Authors: Thomas Foltz
Abstract: This paper focuses on detecting anomalies in surveillance video using keywords by leveraging foundational models' feature representation generalization capabilities. We present a novel, lightweight pipeline for anomaly classification using keyword weights. Our pipeline employs a two-stage process: induction followed by deduction. In induction, descriptions are generated from normal and anomalous frames to identify and assign weights to relevant keywords. In deduction, inference frame descriptions are converted into keyword encodings using induction-derived weights for input into our neural network for anomaly classification. We achieved comparable performance on the three benchmarks UCSD Ped2, Shanghai Tech, and CUHK Avenue, with ROC AUC scores of 0.865, 0.745, and 0.742, respectively. These results are achieved without temporal context, making such a system viable for real-time applications. Our model improves implementation setup, interpretability, and inference speed for surveillance devices on the edge, introducing a performance trade-off against other video anomaly detection systems. As the generalization capabilities of open-source foundational models improve, our model demonstrates that the exclusive use of text for feature representations is a promising direction for efficient real-time interpretable video anomaly detection.
Authors: Khoi Do, Binh-Son Hua
Abstract: Score distillation sampling is an effective technique to generate 3D models from text prompts, utilizing pre-trained large-scale text-to-image diffusion models as guidance. However, the produced 3D assets tend to be over-saturating, over-smoothing, with limited diversity. These issues are results from a reverse Kullback-Leibler (KL) divergence objective, which makes the optimization unstable and results in mode-seeking behavior. In this paper, we derive a bounded score distillation objective based on Jensen-Shannon divergence (JSD), which stabilizes the optimization process and produces high-quality 3D generation. JSD can match well generated and target distribution, therefore mitigating mode seeking. We provide a practical implementation of JSD by utilizing the theory of generative adversarial networks to define an approximate objective function for the generator, assuming the discriminator is well trained. By assuming the discriminator following a log-odds classifier, we propose a minority sampling algorithm to estimate the gradients of our proposed objective, providing a practical implementation for JSD. We conduct both theoretical and empirical studies to validate our method. Experimental results on T3Bench demonstrate that our method can produce high-quality and diversified 3D assets.
Authors: Xiangyu Yin, Jiaxu Liu, Zhen Chen, Jinwei Hu, Yi Dong, Xiaowei Huang, Wenjie Ruan
Abstract: Recent advances in large vision-language models (VLMs) have demonstrated remarkable success across a wide range of visual understanding tasks. However, the robustness of these models against jailbreak attacks remains an open challenge. In this work, we propose a universal certified defence framework to safeguard VLMs rigorously against potential visual jailbreak attacks. First, we proposed a novel distance metric to quantify semantic discrepancies between malicious and intended responses, capturing subtle differences often overlooked by conventional cosine similarity-based measures. Then, we devise a regressed certification approach that employs randomized smoothing to provide formal robustness guarantees against both adversarial and structural perturbations, even under black-box settings. Complementing this, our feature-space defence introduces noise distributions (e.g., Gaussian, Laplacian) into the latent embeddings to safeguard against both pixel-level and structure-level perturbations. Our results highlight the potential of a formally grounded, integrated strategy toward building more resilient and trustworthy VLMs.
Authors: Nitesh Patnaik, Navdeep Nayak, Himani Bansal Agrawal, Moinak Chinmoy Khamaru, Gourav Bal, Saishree Smaranika Panda, Rishi Raj, Vishal Meena, Kartheek Vadlamani
Abstract: The emergence of small vision-language models (sVLMs) marks a critical advancement in multimodal AI, enabling efficient processing of visual and textual data in resource-constrained environments. This survey offers a comprehensive exploration of sVLM development, presenting a taxonomy of architectures - transformer-based, mamba-based, and hybrid - that highlight innovations in compact design and computational efficiency. Techniques such as knowledge distillation, lightweight attention mechanisms, and modality pre-fusion are discussed as enablers of high performance with reduced resource requirements. Through an in-depth analysis of models like TinyGPT-V, MiniGPT-4, and VL-Mamba, we identify trade-offs between accuracy, efficiency, and scalability. Persistent challenges, including data biases and generalization to complex tasks, are critically examined, with proposed pathways for addressing them. By consolidating advancements in sVLMs, this work underscores their transformative potential for accessible AI, setting a foundation for future research into efficient multimodal systems.
Authors: Lehan Yang, Jincen Song, Tianlong Wang, Daiqing Qi, Weili Shi, Yuheng Liu, Sheng Li
Abstract: We propose a new task, video referring matting, which obtains the alpha matte of a specified instance by inputting a referring caption. We treat the dense prediction task of matting as video generation, leveraging the text-to-video alignment prior of video diffusion models to generate alpha mattes that are temporally coherent and closely related to the corresponding semantic instances. Moreover, we propose a new Latent-Constructive loss to further distinguish different instances, enabling more controllable interactive matting. Additionally, we introduce a large-scale video referring matting dataset with 10,000 videos. To the best of our knowledge, this is the first dataset that concurrently contains captions, videos, and instance-level alpha mattes. Extensive experiments demonstrate the effectiveness of our method. The dataset and code are available at https://github.com/Hansxsourse/VRMDiff.
Authors: Jingwen Deng, Zihao Wang, Shaofei Cai, Anji Liu, Yitao Liang
Abstract: Learning skills in open-world environments is essential for developing agents capable of handling a variety of tasks by combining basic skills. Online demonstration videos are typically long but unsegmented, making them difficult to segment and label with skill identifiers. Unlike existing methods that rely on sequence sampling or human labeling, we have developed a self-supervised learning-based approach to segment these long videos into a series of semantic-aware and skill-consistent segments. Drawing inspiration from human cognitive event segmentation theory, we introduce Skill Boundary Detection (SBD), an annotation-free temporal video segmentation algorithm. SBD detects skill boundaries in a video by leveraging prediction errors from a pretrained unconditional action-prediction model. This approach is based on the assumption that a significant increase in prediction error indicates a shift in the skill being executed. We evaluated our method in Minecraft, a rich open-world simulator with extensive gameplay videos available online. Our SBD-generated segments improved the average performance of conditioned policies by 63.7% and 52.1% on short-term atomic skill tasks, and their corresponding hierarchical agents by 11.3% and 20.8% on long-horizon tasks. Our method can leverage the diverse YouTube videos to train instruction-following agents. The project page can be found in https://craftjarvis.github.io/SkillDiscovery.
Authors: Brun\'o B. Englert, Gijs Dubbelman
Abstract: Unsupervised Domain Adaptation (UDA) has shown remarkably strong generalization from a labeled source domain to an unlabeled target domain while requiring relatively little data. At the same time, large-scale pretraining without labels of so-called Vision Foundation Models (VFMs), has also significantly improved downstream generalization. This motivates us to research how UDA can best utilize the benefits of VFMs. The earlier work of VFM-UDA showed that beyond state-of-the-art (SotA) results can be obtained by replacing non-VFM with VFM encoders in SotA UDA methods. In this work, we take it one step further and improve on the UDA architecture and data strategy themselves. We observe that VFM-UDA, the current SotA UDA method, does not use multi-scale inductive biases or feature distillation losses, while it is known that these can improve generalization. We address both limitations in VFM-UDA++ and obtain beyond SotA generalization on standard UDA benchmarks of up to +5.3 mIoU. Inspired by work on VFM fine-tuning, such as Rein, we also explore the benefits of adding more easy-to-generate synthetic source data with easy-to-obtain unlabeled target data and realize a +6.6 mIoU over the current SotA. The improvements of VFM-UDA++ are most significant for smaller models, however, we show that for larger models, the obtained generalization is only 2.8 mIoU from that of fully-supervised learning with all target labels. Based on these strong results, we provide essential insights to help researchers and practitioners advance UDA.
Authors: Anzhe Cheng, Chenzhong Yin, Yu Chang, Heng Ping, Shixuan Li, Shahin Nazarian, Paul Bogdan
Abstract: Low-resolution image segmentation is crucial in real-world applications such as robotics, augmented reality, and large-scale scene understanding, where high-resolution data is often unavailable due to computational constraints. To address this challenge, we propose MaskAttn-UNet, a novel segmentation framework that enhances the traditional U-Net architecture via a mask attention mechanism. Our model selectively emphasizes important regions while suppressing irrelevant backgrounds, thereby improving segmentation accuracy in cluttered and complex scenes. Unlike conventional U-Net variants, MaskAttn-UNet effectively balances local feature extraction with broader contextual awareness, making it particularly well-suited for low-resolution inputs. We evaluate our approach on three benchmark datasets with input images rescaled to 128x128 and demonstrate competitive performance across semantic, instance, and panoptic segmentation tasks. Our results show that MaskAttn-UNet achieves accuracy comparable to state-of-the-art methods at significantly lower computational cost than transformer-based models, making it an efficient and scalable solution for low-resolution segmentation in resource-constrained scenarios.
Authors: Khawar Islam, Naveed Akhtar
Abstract: Generative diffusion models offer a natural choice for data augmentation when training complex vision models. However, ensuring reliability of their generative content as augmentation samples remains an open challenge. Despite a number of techniques utilizing generative images to strengthen model training, it remains unclear how to utilize the combination of natural and generative images as a rich supervisory signal for effective model induction. In this regard, we propose a text-to-image (T2I) data augmentation method, named DiffCoRe-Mix, that computes a set of generative counterparts for a training sample with an explicitly constrained diffusion model that leverages sample-based context and negative prompting for a reliable augmentation sample generation. To preserve key semantic axes, we also filter out undesired generative samples in our augmentation process. To that end, we propose a hard-cosine filtration in the embedding space of CLIP. Our approach systematically mixes the natural and generative images at pixel and patch levels. We extensively evaluate our technique on ImageNet-1K,Tiny ImageNet-200, CIFAR-100, Flowers102, CUB-Birds, Stanford Cars, and Caltech datasets, demonstrating a notable increase in performance across the board, achieving up to $\sim 3\%$ absolute gain for top-1 accuracy over the state-of-the-art methods, while showing comparable computational overhead. Our code is publicly available at https://github.com/khawar-islam/DiffCoRe-Mix
Authors: Qiji Zhou, Yifan Gong, Guangsheng Bao, Hongjie Qiu, Jinqiang Li, Xiangrong Zhu, Huajian Zhang, Yue Zhang
Abstract: Counterfactual reasoning is crucial for robust video understanding but remains underexplored in existing multimodal benchmarks. In this paper, we introduce \textbf{COVER} (\textbf{\underline{CO}}unterfactual \textbf{\underline{V}}id\textbf{\underline{E}}o \textbf{\underline{R}}easoning), a multidimensional multimodal benchmark that systematically evaluates MLLMs across the abstract-concrete and perception-cognition dimensions. Beyond prior multimodal benchmarks, COVER decomposes complex queries into structured sub-questions, enabling fine-grained reasoning analysis. Experiments on commercial and open-source models reveal a strong correlation between sub-question accuracy and counterfactual reasoning performance, highlighting the role of structured inference in video understanding. Furthermore, our results suggest a key insight: enhancing the reasoning capability of models is essential for improving the robustness of video understanding. COVER establishes a new standard for assessing MLLMs' logical reasoning abilities in dynamic environments.
Authors: Yibin Ye, Xichao Teng, Shuo Chen, Zhang Li, Leqi Liu, Qifeng Yu, Tao Tan
Abstract: Absolute Visual Localization (AVL) enables Unmanned Aerial Vehicle (UAV) to determine its position in GNSS-denied environments by establishing geometric relationships between UAV images and geo-tagged reference maps. While many previous works have achieved AVL with image retrieval and matching techniques, research in low-altitude multi-view scenarios still remains limited. Low-altitude Multi-view condition presents greater challenges due to extreme viewpoint changes. To explore the best UAV AVL approach in such condition, we proposed this benchmark. Firstly, a large-scale Low-altitude Multi-view dataset called AnyVisLoc was constructed. This dataset includes 18,000 images captured at multiple scenes and altitudes, along with 2.5D reference maps containing aerial photogrammetry maps and historical satellite maps. Secondly, a unified framework was proposed to integrate the state-of-the-art AVL approaches and comprehensively test their performance. The best combined method was chosen as the baseline and the key factors that influencing localization accuracy are thoroughly analyzed based on it. This baseline achieved a 74.1% localization accuracy within 5m under Low-altitude, Multi-view conditions. In addition, a novel retrieval metric called PDM@K was introduced to better align with the characteristics of the UAV AVL task. Overall, this benchmark revealed the challenges of Low-altitude, Multi-view UAV AVL and provided valuable guidance for future research. The dataset and codes are available at https://github.com/UAV-AVL/Benchmark
Authors: Thuan Than, Nhat-Anh Nguyen-Dang, Dung Nguyen, Salwa K. Al Khatib, Ahmed Elhagry, Hai Phan, Yihui He, Zhiqiang Shen, Marios Savvides, Dang Huynh
Abstract: Semi-Supervised Semantic Segmentation reduces reliance on extensive annotations by using unlabeled data and state-of-the-art models to improve overall performance. Despite the success of deep co-training methods, their underlying mechanisms remain underexplored. This work revisits Cross Pseudo Supervision with dual heterogeneous backbones and introduces Knowledge Consultation (SegKC) to further enhance segmentation performance. The proposed SegKC achieves significant improvements on Pascal and Cityscapes benchmarks, with mIoU scores of 87.1%, 89.2%, and 89.8% on Pascal VOC with the 1/4, 1/2, and full split partition, respectively, while maintaining a compact model architecture.
Authors: Yefei He, Yuanyu He, Shaoxuan He, Feng Chen, Hong Zhou, Kaipeng Zhang, Bohan Zhuang
Abstract: Visual autoregressive models typically adhere to a raster-order ``next-token prediction" paradigm, which overlooks the spatial and temporal locality inherent in visual content. Specifically, visual tokens exhibit significantly stronger correlations with their spatially or temporally adjacent tokens compared to those that are distant. In this paper, we propose Neighboring Autoregressive Modeling (NAR), a novel paradigm that formulates autoregressive visual generation as a progressive outpainting procedure, following a near-to-far ``next-neighbor prediction" mechanism. Starting from an initial token, the remaining tokens are decoded in ascending order of their Manhattan distance from the initial token in the spatial-temporal space, progressively expanding the boundary of the decoded region. To enable parallel prediction of multiple adjacent tokens in the spatial-temporal space, we introduce a set of dimension-oriented decoding heads, each predicting the next token along a mutually orthogonal dimension. During inference, all tokens adjacent to the decoded tokens are processed in parallel, substantially reducing the model forward steps for generation. Experiments on ImageNet$256\times 256$ and UCF101 demonstrate that NAR achieves 2.4$\times$ and 8.6$\times$ higher throughput respectively, while obtaining superior FID/FVD scores for both image and video generation tasks compared to the PAR-4X approach. When evaluating on text-to-image generation benchmark GenEval, NAR with 0.8B parameters outperforms Chameleon-7B while using merely 0.4 of the training data. Code is available at https://github.com/ThisisBillhe/NAR.
Authors: Kaifeng Zou, Xiaoyi Feng, Peng Wang, Tao Huang, Zizhou Huang, Zhang Haihang, Yuntao Zou, Dagang Li
Abstract: Generative models are widely used in visual content creation. However, current text-to-image models often face challenges in practical applications-such as textile pattern design and meme generation-due to the presence of unwanted elements that are difficult to separate with existing methods. Meanwhile, subject-reference generation has emerged as a key research trend, highlighting the need for techniques that can produce clean, high-quality subject images while effectively removing extraneous components. To address this challenge, we introduce a framework for reliable subject-centric image generation. In this work, we propose an entropy-based feature-weighted fusion method to merge the informative cross-attention features obtained from each sampling step of the pretrained text-to-image model FLUX, enabling a precise mask prediction and subject-centric generation. Additionally, we have developed an agent framework based on Large Language Models (LLMs) that translates users' casual inputs into more descriptive prompts, leading to highly detailed image generation. Simultaneously, the agents extract primary elements of prompts to guide the entropy-based feature fusion, ensuring focused primary element generation without extraneous components. Experimental results and user studies demonstrate our methods generates high-quality subject-centric images, outperform existing methods or other possible pipelines, highlighting the effectiveness of our approach.
Authors: Fan Lyu, Tianle Liu, Zhang Zhang, Fuyuan Hu, Liang Wang
Abstract: We introduce Test-Time Discovery (TTD) as a novel task that addresses class shifts during testing, requiring models to simultaneously identify emerging categories while preserving previously learned ones. A key challenge in TTD is distinguishing newly discovered classes from those already identified. To address this, we propose a training-free, hash-based memory mechanism that enhances class discovery through fine-grained comparisons with past test samples. Leveraging the characteristics of unknown classes, our approach introduces hash representation based on feature scale and directions, utilizing Locality-Sensitive Hashing (LSH) for efficient grouping of similar samples. This enables test samples to be easily and quickly compared with relevant past instances. Furthermore, we design a collaborative classification strategy, combining a prototype classifier for known classes with an LSH-based classifier for novel ones. To enhance reliability, we incorporate a self-correction mechanism that refines memory labels through hash-based neighbor retrieval, ensuring more stable and accurate class assignments. Experimental results demonstrate that our method achieves good discovery of novel categories while maintaining performance on known classes, establishing a new paradigm in model testing. Our code is available at https://github.com/fanlyu/ttd.
Authors: Yuhuan You, Xihong Wu, Tianshu Qu
Abstract: As artificial intelligence-generated content (AIGC) continues to evolve, video-to-audio (V2A) generation has emerged as a key area with promising applications in multimedia editing, augmented reality, and automated content creation. While Transformer and Diffusion models have advanced audio generation, a significant challenge persists in extracting precise semantic information from videos, as current models often lose sequential context by relying solely on frame-based features. To address this, we present TA-V2A, a method that integrates language, audio, and video features to improve semantic representation in latent space. By incorporating large language models for enhanced video comprehension, our approach leverages text guidance to enrich semantic expression. Our diffusion model-based system utilizes automated text modulation to enhance inference quality and efficiency, providing personalized control through text-guided interfaces. This integration enhances semantic expression while ensuring temporal alignment, leading to more accurate and coherent video-to-audio generation.
Authors: Yaowu Fan, Jia Wan, Tao Han, Antoni B. Chan, Andy J. Ma
Abstract: Video Individual Counting (VIC) has received increasing attentions recently due to its importance in intelligent video surveillance. Existing works are limited in two aspects, i.e., dataset and method. Previous crowd counting datasets are captured with fixed or rarely moving cameras with relatively sparse individuals, restricting evaluation for a highly varying view and time in crowded scenes. While VIC methods have been proposed based on localization-then-association or localization-then-classification, they may not perform well due to difficulty in accurate localization of crowded and small targets under challenging scenarios. To address these issues, we collect a MovingDroneCrowd Dataset and propose a density map based VIC method. Different from existing datasets, our dataset consists of videos captured by fast-moving drones in crowded scenes under diverse illuminations, shooting heights and angles. Other than localizing individuals, we propose a Depth-wise Cross-Frame Attention (DCFA) module, which directly estimate inflow and outflow density maps through learning shared density maps between consecutive frames. The inflow density maps across frames are summed up to obtain the number of unique pedestrians in a video. Experiments on our datasets and publicly available ones show the superiority of our method over the state of the arts for VIC in highly dynamic and complex crowded scenes. Our dataset and codes will be released publicly.
Authors: Jing Wang, Fengzhuo Zhang, Xiaoli Li, Vincent Y. F. Tan, Tianyu Pang, Chao Du, Aixin Sun, Zhuoran Yang
Abstract: A variety of Auto-Regressive Video Diffusion Models (ARVDM) have achieved remarkable successes in generating realistic long-form videos. However, theoretical analyses of these models remain scant. In this work, we develop theoretical underpinnings for these models and use our insights to improve the performance of existing models. We first develop Meta-ARVDM, a unified framework of ARVDMs that subsumes most existing methods. Using Meta-ARVDM, we analyze the KL-divergence between the videos generated by Meta-ARVDM and the true videos. Our analysis uncovers two important phenomena inherent to ARVDM -- error accumulation and memory bottleneck. By deriving an information-theoretic impossibility result, we show that the memory bottleneck phenomenon cannot be avoided. To mitigate the memory bottleneck, we design various network structures to explicitly use more past frames. We also achieve a significantly improved trade-off between the mitigation of the memory bottleneck and the inference efficiency by compressing the frames. Experimental results on DMLab and Minecraft validate the efficacy of our methods. Our experiments also demonstrate a Pareto-frontier between the error accumulation and memory bottleneck across different methods.
Authors: Haoyuan Gao, Zicong Zhang, Yuqi Wei, Linglan Zhao, Guilin Li, Yexin Li, Linghe Kong, Weiran Huang
Abstract: Vision-Language Models (VLMs) represent a breakthrough in artificial intelligence by integrating visual and textual modalities to achieve impressive zero-shot capabilities. However, VLMs are susceptible to catastrophic forgetting when sequentially fine-tuned on multiple downstream tasks. Existing continual learning methods for VLMs often rely heavily on additional reference datasets, compromise zero-shot performance, or are limited to parameter-efficient fine-tuning scenarios. In this paper, we propose Continual Decoupling-Unifying (ConDU), a novel approach, by introducing model fusion into continual learning for VLMs. ConDU maintains a unified model along with task triggers and prototype sets, employing an iterative process of decoupling task-specific models for previous tasks and unifying them with the model for the newly learned task. Additionally, we introduce an inference strategy for zero-shot scenarios by aggregating predictions from multiple decoupled task-specific models. Extensive experiments across various settings show that ConDU achieves up to a 2\% improvement in average performance across all seen tasks compared to state-of-the-art baselines, while also enhancing zero-shot capabilities relative to the original VLM.
Authors: Minghao Yang, Zhi-An Huang, Zhihang Zheng, Yuqiao Liu, Shichen Zhang, Pengfei Zhang, Hui Xiong, Shaojun Tang
Abstract: Hi-C technology measures genome-wide interaction frequencies, providing a powerful tool for studying the 3D genomic structure within the nucleus. However, high sequencing costs and technical challenges often result in Hi-C data with limited coverage, leading to imprecise estimates of chromatin interaction frequencies. To address this issue, we present a novel deep learning-based method HiCMamba to enhance the resolution of Hi-C contact maps using a state space model. We adopt the UNet-based auto-encoder architecture to stack the proposed holistic scan block, enabling the perception of both global and local receptive fields at multiple scales. Experimental results demonstrate that HiCMamba outperforms state-of-the-art methods while significantly reducing computational resources. Furthermore, the 3D genome structures, including topologically associating domains (TADs) and loops, identified in the contact maps recovered by HiCMamba are validated through associated epigenomic features. Our work demonstrates the potential of a state space model as foundational frameworks in the field of Hi-C resolution enhancement.
Authors: Tsan-Tsung Yang, I-Wei Chen, Kuan-Ting Chen, Shang-Hsuan Chiang, Wen-Chih Peng
Abstract: With the rapid advancement of generative AI, AI-generated images have become increasingly realistic, raising concerns about creativity, misinformation, and content authenticity. Detecting such images and identifying their source models has become a critical challenge in ensuring the integrity of digital media. This paper tackles the detection of AI-generated images and identifying their source models using CNN and CLIP-ViT classifiers. For the CNN-based classifier, we leverage EfficientNet-B0 as the backbone and feed with RGB channels, frequency features, and reconstruction errors, while for CLIP-ViT, we adopt a pretrained CLIP image encoder to extract image features and SVM to perform classification. Evaluated on the Defactify 4 dataset, our methods demonstrate strong performance in both tasks, with CLIP-ViT showing superior robustness to image perturbations. Compared to baselines like AEROBLADE and OCC-CLIP, our approach achieves competitive results. Notably, our method ranked Top-3 overall in the Defactify 4 competition, highlighting its effectiveness and generalizability. All of our implementations can be found in https://github.com/uuugaga/Defactify_4
Authors: Yehang Zhang, Xinli Xu, Xiaojie Xu, Li Liu, Yingcong Chen
Abstract: Video-to-audio synthesis, which generates synchronized audio for visual content, critically enhances viewer immersion and narrative coherence in film and interactive media. However, video-to-audio dubbing for long-form content remains an unsolved challenge due to dynamic semantic shifts, temporal misalignment, and the absence of dedicated datasets. While existing methods excel in short videos, they falter in long scenarios (e.g., movies) due to fragmented synthesis and inadequate cross-scene consistency. We propose LVAS-Agent, a novel multi-agent framework that emulates professional dubbing workflows through collaborative role specialization. Our approach decomposes long-video synthesis into four steps including scene segmentation, script generation, sound design and audio synthesis. Central innovations include a discussion-correction mechanism for scene/script refinement and a generation-retrieval loop for temporal-semantic alignment. To enable systematic evaluation, we introduce LVAS-Bench, the first benchmark with 207 professionally curated long videos spanning diverse scenarios. Experiments demonstrate superior audio-visual alignment over baseline methods.
Authors: Longfei Han, Klaus Kefferp\"utz, J\"urgen Beyerer
Abstract: Object tracking is an essential task for autonomous systems. With the advancement of 3D sensors, these systems can better perceive their surroundings using effective 3D Extended Object Tracking (EOT) methods. Based on the observation that common road users are symmetrical on the right and left sides in the traveling direction, we focus on the side view profile of the object. In order to leverage of the development in 2D EOT and balance the number of parameters of a shape model in the tracking algorithms, we propose a method for 3D extended object tracking (EOT) by describing the side view profile of the object with B-spline curves and forming an extrusion to obtain a 3D extent. The use of B-spline curves exploits their flexible representation power by allowing the control points to move freely. The algorithm is developed into an Extended Kalman Filter (EKF). For a through evaluation of this method, we use simulated traffic scenario of different vehicle models and realworld open dataset containing both radar and lidar data.
Authors: Md Mamunur Rahaman, Ewan K. A. Millar, Erik Meijering
Abstract: Zero-shot learning holds tremendous potential for histopathology image analysis by enabling models to generalize to unseen classes without extensive labeled data. Recent advancements in vision-language models (VLMs) have expanded the capabilities of ZSL, allowing models to perform tasks without task-specific fine-tuning. However, applying VLMs to histopathology presents considerable challenges due to the complexity of histopathological imagery and the nuanced nature of diagnostic tasks. In this paper, we propose a novel framework called Multi-Resolution Prompt-guided Hybrid Embedding (MR-PHE) to address these challenges in zero-shot histopathology image classification. MR-PHE leverages multiresolution patch extraction to mimic the diagnostic workflow of pathologists, capturing both fine-grained cellular details and broader tissue structures critical for accurate diagnosis. We introduce a hybrid embedding strategy that integrates global image embeddings with weighted patch embeddings, effectively combining local and global contextual information. Additionally, we develop a comprehensive prompt generation and selection framework, enriching class descriptions with domain-specific synonyms and clinically relevant features to enhance semantic understanding. A similarity-based patch weighting mechanism assigns attention-like weights to patches based on their relevance to class embeddings, emphasizing diagnostically important regions during classification. Our approach utilizes pretrained VLM, CONCH for ZSL without requiring domain-specific fine-tuning, offering scalability and reducing dependence on large annotated datasets. Experimental results demonstrate that MR-PHE not only significantly improves zero-shot classification performance on histopathology datasets but also often surpasses fully supervised models.
Authors: Shima Shabani, Mohammadsadegh Khoshghiaferezaee, Michael Breu{\ss}
Abstract: In this paper we study the sparse coding problem in the context of sparse dictionary learning for image recovery. To this end, we consider and compare several state-of-the-art sparse optimization methods constructed using the shrinkage operation. As the mathematical setting of these methods, we consider an online approach as algorithmical basis together with the basis pursuit denoising problem that arises by the convex optimization approach to the dictionary learning problem. By a dedicated construction of datasets and corresponding dictionaries, we study the effect of enlarging the underlying learning database on reconstruction quality making use of several error measures. Our study illuminates that the choice of the optimization method may be practically important in the context of availability of training data. In the context of different settings for training data as may be considered part of our study, we illuminate the computational efficiency of the assessed optimization methods.
Authors: Mohammad Mosaffa, Omid Rafieian, Hema Yoganarasimhan
Abstract: Political polarization is a significant issue in American politics, influencing public discourse, policy, and consumer behavior. While studies on polarization in news media have extensively focused on verbal content, non-verbal elements, particularly visual content, have received less attention due to the complexity and high dimensionality of image data. Traditional descriptive approaches often rely on feature extraction from images, leading to biased polarization estimates due to information loss. In this paper, we introduce the Polarization Measurement using Counterfactual Image Generation (PMCIG) method, which combines economic theory with generative models and multi-modal deep learning to fully utilize the richness of image data and provide a theoretically grounded measure of polarization in visual content. Applying this framework to a decade-long dataset featuring 30 prominent politicians across 20 major news outlets, we identify significant polarization in visual content, with notable variations across outlets and politicians. At the news outlet level, we observe significant heterogeneity in visual slant. Outlets such as Daily Mail, Fox News, and Newsmax tend to favor Republican politicians in their visual content, while The Washington Post, USA Today, and The New York Times exhibit a slant in favor of Democratic politicians. At the politician level, our results reveal substantial variation in polarized coverage, with Donald Trump and Barack Obama among the most polarizing figures, while Joe Manchin and Susan Collins are among the least. Finally, we conduct a series of validation tests demonstrating the consistency of our proposed measures with external measures of media slant that rely on non-image-based sources.
