging latent biomarker relationships. Unlike traditional CNN-based models that primarily detect glaucoma from images, GlaLSTM provides deeper interpretability, revealing the key contributing factors and enhancing model transparency. This approach not only improves detection accuracy but also empowers clinicians with actionable insights, facilitating more informed decision-making. Experimental evaluations confirm that GlaLSTM surpasses existing state-of-the-art methods, demonstrating its potential for both advanced biomarker analysis and reliable glaucoma detection.
Authors: Xidong Wang, Dingjie Song, Shunian Chen, Junyin Chen, Zhenyang Cai, Chen Zhang, Lichao Sun, Benyou Wang
Abstract: Expanding the long-context capabilities of Multi-modal Large Language Models~(MLLMs) is critical for advancing video understanding and high-resolution image analysis. Achieving this requires systematic improvements in model architecture, data construction, and training strategies, particularly to address challenges such as performance degradation with increasing image counts and high computational costs. In this paper, we propose a hybrid architecture that integrates Mamba and Transformer blocks, introduce data construction methods that capture both temporal and spatial dependencies, and employ a progressive training strategy. Our released model, LongLLaVA (\textbf{Long}-Context \textbf{L}arge \textbf{L}anguage \textbf{a}nd \textbf{V}ision \textbf{A}ssistant), demonstrates an effective balance between efficiency and performance. LongLLaVA achieves competitive results across various benchmarks while maintaining high throughput and low memory consumption. Notably, it can process nearly one thousand images on a single A100 80GB GPU, underscoring its potential for a wide range of multi-modal applications.
Authors: Zongbo Han, Jialong Yang, Guangyu Wang, Junfan Li, Qianli Xu, Mike Zheng Shou, Changqing Zhang
Abstract: Vision-language foundation models (VLMs), such as CLIP, exhibit remarkable performance across a wide range of tasks. However, deploying these models can be unreliable when significant distribution gaps exist between training and test data, while fine-tuning for diverse scenarios is often costly. Cache-based test-time adapters offer an efficient alternative by storing representative test samples to guide subsequent classifications. Yet, these methods typically employ naive cache management with limited capacity, leading to severe catastrophic forgetting when samples are inevitably dropped during updates. In this paper, we propose DOTA (DistributiOnal Test-time Adaptation), a simple yet effective method addressing this limitation. Crucially, instead of merely memorizing individual test samples, DOTA continuously estimates the underlying distribution of the test data stream. Test-time posterior probabilities are then computed using these dynamically estimated distributions via Bayes' theorem for adaptation. This distribution-centric approach enables the model to continually learn and adapt to the deployment environment. Extensive experiments validate that DOTA significantly mitigates forgetting and achieves state-of-the-art performance compared to existing methods.
Authors: Yedi Hu, Yunzhi Yao, Ningyu Zhang, Huajun Chen, Shumin Deng
Abstract: Model merging has emerged as a key technique for enhancing the capabilities and efficiency of Large Language Models (LLMs). The open-source community has driven model evolution by iteratively merging existing models, yet a principled understanding of the gains and underlying factors in model merging remains limited. In this work, we study model evolution through iterative merging, drawing an analogy to biological evolution, and introduce the concept of model kinship, the degree of similarity or relatedness between LLMs. Through comprehensive empirical analysis, we show that model kinship is closely linked to the performance improvements achieved by merging, providing a useful criterion for selecting candidate models. Building on this insight, we propose a new model merging strategy: Top-k Greedy Merging with Model Kinship, which can improve benchmark performance. Specifically, we discover that incorporating model kinship as a guiding criterion enables continuous merging while mitigating performance degradation caused by local optima, thereby facilitating more effective model evolution. Code is available at https://github.com/zjunlp/ModelKinship.
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 often fail to generalize these updates to multi-hop reasoning tasks that rely on the modified knowledge. Through an analysis of reasoning circuits -- the neural pathways LLMs use for knowledge-based inference, we find that current layer-localized KE approaches (e.g., MEMIT, WISE), which edit only single or a few model layers, inadequately integrate updated knowledge into these reasoning pathways. To address this limitation, we present CaKE (Circuit-aware Knowledge Editing), a novel method that enhances the effective integration of updated knowledge in LLMs. By only leveraging a few curated data samples guided by our circuit-based analysis, CaKE stimulates the model to develop appropriate reasoning circuits for newly incorporated knowledge. Experiments show that CaKE enables more accurate and consistent use of edited knowledge across related reasoning tasks, achieving an average improvement of 20% in multi-hop reasoning accuracy on the MQuAKE dataset while requiring less memory than existing KE methods. We release the code and data in https://github.com/zjunlp/CaKE.