Authors: Jeimin Jeon, Youngmin Oh, Junghyup Lee, Donghyeon Baek, Dohyung Kim, Chanho Eom, Bumsub Ham
Abstract: N-shot neural architecture search (NAS) exploits a supernet containing all candidate subnets for a given search space. The subnets are typically trained with a static training strategy (e.g., using the same learning rate (LR) scheduler and optimizer for all subnets). This, however, does not consider that individual subnets have distinct characteristics, leading to two problems: (1) The supernet training is biased towards the low-complexity subnets (unfairness); (2) the momentum update in the supernet is noisy (noisy momentum). We present a dynamic supernet training technique to address these problems by adjusting the training strategy adaptive to the subnets. Specifically, we introduce a complexity-aware LR scheduler (CaLR) that controls the decay ratio of LR adaptive to the complexities of subnets, which alleviates the unfairness problem. We also present a momentum separation technique (MS). It groups the subnets with similar structural characteristics and uses a separate momentum for each group, avoiding the noisy momentum problem. Our approach can be applicable to various N-shot NAS methods with marginal cost, while improving the search performance drastically. We validate the effectiveness of our approach on various search spaces (e.g., NAS-Bench-201, Mobilenet spaces) and datasets (e.g., CIFAR-10/100, ImageNet).
Authors: Ayush Jain, Alexander Swerdlow, Yuzhou Wang, Sergio Arnaud, Ada Martin, Alexander Sax, Franziska Meier, Katerina Fragkiadaki
Abstract: Progress in 3D vision-language learning has been hindered by the scarcity of large-scale 3D datasets. We introduce UniVLG, a unified architecture for 2D and 3D vision-language understanding that bridges the gap between existing 2D-centric models and the rich 3D sensory data available in embodied systems. Our approach initializes most model weights from pre-trained 2D models and trains on both 2D and 3D vision-language data. We propose a novel language-conditioned mask decoder shared across 2D and 3D modalities to ground objects effectively in both RGB and RGB-D images, outperforming box-based approaches. To further reduce the domain gap between 2D and 3D, we incorporate 2D-to-3D lifting strategies, enabling UniVLG to utilize 2D data to enhance 3D performance. With these innovations, our model achieves state-of-the-art performance across multiple 3D vision-language grounding tasks, demonstrating the potential of transferring advances from 2D vision-language learning to the data-constrained 3D domain. Furthermore, co-training on both 2D and 3D data enhances performance across modalities without sacrificing 2D capabilities. By removing the reliance on 3D mesh reconstruction and ground-truth object proposals, UniVLG sets a new standard for realistic, embodied-aligned evaluation. Code and additional visualizations are available at $\href{https://univlg.github.io}{univlg.github.io}$.
URLs: https://univlg.github.io
Authors: Asaf Joseph, Shmuel Peleg
Abstract: Clothes-Changing Person Re-Identification (ReID) aims to recognize the same individual across different videos captured at various times and locations. This task is particularly challenging due to changes in appearance, such as clothing, hairstyle, and accessories. We propose a Clothes-Changing ReID method that uses only skeleton data and does not use appearance features. Traditional ReID methods often depend on appearance features, leading to decreased accuracy when clothing changes. Our approach utilizes a spatio-temporal Graph Convolution Network (GCN) encoder to generate a skeleton-based descriptor for each individual. During testing, we improve accuracy by aggregating predictions from multiple segments of a video clip. Evaluated on the CCVID dataset with several different pose estimation models, our method achieves state-of-the-art performance, offering a robust and efficient solution for Clothes-Changing ReID.
Authors: Ju He, Qihang Yu, Qihao Liu, Liang-Chieh Chen
Abstract: Bridging different modalities lies at the heart of cross-modality generation. While conventional approaches treat the text modality as a conditioning signal that gradually guides the denoising process from Gaussian noise to the target image modality, we explore a much simpler paradigm-directly evolving between text and image modalities through flow matching. This requires projecting both modalities into a shared latent space, which poses a significant challenge due to their inherently different representations: text is highly semantic and encoded as 1D tokens, whereas images are spatially redundant and represented as 2D latent embeddings. To address this, we introduce FlowTok, a minimal framework that seamlessly flows across text and images by encoding images into a compact 1D token representation. Compared to prior methods, this design reduces the latent space size by 3.3x at an image resolution of 256, eliminating the need for complex conditioning mechanisms or noise scheduling. Moreover, FlowTok naturally extends to image-to-text generation under the same formulation. With its streamlined architecture centered around compact 1D tokens, FlowTok is highly memory-efficient, requires significantly fewer training resources, and achieves much faster sampling speeds-all while delivering performance comparable to state-of-the-art models. Code will be available at https://github.com/bytedance/1d-tokenizer.
Authors: Zhang Zhang, Chao Sun, Chao Yue, Da Wen, Yujie Chen, Tianze Wang, Jianghao Leng
Abstract: Roadside vision centric 3D object detection has received increasing attention in recent years. It expands the perception range of autonomous vehicles, enhances the road safety. Previous methods focused on predicting per-pixel height rather than depth, making significant gains in roadside visual perception. While it is limited by the perspective property of near-large and far-small on image features, making it difficult for network to understand real dimension of objects in the 3D world. BEV features and voxel features present the real distribution of objects in 3D world compared to the image features. However, BEV features tend to lose details due to the lack of explicit height information, and voxel features are computationally expensive. Inspired by this insight, an efficient framework learning height prediction in voxel features via transformer is proposed, dubbed HeightFormer. It groups the voxel features into local height sequences, and utilize attention mechanism to obtain height distribution prediction. Subsequently, the local height sequences are reassembled to generate accurate 3D features. The proposed method is applied to two large-scale roadside benchmarks, DAIR-V2X-I and Rope3D. Extensive experiments are performed and the HeightFormer outperforms the state-of-the-art methods in roadside vision centric 3D object detection task.
Authors: Mir Rayat Imtiaz Hossain, Mennatullah Siam, Leonid Sigal, James J. Little
Abstract: Large-scale vision-language models (VLMs), trained on extensive datasets of image-text pairs, exhibit strong multimodal understanding capabilities by implicitly learning associations between textual descriptions and image regions. This emergent ability enables zero-shot object detection and segmentation, using techniques that rely on text-image attention maps, without necessarily training on abundant labeled segmentation datasets. However, performance of such methods depends heavily on prompt engineering and manually selected layers or head choices for the attention layers. In this work, we demonstrate that, rather than relying solely on textual prompts, providing a single visual example for each category and fine-tuning the text-to-image attention layers and embeddings significantly improves the performance. Additionally, we propose learning an ensemble through few-shot fine-tuning across multiple layers and/or prompts. An entropy-based ranking and selection mechanism for text-to-image attention layers is proposed to identify the top-performing layers without the need for segmentation labels. This eliminates the need for hyper-parameter selection of text-to-image attention layers, providing a more flexible and scalable solution for open-vocabulary segmentation. We show that this approach yields strong zero-shot performance, further enhanced through fine-tuning with a single visual example. Moreover, we demonstrate that our method and findings are general and can be applied across various vision-language models (VLMs).
Authors: Evangelos Kazakos, Cordelia Schmid, Josef Sivic
Abstract: We propose a novel approach for captioning and object grounding in video, where the objects in the caption are grounded in the video via temporally dense bounding boxes. We introduce the following contributions. First, we present a large-scale automatic annotation method that aggregates captions grounded with bounding boxes across individual frames into temporally dense and consistent bounding box annotations. We apply this approach on the HowTo100M dataset to construct a large-scale pre-training dataset, named HowToGround1M. We also introduce a Grounded Video Caption Generation model, dubbed GROVE, and pre-train the model on HowToGround1M. Second, we introduce a new dataset, called iGround, of 3500 videos with manually annotated captions and dense spatio-temporally grounded bounding boxes. This allows us to measure progress on this challenging problem, as well as to fine-tune our model on this small-scale but high-quality data. Third, we demonstrate that our approach achieves state-of-the-art results on the proposed iGround dataset compared to a number of baselines, as well as on the VidSTG and ActivityNet-Entities datasets. We perform extensive ablations that demonstrate the importance of pre-training using our automatically annotated HowToGround1M dataset followed by fine-tuning on the manually annotated iGround dataset and validate the key technical contributions of our model.
Authors: Parisa Boodaghi Malidarreh, Jillur Rahman Saurav, Thuong Le Hoai Pham, Amir Hajighasemi, Anahita Samadi, Saurabh Shrinivas Maydeo, Mohammad Sadegh Nasr, Jacob M. Luber
Abstract: Vector Quantization (VQ) techniques face significant challenges in codebook utilization, limiting reconstruction fidelity in image modeling. We introduce a Dual Codebook mechanism that effectively addresses this limitation by partitioning the representation into complementary global and local components. The global codebook employs a lightweight transformer for concurrent updates of all code vectors, while the local codebook maintains precise feature representation through deterministic selection. This complementary approach is trained from scratch without requiring pre-trained knowledge. Experimental evaluation across multiple standard benchmark datasets demonstrates state-of-the-art reconstruction quality while using a compact codebook of size 512 - half the size of previous methods that require pre-training. Our approach achieves significant FID improvements across diverse image domains, particularly excelling in scene and face reconstruction tasks. These results establish Dual Codebook VQ as an efficient paradigm for high-fidelity image reconstruction with significantly reduced computational requirements.
Authors: Avinash Paliwal, Xilong Zhou, Wei Ye, Jinhui Xiong, Rakesh Ranjan, Nima Khademi Kalantari
Abstract: In this paper, we propose RI3D, a novel 3DGS-based approach that harnesses the power of diffusion models to reconstruct high-quality novel views given a sparse set of input images. Our key contribution is separating the view synthesis process into two tasks of reconstructing visible regions and hallucinating missing regions, and introducing two personalized diffusion models, each tailored to one of these tasks. Specifically, one model ('repair') takes a rendered image as input and predicts the corresponding high-quality image, which in turn is used as a pseudo ground truth image to constrain the optimization. The other model ('inpainting') primarily focuses on hallucinating details in unobserved areas. To integrate these models effectively, we introduce a two-stage optimization strategy: the first stage reconstructs visible areas using the repair model, and the second stage reconstructs missing regions with the inpainting model while ensuring coherence through further optimization. Moreover, we augment the optimization with a novel Gaussian initialization method that obtains per-image depth by combining 3D-consistent and smooth depth with highly detailed relative depth. We demonstrate that by separating the process into two tasks and addressing them with the repair and inpainting models, we produce results with detailed textures in both visible and missing regions that outperform state-of-the-art approaches on a diverse set of scenes with extremely sparse inputs.
Authors: Xiangyu Yin, Yi Qi, Jinwei Hu, Zhen Chen, Yi Dong, Xingyu Zhao, Xiaowei Huang, Wenjie Ruan
Abstract: Vision Language Models (VLMs) have demonstrated impressive inference capabilities, but remain vulnerable to jailbreak attacks that can induce harmful or unethical responses. Existing defence methods are predominantly white-box approaches that require access to model parameters and extensive modifications, making them costly and impractical for many real-world scenarios. Although some black-box defences have been proposed, they often impose input constraints or require multiple queries, limiting their effectiveness in safety-critical tasks such as autonomous driving. To address these challenges, we propose a novel black-box defence framework called \textbf{T}extual \textbf{A}nchoring for \textbf{I}mmunizing \textbf{J}ailbreak \textbf{I}mages (\textbf{TAIJI}). TAIJI leverages key phrase-based textual anchoring to enhance the model's ability to assess and mitigate the harmful content embedded within both visual and textual prompts. Unlike existing methods, TAIJI operates effectively with a single query during inference, while preserving the VLM's performance on benign tasks. Extensive experiments demonstrate that TAIJI significantly enhances the safety and reliability of VLMs, providing a practical and efficient solution for real-world deployment.
Authors: Hai-Vy Nguyen, Fabrice Gamboa, Sixin Zhang, Reda Chhaibi, Serge Gratton, Thierry Giaccone
Abstract: In this paper, we introduce a novel spatial attention module, that can be integrated to any convolutional network. This module guides the model to pay attention to the most discriminative part of an image. This enables the model to attain a better performance by an end-to-end training. In standard approaches, a spatial attention map is generated in a position-wise fashion. We observe that this results in very irregular boundaries. This could make it difficult to generalize to new samples. In our method, the attention region is constrained to be rectangular. This rectangle is parametrized by only 5 parameters, allowing for a better stability and generalization to new samples. In our experiments, our method systematically outperforms the position-wise counterpart. Thus, this provides us a novel useful spatial attention mechanism for convolutional models. Besides, our module also provides the interpretability concerning the ``where to look" question, as it helps to know the part of the input on which the model focuses to produce the prediction.
Authors: Nathaniel Lesperance, Sujeevan Ratnasingham, Graham W. Taylor
Abstract: In the context of pressing climate change challenges and the significant biodiversity loss among arthropods, automated taxonomic classification from organismal images is a subject of intense research. However, traditional AI pipelines based on deep neural visual architectures such as CNNs or ViTs face limitations such as degraded performance on the long-tail of classes and the inability to reason about their predictions. We integrate image captioning and retrieval-augmented generation (RAG) with large language models (LLMs) to enhance biodiversity monitoring, showing particular promise for characterizing rare and unknown arthropod species. While a naive Vision-Language Model (VLM) excels in classifying images of common species, the RAG model enables classification of rarer taxa by matching explicit textual descriptions of taxonomic features to contextual biodiversity text data from external sources. The RAG model shows promise in reducing overconfidence and enhancing accuracy relative to naive LLMs, suggesting its viability in capturing the nuances of taxonomic hierarchy, particularly at the challenging family and genus levels. Our findings highlight the potential for modern vision-language AI pipelines to support biodiversity conservation initiatives, emphasizing the role of comprehensive data curation and collaboration with citizen science platforms to improve species identification, unknown species characterization and ultimately inform conservation strategies.
Authors: Yizhou Huang, Yihua Cheng, Kezhi Wang
Abstract: Motion prediction is crucial for autonomous driving, as it enables accurate forecasting of future vehicle trajectories based on historical inputs. This paper introduces Trajectory Mamba, a novel efficient trajectory prediction framework based on the selective state-space model (SSM). Conventional attention-based models face the challenge of computational costs that grow quadratically with the number of targets, hindering their application in highly dynamic environments. In response, we leverage the SSM to redesign the self-attention mechanism in the encoder-decoder architecture, thereby achieving linear time complexity. To address the potential reduction in prediction accuracy resulting from modifications to the attention mechanism, we propose a joint polyline encoding strategy to better capture the associations between static and dynamic contexts, ultimately enhancing prediction accuracy. Additionally, to balance prediction accuracy and inference speed, we adopted the decoder that differs entirely from the encoder. Through cross-state space attention, all target agents share the scene context, allowing the SSM to interact with the shared scene representation during decoding, thus inferring different trajectories over the next prediction steps. Our model achieves state-of-the-art results in terms of inference speed and parameter efficiency on both the Argoverse 1 and Argoverse 2 datasets. It demonstrates a four-fold reduction in FLOPs compared to existing methods and reduces parameter count by over 40% while surpassing the performance of the vast majority of previous methods. These findings validate the effectiveness of Trajectory Mamba in trajectory prediction tasks.
Authors: Mahshid Shiri, Alessandro Bruno, Daniele Loiacono
Abstract: Generative Adversarial Networks (GANs) have many potential medical imaging applications. Due to the limited memory of Graphical Processing Units (GPUs), most current 3D GAN models are trained on low-resolution medical images, these models cannot scale to high-resolution or are susceptible to patchy artifacts. In this work, we propose an end-to-end novel GAN architecture that uses Conditional Random field (CRF) to model dependencies so that it can generate consistent 3D medical Images without exploiting memory. To achieve this purpose, the generator is divided into two parts during training, the first part produces an intermediate representation and CRF is applied to this intermediate representation to capture correlations. The second part of the generator produces a random sub-volume of image using a subset of the intermediate representation. This structure has two advantages: first, the correlations are modeled by using the features that the generator is trying to optimize. Second, the generator can generate full high-resolution images during inference. Experiments on Lung CTs and Brain MRIs show that our architecture outperforms state-of-the-art while it has lower memory usage and less complexity.
Authors: Linfeng Ye
Abstract: Deep Neural Networks (DNNs) have become an integral part of our daily lives, especially in vision-related applications. However, the conventional lossy image compression algorithms are primarily designed for the Human Vision System (HVS), which can non-trivially compromise the DNNs' validation accuracy after compression, as noted in \cite{liu2018deepn}. Thus developing an image compression algorithm for both human and machine (DNNs) is on the horizon. To address the challenge mentioned above, in this paper, we first formulate the image compression as a multi-objective optimization problem which take both human and machine prespectives into account, then we solve it by linear combination, and proposed a novel distortion measure for both human and machine, dubbed Human and Machine-Oriented Error (HMOE). After that, we develop Human And Machine Oriented Soft Decision Quantization (HMOSDQ) based on HMOE, a lossy image compression algorithm for both human and machine (DNNs), and fully complied with JPEG format. In order to evaluate the performance of HMOSDQ, finally we conduct the experiments for two pre-trained well-known DNN-based image classifiers named Alexnet \cite{Alexnet} and VGG-16 \cite{simonyan2014VGG} on two subsets of the ImageNet \cite{deng2009imagenet} validation set: one subset included images with shorter side in the range of 496 to 512, while the other included images with shorter side in the range of 376 to 384. Our results demonstrate that HMOSDQ outperforms the default JPEG algorithm in terms of rate-accuracy and rate-distortion performance. For the Alexnet comparing with the default JPEG algorithm, HMOSDQ can improve the validation accuracy by more than $0.81\%$ at $0.61$ BPP, or equivalently reduce the compression rate of default JPEG by $9.6\times$ while maintaining the same validation accuracy.
Authors: Chaikal Amrullah, Daniel Panangian, Ksenia Bittner
Abstract: The growing demand for detailed building roof data has driven the development of automated extraction methods to overcome the inefficiencies of traditional approaches, particularly in handling complex variations in building geometries. Re:PolyWorld, which integrates point detection with graph neural networks, presents a promising solution for reconstructing high-detail building roof vector data. This study enhances Re:PolyWorld's performance on complex urban residential structures by incorporating attention-based backbones and additional area segmentation loss. Despite dataset limitations, our experiments demonstrated improvements in point position accuracy (1.33 pixels) and line distance accuracy (14.39 pixels), along with a notable increase in the reconstruction score to 91.99%. These findings highlight the potential of advanced neural network architectures in addressing the challenges of complex urban residential geometries.
Authors: Anirudh Nanduri, Siyuan Huang, Rama Chellappa
Abstract: Biometric recognition becomes increasingly challenging as we move away from the visible spectrum to infrared imagery, where domain discrepancies significantly impact identification performance. In this paper, we show that body embeddings perform better than face embeddings for cross-spectral person identification in medium-wave infrared (MWIR) and long-wave infrared (LWIR) domains. Due to the lack of multi-domain datasets, previous research on cross-spectral body identification - also known as Visible-Infrared Person Re-Identification (VI-ReID) - has primarily focused on individual infrared bands, such as near-infrared (NIR) or LWIR, separately. We address the multi-domain body recognition problem using the IARPA Janus Benchmark Multi-Domain Face (IJB-MDF) dataset, which enables matching of short-wave infrared (SWIR), MWIR, and LWIR images against RGB (VIS) images. We leverage a vision transformer architecture to establish benchmark results on the IJB-MDF dataset and, through extensive experiments, provide valuable insights into the interrelation of infrared domains, the adaptability of VIS-pretrained models, the role of local semantic features in body-embeddings, and effective training strategies for small datasets. Additionally, we show that finetuning a body model, pretrained exclusively on VIS data, with a simple combination of cross-entropy and triplet losses achieves state-of-the-art mAP scores on the LLCM dataset.
Authors: Haoyu Zhang, Raghavendra Ramachandra, Kiran Raja, Christoph Busch
Abstract: Face Recognition Systems (FRS) are increasingly vulnerable to face-morphing attacks, prompting the development of Morphing Attack Detection (MAD) algorithms. However, a key challenge in MAD lies in its limited generalizability to unseen data and its lack of explainability-critical for practical application environments such as enrolment stations and automated border control systems. Recognizing that most existing MAD algorithms rely on supervised learning paradigms, this work explores a novel approach to MAD using zero-shot learning leveraged on Large Language Models (LLMs). We propose two types of zero-shot MAD algorithms: one leveraging general vision models and the other utilizing multimodal LLMs. For general vision models, we address the MAD task by computing the mean support embedding of an independent support set without using morphed images. For the LLM-based approach, we employ the state-of-the-art GPT-4 Turbo API with carefully crafted prompts. To evaluate the feasibility of zero-shot MAD and the effectiveness of the proposed methods, we constructed a print-scan morph dataset featuring various unseen morphing algorithms, simulating challenging real-world application scenarios. Experimental results demonstrated notable detection accuracy, validating the applicability of zero-shot learning for MAD tasks. Additionally, our investigation into LLM-based MAD revealed that multimodal LLMs, such as ChatGPT, exhibit remarkable generalizability to untrained MAD tasks. Furthermore, they possess a unique ability to provide explanations and guidance, which can enhance transparency and usability for end-users in practical applications.
Authors: Muhammad Waseem, Chung-Hsuan Huang, Muhammad Muzzammil Sajjad, Laraib Haider Naqvi, Yaqoob Majeed, Tanzeel Ur Rehman, Tayyaba Nadeem
Abstract: Tomato maturity plays a pivotal role in optimizing harvest timing and ensuring product quality, but current methods struggle to achieve high accuracy along computational efficiency simultaneously. Existing deep learning approaches, while accurate, are often too computationally demanding for practical use in resource-constrained agricultural settings. In contrast, simpler techniques fail to capture the nuanced features needed for precise classification. This study aims to develop a computationally efficient tomato classification model using the ResNet-18 architecture optimized through transfer learning, pruning, and quantization techniques. Our objective is to address the dual challenge of maintaining high accuracy while enabling real-time performance on low-power edge devices. Then, these models were deployed on an edge device to investigate their performance for tomato maturity classification. The quantized model achieved an accuracy of 97.81%, with an average classification time of 0.000975 seconds per image. The pruned and auto-tuned model also demonstrated significant improvements in deployment metrics, further highlighting the benefits of optimization techniques. These results underscore the potential for a balanced solution that meets the accuracy and efficiency demands of modern agricultural production, paving the way for practical, real-world deployment in resource-limited environments.
Authors: Akshat Ramachandran, Mingyu Lee, Huan Xu, Souvik Kundu, Tushar Krishna
Abstract: We present OuroMamba, the first data-free post-training quantization (DFQ) method for vision Mamba-based models (VMMs). We identify two key challenges in enabling DFQ for VMMs, (1) VMM's recurrent state transitions restricts capturing of long-range interactions and leads to semantically weak synthetic data, (2) VMM activations exhibit dynamic outlier variations across time-steps, rendering existing static PTQ techniques ineffective. To address these challenges, OuroMamba presents a two-stage framework: (1) OuroMamba-Gen to generate semantically rich and meaningful synthetic data. It applies contrastive learning on patch level VMM features generated through neighborhood interactions in the latent state space, (2) OuroMamba-Quant to employ mixed-precision quantization with lightweight dynamic outlier detection during inference. In specific, we present a thresholding based outlier channel selection strategy for activations that gets updated every time-step. Extensive experiments across vision and generative tasks show that our data-free OuroMamba surpasses existing data-driven PTQ techniques, achieving state-of-the-art performance across diverse quantization settings. Additionally, we implement efficient GPU kernels to achieve practical latency speedup of up to 2.36x. Code will be released soon.
Authors: Fawaz Sammani, Jonas Fischer, Nikos Deligiannis
Abstract: Visual classifiers offer high-dimensional feature representations that are challenging to interpret and analyze. Text, in contrast, provides a more expressive and human-friendly interpretable medium for understanding and analyzing model behavior. We propose a simple, yet powerful method for reformulating any visual classifier so that it can be accessed with open-set text queries without compromising its original performance. Our approach is label-free, efficient, and preserves the underlying classifier's distribution and reasoning processes. We thus unlock several text-based interpretability applications for any classifier. We apply our method on 40 visual classifiers and demonstrate two primary applications: 1) building both label-free and zero-shot concept bottleneck models and therefore converting any classifier to be inherently-interpretable and 2) zero-shot decoding of visual features into natural language. In both applications, we achieve state-of-the-art results, greatly outperforming existing works. Our method enables text approaches for interpreting visual classifiers.
Authors: Reef Alturki, Adrian Hilton, Jean-Yves Guillemaut
Abstract: Occlusion poses a significant challenge in pedestrian detection from a single view. To address this, multi-view detection systems have been utilized to aggregate information from multiple perspectives. Recent advances in multi-view detection utilized an early-fusion strategy that strategically projects the features onto the ground plane, where detection analysis is performed. A promising approach in this context is the use of 3D feature-pulling technique, which constructs a 3D feature volume of the scene by sampling the corresponding 2D features for each voxel. However, it creates a 3D feature volume of the whole scene without considering the potential locations of pedestrians. In this paper, we introduce a novel model that efficiently leverages traditional 3D reconstruction techniques to enhance deep multi-view pedestrian detection. This is accomplished by complementing the 3D feature volume with probabilistic occupancy volume, which is constructed using the visual hull technique. The probabilistic occupancy volume focuses the model's attention on regions occupied by pedestrians and improves detection accuracy. Our model outperforms state-of-the-art models on the MultiviewX dataset, with an MODA of 97.3%, while achieving competitive performance on the Wildtrack dataset.
Authors: Yanjie Xu, Handing Xu, Tianmu Wang, Yaguan Li, Yunzhi Chen, Zhenguo Nie
Abstract: Rotation-invariant recognition of shapes is a common challenge in computer vision. Recent approaches have significantly improved the accuracy of rotation-invariant recognition by encoding the rotational invariance of shapes as hand-crafted image features and introducing deep neural networks. However, the methods based on pixels have too much redundant information, and the critical geometric information is prone to early leakage, resulting in weak rotation-invariant recognition of fine-grained shapes. In this paper, we reconsider the shape recognition problem from the perspective of contour points rather than pixels. We propose an anti-noise rotation-invariant convolution module based on contour geometric aware for fine-grained shape recognition. The module divides the shape contour into multiple local geometric regions(LGA), where we implement finer-grained rotation-invariant coding in terms of point topological relations. We provide a deep network composed of five such cascaded modules for classification and retrieval experiments. The results show that our method exhibits excellent performance in rotation-invariant recognition of fine-grained shapes. In addition, we demonstrate that our method is robust to contour noise and the rotation centers. The source code is available at https://github.com/zhenguonie/ANRICN_CGA.
Authors: Jiangning Wei, Lixiong Qin, Bo Yu, Tianjian Zou, Chuhan Yan, Dandan Xiao, Yang Yu, Lan Yang, Ke Li, Jun Liu
Abstract: Action recognition is a crucial task in artificial intelligence, with significant implications across various domains. We initially perform a comprehensive analysis of seven prominent action recognition methods across five widely-used datasets. This analysis reveals a critical, yet previously overlooked, observation: as the velocity of actions increases, the performance of these methods variably declines, undermining their robustness. This decline in performance poses significant challenges for their application in real-world scenarios. Building on these findings, we introduce the Velocity-Aware Action Recognition (VA-AR) framework to obtain robust action representations across different velocities. Our principal insight is that rapid actions (e.g., the giant circle backward in uneven bars or a smash in badminton) occur within short time intervals, necessitating smaller temporal attention windows to accurately capture intricate changes. Conversely, slower actions (e.g., drinking water or wiping face) require larger windows to effectively encompass the broader context. VA-AR employs a Mixture of Window Attention (MoWA) strategy, dynamically adjusting its attention window size based on the action's velocity. This adjustment enables VA-AR to obtain a velocity-aware representation, thereby enhancing the accuracy of action recognition. Extensive experiments confirm that VA-AR achieves state-of-the-art performance on the same five datasets, demonstrating VA-AR's effectiveness across a broad spectrum of action recognition scenarios.