Authors: Kangwei Liu, Mengru Wang, Yujie Luo, Yuan Lin, Mengshu Sun, Lei Liang, Zhiqiang Zhang, Jun Zhou, Bryan Hooi, Shumin Deng
Abstract: Fine-tuning enables large language models (LLMs) to adapt to specific domains, but often compromises their previously established safety alignment. To mitigate the degradation of model safety during fine-tuning, we introduce LookAhead Tuning, a lightweight and effective data-driven approach that preserves safety during fine-tuning. The method introduces two simple strategies that modify training data by previewing partial answer prefixes, thereby minimizing perturbations to the model's initial token distributions and maintaining its built-in safety mechanisms. Comprehensive experiments demonstrate that LookAhead Tuning effectively maintains model safety without sacrificing robust performance on downstream tasks. Our findings position LookAhead Tuning as a reliable and efficient solution for the safe and effective adaptation of LLMs.
Authors: Sumeyye Meryem Tasyurek, Tugce Kiziltepe, Hacer Yalim Keles
Abstract: In this work, we propose DARSLP, a simple gloss-free, transformer-based sign language production (SLP) framework that directly maps spoken-language text to sign pose sequences. We first train a pose autoencoder that encodes sign poses into a compact latent space using an articulator-based disentanglement strategy, where features corresponding to the face, right hand, left hand, and body are modeled separately to promote structured and interpretable representation learning. Next, a non-autoregressive transformer decoder is trained to predict these latent representations from word-level text embeddings of the input sentence. To guide this process, we apply channel-aware regularization by aligning predicted latent distributions with priors extracted from the ground-truth encodings using a KL divergence loss. The contribution of each channel to the loss is weighted according to its associated articulator region, enabling the model to account for the relative importance of different articulators during training. Our approach does not rely on gloss supervision or pretrained models, and achieves state-of-the-art results on the PHOENIX14T and CSL-Daily datasets.
Authors: DongGeon Lee, Joonwon Jang, Jihae Jeong, Hwanjo Yu
Abstract: Rapid deployment of vision-language models (VLMs) magnifies safety risks, yet most evaluations rely on artificial images. This study asks: How safe are current VLMs when confronted with meme images that ordinary users share? To investigate this question, we introduce MemeSafetyBench, a 50,430-instance benchmark pairing real meme images with both harmful and benign instructions. Using a comprehensive safety taxonomy and LLM-based instruction generation, we assess multiple VLMs across single and multi-turn interactions. We investigate how real-world memes influence harmful outputs, the mitigating effects of conversational context, and the relationship between model scale and safety metrics. Our findings demonstrate that VLMs are more vulnerable to meme-based harmful prompts than to synthetic or typographic images. Memes significantly increase harmful responses and decrease refusals compared to text-only inputs. Though multi-turn interactions provide partial mitigation, elevated vulnerability persists. These results highlight the need for ecologically valid evaluations and stronger safety mechanisms. MemeSafetyBench is publicly available at https://github.com/oneonlee/Meme-Safety-Bench.
Authors: Prateek Verma, Mert Pilanci
Abstract: This paper presents a fascinating find: By training an auto-regressive LLM model on text tokens, the text model inherently develops internally an ability to understand images and audio, thereby developing the ability to see and hear just by reading. Popular audio and visual LLM models fine-tune text LLM models to give text output conditioned on images and audio embeddings. On the other hand, our architecture takes in patches of images, audio waveforms or tokens as input. It gives us the embeddings or category labels typical of a classification pipeline. We show the generality of text weights in aiding audio classification for datasets FSD-50K and GTZAN. Further, we show this working for image classification on CIFAR-10 and Fashion-MNIST, as well on image patches. This pushes the notion of text-LLMs learning powerful internal circuits that can be utilized by activating necessary connections for various applications rather than training models from scratch every single time.
Authors: Luohong Wu, Matthias Seibold, Nicola A. Cavalcanti, Giuseppe Loggia, Lisa Reissner, Bastian Sigrist, Jonas Hein, Lilian Calvet, Arnd Vieh\"ofer, Philipp F\"urnstahl
Abstract: Bone surface reconstruction is an essential component of computer-assisted orthopedic surgery (CAOS), forming the foundation for preoperative planning and intraoperative guidance. Compared to traditional imaging modalities such as CT and MRI, ultrasound provides a radiation-free, and cost-effective alternative. While ultrasound offers new opportunities in CAOS, technical shortcomings continue to hinder its translation into surgery. In particular, due to the inherent limitations of ultrasound imaging, B-mode ultrasound typically capture only partial bone surfaces, posing major challenges for surface reconstruction. Existing reconstruction methods struggle with such incomplete data, leading to increased reconstruction errors and artifacts. Effective techniques for accurately reconstructing open bone surfaces from real-world 3D ultrasound volumes remain lacking. We propose UltraBoneUDF, a self-supervised framework specifically designed for reconstructing open bone surfaces from ultrasound data using neural unsigned distance functions (UDFs). In addition, we present a novel loss function based on local tangent plane optimization that substantially improves surface reconstruction quality. UltraBoneUDF and competing models are benchmarked on three open-source datasets and further evaluated through ablation studies. Results: Qualitative results highlight the limitations of the state-of-the-art methods for open bone surface reconstruction and demonstrate the effectiveness of UltraBoneUDF. Quantitatively, UltraBoneUDF significantly outperforms competing methods across all evaluated datasets for both open and closed bone surface reconstruction in terms of mean Chamfer distance error: 0.96 mm on the UltraBones100k dataset (28.9% improvement compared to the state-of-the-art), 0.21 mm on the OpenBoneCT dataset (40.0% improvement), and 0.18 mm on the ClosedBoneCT dataset (63.3% improvement).
Authors: Muhammad Adnan, Nithesh Kurella, Akhil Arunkumar, Prashant J. Nair
Abstract: Diffusion Transformers (DiTs) achieve state-of-the-art results in text-to-image, text-to-video generation, and editing. However, their large model size and the quadratic cost of spatial-temporal attention over multiple denoising steps make video generation computationally expensive. Static caching mitigates this by reusing features across fixed steps but fails to adapt to generation dynamics, leading to suboptimal trade-offs between speed and quality. We propose Foresight, an adaptive layer-reuse technique that reduces computational redundancy across denoising steps while preserving baseline performance. Foresight dynamically identifies and reuses DiT block outputs for all layers across steps, adapting to generation parameters such as resolution and denoising schedules to optimize efficiency. Applied to OpenSora, Latte, and CogVideoX, Foresight achieves up to \latencyimprv end-to-end speedup, while maintaining video quality. The source code of Foresight is available at \href{https://github.com/STAR-Laboratory/foresight}{https://github.com/STAR-Laboratory/foresight}.
URLs: https://github.com/STAR-Laboratory/foresight, https://github.com/STAR-Laboratory/foresight
Authors: Ho Yin 'Sam' Ng, Ting-Yao Hsu, Aashish Anantha Ramakrishnan, Branislav Kveton, Nedim Lipka, Franck Dernoncourt, Dongwon Lee, Tong Yu, Sungchul Kim, Ryan A. Rossi, Ting-Hao 'Kenneth' Huang
Abstract: Figure captions are crucial for helping readers understand and remember a figure's key message. Many models have been developed to generate these captions, helping authors compose better quality captions more easily. Yet, authors almost always need to revise generic AI-generated captions to match their writing style and the domain's style, highlighting the need for personalization. Despite language models' personalization (LaMP) advances, these technologies often focus on text-only settings and rarely address scenarios where both inputs and profiles are multimodal. This paper introduces LaMP-Cap, a dataset for personalized figure caption generation with multimodal figure profiles. For each target figure, LaMP-Cap provides not only the needed inputs, such as figure images, but also up to three other figures from the same document--each with its image, caption, and figure-mentioning paragraphs--as a profile to characterize the context. Experiments with four LLMs show that using profile information consistently helps generate captions closer to the original author-written ones. Ablation studies reveal that images in the profile are more helpful than figure-mentioning paragraphs, highlighting the advantage of using multimodal profiles over text-only ones.
Authors: Shuai Wang, Zhenhua Liu, Jiaheng Wei, Xuanwu Yin, Dong Li, Emad Barsoum
Abstract: We present Athena-PRM, a multimodal process reward model (PRM) designed to evaluate the reward score for each step in solving complex reasoning problems. Developing high-performance PRMs typically demands significant time and financial investment, primarily due to the necessity for step-level annotations of reasoning steps. Conventional automated labeling methods, such as Monte Carlo estimation, often produce noisy labels and incur substantial computational costs. To efficiently generate high-quality process-labeled data, we propose leveraging prediction consistency between weak and strong completers as a criterion for identifying reliable process labels. Remarkably, Athena-PRM demonstrates outstanding effectiveness across various scenarios and benchmarks with just 5,000 samples. Furthermore, we also develop two effective strategies to improve the performance of PRMs: ORM initialization and up-sampling for negative data. We validate our approach in three specific scenarios: verification for test time scaling, direct evaluation of reasoning step correctness, and reward ranked fine-tuning. Our Athena-PRM consistently achieves superior performance across multiple benchmarks and scenarios. Notably, when using Qwen2.5-VL-7B as the policy model, Athena-PRM enhances performance by 10.2 points on WeMath and 7.1 points on MathVista for test time scaling. Furthermore, Athena-PRM sets the state-of-the-art (SoTA) results in VisualProcessBench and outperforms the previous SoTA by 3.9 F1-score, showcasing its robust capability to accurately assess the correctness of the reasoning step. Additionally, utilizing Athena-PRM as the reward model, we develop Athena-7B with reward ranked fine-tuning and outperforms baseline with a significant margin on five benchmarks.