Authors: Chuhan Zhang, Chaoyang Zhu, Pingcheng Dong, Long Chen, Dong Zhang
Abstract: In pursuit of detecting unstinted objects that extend beyond predefined categories, prior arts of open-vocabulary object detection (OVD) typically resort to pretrained vision-language models (VLMs) for base-to-novel category generalization. However, to mitigate the misalignment between upstream image-text pretraining and downstream region-level perception, additional supervisions are indispensable, eg, image-text pairs or pseudo annotations generated via self-training strategies. In this work, we propose CCKT-Det trained without any extra supervision. The proposed framework constructs a cyclic and dynamic knowledge transfer from language queries and visual region features extracted from VLMs, which forces the detector to closely align with the visual-semantic space of VLMs. Specifically, 1) we prefilter and inject semantic priors to guide the learning of queries, and 2) introduce a regional contrastive loss to improve the awareness of queries on novel objects. CCKT-Det can consistently improve performance as the scale of VLMs increases, all while requiring the detector at a moderate level of computation overhead. Comprehensive experimental results demonstrate that our method achieves performance gain of +2.9% and +10.2% AP50 over previous state-of-the-arts on the challenging COCO benchmark, both without and with a stronger teacher model. The code is provided at https://github.com/ZCHUHan/CCKT-Det.
Authors: Yifan Xie, Binkai Ou, Fei Ma, Yaohua Liu
Abstract: Vision and Language Navigation (VLN) requires an agent to navigate through environments following natural language instructions. However, existing methods often struggle with effectively integrating visual observations and instruction details during navigation, leading to suboptimal path planning and limited success rates. In this paper, we propose OIKG (Observation-graph Interaction and Key-detail Guidance), a novel framework that addresses these limitations through two key components: (1) an observation-graph interaction module that decouples angular and visual information while strengthening edge representations in the navigation space, and (2) a key-detail guidance module that dynamically extracts and utilizes fine-grained location and object information from instructions. By enabling more precise cross-modal alignment and dynamic instruction interpretation, our approach significantly improves the agent's ability to follow complex navigation instructions. Extensive experiments on the R2R and RxR datasets demonstrate that OIKG achieves state-of-the-art performance across multiple evaluation metrics, validating the effectiveness of our method in enhancing navigation precision through better observation-instruction alignment.
Authors: Gian Antariksa, Rohir Chakraborty, Shriyank Somvanshi, Subasish Das, Mohammad Jalayer, Deep Rameshkumar Patel, David Mills
Abstract: Visual object detection utilizing deep learning plays a vital role in computer vision and has extensive applications in transportation engineering. This paper focuses on detecting pavement marking quality during daytime using the You Only Look Once (YOLO) model, leveraging its advanced architectural features to enhance road safety through precise and real-time assessments. Utilizing image data from New Jersey, this study employed three YOLOv8 variants: YOLOv8m, YOLOv8n, and YOLOv8x. The models were evaluated based on their prediction accuracy for classifying pavement markings into good, moderate, and poor visibility categories. The results demonstrated that YOLOv8n provides the best balance between accuracy and computational efficiency, achieving the highest mean Average Precision (mAP) for objects with good visibility and demonstrating robust performance across various Intersections over Union (IoU) thresholds. This research enhances transportation safety by offering an automated and accurate method for evaluating the quality of pavement markings.
Authors: Jiaqi Jin, Siwei Wang, Zhibin Dong, Xihong Yang, Xinwang Liu, En Zhu, Kunlun He
Abstract: Multi-view clustering leverages complementary representations from diverse sources to enhance performance. However, real-world data often suffer incomplete cases due to factors like privacy concerns and device malfunctions. A key challenge is effectively utilizing available instances to recover missing views. Existing methods frequently overlook the heterogeneity among views during recovery, leading to significant distribution discrepancies between recovered and true data. Additionally, many approaches focus on cross-view correlations, neglecting insights from intra-view reliable structure and cross-view clustering structure. To address these issues, we propose BURG, a novel method for incomplete multi-view clustering with distriBution dUal-consistency Recovery Guidance. We treat each sample as a distinct category and perform cross-view distribution transfer to predict the distribution space of missing views. To compensate for the lack of reliable category information, we design a dual-consistency guided recovery strategy that includes intra-view alignment guided by neighbor-aware consistency and cross-view alignment guided by prototypical consistency. Extensive experiments on benchmarks demonstrate the superiority of BURG in the incomplete multi-view scenario.
Authors: Yixuan Zhang, Qing Chang, Yuxi Wang, Guang Chen, Zhaoxiang Zhang, Junran Peng
Abstract: Speech-driven 3D facial animation seeks to produce lifelike facial expressions that are synchronized with the speech content and its emotional nuances, finding applications in various multimedia fields. However, previous methods often overlook emotional facial expressions or fail to disentangle them effectively from the speech content. To address these challenges, we present EmoDiffusion, a novel approach that disentangles different emotions in speech to generate rich 3D emotional facial expressions. Specifically, our method employs two Variational Autoencoders (VAEs) to separately generate the upper face region and mouth region, thereby learning a more refined representation of the facial sequence. Unlike traditional methods that use diffusion models to connect facial expression sequences with audio inputs, we perform the diffusion process in the latent space. Furthermore, we introduce an Emotion Adapter to evaluate upper face movements accurately. Given the paucity of 3D emotional talking face data in the animation industry, we capture facial expressions under the guidance of animation experts using LiveLinkFace on an iPhone. This effort results in the creation of an innovative 3D blendshape emotional talking face dataset (3D-BEF) used to train our network. Extensive experiments and perceptual evaluations validate the effectiveness of our approach, confirming its superiority in generating realistic and emotionally rich facial animations.
Authors: Ming Deng, Sijin Sun, Zihao Li, Xiaochuan Hu, Xing Wu
Abstract: Camouflaged Object Detection (COD) is challenging due to the strong similarity between camouflaged objects and their surroundings, which complicates identification. Existing methods mainly rely on spatial local features, failing to capture global information, while Transformers increase computational costs.To address this, the Frequency-Assisted Mamba-Like Linear Attention Network (FMNet) is proposed, which leverages frequency-domain learning to efficiently capture global features and mitigate ambiguity between objects and the background. FMNet introduces the Multi-Scale Frequency-Assisted Mamba-Like Linear Attention (MFM) module, integrating frequency and spatial features through a multi-scale structure to handle scale variations while reducing computational complexity. Additionally, the Pyramidal Frequency Attention Extraction (PFAE) module and the Frequency Reverse Decoder (FRD) enhance semantics and reconstruct features. Experimental results demonstrate that FMNet outperforms existing methods on multiple COD datasets, showcasing its advantages in both performance and efficiency. Code available at https://anonymous.4open.science/r/FMNet-3CE5.
Authors: Lilin Zhang, Chengpei Wu, Ning Yang
Abstract: Existing adversarial training (AT) methods often suffer from incomplete perturbation, meaning that not all non-robust features are perturbed when generating adversarial examples (AEs). This results in residual correlations between non-robust features and labels, leading to suboptimal learning of robust features. However, achieving complete perturbation, i.e., perturbing as many non-robust features as possible, is challenging due to the difficulty in distinguishing robust and non-robust features and the sparsity of labeled data. To address these challenges, we propose a novel approach called Weakly Supervised Contrastive Adversarial Training (WSCAT). WSCAT ensures complete perturbation for improved learning of robust features by disrupting correlations between non-robust features and labels through complete AE generation over partially labeled data, grounded in information theory. Extensive theoretical analysis and comprehensive experiments on widely adopted benchmarks validate the superiority of WSCAT.
Authors: Mingjie Wei, Xuemei Xie, Guangming Shi
Abstract: Attributes such as style, fine-grained text, and trajectory are specific conditions for describing motion. However, existing methods often lack precise user control over motion attributes and suffer from limited generalizability to unseen motions. This work introduces an Attribute Controllable Motion generation architecture, to address these challenges via decouple any conditions and control them separately. Firstly, we explored the Attribute Diffusion Model to imporve text-to-motion performance via decouple text and motion learning, as the controllable model relies heavily on the pre-trained model. Then, we introduce Motion Adpater to quickly finetune previously unseen motion patterns. Its motion prompts inputs achieve multimodal text-to-motion generation that captures user-specified styles. Finally, we propose a LLM Planner to bridge the gap between unseen attributes and dataset-specific texts via local knowledage for user-friendly interaction. Our approach introduces the capability for motion prompts for stylize generation, enabling fine-grained and user-friendly attribute control while providing performance comparable to state-of-the-art methods. Project page: https://mjwei3d.github.io/ACMo/
Authors: Hasan Iqbal, Nazmul Karim, Umar Khalid, Azib Farooq, Zichun Zhong, Jing Hua, Chen Chen
Abstract: Instruction-guided generative models, especially those using text-to-image (T2I) and text-to-video (T2V) diffusion frameworks, have advanced the field of content editing in recent years. To extend these capabilities to 4D scene, we introduce a progressive sampling framework for 4D editing (PSF-4D) that ensures temporal and multi-view consistency by intuitively controlling the noise initialization during forward diffusion. For temporal coherence, we design a correlated Gaussian noise structure that links frames over time, allowing each frame to depend meaningfully on prior frames. Additionally, to ensure spatial consistency across views, we implement a cross-view noise model, which uses shared and independent noise components to balance commonalities and distinct details among different views. To further enhance spatial coherence, PSF-4D incorporates view-consistent iterative refinement, embedding view-aware information into the denoising process to ensure aligned edits across frames and views. Our approach enables high-quality 4D editing without relying on external models, addressing key challenges in previous methods. Through extensive evaluation on multiple benchmarks and multiple editing aspects (e.g., style transfer, multi-attribute editing, object removal, local editing, etc.), we show the effectiveness of our proposed method. Experimental results demonstrate that our proposed method outperforms state-of-the-art 4D editing methods in diverse benchmarks.
Authors: Jieyi Tan, Chengwei Zhang, Bo Dang, Yansheng Li
Abstract: Traditional Remote Sensing Foundation models (RSFMs) are pre-trained with a data-centralized paradigm, through self-supervision on large-scale curated remote sensing data. For each institution, however, pre-training RSFMs with limited data in a standalone manner may lead to suboptimal performance, while aggregating remote sensing data from multiple institutions for centralized pre-training raises privacy concerns. Seeking for collaboration is a promising solution to resolve this dilemma, where multiple institutions can collaboratively train RSFMs without sharing private data. In this paper, we propose a novel privacy-preserved pre-training framework (FedSense), which enables multiple institutions to collaboratively train RSFMs without sharing private data. However, it is a non-trivial task hindered by a vicious cycle, which results from model drift by remote sensing data heterogeneity and high communication overhead. To break this vicious cycle, we introduce Federated Mutual-guidance Learning. Specifically, we propose a Server-to-Clients Guidance (SCG) mechanism to guide clients updates towards global-flatness optimal solutions. Additionally, we propose a Clients-to-Server Guidance (CSG) mechanism to inject local knowledge into the server by low-bit communication. Extensive experiments on four downstream tasks demonstrate the effectiveness of our FedSense in both full-precision and communication-reduced scenarios, showcasing remarkable communication efficiency and performance gains.
Authors: Kyle Sargent, Kyle Hsu, Justin Johnson, Li Fei-Fei, Jiajun Wu
Abstract: Since the advent of popular visual generation frameworks like VQGAN and latent diffusion models, state-of-the-art image generation systems have generally been two-stage systems that first tokenize or compress visual data into a lower-dimensional latent space before learning a generative model. Tokenizer training typically follows a standard recipe in which images are compressed and reconstructed subject to a combination of MSE, perceptual, and adversarial losses. Diffusion autoencoders have been proposed in prior work as a way to learn end-to-end perceptually-oriented image compression, but have not yet shown state-of-the-art performance on the competitive task of ImageNet-1K reconstruction. We propose FlowMo, a transformer-based diffusion autoencoder that achieves a new state-of-the-art for image tokenization at multiple compression rates without using convolutions, adversarial losses, spatially-aligned two-dimensional latent codes, or distilling from other tokenizers. Our key insight is that FlowMo training should be broken into a mode-matching pre-training stage and a mode-seeking post-training stage. In addition, we conduct extensive analyses and explore the training of generative models atop the FlowMo tokenizer. Our code and models will be available at http://kylesargent.github.io/flowmo .
Authors: Heng Wang, Yotaro Shimose, Shingo Takamatsu
Abstract: Advertising banners are critical for capturing user attention and enhancing advertising campaign effectiveness. Creating aesthetically pleasing banner designs while conveying the campaign messages is challenging due to the large search space involving multiple design elements. Additionally, advertisers need multiple sizes for different displays and various versions to target different sectors of audiences. Since design is intrinsically an iterative and subjective process, flexible editability is also in high demand for practical usage. While current models have served as assistants to human designers in various design tasks, they typically handle only segments of the creative design process or produce pixel-based outputs that limit editability. This paper introduces a training-free framework for fully automated banner ad design creation, enabling frontier multimodal large language models (MLLMs) to streamline the production of effective banners with minimal manual effort across diverse marketing contexts. We present BannerAgency, an MLLM agent system that collaborates with advertisers to understand their brand identity and banner objectives, generates matching background images, creates blueprints for foreground design elements, and renders the final creatives as editable components in Figma or SVG formats rather than static pixels. To facilitate evaluation and future research, we introduce BannerRequest400, a benchmark featuring 100 unique logos paired with 400 diverse banner requests. Through quantitative and qualitative evaluations, we demonstrate the framework's effectiveness, emphasizing the quality of the generated banner designs, their adaptability to various banner requests, and their strong editability enabled by this component-based approach.
Authors: Wenhao Jiang, Duo Li, Menghan Hu, Chao Ma, Ke Wang, Zhipeng Zhang
Abstract: In the field of autonomous driving, end-to-end deep learning models show great potential by learning driving decisions directly from sensor data. However, training these models requires large amounts of labeled data, which is time-consuming and expensive. Considering that the real-world driving data exhibits a long-tailed distribution where simple scenarios constitute a majority part of the data, we are thus inspired to identify the most challenging scenarios within it. Subsequently, we can efficiently improve the performance of the model by training with the selected data of the highest value. Prior research has focused on the selection of valuable data by empirically designed strategies. However, manually designed methods suffer from being less generalizable to new data distributions. Observing that the BEV (Bird's Eye View) features in end-to-end models contain all the information required to represent the scenario, we propose an active learning framework that relies on these vectorized scene-level features, called SEAD. The framework selects initial data based on driving-environmental information and incremental data based on BEV features. Experiments show that we only need 30\% of the nuScenes training data to achieve performance close to what can be achieved with the full dataset. The source code will be released.
Authors: Kelu Yao, Nuo Xu, Rong Yang, Yingying Xu, Zhuoyan Gao, Titinunt Kitrungrotsakul, Yi Ren, Pu Zhang, Jin Wang, Ning Wei, Chao Li
Abstract: This paper introduces a holistic vision-language foundation model tailored for remote sensing, named Falcon. Falcon offers a unified, prompt-based paradigm that effectively executes comprehensive and complex remote sensing tasks. Falcon demonstrates powerful understanding and reasoning abilities at the image, region, and pixel levels. Specifically, given simple natural language instructions and remote sensing images, Falcon can produce impressive results in text form across 14 distinct tasks, i.e., image classification, object detection, segmentation, image captioning, and etc. To facilitate Falcon's training and empower its representation capacity to encode rich spatial and semantic information, we developed Falcon_SFT, a large-scale, multi-task, instruction-tuning dataset in the field of remote sensing. The Falcon_SFT dataset consists of approximately 78 million high-quality data samples, covering 5.6 million multi-spatial resolution and multi-view remote sensing images with diverse instructions. It features hierarchical annotations and undergoes manual sampling verification to ensure high data quality and reliability. Extensive comparative experiments are conducted, which verify that Falcon achieves remarkable performance over 67 datasets and 14 tasks, despite having only 0.7B parameters. We release the complete dataset, code, and model weights at https://github.com/TianHuiLab/Falcon, hoping to help further develop the open-source community.
Authors: Zhenguang Liu, Chao Shuai, Shaojing Fan, Ziping Dong, Jinwu Hu, Zhongjie Ba, Kui Ren
Abstract: Diffusion models have achieved remarkable success in novel view synthesis, but their reliance on large, diverse, and often untraceable Web datasets has raised pressing concerns about image copyright protection. Current methods fall short in reliably identifying unauthorized image use, as they struggle to generalize across varied generation tasks and fail when the training dataset includes images from multiple sources with few identifiable (watermarked or poisoned) samples. In this paper, we present novel evidence that diffusion-generated images faithfully preserve the statistical properties of their training data, particularly reflected in their spectral features. Leveraging this insight, we introduce \emph{CoprGuard}, a robust frequency domain watermarking framework to safeguard against unauthorized image usage in diffusion model training and fine-tuning. CoprGuard demonstrates remarkable effectiveness against a wide range of models, from naive diffusion models to sophisticated text-to-image models, and is robust even when watermarked images comprise a mere 1\% of the training dataset. This robust and versatile approach empowers content owners to protect their intellectual property in the era of AI-driven image generation.
Authors: Hongyang Wei, Shuaizheng Liu, Chun Yuan, Lei Zhang
Abstract: By leveraging the generative priors from pre-trained text-to-image diffusion models, significant progress has been made in real-world image super-resolution (Real-ISR). However, these methods tend to generate inaccurate and unnatural reconstructions in complex and/or heavily degraded scenes, primarily due to their limited perception and understanding capability of the input low-quality image. To address these limitations, we propose, for the first time to our knowledge, to adapt the pre-trained autoregressive multimodal model such as Lumina-mGPT into a robust Real-ISR model, namely PURE, which Perceives and Understands the input low-quality image, then REstores its high-quality counterpart. Specifically, we implement instruction tuning on Lumina-mGPT to perceive the image degradation level and the relationships between previously generated image tokens and the next token, understand the image content by generating image semantic descriptions, and consequently restore the image by generating high-quality image tokens autoregressively with the collected information. In addition, we reveal that the image token entropy reflects the image structure and present a entropy-based Top-k sampling strategy to optimize the local structure of the image during inference. Experimental results demonstrate that PURE preserves image content while generating realistic details, especially in complex scenes with multiple objects, showcasing the potential of autoregressive multimodal generative models for robust Real-ISR. The model and code will be available at https://github.com/nonwhy/PURE.
Authors: Taehwan Lee, Kyeongkook Seo, Jaejun Yoo, Sung Whan Yoon
Abstract: Flat minima, known to enhance generalization and robustness in supervised learning, remain largely unexplored in generative models. In this work, we systematically investigate the role of loss surface flatness in generative models, both theoretically and empirically, with a particular focus on diffusion models. We establish a theoretical claim that flatter minima improve robustness against perturbations in target prior distributions, leading to benefits such as reduced exposure bias -- where errors in noise estimation accumulate over iterations -- and significantly improved resilience to model quantization, preserving generative performance even under strong quantization constraints. We further observe that Sharpness-Aware Minimization (SAM), which explicitly controls the degree of flatness, effectively enhances flatness in diffusion models, whereas other well-known methods such as Stochastic Weight Averaging (SWA) and Exponential Moving Average (EMA), which promote flatness indirectly via ensembling, are less effective. Through extensive experiments on CIFAR-10, LSUN Tower, and FFHQ, we demonstrate that flat minima in diffusion models indeed improves not only generative performance but also robustness.
Authors: Yifan Liu, Xun Xu, Shijie Li, Jingyi Liao, Xulei Yang
Abstract: Multi-camera systems provide richer contextual information for industrial anomaly detection. However, traditional methods process each view independently, disregarding the complementary information across viewpoints. Existing multi-view anomaly detection approaches typically employ data-driven cross-view attention for feature fusion but fail to leverage the unique geometric properties of multi-camera setups. In this work, we introduce an epipolar geometry-constrained attention module to guide cross-view fusion, ensuring more effective information aggregation. To further enhance the potential of cross-view attention, we propose a pretraining strategy inspired by memory bank-based anomaly detection. This approach encourages normal feature representations to form multiple local clusters and incorporate multi-view aware negative sample synthesis to regularize pretraining. We demonstrate that our epipolar guided multi-view anomaly detection framework outperforms existing methods on the state-of-the-art multi-view anomaly detection dataset.
Authors: Ganlong Zhao, Guanbin Li, Jia Pan, Yizhou Yu
Abstract: Aerial Vision-and-Language Navigation (Aerial VLN) aims to obtain an unmanned aerial vehicle agent to navigate aerial 3D environments following human instruction. Compared to ground-based VLN, aerial VLN requires the agent to decide the next action in both horizontal and vertical directions based on the first-person view observations. Previous methods struggle to perform well due to the longer navigation path, more complicated 3D scenes, and the neglect of the interplay between vertical and horizontal actions. In this paper, we propose a novel grid-based view selection framework that formulates aerial VLN action prediction as a grid-based view selection task, incorporating vertical action prediction in a manner that accounts for the coupling with horizontal actions, thereby enabling effective altitude adjustments. We further introduce a grid-based bird's eye view map for aerial space to fuse the visual information in the navigation history, provide contextual scene information, and mitigate the impact of obstacles. Finally, a cross-modal transformer is adopted to explicitly align the long navigation history with the instruction. We demonstrate the superiority of our method in extensive experiments.
Authors: Yuan Liu, Saihui Hou, Saijie Hou, Jiabao Du, Shibei Meng, Yongzhen Huang
Abstract: Image Difference Captioning (IDC) aims to generate natural language descriptions of subtle differences between image pairs, requiring both precise visual change localization and coherent semantic expression. Despite recent advancements, existing datasets often lack breadth and depth, limiting their applicability in complex and dynamic environments: (1) from a breadth perspective, current datasets are constrained to limited variations of objects in specific scenes, and (2) from a depth perspective, prior benchmarks often provide overly simplistic descriptions. To address these challenges, we introduce OmniDiff, a comprehensive dataset comprising 324 diverse scenarios-spanning real-world complex environments and 3D synthetic settings-with fine-grained human annotations averaging 60 words in length and covering 12 distinct change types. Building on this foundation, we propose M$^3$Diff, a MultiModal large language model enhanced by a plug-and-play Multi-scale Differential Perception (MDP) module. This module improves the model's ability to accurately identify and describe inter-image differences while maintaining the foundational model's generalization capabilities. With the addition of the OmniDiff dataset, M$^3$Diff achieves state-of-the-art performance across multiple benchmarks, including Spot-the-Diff, IEdit, CLEVR-Change, CLEVR-DC, and OmniDiff, demonstrating significant improvements in cross-scenario difference recognition accuracy compared to existing methods. The dataset, code, and models will be made publicly available to support further research.
Authors: Weichen Zhan, Zile Zhou, Zhiheng Zheng, Chen Gao, Jinqiang Cui, Yong Li, Xinlei Chen, Xiao-Ping Zhang
Abstract: Spatial reasoning is a fundamental capability of embodied agents and has garnered widespread attention in the field of multimodal large language models (MLLMs). In this work, we propose a novel benchmark, Open3DVQA, to comprehensively evaluate the spatial reasoning capacities of current state-of-the-art (SOTA) foundation models in open 3D space. Open3DVQA consists of 9k VQA samples, collected using an efficient semi-automated tool in a high-fidelity urban simulator. We evaluate several SOTA MLLMs across various aspects of spatial reasoning, such as relative and absolute spatial relationships, situational reasoning, and object-centric spatial attributes. Our results reveal that: 1) MLLMs perform better at answering questions regarding relative spatial relationships than absolute spatial relationships, 2) MLLMs demonstrate similar spatial reasoning abilities for both egocentric and allocentric perspectives, and 3) Fine-tuning large models significantly improves their performance across different spatial reasoning tasks. We believe that our open-source data collection tools and in-depth analyses will inspire further research on MLLM spatial reasoning capabilities. The benchmark is available at https://github.com/WeichenZh/Open3DVQA.
Authors: He Zhang, Xinyi Fu, John M. Carroll
Abstract: Traditional image annotation tasks rely heavily on human effort for object selection and label assignment, making the process time-consuming and prone to decreased efficiency as annotators experience fatigue after extensive work. This paper introduces a novel framework that leverages the visual understanding capabilities of large multimodal models (LMMs), particularly GPT, to assist annotation workflows. In our proposed approach, human annotators focus on selecting objects via bounding boxes, while the LMM autonomously generates relevant labels. This human-AI collaborative framework enhances annotation efficiency by reducing the cognitive and time burden on human annotators. By analyzing the system's performance across various types of annotation tasks, we demonstrate its ability to generalize to tasks such as object recognition, scene description, and fine-grained categorization. Our proposed framework highlights the potential of this approach to redefine annotation workflows, offering a scalable and efficient solution for large-scale data labeling in computer vision. Finally, we discuss how integrating LMMs into the annotation pipeline can advance bidirectional human-AI alignment, as well as the challenges of alleviating the "endless annotation" burden in the face of information overload by shifting some of the work to AI.
Authors: Wenbang Deng, Xieyuanli Chen, Qinghua Yu, Yunze He, Junhao Xiao, Huimin Lu
Abstract: Semantic segmentation is a key technique that enables mobile robots to understand and navigate surrounding environments autonomously. However, most existing works focus on segmenting known objects, overlooking the identification of unknown classes, which is common in real-world applications. In this paper, we propose a feature-oriented framework for open-set semantic segmentation on LiDAR data, capable of identifying unknown objects while retaining the ability to classify known ones. We design a decomposed dual-decoder network to simultaneously perform closed-set semantic segmentation and generate distinctive features for unknown objects. The network is trained with multi-objective loss functions to capture the characteristics of known and unknown objects. Using the extracted features, we introduce an anomaly detection mechanism to identify unknown objects. By integrating the results of close-set semantic segmentation and anomaly detection, we achieve effective feature-driven LiDAR open-set semantic segmentation. Evaluations on both SemanticKITTI and nuScenes datasets demonstrate that our proposed framework significantly outperforms state-of-the-art methods. The source code will be made publicly available at https://github.com/nubot-nudt/DOSS.
Authors: Asifullah Khan, Laiba Asmatullah, Anza Malik, Shahzaib Khan, Hamna Asif
Abstract: Self-supervised learning is a machine learning approach that generates implicit labels by learning underlined patterns and extracting discriminative features from unlabeled data without manual labelling. Contrastive learning introduces the concept of "positive" and "negative" samples, where positive pairs (e.g., variation of the same image/object) are brought together in the embedding space, and negative pairs (e.g., views from different images/objects) are pushed farther away. This methodology has shown significant improvements in image understanding and image text analysis without much reliance on labeled data. In this paper, we comprehensively discuss the terminologies, recent developments and applications of contrastive learning with respect to text-image models. Specifically, we provide an overview of the approaches of contrastive learning in text-image models in recent years. Secondly, we categorize the approaches based on different model structures. Thirdly, we further introduce and discuss the latest advances of the techniques used in the process such as pretext tasks for both images and text, architectural structures, and key trends. Lastly, we discuss the recent state-of-art applications of self-supervised contrastive learning Text-Image based models.
Authors: Avinash Madasu, Vasudev Lal, Phillip Howard
Abstract: CLIP is one of the most popular foundational models and is heavily used for many vision-language tasks. However, little is known about the inner workings of CLIP. While recent work has proposed decomposition-based interpretability methods for identifying textual descriptions of attention heads in CLIP, the implications of conceptual consistency in these text labels on interpretability and model performance has not been explored. To bridge this gap, we study the conceptual consistency of text descriptions for attention heads in CLIP-like models. We conduct extensive experiments on six different models from OpenAI and OpenCLIP which vary by size, type of pre-training data and patch size. We propose Concept Consistency Score (CCS), a novel interpretability metric that measures how consistently individual attention heads in CLIP models align with specific concepts. To assign concept labels to heads, we use in-context learning with ChatGPT, guided by a few manually-curated examples, and validate these labels using an LLM-as-a-judge approach. Our soft-pruning experiments reveal that high CCS heads are critical for preserving model performance, as pruning them leads to a significantly larger performance drop than pruning random or low CCS heads. Notably, we find that high CCS heads capture essential concepts and play a key role in out-of-domain detection, concept-specific reasoning, and video-language understanding. These results position CCS as a powerful interpretability metric for analyzing CLIP-like models.
Authors: Jun Yu, Yunxiang Zhang, Xilong Lu, Yang Zheng, Yongqi Wang, Lingsi Zhu
Abstract: In this report, we present our solution for the Action Unit (AU) Detection Challenge, in 8th Competition on Affective Behavior Analysis in-the-wild. In order to achieve robust and accurate classification of facial action unit in the wild environment, we introduce an innovative method that leverages audio-visual multimodal data. Our method employs ConvNeXt as the image encoder and uses Whisper to extract Mel spectrogram features. For these features, we utilize a Transformer encoder-based feature fusion module to integrate the affective information embedded in audio and image features. This ensures the provision of rich high-dimensional feature representations for the subsequent multilayer perceptron (MLP) trained on the Aff-Wild2 dataset, enhancing the accuracy of AU detection.