Authors: Shuning Jiang, Wei-Lun Chao, Daniel Haehn, Hanspeter Pfister, Jian Chen
Abstract: We present a data-domain sampling regime for quantifying CNNs' graphic perception behaviors. This regime lets us evaluate CNNs' ratio estimation ability in bar charts from three perspectives: sensitivity to training-test distribution discrepancies, stability to limited samples, and relative expertise to human observers. After analyzing 16 million trials from 800 CNNs models and 6,825 trials from 113 human participants, we arrived at a simple and actionable conclusion: CNNs can outperform humans and their biases simply depend on the training-test distance. We show evidence of this simple, elegant behavior of the machines when they interpret visualization images. osf.io/gfqc3 provides registration, the code for our sampling regime, and experimental results.
Authors: Jieren Deng, Zhizhang Hu, Ziyan He, Aleksandar Cvetkovic, Pak Kiu Chung, Dragomir Yankov, Chiqun Zhang
Abstract: Map applications are still largely point-and-click, making it difficult to ask map-centric questions or connect what a camera sees to the surrounding geospatial context with view-conditioned inputs. We introduce IMAIA, an interactive Maps AI Assistant that enables natural-language interaction with both vector (street) maps and satellite imagery, and augments camera inputs with geospatial intelligence to help users understand the world. IMAIA comprises two complementary components. Maps Plus treats the map as first-class context by parsing tiled vector/satellite views into a grid-aligned representation that a language model can query to resolve deictic references (e.g., ``the flower-shaped building next to the park in the top-right''). Places AI Smart Assistant (PAISA) performs camera-aware place understanding by fusing image--place embeddings with geospatial signals (location, heading, proximity) to ground a scene, surface salient attributes, and generate concise explanations. A lightweight multi-agent design keeps latency low and exposes interpretable intermediate decisions. Across map-centric QA and camera-to-place grounding tasks, IMAIA improves accuracy and responsiveness over strong baselines while remaining practical for user-facing deployments. By unifying language, maps, and geospatial cues, IMAIA moves beyond scripted tools toward conversational mapping that is both spatially grounded and broadly usable.
Authors: Zhuang Qi, Ying-Peng Tang, Lei Meng, Han Yu, Xiaoxiao Li, Xiangxu Meng
Abstract: Federated Class Incremental Learning (FCIL) aims to collaboratively process continuously increasing incoming tasks across multiple clients. Among various approaches, data replay has become a promising solution, which can alleviate forgetting by reintroducing representative samples from previous tasks. However, their performance is typically limited by class imbalance, both within the replay buffer due to limited global awareness and between replayed and newly arrived classes. To address this issue, we propose a class wise balancing data replay method for FCIL (FedCBDR), which employs a global coordination mechanism for class-level memory construction and reweights the learning objective to alleviate the aforementioned imbalances. Specifically, FedCBDR has two key components: 1) the global-perspective data replay module reconstructs global representations of prior task in a privacy-preserving manner, which then guides a class-aware and importance-sensitive sampling strategy to achieve balanced replay; 2) Subsequently, to handle class imbalance across tasks, the task aware temperature scaling module adaptively adjusts the temperature of logits at both class and instance levels based on task dynamics, which reduces the model's overconfidence in majority classes while enhancing its sensitivity to minority classes. Experimental results verified that FedCBDR achieves balanced class-wise sampling under heterogeneous data distributions and improves generalization under task imbalance between earlier and recent tasks, yielding a 2%-15% Top-1 accuracy improvement over six state-of-the-art methods.