Authors: Kaixuan Jiang, Yang Liu, Weixing Chen, Jingzhou Luo, Ziliang Chen, Ling Pan, Guanbin Li, Liang Lin
Abstract: Embodied Question Answering (EQA) is a challenging task in embodied intelligence that requires agents to dynamically explore 3D environments, actively gather visual information, and perform multi-step reasoning to answer questions. However, current EQA approaches suffer from critical limitations in exploration efficiency, dataset design, and evaluation metrics. Moreover, existing datasets often introduce biases or prior knowledge, leading to disembodied reasoning, while frontier-based exploration strategies struggle in cluttered environments and fail to ensure fine-grained exploration of task-relevant areas. To address these challenges, we construct the EXPloration-awaRe Embodied queStion anSwering Benchmark (EXPRESS-Bench), the largest dataset designed specifically to evaluate both exploration and reasoning capabilities. EXPRESS-Bench consists of 777 exploration trajectories and 2,044 question-trajectory pairs. To improve exploration efficiency, we propose Fine-EQA, a hybrid exploration model that integrates frontier-based and goal-oriented navigation to guide agents toward task-relevant regions more effectively. Additionally, we introduce a novel evaluation metric, Exploration-Answer Consistency (EAC), which ensures faithful assessment by measuring the alignment between answer grounding and exploration reliability. Extensive experimental comparisons with state-of-the-art EQA models demonstrate the effectiveness of our EXPRESS-Bench in advancing embodied exploration and question reasoning.
Authors: Hongbin Lin, Zilu Guo, Yifan Zhang, Shuaicheng Niu, Yafeng Li, Ruimao Zhang, Shuguang Cui, Zhen Li
Abstract: In autonomous driving, vision-centric 3D detection aims to identify 3D objects from images. However, high data collection costs and diverse real-world scenarios limit the scale of training data. Once distribution shifts occur between training and test data, existing methods often suffer from performance degradation, known as Out-of-Distribution (OOD) problems. To address this, controllable Text-to-Image (T2I) diffusion offers a potential solution for training data enhancement, which is required to generate diverse OOD scenarios with precise 3D object geometry. Nevertheless, existing controllable T2I approaches are restricted by the limited scale of training data or struggle to preserve all annotated 3D objects. In this paper, we present DriveGEN, a method designed to improve the robustness of 3D detectors in Driving via Training-Free Controllable Text-to-Image Diffusion Generation. Without extra diffusion model training, DriveGEN consistently preserves objects with precise 3D geometry across diverse OOD generations, consisting of 2 stages: 1) Self-Prototype Extraction: We empirically find that self-attention features are semantic-aware but require accurate region selection for 3D objects. Thus, we extract precise object features via layouts to capture 3D object geometry, termed self-prototypes. 2) Prototype-Guided Diffusion: To preserve objects across various OOD scenarios, we perform semantic-aware feature alignment and shallow feature alignment during denoising. Extensive experiments demonstrate the effectiveness of DriveGEN in improving 3D detection. The code is available at https://github.com/Hongbin98/DriveGEN.
Authors: Yijia Xu, Jianzhong Ju, Jian Luan, Jinshi Cui
Abstract: The raster-ordered image token sequence exhibits a significant Euclidean distance between index-adjacent tokens at line breaks, making it unsuitable for autoregressive generation. To address this issue, this paper proposes Direction-Aware Diagonal Autoregressive Image Generation (DAR) method, which generates image tokens following a diagonal scanning order. The proposed diagonal scanning order ensures that tokens with adjacent indices remain in close proximity while enabling causal attention to gather information from a broader range of directions. Additionally, two direction-aware modules: 4D-RoPE and direction embeddings are introduced, enhancing the model's capability to handle frequent changes in generation direction. To leverage the representational capacity of the image tokenizer, we use its codebook as the image token embeddings. We propose models of varying scales, ranging from 485M to 2.0B. On the 256$\times$256 ImageNet benchmark, our DAR-XL (2.0B) outperforms all previous autoregressive image generators, achieving a state-of-the-art FID score of 1.37.
Authors: Hao Liu, Pengyu Guo, Siyuan Yang, Zeqing Jiang, Qinglei Hu, Dongyu Li
Abstract: With the continuous advancement of human exploration into deep space, intelligent perception and high-precision segmentation technology for on-orbit multi-spacecraft targets have become critical factors for ensuring the success of modern space missions. However, the complex deep space environment, diverse imaging conditions, and high variability in spacecraft morphology pose significant challenges to traditional segmentation methods. This paper proposes SpaceSeg, an innovative vision foundation model-based segmentation framework with four core technical innovations: First, the Multi-Scale Hierarchical Attention Refinement Decoder (MSHARD) achieves high-precision feature decoding through cross-resolution feature fusion via hierarchical attention. Second, the Multi-spacecraft Connected Component Analysis (MS-CCA) effectively resolves topological structure confusion in dense targets. Third, the Spatial Domain Adaptation Transform framework (SDAT) eliminates cross-domain disparities and resist spatial sensor perturbations through composite enhancement strategies. Finally, a custom Multi-Spacecraft Segmentation Task Loss Function is created to significantly improve segmentation robustness in deep space scenarios. To support algorithm validation, we construct the first multi-scale on-orbit multi-spacecraft semantic segmentation dataset SpaceES, which covers four types of spatial backgrounds and 17 typical spacecraft targets. In testing, SpaceSeg achieves state-of-the-art performance with 89.87$\%$ mIoU and 99.98$\%$ mAcc, surpassing existing best methods by 5.71 percentage points. The dataset and code are open-sourced at https://github.com/Akibaru/SpaceSeg to provide critical technical support for next-generation space situational awareness systems.
Authors: Lexin Fang, Yunyang Xu, Xiang Ma, Xuemei Li, Caiming Zhang
Abstract: Deep learning has achieved significant advancements in medical image segmentation, but existing models still face challenges in accurately segmenting lesion regions. The main reason is that some lesion regions in medical images have unclear boundaries, irregular shapes, and small tissue density differences, leading to label ambiguity. However, the existing model treats all data equally without taking quality differences into account in the training process, resulting in noisy labels negatively impacting model training and unstable feature representations. In this paper, a data-driven alternating learning (DALE) paradigm is proposed to optimize the model's training process, achieving stable and high-precision segmentation. The paradigm focuses on two key points: (1) reducing the impact of noisy labels, and (2) calibrating unstable representations. To mitigate the negative impact of noisy labels, a loss consistency-based collaborative optimization method is proposed, and its effectiveness is theoretically demonstrated. Specifically, the label confidence parameters are introduced to dynamically adjust the influence of labels of different confidence levels during model training, thus reducing the influence of noise labels. To calibrate the learning bias of unstable representations, a distribution alignment method is proposed. This method restores the underlying distribution of unstable representations, thereby enhancing the discriminative capability of fuzzy region representations. Extensive experiments on various benchmarks and model backbones demonstrate the superiority of the DALE paradigm, achieving an average performance improvement of up to 7.16%.
Authors: Zichen Tang, Yuan Yao, Miaomiao Cui, Liefeng Bo, Hongyu Yang
Abstract: Text-guided 3D human generation has advanced with the development of efficient 3D representations and 2D-lifting methods like Score Distillation Sampling (SDS). However, current methods suffer from prolonged training times and often produce results that lack fine facial and garment details. In this paper, we propose GaussianIP, an effective two-stage framework for generating identity-preserving realistic 3D humans from text and image prompts. Our core insight is to leverage human-centric knowledge to facilitate the generation process. In stage 1, we propose a novel Adaptive Human Distillation Sampling (AHDS) method to rapidly generate a 3D human that maintains high identity consistency with the image prompt and achieves a realistic appearance. Compared to traditional SDS methods, AHDS better aligns with the human-centric generation process, enhancing visual quality with notably fewer training steps. To further improve the visual quality of the face and clothes regions, we design a View-Consistent Refinement (VCR) strategy in stage 2. Specifically, it produces detail-enhanced results of the multi-view images from stage 1 iteratively, ensuring the 3D texture consistency across views via mutual attention and distance-guided attention fusion. Then a polished version of the 3D human can be achieved by directly perform reconstruction with the refined images. Extensive experiments demonstrate that GaussianIP outperforms existing methods in both visual quality and training efficiency, particularly in generating identity-preserving results. Our code is available at: https://github.com/silence-tang/GaussianIP.
Authors: Junbiao Pang, Tianyang Cai
Abstract: Quantization-Aware Training (QAT) is one of the prevailing neural network compression solutions. However, its stability has been challenged for yielding deteriorating performances as the quantization error is inevitable. We find that the sharp landscape of loss, which leads to a dramatic performance drop, is an essential factor that causes instability. Theoretically, we have discovered that the perturbations in the feature would bring a flat local minima. However, simply adding perturbations into either weight or feature empirically deteriorates the performance of the Full Precision (FP) model. In this paper, we propose Feature-Perturbed Quantization (FPQ) to stochastically perturb the feature and employ the feature distillation method to the quantized model. Our method generalizes well to different network architectures and various QAT methods. Furthermore, we mathematically show that FPQ implicitly regularizes the Hessian norm, which calibrates the smoothness of a loss landscape. Extensive experiments demonstrate that our approach significantly outperforms the current State-Of-The-Art (SOTA) QAT methods and even the FP counterparts.
Authors: Haonan Wang, Qixiang Zhang, Lehan Wang, Xuanqi Huang, Xiaomeng Li
Abstract: Decoding visual stimuli from neural activity is essential for understanding the human brain. While fMRI methods have successfully reconstructed static images, fMRI-to-video reconstruction faces challenges due to the need for capturing spatiotemporal dynamics like motion and scene transitions. Recent approaches have improved semantic and perceptual alignment but struggle to integrate coarse fMRI data with detailed visual features. Inspired by the hierarchical organization of the visual system, we propose NEURONS, a novel framework that decouples learning into four correlated sub-tasks: key object segmentation, concept recognition, scene description, and blurry video reconstruction. This approach simulates the visual cortex's functional specialization, allowing the model to capture diverse video content. In the inference stage, NEURONS generates robust conditioning signals for a pre-trained text-to-video diffusion model to reconstruct the videos. Extensive experiments demonstrate that NEURONS outperforms state-of-the-art baselines, achieving solid improvements in video consistency (26.6%) and semantic-level accuracy (19.1%). Notably, NEURONS shows a strong functional correlation with the visual cortex, highlighting its potential for brain-computer interfaces and clinical applications. Code and model weights will be available at: https://github.com/xmed-lab/NEURONS.
Authors: Zhen Tan, Xieyuanli Chen, Jinpu Zhang, Lei Feng, Dewen Hu
Abstract: 3D Gaussian Splatting (3DGS) has achieved impressive rendering performance in novel view synthesis. However, its efficacy diminishes considerably in sparse image sequences, where inherent data sparsity amplifies geometric uncertainty during optimization. This often leads to convergence at suboptimal local minima, resulting in noticeable structural artifacts in the reconstructed scenes.To mitigate these issues, we propose Uncertainty-aware Normal-Guided Gaussian Splatting (UNG-GS), a novel framework featuring an explicit Spatial Uncertainty Field (SUF) to quantify geometric uncertainty within the 3DGS pipeline. UNG-GS enables high-fidelity rendering and achieves high-precision reconstruction without relying on priors. Specifically, we first integrate Gaussian-based probabilistic modeling into the training of 3DGS to optimize the SUF, providing the model with adaptive error tolerance. An uncertainty-aware depth rendering strategy is then employed to weight depth contributions based on the SUF, effectively reducing noise while preserving fine details. Furthermore, an uncertainty-guided normal refinement method adjusts the influence of neighboring depth values in normal estimation, promoting robust results. Extensive experiments demonstrate that UNG-GS significantly outperforms state-of-the-art methods in both sparse and dense sequences. The code will be open-source.
Authors: Yini Li, Nantheera Anantrasirichai
Abstract: Low-light and underwater videos suffer from poor visibility, low contrast, and high noise, necessitating enhancements in visual quality. However, existing approaches typically rely on paired ground truth, which limits their practicality and often fails to maintain temporal consistency. To overcome these obstacles, this paper introduces a novel zero-shot learning approach named Zero-TIG, leveraging the Retinex theory and optical flow techniques. The proposed network consists of an enhancement module and a temporal feedback module. The enhancement module comprises three subnetworks: low-light image denoising, illumination estimation, and reflection denoising. The temporal enhancement module ensures temporal consistency by incorporating histogram equalization, optical flow computation, and image warping to align the enhanced previous frame with the current frame, thereby maintaining continuity. Additionally, we address color distortion in underwater data by adaptively balancing RGB channels. The experimental results demonstrate that our method achieves low-light video enhancement without the need for paired training data, making it a promising and applicable method for real-world scenario enhancement.
Authors: Luca Martini, Daniele Zolezzi, Saverio Iacono, Gianni Viardo Vercelli
Abstract: The reconstruction of low-resolution football broadcast images presents a significant challenge in sports broadcasting, where detailed visuals are essential for analysis and audience engagement. This study introduces a multi-stage generative upscaling framework leveraging Diffusion Models to enhance degraded images, transforming inputs as small as $64 \times 64$ pixels into high-fidelity $1024 \times 1024$ outputs. By integrating an image-to-image pipeline, ControlNet conditioning, and LoRA fine-tuning, our approach surpasses traditional upscaling methods in restoring intricate textures and domain-specific elements such as player details and jersey logos. The custom LoRA is trained on a custom football dataset, ensuring adaptability to sports broadcast needs. Experimental results demonstrate substantial improvements over conventional models, with ControlNet refining fine details and LoRA enhancing task-specific elements. These findings highlight the potential of diffusion-based image reconstruction in sports media, paving the way for future applications in automated video enhancement and real-time sports analytics.
Authors: Leideng Shi, Juan Zhang
Abstract: Referring remote sensing image segmentation (RRSIS) is a novel visual task in remote sensing images segmentation, which aims to segment objects based on a given text description, with great significance in practical application. Previous studies fuse visual and linguistic modalities by explicit feature interaction, which fail to effectively excavate useful multimodal information from dual-branch encoder. In this letter, we design a multimodal-aware fusion network (MAFN) to achieve fine-grained alignment and fusion between the two modalities. We propose a correlation fusion module (CFM) to enhance multi-scale visual features by introducing adaptively noise in transformer, and integrate cross-modal aware features. In addition, MAFN employs multi-scale refinement convolution (MSRC) to adapt to the various orientations of objects at different scales to boost their representation ability to enhances segmentation accuracy. Extensive experiments have shown that MAFN is significantly more effective than the state of the art on RRSIS-D datasets. The source code is available at https://github.com/Roaxy/MAFN.
Authors: Leqi Shen, Guoqiang Gong, Tao He, Yifeng Zhang, Pengzhang Liu, Sicheng Zhao, Guiguang Ding
Abstract: Video Large Language Models have shown impressive capabilities in video comprehension, yet their practical deployment is hindered by substantial inference costs caused by redundant video tokens. Existing pruning techniques fail to fully exploit the spatiotemporal redundancy inherent in video data. To bridge this gap, we perform a systematic analysis of video redundancy from two perspectives: temporal context and visual context. Leveraging this insight, we propose Dynamic Density Pruning for Fast Video LLMs termed FastVID. Specifically, FastVID dynamically partitions videos into temporally ordered segments to preserve temporal structure and applies a density-based token pruning strategy to maintain essential visual information. Our method significantly reduces computational overhead while maintaining temporal and visual integrity. Extensive evaluations show that FastVID achieves state-of-the-art performance across various short- and long-video benchmarks on leading Video LLMs, including LLaVA-OneVision and LLaVA-Video. Notably, FastVID effectively prunes 90% of video tokens while retaining 98.0% of LLaVA-OneVision's original performance. The code is available at https://github.com/LunarShen/FastVID.
Authors: Qiuxia Lin, Kerui Gu, Linlin Yang, Angela Yao
Abstract: Online test-time adaptation for 3D human pose estimation is used for video streams that differ from training data. Ground truth 2D poses are used for adaptation, but only estimated 2D poses are available in practice. This paper addresses adapting models to streaming videos with estimated 2D poses. Comparing adaptations reveals the challenge of limiting estimation errors while preserving accurate pose information. To this end, we propose adaptive aggregation, a two-stage optimization, and local augmentation for handling varying levels of estimated pose error. First, we perform adaptive aggregation across videos to initialize the model state with labeled representative samples. Within each video, we use a two-stage optimization to benefit from 2D fitting while minimizing the impact of erroneous updates. Second, we employ local augmentation, using adjacent confident samples to update the model before adapting to the current non-confident sample. Our method surpasses state-of-the-art by a large margin, advancing adaptation towards more practical settings of using estimated 2D poses.
Authors: Shree Singhi, Aayan Yadav, Aayush Gupta, Shariar Ebrahimi, Parisa Hassanizadeh
Abstract: As AI-generated sensitive images become more prevalent, identifying their source is crucial for distinguishing them from real images. Conventional image watermarking methods are vulnerable to common transformations like filters, lossy compression, and screenshots, often applied during social media sharing. Watermarks can also be faked or removed if models are open-sourced or leaked since images can be rewatermarked. We have developed a three-part framework for secure, transformation-resilient AI content provenance detection, to address these limitations. We develop an adversarially robust state-of-the-art perceptual hashing model, DinoHash, derived from DINOV2, which is robust to common transformations like filters, compression, and crops. Additionally, we integrate a Multi-Party Fully Homomorphic Encryption~(MP-FHE) scheme into our proposed framework to ensure the protection of both user queries and registry privacy. Furthermore, we improve previous work on AI-generated media detection. This approach is useful in cases where the content is absent from our registry. DinoHash significantly improves average bit accuracy by 12% over state-of-the-art watermarking and perceptual hashing methods while maintaining superior true positive rate (TPR) and false positive rate (FPR) tradeoffs across various transformations. Our AI-generated media detection results show a 25% improvement in classification accuracy on commonly used real-world AI image generators over existing algorithms. By combining perceptual hashing, MP-FHE, and an AI content detection model, our proposed framework provides better robustness and privacy compared to previous work.
Authors: Li Cui, Yang Ding, Richard Hartley, Zirui Xie, Laurent Kneip, Zhenghua Yu
Abstract: We propose a novel, vision-only object-level SLAM framework for automotive applications representing 3D shapes by implicit signed distance functions. Our key innovation consists of augmenting the standard neural representation by a normalizing flow network. As a result, achieving strong representation power on the specific class of road vehicles is made possible by compact networks with only 16-dimensional latent codes. Furthermore, the newly proposed architecture exhibits a significant performance improvement in the presence of only sparse and noisy data, which is demonstrated through comparative experiments on synthetic data. The module is embedded into the back-end of a stereo-vision based framework for joint, incremental shape optimization. The loss function is given by a combination of a sparse 3D point-based SDF loss, a sparse rendering loss, and a semantic mask-based silhouette-consistency term. We furthermore leverage semantic information to determine keypoint extraction density in the front-end. Finally, experimental results on real-world data reveal accurate and reliable performance comparable to alternative frameworks that make use of direct depth readings. The proposed method performs well with only sparse 3D points obtained from bundle adjustment, and eventually continues to deliver stable results even under exclusive use of the mask-consistency term.
Authors: Leqi Shen, Tao He, Guoqiang Gong, Fan Yang, Yifeng Zhang, Pengzhang Liu, Sicheng Zhao, Guiguang Ding
Abstract: Training-free video large language models (LLMs) leverage pretrained Image LLMs to process video content without the need for further training. A key challenge in such approaches is the difficulty of retaining essential visual and temporal information, constrained by the token limits in Image LLMs. To address this, we propose a two-stage method for selecting query-relevant tokens based on the LLM attention scores: compressing the video sequence and then expanding the sequence. However, during the compression stage, Image LLMs often exhibit a positional attention bias in video sequences, where attention is overly concentrated on later frames, causing early-frame information to be underutilized. To alleviate this attention bias during sequence compression, we propose Gridded Attention Pooling for preserving spatiotemporal structure. Additionally, we introduce Visual Summarization Tail to effectively utilize this bias, facilitating overall video understanding during sequence expansion. In this way, our method effectively Mitigates and Leverages attention Bias (LLaVA-MLB), enabling the frozen Image LLM for detailed video understanding. Experiments on several benchmarks demonstrate that our approach outperforms state-of-the-art methods, achieving superior performance in both efficiency and accuracy. Our code will be released.
Authors: Fengchen He, Dayang Zhao, Hao Xu, Tingwei Quan, Shaoqun Zeng
Abstract: Many studies utilize dual-pixel (DP) sensor phase characteristics for various applications, such as depth estimation and deblurring. However, since the DP image features are entirely determined by the camera hardware, DP-depth paired datasets are very scarce, especially when performing depth estimation on customized cameras. To overcome this, studies simulate DP images using ideal optical system models. However, these simulations often violate real optical propagation laws,leading to poor generalization to real DP data. To address this, we investigate the domain gap between simulated and real DP data, and propose solutions using the Simulating DP images from ray tracing (Sdirt) scheme. The Sdirt generates realistic DP images via ray tracing and integrates them into the depth estimation training pipeline. Experimental results show that models trained with Sdirt-simulated images generalize better to real DP data.
Authors: Andong Lu, Mai Wen, Jinhu Wang, Yuanzhi Guo, Chenglong Li, Jin Tang, Bin Luo
Abstract: Existing multimodal tracking studies focus on bi-modal scenarios such as RGB-Thermal, RGB-Event, and RGB-Language. Although promising tracking performance is achieved through leveraging complementary cues from different sources, it remains challenging in complex scenes due to the limitations of bi-modal scenarios. In this work, we introduce a general multimodal visual tracking task that fully exploits the advantages of four modalities, including RGB, thermal infrared, event, and language, for robust tracking under challenging conditions. To provide a comprehensive evaluation platform for general multimodal visual tracking, we construct QuadTrack600, a large-scale, high-quality benchmark comprising 600 video sequences (totaling 384.7K high-resolution (640x480) frame groups). In each frame group, all four modalities are spatially aligned and meticulously annotated with bounding boxes, while 21 sequence-level challenge attributes are provided for detailed performance analysis. Despite quad-modal data provides richer information, the differences in information quantity among modalities and the computational burden from four modalities are two challenging issues in fusing four modalities. To handle these issues, we propose a novel approach called QuadFusion, which incorporates an efficient Multiscale Fusion Mamba with four different scanning scales to achieve sufficient interactions of the four modalities while overcoming the exponential computational burden, for general multimodal visual tracking. Extensive experiments on the QuadTrack600 dataset and three bi-modal tracking datasets, including LasHeR, VisEvent, and TNL2K, validate the effectiveness of our QuadFusion.
Authors: Yansheng Li, Yuning Wu, Gong Cheng, Chao Tao, Bo Dang, Yu Wang, Jiahao Zhang, Chuge Zhang, Yiting Liu, Xu Tang, Jiayi Ma, Yongjun Zhang
Abstract: Accurate fine-grained geospatial scene classification using remote sensing imagery is essential for a wide range of applications. However, existing approaches often rely on manually zooming remote sensing images at different scales to create typical scene samples. This approach fails to adequately support the fixed-resolution image interpretation requirements in real-world scenarios. To address this limitation, we introduce the Million-scale finE-grained geospatial scEne classification dataseT (MEET), which contains over 1.03 million zoom-free remote sensing scene samples, manually annotated into 80 fine-grained categories. In MEET, each scene sample follows a scene-inscene layout, where the central scene serves as the reference, and auxiliary scenes provide crucial spatial context for finegrained classification. Moreover, to tackle the emerging challenge of scene-in-scene classification, we present the Context-Aware Transformer (CAT), a model specifically designed for this task, which adaptively fuses spatial context to accurately classify the scene samples. CAT adaptively fuses spatial context to accurately classify the scene samples by learning attentional features that capture the relationships between the center and auxiliary scenes. Based on MEET, we establish a comprehensive benchmark for fine-grained geospatial scene classification, evaluating CAT against 11 competitive baselines. The results demonstrate that CAT significantly outperforms these baselines, achieving a 1.88% higher balanced accuracy (BA) with the Swin-Large backbone, and a notable 7.87% improvement with the Swin-Huge backbone. Further experiments validate the effectiveness of each module in CAT and show the practical applicability of CAT in the urban functional zone mapping. The source code and dataset will be publicly available at https://jerrywyn.github.io/project/MEET.html.
Authors: Du Chen, Tianhe Wu, Kede Ma, Lei Zhang
Abstract: Full-reference image quality assessment (FR-IQA) generally assumes that reference images are of perfect quality. However, this assumption is flawed due to the sensor and optical limitations of modern imaging systems. Moreover, recent generative enhancement methods are capable of producing images of higher quality than their original. All of these challenge the effectiveness and applicability of current FR-IQA models. To relax the assumption of perfect reference image quality, we build a large-scale IQA database, namely DiffIQA, containing approximately 180,000 images generated by a diffusion-based image enhancer with adjustable hyper-parameters. Each image is annotated by human subjects as either worse, similar, or better quality compared to its reference. Building on this, we present a generalized FR-IQA model, namely Adaptive Fidelity-Naturalness Evaluator (A-FINE), to accurately assess and adaptively combine the fidelity and naturalness of a test image. A-FINE aligns well with standard FR-IQA when the reference image is much more natural than the test image. We demonstrate by extensive experiments that A-FINE surpasses standard FR-IQA models on well-established IQA datasets and our newly created DiffIQA. To further validate A-FINE, we additionally construct a super-resolution IQA benchmark (SRIQA-Bench), encompassing test images derived from ten state-of-the-art SR methods with reliable human quality annotations. Tests on SRIQA-Bench re-affirm the advantages of A-FINE. The code and dataset are available at https://tianhewu.github.io/A-FINE-page.github.io/.
Authors: Abbaas Alif Mohamed Nishar, Alireza Marefat, Ashwin Ashok
Abstract: Neuromorphic or event cameras, inspired by biological vision systems, capture changes in illumination with high temporal resolution and efficiency, producing streams of events rather than traditional images. In this paper, we explore the use of neuromorphic cameras for passive optical wireless communication (OWC), leveraging their asynchronous detection of illumination changes to decode data transmitted through reflections of light from objects. We propose a novel system that utilizes neuromorphic cameras for passive visible light communication (VLC), extending the concept to Non Line-of-Sight (NLoS) scenarios through passive reflections from everyday objects. Our experiments demonstrate the feasibility and advantages of using neuromorphic cameras for VLC, characterizing the performance of various modulation schemes, including traditional On-Off Keying (OOK) and advanced N-pulse modulation. We introduce an adaptive N-pulse modulation scheme that dynamically adjusts encoding based on the packet's bit composition, achieving higher data rates and robustness in different scenarios. Our results show that lighter-colored, glossy objects are better for NLoS communication, while larger objects and those with matte finishes experience higher error rates due to multipath reflections.
Authors: Zijing Hu, Fengda Zhang, Long Chen, Kun Kuang, Jiahui Li, Kaifeng Gao, Jun Xiao, Xin Wang, Wenwu Zhu
Abstract: Diffusion models have achieved remarkable success in text-to-image generation. However, their practical applications are hindered by the misalignment between generated images and corresponding text prompts. To tackle this issue, reinforcement learning (RL) has been considered for diffusion model fine-tuning. Yet, RL's effectiveness is limited by the challenge of sparse reward, where feedback is only available at the end of the generation process. This makes it difficult to identify which actions during the denoising process contribute positively to the final generated image, potentially leading to ineffective or unnecessary denoising policies. To this end, this paper presents a novel RL-based framework that addresses the sparse reward problem when training diffusion models. Our framework, named $\text{B}^2\text{-DiffuRL}$, employs two strategies: \textbf{B}ackward progressive training and \textbf{B}ranch-based sampling. For one thing, backward progressive training focuses initially on the final timesteps of denoising process and gradually extends the training interval to earlier timesteps, easing the learning difficulty from sparse rewards. For another, we perform branch-based sampling for each training interval. By comparing the samples within the same branch, we can identify how much the policies of the current training interval contribute to the final image, which helps to learn effective policies instead of unnecessary ones. $\text{B}^2\text{-DiffuRL}$ is compatible with existing optimization algorithms. Extensive experiments demonstrate the effectiveness of $\text{B}^2\text{-DiffuRL}$ in improving prompt-image alignment and maintaining diversity in generated images. The code for this work is available.
Authors: Jun Yu, Xilong Lu
Abstract: Compound Expression Recognition (CER) is crucial for understanding human emotions and improving human-computer interaction. However, CER faces challenges due to the complexity of facial expressions and the difficulty of capturing subtle emotional cues. To address these issues, we propose a novel approach leveraging Large Vision-Language Models (LVLMs). Our method employs a two-stage fine-tuning process: first, pre-trained LVLMs are fine-tuned on basic facial expressions to establish foundational patterns; second, the model is further optimized on a compound-expression dataset to refine visual-language feature interactions. Our approach achieves advanced accuracy on the RAF-DB dataset and demonstrates strong zero-shot generalization on the C-EXPR-DB dataset, showcasing its potential for real-world applications in emotion analysis and human-computer interaction.
Authors: Ziwei Shi, Xiaoran Zhang, Yan Xia, Yu Zang, Siqi Shen, Cheng Wang
Abstract: We tackle the challenge of LiDAR-based place recognition, which traditionally depends on costly and time-consuming prior 3D maps. To overcome this, we first construct XA-L&RSI dataset, which encompasses approximately $110,000$ remote sensing submaps and $13,000$ LiDAR point cloud submaps captured in urban scenes, and propose a novel method, L2RSI, for cross-view LiDAR place recognition using high-resolution Remote Sensing Imagery. This approach enables large-scale localization capabilities at a reduced cost by leveraging readily available overhead images as map proxies. L2RSI addresses the dual challenges of cross-view and cross-modal place recognition by learning feature alignment between point cloud submaps and remote sensing submaps in the semantic domain. Additionally, we introduce a novel probability propagation method based on a dynamic Gaussian mixture model to refine position predictions, effectively leveraging temporal and spatial information. This approach enables large-scale retrieval and cross-scene generalization without fine-tuning. Extensive experiments on XA-L&RSI demonstrate that, within a $100km^2$ retrieval range, L2RSI accurately localizes $95.08\%$ of point cloud submaps within a $30m$ radius for top-$1$ retrieved location. We provide a video to more vividly display the place recognition results of L2RSI at https://shizw695.github.io/L2RSI/.
Authors: Andong Lu, Yuanzhi Guo, Wanyu Wang, Chenglong Li, Jin Tang, Bin Luo
Abstract: Current RGBT tracking methods often overlook the impact of fusion location on mitigating modality gap, which is key factor to effective tracking. Our analysis reveals that shallower fusion yields smaller distribution gap. However, the limited discriminative power of shallow networks hard to distinguish task-relevant information from noise, limiting the potential of pixel-level fusion. To break shallow limits, we propose a novel \textbf{T}ask-driven \textbf{P}ixel-level \textbf{F}usion network, named \textbf{TPF}, which unveils the power of pixel-level fusion in RGBT tracking through a progressive learning framework. In particular, we design a lightweight Pixel-level Fusion Adapter (PFA) that exploits Mamba's linear complexity to ensure real-time, low-latency RGBT tracking. To enhance the fusion capabilities of the PFA, our task-driven progressive learning framework first utilizes adaptive multi-expert distillation to inherits fusion knowledge from state-of-the-art image fusion models, establishing robust initialization, and then employs a decoupled representation learning scheme to achieve task-relevant information fusion. Moreover, to overcome appearance variations between the initial template and search frames, we presents a nearest-neighbor dynamic template updating scheme, which selects the most reliable frame closest to the current search frame as the dynamic template. Extensive experiments demonstrate that TPF significantly outperforms existing most of advanced trackers on four public RGBT tracking datasets. The code will be released upon acceptance.
Authors: Haoyang Huang, Guoqing Ma, Nan Duan, Xing Chen, Changyi Wan, Ranchen Ming, Tianyu Wang, Bo Wang, Zhiying Lu, Aojie Li, Xianfang Zeng, Xinhao Zhang, Gang Yu, Yuhe Yin, Qiling Wu, Wen Sun, Kang An, Xin Han, Deshan Sun, Wei Ji, Bizhu Huang, Brian Li, Chenfei Wu, Guanzhe Huang, Huixin Xiong, Jiaxin He, Jianchang Wu, Jianlong Yuan, Jie Wu, Jiashuai Liu, Junjing Guo, Kaijun Tan, Liangyu Chen, Qiaohui Chen, Ran Sun, Shanshan Yuan, Shengming Yin, Sitong Liu, Wei Chen, Yaqi Dai, Yuchu Luo, Zheng Ge, Zhisheng Guan, Xiaoniu Song, Yu Zhou, Binxing Jiao, Jiansheng Chen, Jing Li, Shuchang Zhou, Xiangyu Zhang, Yi Xiu, Yibo Zhu, Heung-Yeung Shum, Daxin Jiang
Abstract: We present Step-Video-TI2V, a state-of-the-art text-driven image-to-video generation model with 30B parameters, capable of generating videos up to 102 frames based on both text and image inputs. We build Step-Video-TI2V-Eval as a new benchmark for the text-driven image-to-video task and compare Step-Video-TI2V with open-source and commercial TI2V engines using this dataset. Experimental results demonstrate the state-of-the-art performance of Step-Video-TI2V in the image-to-video generation task. Both Step-Video-TI2V and Step-Video-TI2V-Eval are available at https://github.com/stepfun-ai/Step-Video-TI2V.
Authors: Liying Lu, Rapha\"el Achddou, Sabine S\"usstrunk
Abstract: Low-light photography produces images with low signal-to-noise ratios due to limited photons. In such conditions, common approximations like the Gaussian noise model fall short, and many denoising techniques fail to remove noise effectively. Although deep-learning methods perform well, they require large datasets of paired images that are impractical to acquire. As a remedy, synthesizing realistic low-light noise has gained significant attention. In this paper, we investigate the ability of diffusion models to capture the complex distribution of low-light noise. We show that a naive application of conventional diffusion models is inadequate for this task and propose three key adaptations that enable high-precision noise generation without calibration or post-processing: a two-branch architecture to better model signal-dependent and signal-independent noise, the incorporation of positional information to capture fixed-pattern noise, and a tailored diffusion noise schedule. Consequently, our model enables the generation of large datasets for training low-light denoising networks, leading to state-of-the-art performance. Through comprehensive analysis, including statistical evaluation and noise decomposition, we provide deeper insights into the characteristics of the generated data.
Authors: Xirui Zhou, Lianlei Shan, Xiaolin Gui
Abstract: Visual Question Answering (VQA) models, which fall under the category of vision-language models, conventionally execute multiple downsampling processes on image inputs to strike a balance between computational efficiency and model performance. Although this approach aids in concentrating on salient features and diminishing computational burden, it incurs the loss of vital detailed information, a drawback that is particularly damaging in end-to-end autonomous driving scenarios. Downsampling can lead to an inadequate capture of distant or small objects such as pedestrians, road signs, or obstacles, all of which are crucial for safe navigation. This loss of features negatively impacts an autonomous driving system's capacity to accurately perceive the environment, potentially escalating the risk of accidents. To tackle this problem, we put forward the Dynamic Resolution Vision Language Model (DynRsl-VLM). DynRsl-VLM incorporates a dynamic resolution image input processing approach that captures all entity feature information within an image while ensuring that the image input remains computationally tractable for the Vision Transformer (ViT). Moreover, we devise a novel image-text alignment module to replace the Q-Former, enabling simple and efficient alignment with text when dealing with dynamic resolution image inputs. Our method enhances the environmental perception capabilities of autonomous driving systems without overstepping computational constraints.
Authors: Jonas Utz, Stefan Vocht, Anne Tjorven Buessen, Dennis Possart, Fabian Wagner, Mareike Thies, Mingxuan Gu, Stefan Uderhardt, Katharina Breininger
Abstract: In recent years, numerous neural network architectures specifically designed for the instance segmentation of nuclei in microscopic images have been released. These models embed nuclei-specific priors to outperform generic architectures like U-Nets; however, they require large annotated datasets, which are often not available. Generative models (GANs, diffusion models) have been used to compensate for this by synthesizing training data. These two-stage approaches are computationally expensive, as first a generative model and then a segmentation model has to be trained. We propose CyclePose, a hybrid framework integrating synthetic data generation and segmentation training. CyclePose builds on a CycleGAN architecture, which allows unpaired translation between microscopy images and segmentation masks. We embed a segmentation model into CycleGAN and leverage a cycle consistency loss for self-supervision. Without annotated data, CyclePose outperforms other weakly or unsupervised methods on two public datasets. Code is available at https://github.com/jonasutz/CyclePose
Authors: Qi Mao, Haobo Hu, Yujie He, Difei Gao, Haokun Chen, Libiao Jin
Abstract: Affective Image Manipulation (AIM) aims to alter an image's emotional impact by adjusting multiple visual elements to evoke specific feelings.Effective AIM is inherently complex, necessitating a collaborative approach that involves identifying semantic cues within source images, manipulating these elements to elicit desired emotional responses, and verifying that the combined adjustments successfully evoke the target emotion.To address these challenges, we introduce EmoAgent, the first multi-agent collaboration framework for AIM. By emulating the cognitive behaviors of a human painter, EmoAgent incorporates three specialized agents responsible for planning, editing, and critical evaluation. Furthermore, we develop an emotion-factor knowledge retriever, a decision-making tree space, and a tool library to enhance EmoAgent's effectiveness in handling AIM. Experiments demonstrate that the proposed multi-agent framework outperforms existing methods, offering more reasonable and effective emotional expression.
Authors: Yuhao Du, Hui Liu, Haoxiang Peng, Xinyuan Chen, Chenrong Wu, Jiankai Zhang
Abstract: Recent years, weather forecasting has gained significant attention. However, accurately predicting weather remains a challenge due to the rapid variability of meteorological data and potential teleconnections. Current spatiotemporal forecasting models primarily rely on convolution operations or sliding windows for feature extraction. These methods are limited by the size of the convolutional kernel or sliding window, making it difficult to capture and identify potential teleconnection features in meteorological data. Additionally, weather data often involve non-rigid bodies, whose motion processes are accompanied by unpredictable deformations, further complicating the forecasting task. In this paper, we propose the GMG model to address these two core challenges. The Global Focus Module, a key component of our model, enhances the global receptive field, while the Motion Guided Module adapts to the growth or dissipation processes of non-rigid bodies. Through extensive evaluations, our method demonstrates competitive performance across various complex tasks, providing a novel approach to improving the predictive accuracy of complex spatiotemporal data.
Authors: Jeong Hun Yeo, Hyeongseop Rha, Se Jin Park, Yong Man Ro
Abstract: Audio-Visual Speech Recognition (AVSR) achieves robust speech recognition in noisy environments by combining auditory and visual information. However, recent Large Language Model (LLM) based AVSR systems incur high computational costs due to the high temporal resolution of audio-visual speech processed by LLMs. In this work, we introduce an efficient multimodal speech LLM framework that minimizes token length while preserving essential linguistic content. Our approach employs an early av-fusion module for streamlined feature integration, an audio-visual speech Q-Former that dynamically allocates tokens based on input duration, and a refined query allocation strategy with a speech rate predictor to adjust token allocation according to speaking speed of each audio sample. Extensive experiments on the LRS3 dataset show that our method achieves state-of-the-art performance with a WER of 0.74% while using only 3.5 tokens per second. Moreover, our approach not only reduces token usage by 86% compared to the previous multimodal speech LLM framework, but also improves computational efficiency by reducing FLOPs by 35.7%.
Authors: Joona Kareinen, Annaliina Skytt\"a, Tuomas Eerola, Kaisa Kraft, Lasse Lensu, Sanna Suikkanen, Maiju Lehtiniemi, Heikki K\"alvi\"ainen
Abstract: This paper considers open-set recognition (OSR) of plankton images. Plankton include a diverse range of microscopic aquatic organisms that have an important role in marine ecosystems as primary producers and as a base of food webs. Given their sensitivity to environmental changes, fluctuations in plankton populations offer valuable information about oceans' health and climate change motivating their monitoring. Modern automatic plankton imaging devices enable the collection of large-scale plankton image datasets, facilitating species-level analysis. Plankton species recognition can be seen as an image classification task and is typically solved using deep learning-based image recognition models. However, data collection in real aquatic environments results in imaging devices capturing a variety of non-plankton particles and plankton species not present in the training set. This creates a challenging fine-grained OSR problem, characterized by subtle differences between taxonomically close plankton species. We address this challenge by conducting extensive experiments on three OSR approaches using both phyto- and zooplankton images analyzing also on the effect of the rejection thresholds for OSR. The results demonstrate that high OSR accuracy can be obtained promoting the use of these methods in operational plankton research. We have made the data publicly available to the research community.
Authors: Lingyu Zhu, Xiangrui Zeng, Bolin Chen, Peilin Chen, Yung-Hui Li, Shiqi Wang
Abstract: By optimizing the rate-distortion-realism trade-off, generative image compression approaches produce detailed, realistic images instead of the only sharp-looking reconstructions produced by rate-distortion-optimized models. In this paper, we propose a novel deep learning-based generative image compression method injected with diffusion knowledge, obtaining the capacity to recover more realistic textures in practical scenarios. Efforts are made from three perspectives to navigate the rate-distortion-realism trade-off in the generative image compression task. First, recognizing the strong connection between image texture and frequency-domain characteristics, we design a Fractal Frequency-Aware Band Image Compression (FFAB-IC) network to effectively capture the directional frequency components inherent in natural images. This network integrates commonly used fractal band feature operations within a neural non-linear mapping design, enhancing its ability to retain essential given information and filter out unnecessary details. Then, to improve the visual quality of image reconstruction under limited bandwidth, we integrate diffusion knowledge into the encoder and implement diffusion iterations into the decoder process, thus effectively recovering lost texture details. Finally, to fully leverage the spatial and frequency intensity information, we incorporate frequency- and content-aware regularization terms to regularize the training of the generative image compression network. Extensive experiments in quantitative and qualitative evaluations demonstrate the superiority of the proposed method, advancing the boundaries of achievable distortion-realism pairs, i.e., our method achieves better distortions at high realism and better realism at low distortion than ever before.
Authors: Ruiqian Li, Siyuan Shen, Suan Xia, Ziheng Wang, Xingyue Peng, Chengxuan Song, Yingsheng Zhu, Tao Wu, Shiying Li, Jingyi Yu
Abstract: High quality and high speed videography using Non-Line-of-Sight (NLOS) imaging benefit autonomous navigation, collision prevention, and post-disaster search and rescue tasks. Current solutions have to balance between the frame rate and image quality. High frame rates, for example, can be achieved by reducing either per-point scanning time or scanning density, but at the cost of lowering the information density at individual frames. Fast scanning process further reduces the signal-to-noise ratio and different scanning systems exhibit different distortion characteristics. In this work, we design and employ a new Transient Transformer architecture called TransiT to achieve real-time NLOS recovery under fast scans. TransiT directly compresses the temporal dimension of input transients to extract features, reducing computation costs and meeting high frame rate requirements. It further adopts a feature fusion mechanism as well as employs a spatial-temporal Transformer to help capture features of NLOS transient videos. Moreover, TransiT applies transfer learning to bridge the gap between synthetic and real-measured data. In real experiments, TransiT manages to reconstruct from sparse transients of $16 \times 16$ measured at an exposure time of 0.4 ms per point to NLOS videos at a $64 \times 64$ resolution at 10 frames per second. We will make our code and dataset available to the community.
Authors: Marvin Kahra, Michael Breu{\ss}
Abstract: Mathematical morphology, a field within image processing, includes various filters that either highlight, modify, or eliminate certain information in images based on an application's needs. Key operations in these filters are dilation and erosion, which determine the supremum or infimum for each pixel with respect to an order of the tonal values over a subset of the image surrounding the pixel. This subset is formed by a structuring element at the specified pixel, which weighs the tonal values. Unlike grey-scale morphology, where tonal order is clearly defined, colour morphology lacks a definitive total order. As no method fully meets all desired properties for colour, because of this difficulty, some limitations are always present. This paper shows how to combine the theory of the log-exp-supremum of colour matrices that employs the Loewner semi-order with a well-known colour distance approach in the form of a pre-ordering. The log-exp-supremum will therefore serve as the reference colour for determining the colour distance. To the resulting pre-ordering with respect to these distance values, we add a lexicographic cascade to ensure a total order and a unique result. The objective of this approach is to identify the original colour within the structuring element that most closely resembles a supremum, which fulfils a number of desired properties. Consequently, this approach avoids the false-colour problem. The behaviour of the introduced operators is illustrated by application examples of dilation and closing for synthetic and natural images.
Authors: Moein Sorkhei, Emir Konuk, Kevin Smith, Christos Matsoukas
Abstract: Existing adaptation techniques typically require architectural modifications or added parameters, leading to high computational costs and complexity. We introduce Attention Projection Layer Adaptation (APLA), a simple approach to adapt vision transformers (ViTs) without altering the architecture or adding parameters. Through a systematic analysis, we find that the layer immediately after the attention mechanism is crucial for adaptation. By updating only this projection layer, or even just a random subset of this layer's weights, APLA achieves state-of-the-art performance while reducing GPU memory usage by up to 52.63% and training time by up to 43.0%, with no extra cost at inference. Across 46 datasets covering a variety of tasks including scene classification, medical imaging, satellite imaging, and fine-grained classification, APLA consistently outperforms 17 other leading adaptation methods, including full fine-tuning, on classification, segmentation, and detection tasks. The code is available at https://github.com/MoeinSorkhei/APLA.
Authors: Joona Kareinen, Tuomas Eerola, Kaisa Kraft, Lasse Lensu, Sanna Suikkanen, Heikki K\"alvi\"ainen
Abstract: Plankton recognition is an important computer vision problem due to plankton's essential role in ocean food webs and carbon capture, highlighting the need for species-level monitoring. However, this task is challenging due to its fine-grained nature and dataset shifts caused by different imaging instruments and varying species distributions. As new plankton image datasets are collected at an increasing pace, there is a need for general plankton recognition models that require minimal expert effort for data labeling. In this work, we study large-scale self-supervised pretraining for fine-grained plankton recognition. We first employ masked autoencoding and a large volume of diverse plankton image data to pretrain a general-purpose plankton image encoder. Then we utilize fine-tuning to obtain accurate plankton recognition models for new datasets with a very limited number of labeled training images. Our experiments show that self-supervised pretraining with diverse plankton data clearly increases plankton recognition accuracy compared to standard ImageNet pretraining when the amount of training data is limited. Moreover, the accuracy can be further improved when unlabeled target data is available and utilized during the pretraining.
Authors: Yibing Weng, Yu Gu, Fuji Ren
Abstract: Road rage, triggered by driving-related stimuli such as traffic congestion and aggressive driving, poses a significant threat to road safety. Previous research on road rage regulation has primarily focused on response suppression, lacking proactive prevention capabilities. With the advent of Vision-Language Models (VLMs), it has become possible to reason about trigger events visually and then engage in dialog-based comforting before drivers' anger escalates. To this end, we propose the road rage reasoning task, along with a finely annotated test dataset and evaluation metrics, to assess the capabilities of current mainstream VLMs in scene understanding, event recognition, and road rage reasoning. The results indicate that current VLMs exhibit significant shortcomings in scene understanding within the visual modality, as well as in comprehending the spatial relationships between objects in the textual modality. Improving VLMs' performance in these areas will greatly benefit downstream tasks like antecedent-focused road rage regulation.
Authors: Di Li, Jie Feng, Jiahao Chen, Weisheng Dong, Guanbin Li, Guangming Shi, Licheng Jiao
Abstract: Egocentric scenes exhibit frequent occlusions, varied viewpoints, and dynamic interactions compared to typical scene understanding tasks. Occlusions and varied viewpoints can lead to multi-view semantic inconsistencies, while dynamic objects may act as transient distractors, introducing artifacts into semantic feature modeling. To address these challenges, we propose EgoSplat, a language-embedded 3D Gaussian Splatting framework for open-vocabulary egocentric scene understanding. A multi-view consistent instance feature aggregation method is designed to leverage the segmentation and tracking capabilities of SAM2 to selectively aggregate complementary features across views for each instance, ensuring precise semantic representation of scenes. Additionally, an instance-aware spatial-temporal transient prediction module is constructed to improve spatial integrity and temporal continuity in predictions by incorporating spatial-temporal associations across multi-view instances, effectively reducing artifacts in the semantic reconstruction of egocentric scenes. EgoSplat achieves state-of-the-art performance in both localization and segmentation tasks on two datasets, outperforming existing methods with a 8.2% improvement in localization accuracy and a 3.7% improvement in segmentation mIoU on the ADT dataset, and setting a new benchmark in open-vocabulary egocentric scene understanding. The code will be made publicly available.
Authors: Mayank Nautiyal, Stela Arranz Gheorghe, Kristiana Stefa, Li Ju, Ida-Maria Sintorn, Prashant Singh
Abstract: Language-guided attention frameworks have significantly enhanced both interpretability and performance in image classification; however, the reliance on deterministic embeddings from pre-trained vision-language foundation models to generate reference attention maps frequently overlooks the intrinsic multivaluedness and ill-posed characteristics of cross-modal mappings. To address these limitations, we introduce PARIC, a probabilistic framework for guiding visual attention via language specifications. Our approach enables pre-trained vision-language models to generate probabilistic reference attention maps, which align textual and visual modalities more effectively while incorporating uncertainty estimates, as compared to their deterministic counterparts. Experiments on benchmark test problems demonstrate that PARIC enhances prediction accuracy, mitigates bias, ensures consistent predictions, and improves robustness across various datasets.
Authors: Xiaokang Wei, Bowen Zhang, Xianghui Yang, Yuxuan Wang, Chunchao Guo, Xi Zhao, Yan Luximon
Abstract: Generating high-quality physically based rendering (PBR) materials is important to achieve realistic rendering in the downstream tasks, yet it remains challenging due to the intertwined effects of materials and lighting. While existing methods have made breakthroughs by incorporating material decomposition in the 3D generation pipeline, they tend to bake highlights into albedo and ignore spatially varying properties of metallicity and roughness. In this work, we present PBR3DGen, a two-stage mesh generation method with high-quality PBR materials that integrates the novel multi-view PBR material estimation model and a 3D PBR mesh reconstruction model. Specifically, PBR3DGen leverages vision language models (VLM) to guide multi-view diffusion, precisely capturing the spatial distribution and inherent attributes of reflective-metalness material. Additionally, we incorporate view-dependent illumination-aware conditions as pixel-aware priors to enhance spatially varying material properties. Furthermore, our reconstruction model reconstructs high-quality mesh with PBR materials. Experimental results demonstrate that PBR3DGen significantly outperforms existing methods, achieving new state-of-the-art results for PBR estimation and mesh generation. More results and visualization can be found on our project page: https://pbr3dgen1218.github.io/.
Authors: Zengyu Wan, Wei Zhai, Yang Cao, Zhengjun Zha
Abstract: Visual 3D motion estimation aims to infer the motion of 2D pixels in 3D space based on visual cues. The key challenge arises from depth variation induced spatio-temporal motion inconsistencies, disrupting the assumptions of local spatial or temporal motion smoothness in previous motion estimation frameworks. In contrast, event cameras offer new possibilities for 3D motion estimation through continuous adaptive pixel-level responses to scene changes. This paper presents EMoTive, a novel event-based framework that models spatio-temporal trajectories via event-guided non-uniform parametric curves, effectively characterizing locally heterogeneous spatio-temporal motion. Specifically, we first introduce Event Kymograph - an event projection method that leverages a continuous temporal projection kernel and decouples spatial observations to encode fine-grained temporal evolution explicitly. For motion representation, we introduce a density-aware adaptation mechanism to fuse spatial and temporal features under event guidance, coupled with a non-uniform rational curve parameterization framework to adaptively model heterogeneous trajectories. The final 3D motion estimation is achieved through multi-temporal sampling of parametric trajectories, yielding optical flow and depth motion fields. To facilitate evaluation, we introduce CarlaEvent3D, a multi-dynamic synthetic dataset for comprehensive validation. Extensive experiments on both this dataset and a real-world benchmark demonstrate the effectiveness of the proposed method.
Authors: Lukas Kroi{\ss}, Johannes Reschke
Abstract: Fake News and especially deepfakes (generated, non-real image or video content) have become a serious topic over the last years. With the emergence of machine learning algorithms it is now easier than ever before to generate such fake content, even for private persons. This issue of generated fake images is especially critical in the context of politics and public figures. We want to address this conflict by building a model based on a Convolutions Neural Network in order to detect such generated and fake images showing human portraits. As a basis, we use a pre-trained ResNet-50 model due to its effectiveness in terms of classifying images. We then adopted the base model to our task of classifying a single image as authentic/real or fake by adding an fully connected output layer containing a single neuron indicating the authenticity of an image. We applied fine tuning and transfer learning to develop the model and improve its parameters. For the training process we collected the image data set "Diverse Face Fake Dataset" containing a wide range of different image manipulation methods and also diversity in terms of faces visible on the images. With our final model we reached the following outstanding performance metrics: precision = 0.98, recall 0.96, F1-Score = 0.97 and an area-under-curve = 0.99.
Authors: David Gastager, Ghazal Ghazaei, Constantin Patsch
Abstract: Automated surgical workflow analysis is crucial for education, research, and clinical decision-making, but the lack of annotated datasets hinders the development of accurate and comprehensive workflow analysis solutions. We introduce a novel approach for addressing the sparsity and heterogeneity of annotated training data inspired by the human learning procedure of watching experts and understanding their explanations. Our method leverages a video-language model trained on alignment, denoising, and generative tasks to learn short-term spatio-temporal and multimodal representations. A task-specific temporal model is then used to capture relationships across entire videos. To achieve comprehensive video-language understanding in the surgical domain, we introduce a data collection and filtering strategy to construct a large-scale pretraining dataset from educational YouTube videos. We then utilize parameter-efficient fine-tuning by projecting downstream task annotations from publicly available surgical datasets into the language domain. Extensive experiments in two surgical domains demonstrate the effectiveness of our approach, with performance improvements of up to 7% in phase segmentation tasks, 8% in zero-shot phase segmentation, and comparable capabilities to fully-supervised models in few-shot settings. Harnessing our model's capabilities for long-range temporal localization and text generation, we present the first comprehensive solution for dense video captioning (DVC) of surgical videos, addressing this task despite the absence of existing DVC datasets in the surgical domain. We introduce a novel approach to surgical workflow understanding that leverages video-language pretraining, large-scale video pretraining, and optimized fine-tuning. Our method improves performance over state-of-the-art techniques and enables new downstream tasks for surgical video understanding.
Authors: Tin Stribor Sohn, Philipp Reis, Maximilian Dillitzer, Johannes Bach, Jason J. Corso, Eric Sax
Abstract: Multimodal large language models (MLLMs) hold the potential to enhance autonomous driving by combining domain-independent world knowledge with context-specific language guidance. Their integration into autonomous driving systems shows promising results in isolated proof-of-concept applications, while their performance is evaluated on selective singular aspects of perception, reasoning, or planning. To leverage their full potential a systematic framework for evaluating MLLMs in the context of autonomous driving is required. This paper proposes a holistic framework for a capability-driven evaluation of MLLMs in autonomous driving. The framework structures scenario understanding along the four core capability dimensions semantic, spatial, temporal, and physical. They are derived from the general requirements of autonomous driving systems, human driver cognition, and language-based reasoning. It further organises the domain into context layers, processing modalities, and downstream tasks such as language-based interaction and decision-making. To illustrate the framework's applicability, two exemplary traffic scenarios are analysed, grounding the proposed dimensions in realistic driving situations. The framework provides a foundation for the structured evaluation of MLLMs' potential for scenario understanding in autonomous driving.
Authors: Shuaifeng Jiao, Zhiwen Zeng, Zhuoqun Su, Xieyuanli Chen, Zongtan Zhou, Huimin Lu
Abstract: As lunar exploration missions grow increasingly complex, ensuring safe and autonomous rover-based surface exploration has become one of the key challenges in lunar exploration tasks. In this work, we have developed a lunar surface simulation system called the Lunar Exploration Simulator System (LESS) and the LunarSeg dataset, which provides RGB-D data for lunar obstacle segmentation that includes both positive and negative obstacles. Additionally, we propose a novel two-stage segmentation network called LuSeg. Through contrastive learning, it enforces semantic consistency between the RGB encoder from Stage I and the depth encoder from Stage II. Experimental results on our proposed LunarSeg dataset and additional public real-world NPO road obstacle dataset demonstrate that LuSeg achieves state-of-the-art segmentation performance for both positive and negative obstacles while maintaining a high inference speed of approximately 57\,Hz. We have released the implementation of our LESS system, LunarSeg dataset, and the code of LuSeg at:https://github.com/nubot-nudt/LuSeg.
Authors: Shiyuan Yang, Zheng Gu, Liang Hou, Xin Tao, Pengfei Wan, Xiaodong Chen, Jing Liao
Abstract: Video inpainting involves modifying local regions within a video, ensuring spatial and temporal consistency. Most existing methods focus primarily on scene completion (i.e., filling missing regions) and lack the capability to insert new objects into a scene in a controllable manner. Fortunately, recent advancements in text-to-video (T2V) diffusion models pave the way for text-guided video inpainting. However, directly adapting T2V models for inpainting remains limited in unifying completion and insertion tasks, lacks input controllability, and struggles with long videos, thereby restricting their applicability and flexibility. To address these challenges, we propose MTV-Inpaint, a unified multi-task video inpainting framework capable of handling both traditional scene completion and novel object insertion tasks. To unify these distinct tasks, we design a dual-branch spatial attention mechanism in the T2V diffusion U-Net, enabling seamless integration of scene completion and object insertion within a single framework. In addition to textual guidance, MTV-Inpaint supports multimodal control by integrating various image inpainting models through our proposed image-to-video (I2V) inpainting mode. Additionally, we propose a two-stage pipeline that combines keyframe inpainting with in-between frame propagation, enabling MTV-Inpaint to effectively handle long videos with hundreds of frames. Extensive experiments demonstrate that MTV-Inpaint achieves state-of-the-art performance in both scene completion and object insertion tasks. Furthermore, it demonstrates versatility in derived applications such as multi-modal inpainting, object editing, removal, image object brush, and the ability to handle long videos. Project page: https://mtv-inpaint.github.io/.