Authors: Santosh Thoduka, Sebastian Houben, Juergen Gall, Paul G. Pl\"oger
Abstract: Robust robotic task execution hinges on the reliable detection of execution failures in order to trigger safe operation modes, recovery strategies, or task replanning. However, many failure detection methods struggle to provide meaningful performance when applied to a variety of real-world scenarios. In this paper, we propose a video-based failure detection approach that uses spatio-temporal knowledge in the form of the actions the robot performs and task-relevant objects within the field of view. Both pieces of information are available in most robotic scenarios and can thus be readily obtained. We demonstrate the effectiveness of our approach on three datasets that we amend, in part, with additional annotations of the aforementioned task-relevant knowledge. In light of the results, we also propose a data augmentation method that improves performance by applying variable frame rates to different parts of the video. We observe an improvement from 77.9 to 80.0 in F1 score on the ARMBench dataset without additional computational expense and an additional increase to 81.4 with test-time augmentation. The results emphasize the importance of spatio-temporal information during failure detection and suggest further investigation of suitable heuristics in future implementations. Code and annotations are available.
Authors: Huaiyu Chen, Fahed Hassanat, Robert Laganiere, Martin Bouchard
Abstract: Frequency-modulated continuous wave radars have gained increasing popularity in the automotive industry. Its robustness against adverse weather conditions makes it a suitable choice for radar object detection in advanced driver assistance systems. These real-time embedded systems have requirements for the compactness and efficiency of the model, which have been largely overlooked in previous work. In this work, we propose mRadNet, a novel radar object detection model with compactness in mind. mRadNet employs a U-net style architecture with MetaFormer blocks, in which separable convolution and attention token mixers are used to capture both local and global features effectively. More efficient token embedding and merging strategies are introduced to further facilitate the lightweight design. The performance of mRadNet is validated on the CRUW dataset, improving state-of-the-art performance with the least number of parameters and FLOPs.
Authors: Qishuai Wen, Zhiyuan Huang, Chun-Guang Li
Abstract: Attention mechanisms in Transformers have gained significant empirical success. Nonetheless, the optimization objectives underlying their forward pass are still unclear. Additionally, the quadratic complexity of self-attention is increasingly prohibitive. Unlike the prior work on addressing the interpretability or efficiency issue separately, we propose a unified optimization objective to alleviate both issues simultaneously. By unrolling the optimization over the objective, we derive an inherently interpretable and efficient attention mechanism, which compresses all tokens into low-dimensional structures by contracting a few representative tokens and then broadcasting the contractions back. This Contract-and-Broadcast Self-Attention (CBSA) mechanism can not only scale linearly but also generalize existing attention mechanisms as its special cases. Experiments further demonstrate comparable performance and even superior advantages of CBSA on several visual tasks. Code is available at this https URL.
Authors: Haojun Yu, Youcheng Li, Zihan Niu, Nan Zhang, Xuantong Gong, Huan Li, Zhiying Zou, Haifeng Qi, Zhenxiao Cao, Zijie Lan, Xingjian Yuan, Jiating He, Haokai Zhang, Shengtao Zhang, Zicheng Wang, Dong Wang, Ziwei Zhao, Congying Chen, Yong Wang, Wangyan Qin, Qingli Zhu, Liwei Wang
Abstract: Breast ultrasound (BUS) is an essential tool for diagnosing breast lesions, with millions of examinations per year. However, publicly available high-quality BUS benchmarks for AI development are limited in data scale and annotation richness. In this work, we present BUS-CoT, a BUS dataset for chain-of-thought (CoT) reasoning analysis, which contains 11,439 images of 10,019 lesions from 4,838 patients and covers all 99 histopathology types. To facilitate research on incentivizing CoT reasoning, we construct the reasoning processes based on observation, feature, diagnosis and pathology labels, annotated and verified by experienced experts. Moreover, by covering lesions of all histopathology types, we aim to facilitate robust AI systems in rare cases, which can be error-prone in clinical practice.
Authors: Yunchu Han, Zhaojun Nan, Sheng Zhou, Zhisheng Niu
Abstract: Deep neural networks (DNNs) have been widely applied in diverse applications, but the problems of high latency and energy overhead are inevitable on resource-constrained devices. To address this challenge, most researchers focus on the dynamic voltage and frequency scaling (DVFS) technique to balance the latency and energy consumption by changing the computing frequency of processors. However, the adjustment of memory frequency is usually ignored and not fully utilized to achieve efficient DNN inference, which also plays a significant role in the inference time and energy consumption. In this paper, we first investigate the impact of joint memory frequency and computing frequency scaling on the inference time and energy consumption with a model-based and data-driven method. Then by combining with the fitting parameters of different DNN models, we give a preliminary analysis for the proposed model to see the effects of adjusting memory frequency and computing frequency simultaneously. Finally, simulation results in local inference and cooperative inference cases further validate the effectiveness of jointly scaling the memory frequency and computing frequency to reduce the energy consumption of devices.