Authors: Hongxiang Zhao, Xingchen Liu, Mutian Xu, Yiming Hao, Weikai Chen, Xiaoguang Han
Abstract: We address key limitations in existing datasets and models for task-oriented hand-object interaction video generation, a critical approach of generating video demonstrations for robotic imitation learning. Current datasets, such as Ego4D, often suffer from inconsistent view perspectives and misaligned interactions, leading to reduced video quality and limiting their applicability for precise imitation learning tasks. Towards this end, we introduce TASTE-Rob -- a pioneering large-scale dataset of 100,856 ego-centric hand-object interaction videos. Each video is meticulously aligned with language instructions and recorded from a consistent camera viewpoint to ensure interaction clarity. By fine-tuning a Video Diffusion Model (VDM) on TASTE-Rob, we achieve realistic object interactions, though we observed occasional inconsistencies in hand grasping postures. To enhance realism, we introduce a three-stage pose-refinement pipeline that improves hand posture accuracy in generated videos. Our curated dataset, coupled with the specialized pose-refinement framework, provides notable performance gains in generating high-quality, task-oriented hand-object interaction videos, resulting in achieving superior generalizable robotic manipulation. The TASTE-Rob dataset will be made publicly available upon publication to foster further advancements in the field.
Authors: Sanghyun Jo, Seo Jin Lee, Seungwoo Lee, Seohyung Hong, Hyungseok Seo, Kyungsu Kim
Abstract: Cell instance segmentation (CIS) is crucial for identifying individual cell morphologies in histopathological images, providing valuable insights for biological and medical research. While unsupervised CIS (UCIS) models aim to reduce the heavy reliance on labor-intensive image annotations, they fail to accurately capture cell boundaries, causing missed detections and poor performance. Recognizing the absence of error-free instances as a key limitation, we present COIN (COnfidence score-guided INstance distillation), a novel annotation-free framework with three key steps: (1) Increasing the sensitivity for the presence of error-free instances via unsupervised semantic segmentation with optimal transport, leveraging its ability to discriminate spatially minor instances, (2) Instance-level confidence scoring to measure the consistency between model prediction and refined mask and identify highly confident instances, offering an alternative to ground truth annotations, and (3) Progressive expansion of confidence with recursive self-distillation. Extensive experiments across six datasets show COIN outperforming existing UCIS methods, even surpassing semi- and weakly-supervised approaches across all metrics on the MoNuSeg and TNBC datasets. The code is available at https://github.com/shjo-april/COIN.
Authors: Hang Shao, Lei Luo, Jianjun Qian, Mengkai Yan, Shuo Chen, Jian Yang
Abstract: Physiological activities can be manifested by the sensitive changes in facial imaging. While they are barely observable to our eyes, computer vision manners can, and the derived remote photoplethysmography (rPPG) has shown considerable promise. However, existing studies mainly rely on spatial skin recognition and temporal rhythmic interactions, so they focus on identifying explicit features under ideal light conditions, but perform poorly in-the-wild with intricate obstacles and extreme illumination exposure. In this paper, we propose an end-to-end video transformer model for rPPG. It strives to eliminate complex and unknown external time-varying interferences, whether they are sufficient to occupy subtle biosignal amplitudes or exist as periodic perturbations that hinder network training. In the specific implementation, we utilize global interference sharing, subject background reference, and self-supervised disentanglement to eliminate interference, and further guide learning based on spatiotemporal filtering, reconstruction guidance, and frequency domain and biological prior constraints to achieve effective rPPG. To the best of our knowledge, this is the first robust rPPG model for real outdoor scenarios based on natural face videos, and is lightweight to deploy. Extensive experiments show the competitiveness and performance of our model in rPPG prediction across datasets and scenes.
Authors: Seyed Mohammad Hadi Hosseini, Amir Mohammad Izadi, Ali Abdollahi, Armin Saghafian, Mahdieh Soleymani Baghshah
Abstract: Although recent text-to-image generative models have achieved impressive performance, they still often struggle with capturing the compositional complexities of prompts including attribute binding, and spatial relationships between different entities. This misalignment is not revealed by common evaluation metrics such as CLIPScore. Recent works have proposed evaluation metrics that utilize Visual Question Answering (VQA) by decomposing prompts into questions about the generated image for more robust compositional evaluation. Although these methods align better with human evaluations, they still fail to fully cover the compositionality within the image. To address this, we propose a novel metric that breaks down images into components, and texts into fine-grained questions about the generated image for evaluation. Our method outperforms previous state-of-the-art metrics, demonstrating its effectiveness in evaluating text-to-image generative models. Code is available at https://github.com/hadi-hosseini/ T2I-FineEval.
Authors: Zixu Cheng, Jian Hu, Ziquan Liu, Chenyang Si, Wei Li, Shaogang Gong
Abstract: Human processes video reasoning in a sequential spatio-temporal reasoning logic, we first identify the relevant frames ("when") and then analyse the spatial relationships ("where") between key objects, and finally leverage these relationships to draw inferences ("what"). However, can Video Large Language Models (Video-LLMs) also "reason through a sequential spatio-temporal logic" in videos? Existing Video-LLM benchmarks primarily focus on assessing object presence, neglecting relational reasoning. Consequently, it is difficult to measure whether a model truly comprehends object interactions (actions/events) in videos or merely relies on pre-trained "memory" of co-occurrences as biases in generating answers. In this work, we introduce a Video Spatio-Temporal Reasoning (V-STaR) benchmark to address these shortcomings. The key idea is to decompose video understanding into a Reverse Spatio-Temporal Reasoning (RSTR) task that simultaneously evaluates what objects are present, when events occur, and where they are located while capturing the underlying Chain-of-thought (CoT) logic. To support this evaluation, we construct a dataset to elicit the spatial-temporal reasoning process of Video-LLMs. It contains coarse-to-fine CoT questions generated by a semi-automated GPT-4-powered pipeline, embedding explicit reasoning chains to mimic human cognition. Experiments from 14 Video-LLMs on our V-STaR reveal significant gaps between current Video-LLMs and the needs for robust and consistent spatio-temporal reasoning.
Authors: Shaofeng Liang, Runwei Guan, Wangwang Lian, Daizong Liu, Xiaolou Sun, Dongming Wu, Yutao Yue, Weiping Ding, Hui Xiong
Abstract: As a significant application of multi-source information fusion in intelligent transportation perception systems, Referring Multi-Object Tracking (RMOT) involves localizing and tracking specific objects in video sequences based on language references. However, existing RMOT approaches often treat language descriptions as holistic embeddings and struggle to effectively integrate the rich semantic information contained in language expressions with visual features. This limitation is especially apparent in complex scenes requiring comprehensive understanding of both static object attributes and spatial motion information. In this paper, we propose a Cognitive Disentanglement for Referring Multi-Object Tracking (CDRMT) framework that addresses these challenges. It adapts the "what" and "where" pathways from human visual processing system to RMOT tasks. Specifically, our framework comprises three collaborative components: (1)The Bidirectional Interactive Fusion module first establishes cross-modal connections while preserving modality-specific characteristics; (2) Building upon this foundation, the Progressive Semantic-Decoupled Query Learning mechanism hierarchically injects complementary information into object queries, progressively refining object understanding from coarse to fine-grained semantic levels; (3) Finally, the Structural Consensus Constraint enforces bidirectional semantic consistency between visual features and language descriptions, ensuring that tracked objects faithfully reflect the referring expression. Extensive experiments on different benchmark datasets demonstrate that CDRMT achieves substantial improvements over state-of-the-art methods, with average gains of 6.0% in HOTA score on Refer-KITTI and 3.2% on Refer-KITTI-V2. Our approach advances the state-of-the-art in RMOT while simultaneously providing new insights into multi-source information fusion.
Authors: Sl\'avek Zbirovsk\'y, V\'aclav Ne\v{z}erka
Abstract: Building Information Modeling (BIM) is an essential component in the sustainable reconstruction and revitalization of ageing structures. However, model creation usually relies on laborious manual transformation of the unstructured point cloud data provided by laser scans or photogrammetry. This paper presents Cloud2BIM, an open-source software tool designed to automate the conversion of point clouds into BIM models compliant with the Industry Foundation Classes (IFC) standard. Cloud2BIM integrates advanced algorithms for wall and slab segmentation, opening detection, and room zoning based on real wall surfaces, resulting in a comprehensive and fully automated workflow. Unlike existing tools, it avoids computationally- and calibration-intensive techniques such as RANSAC, supports non-orthogonal geometries, and provides unprecedented processing speed-achieving results up to seven times faster than fastest competing solutions. Systematic validation using benchmark datasets confirms that Cloud2BIM is an easy-to-use, efficient, and scalable solution for generating accurate BIM models, capable of converting extensive point cloud datasets for entire buildings into IFC format with minimal user input.
Authors: Ziqin Zhou, Yifan Yang, Yuqing Yang, Tianyu He, Houwen Peng, Kai Qiu, Qi Dai, Lili Qiu, Chong Luo, Lingqiao Liu
Abstract: Text-to-video generation poses significant challenges due to the inherent complexity of video data, which spans both temporal and spatial dimensions. It introduces additional redundancy, abrupt variations, and a domain gap between language and vision tokens while generation. Addressing these challenges requires an effective video tokenizer that can efficiently encode video data while preserving essential semantic and spatiotemporal information, serving as a critical bridge between text and vision. Inspired by the observation in VQ-VAE-2 and workflows of traditional animation, we propose HiTVideo for text-to-video generation with hierarchical tokenizers. It utilizes a 3D causal VAE with a multi-layer discrete token framework, encoding video content into hierarchically structured codebooks. Higher layers capture semantic information with higher compression, while lower layers focus on fine-grained spatiotemporal details, striking a balance between compression efficiency and reconstruction quality. Our approach efficiently encodes longer video sequences (e.g., 8 seconds, 64 frames), reducing bits per pixel (bpp) by approximately 70\% compared to baseline tokenizers, while maintaining competitive reconstruction quality. We explore the trade-offs between compression and reconstruction, while emphasizing the advantages of high-compressed semantic tokens in text-to-video tasks. HiTVideo aims to address the potential limitations of existing video tokenizers in text-to-video generation tasks, striving for higher compression ratios and simplify LLMs modeling under language guidance, offering a scalable and promising framework for advancing text to video generation. Demo page: https://ziqinzhou66.github.io/project/HiTVideo.
Authors: Hao Cheng, Erjia Xiao, Yichi Wang, Kaidi Xu, Mengshu Sun, Jindong Gu, Renjing Xu
Abstract: Current Cross-Modality Generation Models (GMs) demonstrate remarkable capabilities in various generative tasks. Given the ubiquity and information richness of vision modality inputs in real-world scenarios, Cross-vision, encompassing Vision-Language Perception (VLP) and Image-to-Image (I2I), tasks have attracted significant attention. Large Vision Language Models (LVLMs) and I2I GMs are employed to handle VLP and I2I tasks, respectively. Previous research indicates that printing typographic words into input images significantly induces LVLMs and I2I GMs to generate disruptive outputs semantically related to those words. Additionally, visual prompts, as a more sophisticated form of typography, are also revealed to pose security risks to various applications of VLP tasks when injected into images. In this paper, we comprehensively investigate the performance impact induced by Typographic Visual Prompt Injection (TVPI) in various LVLMs and I2I GMs. To better observe performance modifications and characteristics of this threat, we also introduce the TVPI Dataset. Through extensive explorations, we deepen the understanding of the underlying causes of the TVPI threat in various GMs and offer valuable insights into its potential origins.
Authors: Ayush Paliwal, Oliver Schlenczek, Birte Thiede, Manuel Santos Pereira, Katja Stieger, Eberhard Bodenschatz, Gholamhossein Bagheri, Alexander Ecker
Abstract: Reconstructing the 3D location and size of microparticles from diffraction images - holograms - is a computationally expensive inverse problem that has traditionally been solved using physics-based reconstruction methods. More recently, researchers have used machine learning methods to speed up the process. However, for small particles in large sample volumes the performance of these methods falls short of standard physics-based reconstruction methods. Here we designed a two-stage neural network architecture, FLASH$\mu$, to detect small particles (6-100$\mu$m) from holograms with large sample depths up to 20cm. Trained only on synthetic data with added physical noise, our method reliably detects particles of at least 9$\mu$m diameter in real holograms, comparable to the standard reconstruction-based approaches while operating on smaller crops, at quarter of the original resolution and providing roughly a 600-fold speedup. In addition to introducing a novel approach to a non-local object detection or signal demixing problem, our work could enable low-cost, real-time holographic imaging setups.
Authors: Parsa Rahimi, Damien Teney, Sebastien Marcel
Abstract: The increasing dependence on large-scale datasets in machine learning introduces significant privacy and ethical challenges. Synthetic data generation offers a promising solution; however, most current methods rely on external datasets or pre-trained models, which add complexity and escalate resource demands. In this work, we introduce a novel self-contained synthetic augmentation technique that strategically samples from a conditional generative model trained exclusively on the target dataset. This approach eliminates the need for auxiliary data sources. Applied to face recognition datasets, our method achieves 1--12\% performance improvements on the IJB-C and IJB-B benchmarks. It outperforms models trained solely on real data and exceeds the performance of state-of-the-art synthetic data generation baselines. Notably, these enhancements often surpass those achieved through architectural improvements, underscoring the significant impact of synthetic augmentation in data-scarce environments. These findings demonstrate that carefully integrated synthetic data not only addresses privacy and resource constraints but also substantially boosts model performance. Project page https://parsa-ra.github.io/auggen
Authors: Ahmadreza Jeddi, Negin Baghbanzadeh, Elham Dolatabadi, Babak Taati
Abstract: The computational demands of Vision Transformers (ViTs) and Vision-Language Models (VLMs) remain a significant challenge due to the quadratic complexity of self-attention. While token pruning offers a promising solution, existing methods often introduce training overhead or fail to adapt dynamically across layers. We present SAINT, a training-free token pruning framework that leverages token similarity and a graph-based formulation to dynamically optimize pruning rates and redundancy thresholds. Through systematic analysis, we identify a universal three-stage token evolution process (aligner-explorer-aggregator) in transformers, enabling aggressive pruning in early stages without sacrificing critical information. For ViTs, SAINT doubles the throughput of ViT-H/14 at 224px with only 0.6% accuracy loss on ImageNet-1K, surpassing the closest competitor by 0.8%. For VLMs, we apply SAINT in three modes: ViT-only, LLM-only, and hybrid. SAINT reduces LLaVA-13B's tokens by 75%, achieving latency comparable to LLaVA-7B with less than 1% performance loss across benchmarks. Our work establishes a unified, practical framework for efficient inference in ViTs and VLMs.
Authors: Jing Bi, Junjia Guo, Susan Liang, Guangyu Sun, Luchuan Song, Yunlong Tang, Jinxi He, Jiarui Wu, Ali Vosoughi, Chen Chen, Chenliang Xu
Abstract: Visual reasoning is central to human cognition, enabling individuals to interpret and abstractly understand their environment. Although recent Multimodal Large Language Models (MLLMs) have demonstrated impressive performance across language and vision-language tasks, existing benchmarks primarily measure recognition-based skills and inadequately assess true visual reasoning capabilities. To bridge this critical gap, we introduce VERIFY, a benchmark explicitly designed to isolate and rigorously evaluate the visual reasoning capabilities of state-of-the-art MLLMs. VERIFY compels models to reason primarily from visual information, providing minimal textual context to reduce reliance on domain-specific knowledge and linguistic biases. Each problem is accompanied by a human-annotated reasoning path, making it the first to provide in-depth evaluation of model decision-making processes. Additionally, we propose novel metrics that assess visual reasoning fidelity beyond mere accuracy, highlighting critical imbalances in current model reasoning patterns. Our comprehensive benchmarking of leading MLLMs uncovers significant limitations, underscoring the need for a balanced and holistic approach to both perception and reasoning. For more teaser and testing, visit our project page (https://verify-eqh.pages.dev/).
Authors: David Emukpere, Romain Deffayet, Bingbing Wu, Romain Br\'egier, Michael Niemaz, Jean-Luc Meunier, Denys Proux, Jean-Michel Renders, Seungsu Kim
Abstract: Learning robotic manipulation skills from vision is a promising approach for developing robotics applications that can generalize broadly to real-world scenarios. As such, many approaches to enable this vision have been explored with fruitful results. Particularly, object-centric representation methods have been shown to provide better inductive biases for skill learning, leading to improved performance and generalization. Nonetheless, we show that object-centric methods can struggle to learn simple manipulation skills in multi-object environments. Thus, we propose DOCIR, an object-centric framework that introduces a disentangled representation for objects of interest, obstacles, and robot embodiment. We show that this approach leads to state-of-the-art performance for learning pick and place skills from visual inputs in multi-object environments and generalizes at test time to changing objects of interest and distractors in the scene. Furthermore, we show its efficacy both in simulation and zero-shot transfer to the real world.
Authors: Tianrui Pan, Lin Liu, Jie Liu, Xiaopeng Zhang, Jie Tang, Gangshan Wu, Qi Tian
Abstract: Portrait video editing focuses on modifying specific attributes of portrait videos, guided by audio or video streams. Previous methods typically either concentrate on lip-region reenactment or require training specialized models to extract keypoints for motion transfer to a new identity. In this paper, we introduce a training-free universal portrait video editing framework that provides a versatile and adaptable editing strategy. This framework supports portrait appearance editing conditioned on the changed first reference frame, as well as lip editing conditioned on varied speech, or a combination of both. It is based on a Unified Animation Control (UAC) mechanism with source inversion latents to edit the entire portrait, including visual-driven shape control, audio-driven speaking control, and inter-frame temporal control. Furthermore, our method can be adapted to different scenarios by adjusting the initial reference frame, enabling detailed editing of portrait videos with specific head rotations and facial expressions. This comprehensive approach ensures a holistic and flexible solution for portrait video editing. The experimental results show that our model can achieve more accurate and synchronized lip movements for the lip editing task, as well as more flexible motion transfer for the appearance editing task. Demo is available at https://alice01010101.github.io/RASA/.
Authors: Ahmed Nassar, Andres Marafioti, Matteo Omenetti, Maksym Lysak, Nikolaos Livathinos, Christoph Auer, Lucas Morin, Rafael Teixeira de Lima, Yusik Kim, A. Said Gurbuz, Michele Dolfi, Miquel Farr\'e, Peter W. J. Staar
Abstract: We introduce SmolDocling, an ultra-compact vision-language model targeting end-to-end document conversion. Our model comprehensively processes entire pages by generating DocTags, a new universal markup format that captures all page elements in their full context with location. Unlike existing approaches that rely on large foundational models, or ensemble solutions that rely on handcrafted pipelines of multiple specialized models, SmolDocling offers an end-to-end conversion for accurately capturing content, structure and spatial location of document elements in a 256M parameters vision-language model. SmolDocling exhibits robust performance in correctly reproducing document features such as code listings, tables, equations, charts, lists, and more across a diverse range of document types including business documents, academic papers, technical reports, patents, and forms -- significantly extending beyond the commonly observed focus on scientific papers. Additionally, we contribute novel publicly sourced datasets for charts, tables, equations, and code recognition. Experimental results demonstrate that SmolDocling competes with other Vision Language Models that are up to 27 times larger in size, while reducing computational requirements substantially. The model is currently available, datasets will be publicly available soon.
Authors: Weiming Ren, Wentao Ma, Huan Yang, Cong Wei, Ge Zhang, Wenhu Chen
Abstract: State-of-the-art transformer-based large multimodal models (LMMs) struggle to handle hour-long video inputs due to the quadratic complexity of the causal self-attention operations, leading to high computational costs during training and inference. Existing token compression-based methods reduce the number of video tokens but often incur information loss and remain inefficient for extremely long sequences. In this paper, we explore an orthogonal direction to build a hybrid Mamba-Transformer model (VAMBA) that employs Mamba-2 blocks to encode video tokens with linear complexity. Without any token reduction, VAMBA can encode more than 1024 frames (640$\times$360) on a single GPU, while transformer-based models can only encode 256 frames. On long video input, VAMBA achieves at least 50% reduction in GPU memory usage during training and inference, and nearly doubles the speed per training step compared to transformer-based LMMs. Our experimental results demonstrate that VAMBA improves accuracy by 4.3% on the challenging hour-long video understanding benchmark LVBench over prior efficient video LMMs, and maintains strong performance on a broad spectrum of long and short video understanding tasks.
Authors: Xuanqi Zhang, Jieun Lee, Chris Joslin, Wonsook Lee
Abstract: We present a novel framework for enhancing the visual fidelity and consistency of text-guided 3D Gaussian Splatting (3DGS) editing. Existing editing approaches face two critical challenges: inconsistent geometric reconstructions across multiple viewpoints, particularly in challenging camera positions, and ineffective utilization of depth information during image manipulation, resulting in over-texture artifacts and degraded object boundaries. To address these limitations, we introduce: 1) A complementary information mutual learning network that enhances depth map estimation from 3DGS, enabling precise depth-conditioned 3D editing while preserving geometric structures. 2) A wavelet consensus attention mechanism that effectively aligns latent codes during the diffusion denoising process, ensuring multi-view consistency in the edited results. Through extensive experimentation, our method demonstrates superior performance in rendering quality and view consistency compared to state-of-the-art approaches. The results validate our framework as an effective solution for text-guided editing of 3D scenes.
Authors: Matteo Farina, Massimiliano Mancini, Giovanni Iacca, Elisa Ricci
Abstract: An old-school recipe for training a classifier is to (i) learn a good feature extractor and (ii) optimize a linear layer atop. When only a handful of samples are available per category, as in Few-Shot Adaptation (FSA), data are insufficient to fit a large number of parameters, rendering the above impractical. This is especially true with large pre-trained Vision-Language Models (VLMs), which motivated successful research at the intersection of Parameter-Efficient Fine-tuning (PEFT) and FSA. In this work, we start by analyzing the learning dynamics of PEFT techniques when trained on few-shot data from only a subset of categories, referred to as the ``base'' classes. We show that such dynamics naturally splits into two distinct phases: (i) task-level feature extraction and (ii) specialization to the available concepts. To accommodate this dynamic, we then depart from prompt- or adapter-based methods and tackle FSA differently. Specifically, given a fixed computational budget, we split it to (i) learn a task-specific feature extractor via PEFT and (ii) train a linear classifier on top. We call this scheme Two-Stage Few-Shot Adaptation (2SFS). Differently from established methods, our scheme enables a novel form of selective inference at a category level, i.e., at test time, only novel categories are embedded by the adapted text encoder, while embeddings of base categories are available within the classifier. Results with fixed hyperparameters across two settings, three backbones, and eleven datasets, show that 2SFS matches or surpasses the state-of-the-art, while established methods degrade significantly across settings.
Authors: Hongyu Wen, Yiming Zuo, Venkat Subramanian, Patrick Chen, Jia Deng
Abstract: Transparent objects are common in daily life, and understanding their multi-layer depth information -- perceiving both the transparent surface and the objects behind it -- is crucial for real-world applications that interact with transparent materials. In this paper, we introduce LayeredDepth, the first dataset with multi-layer depth annotations, including a real-world benchmark and a synthetic data generator, to support the task of multi-layer depth estimation. Our real-world benchmark consists of 1,500 images from diverse scenes, and evaluating state-of-the-art depth estimation methods on it reveals that they struggle with transparent objects. The synthetic data generator is fully procedural and capable of providing training data for this task with an unlimited variety of objects and scene compositions. Using this generator, we create a synthetic dataset with 15,300 images. Baseline models training solely on this synthetic dataset produce good cross-domain multi-layer depth estimation. Fine-tuning state-of-the-art single-layer depth models on it substantially improves their performance on transparent objects, with quadruplet accuracy on our benchmark increased from 55.14% to 75.20%. All images and validation annotations are available under CC0 at https://layereddepth.cs.princeton.edu.
Authors: Jianhong Bai, Menghan Xia, Xiao Fu, Xintao Wang, Lianrui Mu, Jinwen Cao, Zuozhu Liu, Haoji Hu, Xiang Bai, Pengfei Wan, Di Zhang
Abstract: Camera control has been actively studied in text or image conditioned video generation tasks. However, altering camera trajectories of a given video remains under-explored, despite its importance in the field of video creation. It is non-trivial due to the extra constraints of maintaining multiple-frame appearance and dynamic synchronization. To address this, we present ReCamMaster, a camera-controlled generative video re-rendering framework that reproduces the dynamic scene of an input video at novel camera trajectories. The core innovation lies in harnessing the generative capabilities of pre-trained text-to-video models through a simple yet powerful video conditioning mechanism -- its capability often overlooked in current research. To overcome the scarcity of qualified training data, we construct a comprehensive multi-camera synchronized video dataset using Unreal Engine 5, which is carefully curated to follow real-world filming characteristics, covering diverse scenes and camera movements. It helps the model generalize to in-the-wild videos. Lastly, we further improve the robustness to diverse inputs through a meticulously designed training strategy. Extensive experiments tell that our method substantially outperforms existing state-of-the-art approaches and strong baselines. Our method also finds promising applications in video stabilization, super-resolution, and outpainting. Project page: https://jianhongbai.github.io/ReCamMaster/
Authors: Jianyuan Wang, Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, David Novotny
Abstract: We present VGGT, a feed-forward neural network that directly infers all key 3D attributes of a scene, including camera parameters, point maps, depth maps, and 3D point tracks, from one, a few, or hundreds of its views. This approach is a step forward in 3D computer vision, where models have typically been constrained to and specialized for single tasks. It is also simple and efficient, reconstructing images in under one second, and still outperforming alternatives that require post-processing with visual geometry optimization techniques. The network achieves state-of-the-art results in multiple 3D tasks, including camera parameter estimation, multi-view depth estimation, dense point cloud reconstruction, and 3D point tracking. We also show that using pretrained VGGT as a feature backbone significantly enhances downstream tasks, such as non-rigid point tracking and feed-forward novel view synthesis. Code and models are publicly available at https://github.com/facebookresearch/vggt.
Authors: Hiroyasu Akada, Jian Wang, Vladislav Golyanik, Christian Theobalt
Abstract: Egocentric 3D human pose estimation has been actively studied using cameras installed in front of a head-mounted device (HMD). While frontal placement is the optimal and the only option for some tasks, such as hand tracking, it remains unclear if the same holds for full-body tracking due to self-occlusion and limited field-of-view coverage. Notably, even the state-of-the-art methods often fail to estimate accurate 3D poses in many scenarios, such as when HMD users tilt their heads upward (a common motion in human activities). A key limitation of existing HMD designs is their neglect of the back of the body, despite its potential to provide crucial 3D reconstruction cues. Hence, this paper investigates the usefulness of rear cameras in the HMD design for full-body tracking. We also show that simply adding rear views to the frontal inputs is not optimal for existing methods due to their dependence on individual 2D joint detectors without effective multi-view integration. To address this issue, we propose a new transformer-based method that refines 2D joint heatmap estimation with multi-view information and heatmap uncertainty, thereby improving 3D pose tracking. Moreover, we introduce two new large-scale datasets, Ego4View-Syn and Ego4View-RW, for a rear-view evaluation. Our experiments show that the new camera configurations with back views provide superior support for 3D pose tracking compared to only frontal placements. The proposed method achieves significant improvement over the current state of the art (>10% on MPJPE). We will release the source code, trained models, and new datasets on our project page https://4dqv.mpi-inf.mpg.de/EgoRear/.
Authors: Weiquan Wang, Jun Xiao, Yueting Zhuang, Long Chen
Abstract: Rendering realistic human-object interactions (HOIs) from sparse-view inputs is challenging due to occlusions and incomplete observations, yet crucial for various real-world applications. Existing methods always struggle with either low rendering qualities (\eg, visual fidelity and physically plausible HOIs) or high computational costs. To address these limitations, we propose HOGS (Human-Object Rendering via 3D Gaussian Splatting), a novel framework for efficient and physically plausible HOI rendering from sparse views. Specifically, HOGS combines 3D Gaussian Splatting with a physics-aware optimization process. It incorporates a Human Pose Refinement module for accurate pose estimation and a Sparse-View Human-Object Contact Prediction module for efficient contact region identification. This combination enables coherent joint rendering of human and object Gaussians while enforcing physically plausible interactions. Extensive experiments on the HODome dataset demonstrate that HOGS achieves superior rendering quality, efficiency, and physical plausibility compared to existing methods. We further show its extensibility to hand-object grasp rendering tasks, presenting its broader applicability to articulated object interactions.
Authors: Yang Xiao, Wang Lu, Jie Ji, Ruimeng Ye, Gen Li, Xiaolong Ma, Bo Hui
Abstract: The design of artificial neural networks (ANNs) is inspired by the structure of the human brain, and in turn, ANNs offer a potential means to interpret and understand brain signals. Existing methods primarily align brain signals with real-world signals using Mean Squared Error (MSE), which solely focuses on local point-wise alignment, and ignores global matching, leading to coarse interpretations and inaccuracies in brain signal decoding. In this paper, we address these issues through optimal transport (OT) and theoretically demonstrate why OT provides a more effective alignment strategy than MSE. Specifically, we construct a transport plan between brain voxel embeddings and image embeddings, enabling more precise matching. By controlling the amount of transport, we mitigate the influence of redundant information. We apply our alignment model directly to the Brain Captioning task by feeding brain siginals into a large language model (LLM) instead of images. Our approach achieves state-of-the-art performance across ten evaluation metrics, surpassing the previous best method by an average of 6.11\% in single-subject training and 3.81\% in cross-subject training. Additionally, we have uncovered several insightful conclusions that align with existing brain research. We unveil the redundancy and synergy of brain information processing through region masking and data dimensionality reduction visualization experiments. We believe our approach paves the way for a more precise understanding of brain signals in the future. The code is available soon.
Authors: Praveen Shastry, Ashok Sharma, Kavya Mohan, Naveen Kumarasami, Anandakumar D, Mounigasri M, Keerthana R, Kishore Prasath Venkatesh, Bargava Subramanian, Kalyan Sivasailam
Abstract: Background: Automated analysis of CT scans for abdominal organ measurement is crucial for improving diagnostic efficiency and reducing inter-observer variability. Manual segmentation and measurement of organs such as the kidneys, liver, spleen, and prostate are time-consuming and subject to inconsistency, underscoring the need for automated approaches. Purpose: The purpose of this study is to develop and validate an automated workflow for the segmentation and measurement of abdominal organs in CT scans using advanced deep learning models, in order to improve accuracy, reliability, and efficiency in clinical evaluations. Methods: The proposed workflow combines nnU-Net, U-Net++ for organ segmentation, followed by a 3D RCNN model for measuring organ volumes and dimensions. The models were trained and evaluated on CT datasets with metrics such as precision, recall, and Mean Squared Error (MSE) to assess performance. Segmentation quality was verified for its adaptability to variations in patient anatomy and scanner settings. Results: The developed workflow achieved high precision and recall values, exceeding 95 for all targeted organs. The Mean Squared Error (MSE) values were low, indicating a high level of consistency between predicted and ground truth measurements. The segmentation and measurement pipeline demonstrated robust performance, providing accurate delineation and quantification of the kidneys, liver, spleen, and prostate. Conclusion: The proposed approach offers an automated, efficient, and reliable solution for abdominal organ measurement in CT scans. By significantly reducing manual intervention, this workflow enhances measurement accuracy and consistency, with potential for widespread clinical implementation. Future work will focus on expanding the approach to other organs and addressing complex pathological cases.
Authors: Kai Zhang, Jianwei Yang, Jeevana Priya Inala, Chandan Singh, Jianfeng Gao, Yu Su, Chenglong Wang
Abstract: Despite the promising results of large multimodal models (LMMs) in complex vision-language tasks that require knowledge, reasoning, and perception abilities together, we surprisingly found that these models struggle with simple tasks on infographics that require perception only. As existing benchmarks primarily focus on end tasks that require various abilities, they provide limited, fine-grained insights into the limitations of the models' perception abilities. To address this gap, we leverage the theory of graphical perception, an approach used to study how humans decode visual information encoded on charts and graphs, to develop an evaluation framework for analyzing gaps in LMMs' perception abilities in charts. With automated task generation and response evaluation designs, our framework enables comprehensive and controlled testing of LMMs' graphical perception across diverse chart types, visual elements, and task types. We apply our framework to evaluate and diagnose the perception capabilities of state-of-the-art LMMs at three granularity levels (chart, visual element, and pixel). Our findings underscore several critical limitations of current state-of-the-art LMMs, including GPT-4o: their inability to (1) generalize across chart types, (2) understand fundamental visual elements, and (3) cross reference values within a chart. These insights provide guidance for future improvements in perception abilities of LMMs. The evaluation framework and labeled data are publicly available at https://github.com/microsoft/lmm-graphical-perception.
URLs: https://github.com/microsoft/lmm-graphical-perception.
Authors: Zhuoyan Xu, Khoi Duc Nguyen, Preeti Mukherjee, Saurabh Bagchi, Somali Chaterji, Yingyu Liang, Yin Li
Abstract: Multimodal Large Language Models (MLLMs) have shown impressive capabilities in reasoning, yet come with substantial computational cost, limiting their deployment in resource-constrained settings. Despite recent efforts on improving the efficiency of MLLMs, prior solutions fall short in responding to varying runtime conditions, in particular changing resource availability (e.g., contention due to the execution of other programs on the device). To bridge this gap, we introduce AdaLLaVA, an adaptive inference framework that learns to dynamically reconfigure operations in an MLLM during inference, accounting for the input data and a latency budget. We conduct extensive experiments across benchmarks involving question-answering, reasoning, and hallucination. Our results show that AdaLLaVA effectively adheres to input latency budget, achieving varying accuracy and latency tradeoffs at runtime. Further, we demonstrate that AdaLLaVA adapts to both input latency and content, can be integrated with token selection for enhanced efficiency, and generalizes across MLLMs.Our project webpage with code release is at https://zhuoyan-xu.github.io/ada-llava/.
Authors: Xingfei Wei, Qiankun Mo, Chi Chen, Mark Bathe, Rigoberto Hernandez
Abstract: Artificial intelligence (AI) models remain an emerging strategy to accelerate materials design and development. We demonstrate that convolutional neural network (CNN) models can characterize DNA origami nanostructures employed in programmable self-assembling, which is important in many applications such as in biomedicine. Specifically, we benchmark the performance of 9 CNN models -- viz. AlexNet, GoogLeNet, VGG16, VGG19, ResNet18, ResNet34, ResNet50, ResNet101, and ResNet152 -- to characterize the ligation number of DNA origami nanostructures in transmission electron microscopy (TEM) images. We first pre-train CNN models using a large image dataset of 720 images from our coarse-grained (CG) molecular dynamics (MD) simulations. Then, we fine-tune the pre-trained CNN models, using a small experimental TEM dataset with 146 TEM images. All CNN models were found to have similar computational time requirements, while their model sizes and performances are different. We use 20 test MD images to demonstrate that among all of the pre-trained CNN models ResNet50 and VGG16 have the highest and second highest accuracies. Among the fine-tuned models, VGG16 was found to have the highest agreement on the test TEM images. Thus, we conclude that fine-tuned VGG16 models can quickly characterize the ligation number of nanostructures in large TEM images.
Authors: Zhicheng Feng, Xieyuanli Chen, Chenghao Shi, Lun Luo, Zhichao Chen, Yun-Hui Liu, Huimin Lu
Abstract: In this paper, we introduce a novel image-goal navigation approach, named RFSG. Our focus lies in leveraging the fine-grained connections between goals, observations, and the environment within limited image data, all the while keeping the navigation architecture simple and lightweight. To this end, we propose the spatial-channel attention mechanism, enabling the network to learn the importance of multi-dimensional features to fuse the goal and observation features. In addition, a selfdistillation mechanism is incorporated to further enhance the feature representation capabilities. Given that the navigation task needs surrounding environmental information for more efficient navigation, we propose an image scene graph to establish feature associations at both the image and object levels, effectively encoding the surrounding scene information. Crossscene performance validation was conducted on the Gibson and HM3D datasets, and the proposed method achieved stateof-the-art results among mainstream methods, with a speed of up to 53.5 frames per second on an RTX3080. This contributes to the realization of end-to-end image-goal navigation in realworld scenarios. The implementation and model of our method have been released at: https://github.com/nubot-nudt/RFSG.
Authors: Aashish Anantha Ramakrishnan, Aadarsh Anantha Ramakrishnan, Dongwon Lee
Abstract: Writing Assistants (e.g., Grammarly, Microsoft Copilot) traditionally generate diverse image captions by employing syntactic and semantic variations to describe image components. However, human-written captions prioritize conveying a central message alongside visual descriptions using pragmatic cues. To enhance pragmatic diversity, it is essential to explore alternative ways of communicating these messages in conjunction with visual content. To address this challenge, we propose RONA, a novel prompting strategy for Multi-modal Large Language Models (MLLM) that leverages Coherence Relations as an axis for variation. We demonstrate that RONA generates captions with better overall diversity and ground-truth alignment, compared to MLLM baselines across multiple domains. Our code is available at: https://github.com/aashish2000/RONA
Authors: Ruochen Hou, Mingzhang Zhu, Hyunwoo Nam, Gabriel I. Fernandez, Dennis W. Hong
Abstract: Accurate robot localization is essential for effective operation. Monte Carlo Localization (MCL) is commonly used with known maps but is computationally expensive due to landmark matching for each particle. Humanoid robots face additional challenges, including sensor noise from locomotion vibrations and a limited field of view (FOV) due to camera placement. This paper proposes a fast and robust localization method via iterative landmark matching (ILM) for humanoid robots. The iterative matching process improves the accuracy of the landmark association so that it does not need MCL to match landmarks to particles. Pose estimation with the outlier removal process enhances its robustness to measurement noise and faulty detections. Furthermore, an additional filter can be utilized to fuse inertial data from the inertial measurement unit (IMU) and pose data from localization. We compared ILM with Iterative Closest Point (ICP), which shows that ILM method is more robust towards the error in the initial guess and easier to get a correct matching. We also compared ILM with the Augmented Monte Carlo Localization (aMCL), which shows that ILM method is much faster than aMCL and even more accurate. The proposed method's effectiveness is thoroughly evaluated through experiments and validated on the humanoid robot ARTEMIS during RoboCup 2024 adult-sized soccer competition.
Authors: Sungwoo Cho, Jeongsoo Choi, Sungnyun Kim, Se-Young Yun
Abstract: Despite recent advances in text-to-speech (TTS) models, audio-visual to audio-visual (AV2AV) translation still faces a critical challenge: maintaining speaker consistency between the original and translated vocal and facial features. To address this issue, we propose a conditional flow matching (CFM) zero-shot audio-visual renderer that utilizes strong dual guidance from both audio and visual modalities. By leveraging multi-modal guidance with CFM, our model robustly preserves speaker-specific characteristics and significantly enhances zero-shot AV2AV translation abilities. For the audio modality, we enhance the CFM process by integrating robust speaker embeddings with x-vectors, which serve to bolster speaker consistency. Additionally, we convey emotional nuances to the face rendering module. The guidance provided by both audio and visual cues remains independent of semantic or linguistic content, allowing our renderer to effectively handle zero-shot translation tasks for monolingual speakers in different languages. We empirically demonstrate that the inclusion of high-quality mel-spectrograms conditioned on facial information not only enhances the quality of the synthesized speech but also positively influences facial generation, leading to overall performance improvements.
Authors: Worameth Chinchuthakun, Tossaporn Saengja, Nontawat Tritrong, Pitchaporn Rewatbowornwong, Pramook Khungurn, Supasorn Suwajanakorn
Abstract: While diffusion models show promising results in image editing given a target prompt, achieving both prompt fidelity and background preservation remains difficult. Recent works have introduced score distillation techniques that leverage the rich generative prior of text-to-image diffusion models to solve this task without additional fine-tuning. However, these methods often struggle with tasks such as object insertion. Our investigation of these failures reveals significant variations in gradient magnitude and spatial distribution, making hyperparameter tuning highly input-specific or unsuccessful. To address this, we propose two simple yet effective modifications: attention-based spatial regularization and gradient filtering-normalization, both aimed at reducing these variations during gradient updates. Experimental results show our method outperforms state-of-the-art score distillation techniques in prompt fidelity, improving successful edits while preserving the background. Users also preferred our method over state-of-the-art techniques across three metrics, and by 58-64% overall.
Authors: Pingrui Zhang, Xianqiang Gao, Yuhan Wu, Kehui Liu, Dong Wang, Zhigang Wang, Bin Zhao, Yan Ding, Xuelong Li
Abstract: In mobile manipulation, navigation and manipulation are often treated as separate problems, resulting in a significant gap between merely approaching an object and engaging with it effectively. Many navigation approaches primarily define success by proximity to the target, often overlooking the necessity for optimal positioning that facilitates subsequent manipulation. To address this, we introduce MoMa-Kitchen, a benchmark dataset comprising over 100k samples that provide training data for models to learn optimal final navigation positions for seamless transition to manipulation. Our dataset includes affordance-grounded floor labels collected from diverse kitchen environments, in which robotic mobile manipulators of different models attempt to grasp target objects amidst clutter. Using a fully automated pipeline, we simulate diverse real-world scenarios and generate affordance labels for optimal manipulation positions. Visual data are collected from RGB-D inputs captured by a first-person view camera mounted on the robotic arm, ensuring consistency in viewpoint during data collection. We also develop a lightweight baseline model, NavAff, for navigation affordance grounding that demonstrates promising performance on the MoMa-Kitchen benchmark. Our approach enables models to learn affordance-based final positioning that accommodates different arm types and platform heights, thereby paving the way for more robust and generalizable integration of navigation and manipulation in embodied AI. Project page: \href{https://momakitchen.github.io/}{https://momakitchen.github.io/}.
URLs: https://momakitchen.github.io/, https://momakitchen.github.io/
Authors: Yi Zhang, Qiang Zhang, Xiaozhu Ju, Zhaoyang Liu, Jilei Mao, Jingkai Sun, Jintao Wu, Shixiong Gao, Shihan Cai, Zhiyuan Qin, Linkai Liang, Jiaxu Wang, Yiqun Duan, Jiahang Cao, Renjing Xu, Jian Tang
Abstract: While multimodal large language models (MLLMs) have made groundbreaking progress in embodied intelligence, they still face significant challenges in spatial reasoning for complex long-horizon tasks. To address this gap, we propose EmbodiedVSR (Embodied Visual Spatial Reasoning), a novel framework that integrates dynamic scene graph-guided Chain-of-Thought (CoT) reasoning to enhance spatial understanding for embodied agents. By explicitly constructing structured knowledge representations through dynamic scene graphs, our method enables zero-shot spatial reasoning without task-specific fine-tuning. This approach not only disentangles intricate spatial relationships but also aligns reasoning steps with actionable environmental dynamics. To rigorously evaluate performance, we introduce the eSpatial-Benchmark, a comprehensive dataset including real-world embodied scenarios with fine-grained spatial annotations and adaptive task difficulty levels. Experiments demonstrate that our framework significantly outperforms existing MLLM-based methods in accuracy and reasoning coherence, particularly in long-horizon tasks requiring iterative environment interaction. The results reveal the untapped potential of MLLMs for embodied intelligence when equipped with structured, explainable reasoning mechanisms, paving the way for more reliable deployment in real-world spatial applications. The codes and datasets will be released soon.
Authors: G\"okhan \"Ozbulak, Oscar Jimenez-del-Toro, Ma\'ira Fatoretto, Lilian Berton, Andr\'e Anjos
Abstract: The evaluation of fairness models in Machine Learning involves complex challenges, such as defining appropriate metrics, balancing trade-offs between utility and fairness, and there are still gaps in this stage. This work presents a novel multi-objective evaluation framework that enables the analysis of utility-fairness trade-offs in Machine Learning systems. The framework was developed using criteria from Multi-Objective Optimization that collect comprehensive information regarding this complex evaluation task. The assessment of multiple Machine Learning systems is summarized, both quantitatively and qualitatively, in a straightforward manner through a radar chart and a measurement table encompassing various aspects such as convergence, system capacity, and diversity. The framework's compact representation of performance facilitates the comparative analysis of different Machine Learning strategies for decision-makers, in real-world applications, with single or multiple fairness requirements. The framework is model-agnostic and flexible to be adapted to any kind of Machine Learning systems, that is, black- or white-box, any kind and quantity of evaluation metrics, including multidimensional fairness criteria. The functionality and effectiveness of the proposed framework is shown with different simulations, and an empirical study conducted on a real-world dataset with various Machine Learning systems.
Authors: Rachel S. Y. Teo, Tan M. Nguyen
Abstract: Large-scale pre-training of deep models, followed by fine-tuning them, has become the cornerstone of natural language processing (NLP). The prevalence of data coupled with computational resources has led to large models with a considerable number of parameters. While the massive size of these models has led to remarkable success in many NLP tasks, a detriment is the expense required to retrain all the base model's parameters for the adaptation to each task or domain. Parameter Efficient Fine-Tuning (PEFT) provides an effective solution for this challenge by minimizing the number of parameters required to be fine-tuned while maintaining the quality of the model. While existing methods have achieved impressive results, they mainly focus on adapting a subset of parameters, weight reparameterization, and prompt engineering. In this paper, we study layers as extractors of different types of linguistic information that are valuable when used in conjunction. We then propose the Mixture of Layer Experts (MoLEx), a novel sparse mixture of experts (SMoE) whose experts are layers in the pre-trained model. It performs a conditional computation of a mixture of layers during fine-tuning to provide the model with more structural knowledge about the data. By providing an avenue for information exchange between layers, MoLEx enables the model to make a more well-informed prediction for the downstream task, leading to better fine-tuning results with the same number of effective parameters. As experts can be processed in parallel, MoLEx introduces minimal additional computational overhead. We empirically corroborate the advantages of MoLEx when combined with popular PEFT baseline methods on a variety of downstream fine-tuning tasks, including the popular GLUE benchmark as well as the End-to-End Challenge (E2E). The code is publicly available at https://github.com/rachtsy/molex.
Authors: Bharath K Rameshbabu, Sumukh S Balakrishna, Brian Flynn, Vinarak Kapoor, Adam Norton, Holly Yanco, Berk Calli
Abstract: We present a benchmarking study of vision-based robotic grasping algorithms with distinct approaches, and provide a comparative analysis. In particular, we compare two machine-learning-based and two analytical algorithms using an existing benchmarking protocol from the literature and determine the algorithm's strengths and weaknesses under different experimental conditions. These conditions include variations in lighting, background textures, cameras with different noise levels, and grippers. We also run analogous experiments in simulations and with real robots and present the discrepancies. Some experiments are also run in two different laboratories using same protocols to further analyze the repeatability of our results. We believe that this study, comprising 5040 experiments, provides important insights into the role and challenges of systematic experimentation in robotic manipulation, and guides the development of new algorithms by considering the factors that could impact the performance. The experiment recordings and our benchmarking software are publicly available.
Authors: Tiantian Li, Qunbing Xia, Yue Li, Ruixiao Guo, Gaobo Yang
Abstract: Learning-based lossless image compression employs pixel-based or subimage-based auto-regression for probability estimation, which achieves desirable performances. However, the existing works only consider context dependencies in one direction, namely, those symbols that appear before the current symbol in raster order. We believe that the dependencies between the current and future symbols should be further considered. In this work, we propose a deep lossless image compression via masked sampling and coarse-to-fine auto-regression. It combines lossy reconstruction and progressive residual compression, which fuses contexts from various directions and is more consistent with human perception. Specifically, the residuals are decomposed via $T$ iterative masked sampling, and each sampling consists of three steps: 1) probability estimation, 2) mask computation, and 3) arithmetic coding. The iterative process progressively refines our prediction and gradually presents a real image. Extensive experimental results show that compared with the existing traditional and learned lossless compression, our method achieves comparable compression performance on extensive datasets with competitive coding speed and more flexibility.
Authors: Quanyuan Ruan, Jiabao Lei, Wenhao Yuan, Yanglin Zhang, Dekun Lu, Guiliang Liu, Kui Jia
Abstract: Differentiable rendering has gained significant attention in the field of robotics, with differentiable robot rendering emerging as an effective paradigm for learning robotic actions from image-space supervision. However, the lack of physical world perception in this approach may lead to potential collisions during action optimization. In this work, we introduce a novel improvement on previous efforts by incorporating physical awareness of collisions through the learning of a neural robotic collision classifier. This enables the optimization of actions that avoid collisions with static, non-interactable environments as well as the robot itself. To facilitate effective gradient optimization with the classifier, we identify the underlying issue and propose leveraging Eikonal regularization to ensure consistent gradients for optimization. Our solution can be seamlessly integrated into existing differentiable robot rendering frameworks, utilizing gradients for optimization and providing a foundation for future applications of differentiable rendering in robotics with improved reliability of interactions with the physical world. Both qualitative and quantitative experiments demonstrate the necessity and effectiveness of our method compared to previous solutions.
Authors: Ziyi Wang, Songbai Tan, Gang Xu, Xuerui Qiu, Hongbin Xu, Xin Meng, Ming Li, Fei Richard Yu
Abstract: With the success of autoregressive learning in large language models, it has become a dominant approach for text-to-image generation, offering high efficiency and visual quality. However, invisible watermarking for visual autoregressive (VAR) models remains underexplored, despite its importance in misuse prevention. Existing watermarking methods, designed for diffusion models, often struggle to adapt to the sequential nature of VAR models. To bridge this gap, we propose Safe-VAR, the first watermarking framework specifically designed for autoregressive text-to-image generation. Our study reveals that the timing of watermark injection significantly impacts generation quality, and watermarks of different complexities exhibit varying optimal injection times. Motivated by this observation, we propose an Adaptive Scale Interaction Module, which dynamically determines the optimal watermark embedding strategy based on the watermark information and the visual characteristics of the generated image. This ensures watermark robustness while minimizing its impact on image quality. Furthermore, we introduce a Cross-Scale Fusion mechanism, which integrates mixture of both heads and experts to effectively fuse multi-resolution features and handle complex interactions between image content and watermark patterns. Experimental results demonstrate that Safe-VAR achieves state-of-the-art performance, significantly surpassing existing counterparts regarding image quality, watermarking fidelity, and robustness against perturbations. Moreover, our method exhibits strong generalization to an out-of-domain watermark dataset QR Codes.
Authors: Masaya Mori, Yuto Omae, Yutaka Koyama, Kazuyuki Hara, Jun Toyotani, Yasuo Okumura, Hiroyuki Hao
Abstract: As the number of patients with heart failure increases, machine learning (ML) has garnered attention in cardiomyopathy diagnosis, driven by the shortage of pathologists. However, endomyocardial biopsy specimens are often small sample size and require techniques such as feature extraction and dimensionality reduction. This study aims to determine whether texture features are effective for feature extraction in the pathological diagnosis of cardiomyopathy. Furthermore, model designs that contribute toward improving generalization performance are examined by applying feature selection (FS) and dimensional compression (DC) to several ML models. The obtained results were verified by visualizing the inter-class distribution differences and conducting statistical hypothesis testing based on texture features. Additionally, they were evaluated using predictive performance across different model designs with varying combinations of FS and DC (applied or not) and decision boundaries. The obtained results confirmed that texture features may be effective for the pathological diagnosis of cardiomyopathy. Moreover, when the ratio of features to the sample size is high, a multi-step process involving FS and DC improved the generalization performance, with the linear kernel support vector machine achieving the best results. This process was demonstrated to be potentially effective for models with reduced complexity, regardless of whether the decision boundaries were linear, curved, perpendicular, or parallel to the axes. These findings are expected to facilitate the development of an effective cardiomyopathy diagnostic model for its rapid adoption in medical practice.
Authors: Leila Bagheriye, Johan Kwisthout
Abstract: Real-time computer-aided diagnosis using artificial intelligence (AI), with images, can help oncologists diagnose cancer with high accuracy and in an early phase. We reviewed real-time AI-based analyzed images for decision-making in different cancer types. This paper provides insights into the present and future potential of real-time imaging and image fusion. It explores various real-time techniques, encompassing technical solutions, AI-based imaging, and image fusion diagnosis across multiple anatomical areas, and electromagnetic needle tracking. To provide a thorough overview, this paper discusses ultrasound image fusion, real-time in vivo cancer diagnosis with different spectroscopic techniques, different real-time optical imaging-based cancer diagnosis techniques, elastography-based cancer diagnosis, cervical cancer detection using neuromorphic architectures, different fluorescence image-based cancer diagnosis techniques, and hyperspectral imaging-based cancer diagnosis. We close by offering a more futuristic overview to solve existing problems in real-time image-based cancer diagnosis.
Authors: M. Ak{\i}n Y{\i}lmaz, Ahmet Bilican, A. Murat Tekalp
Abstract: Video frame prediction remains a fundamental challenge in computer vision with direct implications for autonomous systems, video compression, and media synthesis. We present FG-DFPN, a novel architecture that harnesses the synergy between optical flow estimation and deformable convolutions to model complex spatio-temporal dynamics. By guiding deformable sampling with motion cues, our approach addresses the limitations of fixed-kernel networks when handling diverse motion patterns. The multi-scale design enables FG-DFPN to simultaneously capture global scene transformations and local object movements with remarkable precision. Our experiments demonstrate that FG-DFPN achieves state-of-the-art performance on eight diverse MPEG test sequences, outperforming existing methods by 1dB PSNR while maintaining competitive inference speeds. The integration of motion cues with adaptive geometric transformations makes FG-DFPN a promising solution for next-generation video processing systems that require high-fidelity temporal predictions. The model and instructions to reproduce our results will be released at: https://github.com/KUIS-AI-Tekalp-Research Group/frame-prediction
Authors: Arno Verduyn, Maxim Vochten, Joris De Schutter
Abstract: The ability of robots to recognize human gestures facilitates a natural and accessible human-robot collaboration. However, most work in gesture recognition remains rooted in reference frame-dependent representations. This poses a challenge when reference frames vary due to different work cell layouts, imprecise frame calibrations, or other environmental changes. This paper investigated the use of invariant trajectory descriptors for robust hand palm motion gesture recognition under reference frame changes. First, a novel dataset of recorded Hand Palm Motion (HPM) gestures is introduced. The motion gestures in this dataset were specifically designed to be distinguishable without dependence on specific reference frames or directional cues. Afterwards, multiple invariant trajectory descriptor approaches were benchmarked to assess how their performances generalize to this novel HPM dataset. After this offline benchmarking, the best scoring approach is validated for online recognition by developing a real-time Proof of Concept (PoC). In this PoC, hand palm motion gestures were used to control the real-time movement of a manipulator arm. The PoC demonstrated a high recognition reliability in real-time operation, achieving an $F_1$-score of 92.3%. This work demonstrates the effectiveness of the invariant descriptor approach as a standalone solution. Moreover, we believe that the invariant descriptor approach can also be utilized within other state-of-the-art pattern recognition and learning systems to improve their robustness against reference frame variations.
Authors: Ziyue Wang, Chenghao Shi, Neng Wang, Qinghua Yu, Xieyuanli Chen, Huimin Lu
Abstract: Localization is one of the core parts of modern robotics. Classic localization methods typically follow the retrieve-then-register paradigm, achieving remarkable success. Recently, the emergence of end-to-end localization approaches has offered distinct advantages, including a streamlined system architecture and the elimination of the need to store extensive map data. Although these methods have demonstrated promising results, current end-to-end localization approaches still face limitations in robustness and accuracy. Bird's-Eye-View (BEV) image is one of the most widely adopted data representations in autonomous driving. It significantly reduces data complexity while preserving spatial structure and scale consistency, making it an ideal representation for localization tasks. However, research on BEV-based end-to-end localization remains notably insufficient. To fill this gap, we propose BEVDiffLoc, a novel framework that formulates LiDAR localization as a conditional generation of poses. Leveraging the properties of BEV, we first introduce a specific data augmentation method to significantly enhance the diversity of input data. Then, the Maximum Feature Aggregation Module and Vision Transformer are employed to learn robust features while maintaining robustness against significant rotational view variations. Finally, we incorporate a diffusion model that iteratively refines the learned features to recover the absolute pose. Extensive experiments on the Oxford Radar RobotCar and NCLT datasets demonstrate that BEVDiffLoc outperforms the baseline methods. Our code is available at https://github.com/nubot-nudt/BEVDiffLoc.
Authors: Jonas Thietke, Andreas M\"uller, Denis Lukovnikov, Asja Fischer, Erwin Quiring
Abstract: Semantic watermarking methods enable the direct integration of watermarks into the generation process of latent diffusion models by only modifying the initial latent noise. One line of approaches building on Gaussian Shading relies on cryptographic primitives to steer the sampling process of the latent noise. However, we identify several issues in the usage of cryptographic techniques in Gaussian Shading, particularly in its proof of lossless performance and key management, causing ambiguity in follow-up works, too. In this work, we therefore revisit the cryptographic primitives for semantic watermarking. We introduce a novel, general proof of lossless performance based on IND\$-CPA security for semantic watermarks. We then discuss the configuration of the cryptographic primitives in semantic watermarks with respect to security, efficiency, and generation quality.
Authors: Shida Xu, Kaicheng Zhang, Sen Wang
Abstract: Underwater environments pose significant challenges for visual Simultaneous Localization and Mapping (SLAM) systems due to limited visibility, inadequate illumination, and sporadic loss of structural features in images. Addressing these challenges, this paper introduces a novel, tightly-coupled Acoustic-Visual-Inertial SLAM approach, termed AQUA-SLAM, to fuse a Doppler Velocity Log (DVL), a stereo camera, and an Inertial Measurement Unit (IMU) within a graph optimization framework. Moreover, we propose an efficient sensor calibration technique, encompassing multi-sensor extrinsic calibration (among the DVL, camera and IMU) and DVL transducer misalignment calibration, with a fast linear approximation procedure for real-time online execution. The proposed methods are extensively evaluated in a tank environment with ground truth, and validated for offshore applications in the North Sea. The results demonstrate that our method surpasses current state-of-the-art underwater and visual-inertial SLAM systems in terms of localization accuracy and robustness. The proposed system will be made open-source for the community.
Authors: Jonas Belouadi, Eddy Ilg, Margret Keuper, Hideki Tanaka, Masao Utiyama, Raj Dabre, Steffen Eger, Simone Paolo Ponzetto
Abstract: With the rise of generative AI, synthesizing figures from text captions becomes a compelling application. However, achieving high geometric precision and editability requires representing figures as graphics programs in languages like TikZ, and aligned training data (i.e., graphics programs with captions) remains scarce. Meanwhile, large amounts of unaligned graphics programs and captioned raster images are more readily available. We reconcile these disparate data sources by presenting TikZero, which decouples graphics program generation from text understanding by using image representations as an intermediary bridge. It enables independent training on graphics programs and captioned images and allows for zero-shot text-guided graphics program synthesis during inference. We show that our method substantially outperforms baselines that can only operate with caption-aligned graphics programs. Furthermore, when leveraging caption-aligned graphics programs as a complementary training signal, TikZero matches or exceeds the performance of much larger models, including commercial systems like GPT-4o. Our code, datasets, and select models are publicly available.
Authors: Siva Manohar Reddy Kesu, Neelam Sinha, Hariharan Ramasangu, Thomas Gregor Issac
Abstract: Retinal optical coherence tomography (OCT) images are the biomarkers for neurodegenerative diseases, which are rising in prevalence. Early detection of Alzheimer's disease using retinal OCT is a primary challenging task. This work utilizes advanced deep learning techniques to classify retinal OCT images of subjects with Alzheimer's disease (AD) and healthy controls (CO). The goal is to enhance diagnostic capabilities through efficient image analysis. In the proposed model, Raw OCT images have been preprocessed with ImageJ and given to various deep-learning models to evaluate the accuracy. The best classification architecture is TransNetOCT, which has an average accuracy of 98.18% for input OCT images and 98.91% for segmented OCT images for five-fold cross-validation compared to other models, and the Swin Transformer model has achieved an accuracy of 93.54%. The evaluation accuracy metric demonstrated TransNetOCT and Swin transformer models capability to classify AD and CO subjects reliably, contributing to the potential for improved diagnostic processes in clinical settings.
Authors: Srikar Yellapragada, Alexandros Graikos, Kostas Triaridis, Zilinghan Li, Tarak Nath Nandi, Ravi K Madduri, Prateek Prasanna, Joel Saltz, Dimitris Samaras
Abstract: The growing volume of high-resolution Whole Slide Images in digital histopathology poses significant storage, transmission, and computational efficiency challenges. Standard compression methods, such as JPEG, reduce file sizes but often fail to preserve fine-grained phenotypic details critical for downstream tasks. In this work, we repurpose autoencoders (AEs) designed for Latent Diffusion Models as an efficient learned compression framework for pathology images. We systematically benchmark three AE models with varying compression levels and evaluate their reconstruction ability using pathology foundation models. We introduce a fine-tuning strategy to further enhance reconstruction fidelity that optimizes a pathology-specific learned perceptual metric. We validate our approach on downstream tasks, including segmentation, patch classification, and multiple instance learning, showing that replacing images with AE-compressed reconstructions leads to minimal performance degradation. Additionally, we propose a K-means clustering-based quantization method for AE latents, improving storage efficiency while maintaining reconstruction quality. We provide the weights of the fine-tuned autoencoders at https://huggingface.co/collections/StonyBrook-CVLab/pathology-fine-tuned-aes-67d45f223a659ff2e3402dd0.
Authors: Stefan Lionar, Jiabin Liang, Gim Hee Lee
Abstract: We introduce TreeMeshGPT, an autoregressive Transformer designed to generate high-quality artistic meshes aligned with input point clouds. Instead of the conventional next-token prediction in autoregressive Transformer, we propose a novel Autoregressive Tree Sequencing where the next input token is retrieved from a dynamically growing tree structure that is built upon the triangle adjacency of faces within the mesh. Our sequencing enables the mesh to extend locally from the last generated triangular face at each step, and therefore reduces training difficulty and improves mesh quality. Our approach represents each triangular face with two tokens, achieving a compression rate of approximately 22% compared to the naive face tokenization. This efficient tokenization enables our model to generate highly detailed artistic meshes with strong point cloud conditioning, surpassing previous methods in both capacity and fidelity. Furthermore, our method generates mesh with strong normal orientation constraints, minimizing flipped normals commonly encountered in previous methods. Our experiments show that TreeMeshGPT enhances the mesh generation quality with refined details and normal orientation consistency.
Authors: Chonghao Sima, Kashyap Chitta, Zhiding Yu, Shiyi Lan, Ping Luo, Andreas Geiger, Hongyang Li, Jose M. Alvarez
Abstract: How can we rely on an end-to-end autonomous vehicle's complex decision-making system during deployment? One common solution is to have a ``fallback layer'' that checks the planned trajectory for rule violations and replaces it with a pre-defined safe action if necessary. Another approach involves adjusting the planner's decisions to minimize a pre-defined ``cost function'' using additional system predictions such as road layouts and detected obstacles. However, these pre-programmed rules or cost functions cannot learn and improve with new training data, often resulting in overly conservative behaviors. In this work, we propose Centaur (Cluster Entropy for Test-time trAining using Uncertainty) which updates a planner's behavior via test-time training, without relying on hand-engineered rules or cost functions. Instead, we measure and minimize the uncertainty in the planner's decisions. For this, we develop a novel uncertainty measure, called Cluster Entropy, which is simple, interpretable, and compatible with state-of-the-art planning algorithms. Using data collected at prior test-time time-steps, we perform an update to the model's parameters using a gradient that minimizes the Cluster Entropy. With only this sole gradient update prior to inference, Centaur exhibits significant improvements, ranking first on the navtest leaderboard with notable gains in safety-critical metrics such as time to collision. To provide detailed insights on a per-scenario basis, we also introduce navsafe, a challenging new benchmark, which highlights previously undiscovered failure modes of driving models.
Authors: Marjan Famili, Jola Mirecka, Camila Rangel Smith, Anna Kota\'nska, Nikolai Juraschko, Beatriz Costa-Gomes, Colin M. Palmer, Jeyan Thiyagalingam, Tom Burnley, Mark Basham, Alan R. Lowe
Abstract: Learning compact and interpretable representations of data is a critical challenge in scientific image analysis. Here, we introduce Affinity-VAE, a generative model that enables us to impose our scientific intuition about the similarity of instances in the dataset on the learned representation during training. We demonstrate the utility of the approach in the scientific domain of cryo-electron tomography (cryo-ET) where a significant current challenge is to identify similar molecules within a noisy and low contrast tomographic image volume. This task is distinct from classification in that, at inference time, it is unknown whether an instance is part of the training set or not. We trained affinity-VAE using prior knowledge of protein structure to inform the latent space. Our model is able to create rotationally-invariant, morphologically homogeneous clusters in the latent representation, with improved cluster separation compared to other approaches. It achieves competitive performance on protein classification with the added benefit of disentangling object pose, structural similarity and an interpretable latent representation. In the context of cryo-ET data, affinity-VAE captures the orientation of identified proteins in 3D which can be used as a prior for subsequent scientific experiments. Extracting physical principles from a trained network is of significant importance in scientific imaging where a ground truth training set is not always feasible.
Authors: Vlad Hondru, Radu Tudor Ionescu
Abstract: Diffusion models showcase strong capabilities in image synthesis, being used in many computer vision tasks with great success. To this end, we propose to explore a new use case, namely to copy black-box classification models without having access to the original training data, the architecture, and the weights of the model, i.e. the model is only exposed through an inference API. More specifically, we can only observe the (soft or hard) labels for some image samples passed as input to the model. Furthermore, we consider an additional constraint limiting the number of model calls, mostly focusing our research on few-call model stealing. In order to solve the model extraction task given the applied restrictions, we propose the following framework. As training data, we create a synthetic data set (called proxy data set) by leveraging the ability of diffusion models to generate realistic and diverse images. Given a maximum number of allowed API calls, we pass the respective number of samples through the black-box model to collect labels. Finally, we distill the knowledge of the black-box teacher (attacked model) into a student model (copy of the attacked model), harnessing both labeled and unlabeled data generated by the diffusion model. We employ a novel active self-paced learning framework to make the most of the proxy data during distillation. Our empirical results on three data sets confirm the superiority of our framework over four state-of-the-art methods in the few-call model extraction scenario. We release our code for free non-commercial use at https://github.com/vladhondru25/model-stealing.
Authors: Hu Wang, Congbo Ma, Jianpeng Zhang, Yuan Zhang, Jodie Avery, Louise Hull, Gustavo Carneiro
Abstract: The problem of missing modalities is both critical and non-trivial to be handled in multi-modal models. It is common for multi-modal tasks that certain modalities contribute more compared to other modalities, and if those important modalities are missing, the model performance drops significantly. Such fact remains unexplored by current multi-modal approaches that recover the representation from missing modalities by feature reconstruction or blind feature aggregation from other modalities, instead of extracting useful information from the best performing modalities. In this paper, we propose a Learnable Cross-modal Knowledge Distillation (LCKD) model to adaptively identify important modalities and distil knowledge from them to help other modalities from the cross-modal perspective for solving the missing modality issue. Our approach introduces a teacher election procedure to select the most ``qualified'' teachers based on their single modality performance on certain tasks. Then, cross-modal knowledge distillation is performed between teacher and student modalities for each task to push the model parameters to a point that is beneficial for all tasks. Hence, even if the teacher modalities for certain tasks are missing during testing, the available student modalities can accomplish the task well enough based on the learned knowledge from their automatically elected teacher modalities. Experiments on the Brain Tumour Segmentation Dataset 2018 (BraTS2018) shows that LCKD outperforms other methods by a considerable margin, improving the state-of-the-art performance by 3.61% for enhancing tumour, 5.99% for tumour core, and 3.76% for whole tumour in terms of segmentation Dice score.
Authors: Duy-Kien Nguyen, Martin R. Oswald, Cees G. M. Snoek
Abstract: The ability to detect objects in images at varying scales has played a pivotal role in the design of modern object detectors. Despite considerable progress in removing hand-crafted components and simplifying the architecture with transformers, multi-scale feature maps and pyramid designs remain a key factor for their empirical success. In this paper, we show that shifting the multiscale inductive bias into the attention mechanism can work well, resulting in a plain detector `SimPLR' whose backbone and detection head are both non-hierarchical and operate on single-scale features. We find through our experiments that SimPLR with scale-aware attention is plain and simple architecture, yet competitive with multi-scale vision transformer alternatives. Compared to the multi-scale and single-scale state-of-the-art, our model scales better with bigger capacity (self-supervised) models and more pre-training data, allowing us to report a consistently better accuracy and faster runtime for object detection, instance segmentation, as well as panoptic segmentation. Code is released at https://github.com/kienduynguyen/SimPLR.
Authors: Hui Tian, Kai Xu
Abstract: Point cloud reconstruction from raw point cloud has been an important topic in computer graphics for decades, especially due to its high demand in modeling and rendering applications. An important way to solve this problem is establishing a local geometry to fit the local curve. However, previous methods build either a local plane or polynomial curve. Local plane brings the loss of sharp feature and the boundary artefacts on open surface. Polynomial curve is hard to combine with neural network due to the local coordinate consistent problem. To address this, we propose a novel polyhedral surface to represent local surface. This method provides more flexible to represent sharp feature and surface boundary on open surface. It does not require any local coordinate system, which is important when introducing neural networks. Specifically, we use normals to construct the polyhedral surface, including both dihedral and trihedral surfaces using 2 and 3 normals, respectively. Our method achieves state-of-the-art results on three commonly used datasets (ShapeNetCore, ABC, and ScanNet). Code will be released upon acceptance.
Authors: Kaichen Zhou, Jia-Xing Zhong, Sangyun Shin, Kai Lu, Yiyuan Yang, Andrew Markham, Niki Trigoni
Abstract: The introduction of neural radiance fields has greatly improved the effectiveness of view synthesis for monocular videos. However, existing algorithms face difficulties when dealing with uncontrolled or lengthy scenarios, and require extensive training time specific to each new scenario. To tackle these limitations, we propose DynPoint, an algorithm designed to facilitate the rapid synthesis of novel views for unconstrained monocular videos. Rather than encoding the entirety of the scenario information into a latent representation, DynPoint concentrates on predicting the explicit 3D correspondence between neighboring frames to realize information aggregation. Specifically, this correspondence prediction is achieved through the estimation of consistent depth and scene flow information across frames. Subsequently, the acquired correspondence is utilized to aggregate information from multiple reference frames to a target frame, by constructing hierarchical neural point clouds. The resulting framework enables swift and accurate view synthesis for desired views of target frames. The experimental results obtained demonstrate the considerable acceleration of training time achieved - typically an order of magnitude - by our proposed method while yielding comparable outcomes compared to prior approaches. Furthermore, our method exhibits strong robustness in handling long-duration videos without learning a canonical representation of video content.
Authors: Robert Johanson, Christian Wilms, Ole Johannsen, Simone Frintrop
Abstract: Crop detection is integral for precision agriculture applications such as automated yield estimation or fruit picking. However, crop detection, e.g., apple detection in orchard environments remains challenging due to a lack of large-scale datasets and the small relative size of the crops in the image. In this work, we address these challenges by reformulating the apple detection task in a semi-supervised manner. To this end, we provide the large, high-resolution dataset MAD comprising 105 labeled images with 14,667 annotated apple instances and 4,440 unlabeled images. Utilizing this dataset, we also propose a novel Semi-Supervised Small Apple Detection system S$^3$AD based on contextual attention and selective tiling to improve the challenging detection of small apples, while limiting the computational overhead. We conduct an extensive evaluation on MAD and the MSU dataset, showing that S$^3$AD substantially outperforms strong fully-supervised baselines, including several small object detection systems, by up to $14.9\%$. Additionally, we exploit the detailed annotations of our dataset w.r.t. apple properties to analyze the influence of relative size or level of occlusion on the results of various systems, quantifying current challenges.
Authors: Amin Karimi Monsefi, Payam Karisani, Mengxi Zhou, Stacey Choi, Nathan Doble, Heng Ji, Srinivasan Parthasarathy, Rajiv Ramnath
Abstract: Standard modern machine-learning-based imaging methods have faced challenges in medical applications due to the high cost of dataset construction and, thereby, the limited labeled training data available. Additionally, upon deployment, these methods are usually used to process a large volume of data on a daily basis, imposing a high maintenance cost on medical facilities. In this paper, we introduce a new neural network architecture, termed LoGoNet, with a tailored self-supervised learning (SSL) method to mitigate such challenges. LoGoNet integrates a novel feature extractor within a U-shaped architecture, leveraging Large Kernel Attention (LKA) and a dual encoding strategy to capture both long-range and short-range feature dependencies adeptly. This is in contrast to existing methods that rely on increasing network capacity to enhance feature extraction. This combination of novel techniques in our model is especially beneficial in medical image segmentation, given the difficulty of learning intricate and often irregular body organ shapes, such as the spleen. Complementary, we propose a novel SSL method tailored for 3D images to compensate for the lack of large labeled datasets. The method combines masking and contrastive learning techniques within a multi-task learning framework and is compatible with both Vision Transformer (ViT) and CNN-based models. We demonstrate the efficacy of our methods in numerous tasks across two standard datasets (i.e., BTCV and MSD). Benchmark comparisons with eight state-of-the-art models highlight LoGoNet's superior performance in both inference time and accuracy.
Authors: Sicen Guo, Ziwei Long, Zhiyuan Wu, Qijun Chen, Ioannis Pitas, Rui Fan
Abstract: Despite the impressive performance achieved by data-fusion networks with duplex encoders for visual semantic segmentation, they become ineffective when spatial geometric data are not available. Implicitly infusing the spatial geometric prior knowledge acquired by a data-fusion teacher network into a single-modal student network is a practical, albeit less explored research avenue. This article delves into this topic and resorts to knowledge distillation approaches to address this problem. We introduce the Learning to Infuse ''X'' (LIX) framework, with novel contributions in both logit distillation and feature distillation aspects. We present a mathematical proof that underscores the limitation of using a single, fixed weight in decoupled knowledge distillation and introduce a logit-wise dynamic weight controller as a solution to this issue. Furthermore, we develop an adaptively-recalibrated feature distillation algorithm, including two novel techniques: feature recalibration via kernel regression and in-depth feature consistency quantification via centered kernel alignment. Extensive experiments conducted with intermediate-fusion and late-fusion networks across various public datasets provide both quantitative and qualitative evaluations, demonstrating the superior performance of our LIX framework when compared to other state-of-the-art approaches.
Authors: Stefan Andreas Baumann, Felix Krause, Michael Neumayr, Nick Stracke, Melvin Sevi, Vincent Tao Hu, Bj\"orn Ommer
Abstract: Recent advances in text-to-image (T2I) diffusion models have significantly improved the quality of generated images. However, providing efficient control over individual subjects, particularly the attributes characterizing them, remains a key challenge. While existing methods have introduced mechanisms to modulate attribute expression, they typically provide either detailed, object-specific localization of such a modification or full-scale fine-grained, nuanced control of attributes. No current approach offers both simultaneously, resulting in a gap when trying to achieve precise continuous and subject-specific attribute modulation in image generation. In this work, we demonstrate that token-level directions exist within commonly used CLIP text embeddings that enable fine-grained, subject-specific control of high-level attributes in T2I models. We introduce two methods to identify these directions: a simple, optimization-free technique and a learning-based approach that utilizes the T2I model to characterize semantic concepts more specifically. Our methods allow the augmentation of the prompt text input, enabling fine-grained control over multiple attributes of individual subjects simultaneously, without requiring any modifications to the diffusion model itself. This approach offers a unified solution that fills the gap between global and localized control, providing competitive flexibility and precision in text-guided image generation. Project page: https://compvis.github.io/attribute-control. Code is available at https://github.com/CompVis/attribute-control.
URLs: https://compvis.github.io/attribute-control., https://github.com/CompVis/attribute-control.
Authors: Hao He, Yinghao Xu, Yuwei Guo, Gordon Wetzstein, Bo Dai, Hongsheng Li, Ceyuan Yang
Abstract: Controllability plays a crucial role in video generation, as it allows users to create and edit content more precisely. Existing models, however, lack control of camera pose that serves as a cinematic language to express deeper narrative nuances. To alleviate this issue, we introduce CameraCtrl, enabling accurate camera pose control for video diffusion models. Our approach explores effective camera trajectory parameterization along with a plug-and-play camera pose control module that is trained on top of a video diffusion model, leaving other modules of the base model untouched. Moreover, a comprehensive study on the effect of various training datasets is conducted, suggesting that videos with diverse camera distributions and similar appearance to the base model indeed enhance controllability and generalization. Experimental results demonstrate the effectiveness of CameraCtrl in achieving precise camera control with different video generation models, marking a step forward in the pursuit of dynamic and customized video storytelling from textual and camera pose inputs.
Authors: Jiannan Huang, Jun Hao Liew, Hanshu Yan, Yuyang Yin, Yao Zhao, Humphrey Shi, Yunchao Wei
Abstract: Recent text-to-image customization works have proven successful in generating images of given concepts by fine-tuning diffusion models on a few examples. However, tuning-based methods inherently tend to overfit the concepts, resulting in failure to create the concept under multiple conditions (*e.g.*, headphone is missing when generating "a `dog wearing a headphone"). Interestingly, we notice that the base model before fine-tuning exhibits the capability to compose the base concept with other elements (*e.g.*, "a dog wearing a headphone"), implying that the compositional ability only disappears after personalization tuning. We observe a semantic shift in the customized concept after fine-tuning, indicating that the personalized concept is not aligned with the original concept, and further show through theoretical analyses that this semantic shift leads to increased difficulty in sampling the joint conditional probability distribution, resulting in the loss of the compositional ability. Inspired by this finding, we present **ClassDiffusion**, a technique that leverages a **semantic preservation loss** to explicitly regulate the concept space when learning a new concept. Although simple, this approach effectively prevents semantic drift during the fine-tuning process of the target concepts. Extensive qualitative and quantitative experiments demonstrate that the use of semantic preservation loss effectively improves the compositional abilities of fine-tuning models. Lastly, we also extend our ClassDiffusion to personalized video generation, demonstrating its flexibility.
Authors: Ziyang Wang, Shoubin Yu, Elias Stengel-Eskin, Jaehong Yoon, Feng Cheng, Gedas Bertasius, Mohit Bansal
Abstract: Long-form video understanding is complicated by the high redundancy of video data and the abundance of query-irrelevant information. To tackle these challenges, we propose VideoTree, a training-free framework which builds a query-adaptive and hierarchical video representation for LLM reasoning over long-form videos. First, VideoTree extracts query-relevant information from the input video through an iterative process, progressively refining the selection of keyframes based on their relevance to the query. Furthermore, VideoTree leverages the inherent hierarchical structure of long video data, which is often overlooked by existing LLM-based methods. Specifically, we incorporate multi-granularity information into a tree-based representation, allowing VideoTree to extract query-relevant details from long videos in a coarse-to-fine manner. This enables the model to effectively handle a wide range of video queries with varying levels of detail. Finally, VideoTree aggregates the hierarchical query-relevant information within the tree structure and feeds it into an LLM reasoning model to answer the query. Our experiments show that our method improves both reasoning accuracy and efficiency. Specifically, VideoTree outperforms existing training-free approaches on EgoSchema and NExT-QA with less inference time, achieving 61.1% and 75.6% accuracy on the test set without additional video-specific training. Moreover, on the long split of Video-MME (average 44 minutes), VideoTree achieves better performance than GPT-4V and many other MLLMs that were extensively trained on video data.
Authors: Duy-Kien Nguyen, Mahmoud Assran, Unnat Jain, Martin R. Oswald, Cees G. M. Snoek, Xinlei Chen
Abstract: This work does not introduce a new method. Instead, we present an interesting finding that questions the necessity of the inductive bias of locality in modern computer vision architectures. Concretely, we find that vanilla Transformers can operate by directly treating each individual pixel as a token and achieve highly performant results. This is substantially different from the popular design in Vision Transformer, which maintains the inductive bias from ConvNets towards local neighborhoods (e.g. by treating each 16x16 patch as a token). We showcase the effectiveness of pixels-as-tokens across three well-studied computer vision tasks: supervised learning for classification and regression, self-supervised learning via masked autoencoding, and image generation with diffusion models. Although it's computationally less practical to directly operate on individual pixels, we believe the community must be made aware of this surprising piece of knowledge when devising the next generation of neural network architectures for computer vision.
Authors: Wenbin An, Feng Tian, Sicong Leng, Jiahao Nie, Haonan Lin, QianYing Wang, Ping Chen, Xiaoqin Zhang, Shijian Lu
Abstract: Despite great success across various multimodal tasks, Large Vision-Language Models (LVLMs) often encounter object hallucinations with generated textual responses being inconsistent with the actual objects in images. We examine different LVLMs and pinpoint that one root cause of object hallucinations lies with deficient attention on discriminative image features. Specifically, LVLMs often predominantly attend to prompt-irrelevant global features instead of prompt-relevant local features, undermining their visual grounding capacity and leading to object hallucinations. We propose Assembly of Global and Local Attention (AGLA), a training-free and plug-and-play approach that mitigates hallucinations by assembling global features for response generation and local features for visual discrimination simultaneously. Specifically, we introduce an image-prompt matching scheme that captures prompt-relevant local features from images, leading to an augmented view of the input image where prompt-relevant content is highlighted while irrelevant distractions are suppressed. Hallucinations can thus be mitigated with a calibrated logit distribution that is from generative global features of the original image and discriminative local features of the augmented image. Extensive experiments show the superiority of AGLA in LVLM hallucination mitigation, demonstrating its wide applicability across both discriminative and generative tasks. Our code is available at https://github.com/Lackel/AGLA.
Authors: Qingxuan Wu, Zhiyang Dou, Sirui Xu, Soshi Shimada, Chen Wang, Zhengming Yu, Yuan Liu, Cheng Lin, Zeyu Cao, Taku Komura, Vladislav Golyanik, Christian Theobalt, Wenping Wang, Lingjie Liu
Abstract: Reconstructing 3D hand-face interactions with deformations from a single image is a challenging yet crucial task with broad applications in AR, VR, and gaming. The challenges stem from self-occlusions during single-view hand-face interactions, diverse spatial relationships between hands and face, complex deformations, and the ambiguity of the single-view setting. The first and only method for hand-face interaction recovery, Decaf, introduces a global fitting optimization guided by contact and deformation estimation networks trained on studio-collected data with 3D annotations. However, Decaf suffers from a time-consuming optimization process and limited generalization capability due to its reliance on 3D annotations of hand-face interaction data. To address these issues, we present DICE, the first end-to-end method for Deformation-aware hand-face Interaction reCovEry from a single image. DICE estimates the poses of hands and faces, contacts, and deformations simultaneously using a Transformer-based architecture. It features disentangling the regression of local deformation fields and global mesh vertex locations into two network branches, enhancing deformation and contact estimation for precise and robust hand-face mesh recovery. To improve generalizability, we propose a weakly-supervised training approach that augments the training set using in-the-wild images without 3D ground-truth annotations, employing the depths of 2D keypoints estimated by off-the-shelf models and adversarial priors of poses for supervision. Our experiments demonstrate that DICE achieves state-of-the-art performance on a standard benchmark and in-the-wild data in terms of accuracy and physical plausibility. Additionally, our method operates at an interactive rate (20 fps) on an Nvidia 4090 GPU, whereas Decaf requires more than 15 seconds for a single image. Our code will be publicly available upon publication.
Authors: Sara Mandelli, Edoardo Daniele Cannas, Paolo Bestagini, Stefano Tebaldini, Stefano Tubaro
Abstract: The vast accessibility of Synthetic Aperture Radar (SAR) images through online portals has propelled the research across various fields. This widespread use and easy availability have unfortunately made SAR data susceptible to malicious alterations, such as local editing applied to the images for inserting or covering the presence of sensitive targets. Vulnerability is further emphasized by the fact that most SAR products, despite their original complex nature, are often released as amplitude-only information, allowing even inexperienced attackers to edit and easily alter the pixel content. To contrast malicious manipulations, in the last years the forensic community has begun to dig into the SAR manipulation issue, proposing detectors that effectively localize the tampering traces in amplitude images. Nonetheless, in this paper we demonstrate that an expert practitioner can exploit the complex nature of SAR data to obscure any signs of manipulation within a locally altered amplitude image. We refer to this approach as a counter-forensic attack. To achieve the concealment of manipulation traces, the attacker can simulate a re-acquisition of the manipulated scene by the SAR system that initially generated the pristine image. In doing so, the attacker can obscure any evidence of manipulation, making it appear as if the image was legitimately produced by the system. This attack has unique features that make it both highly generalizable and relatively easy to apply. First, it is a black-box attack, meaning it is not designed to deceive a specific forensic detector. Furthermore, it does not require a training phase and is not based on adversarial operations. We assess the effectiveness of the proposed counter-forensic approach across diverse scenarios, examining various manipulation operations.
Authors: Zhenghao Zhang, Junchao Liao, Menghao Li, Zuozhuo Dai, Bingxue Qiu, Siyu Zhu, Long Qin, Weizhi Wang
Abstract: Recent advancements in Diffusion Transformer (DiT) have demonstrated remarkable proficiency in producing high-quality video content. Nonetheless, the potential of transformer-based diffusion models for effectively generating videos with controllable motion remains an area of limited exploration. This paper introduces Tora, the first trajectory-oriented DiT framework that concurrently integrates textual, visual, and trajectory conditions, thereby enabling scalable video generation with effective motion guidance. Specifically, Tora consists of a Trajectory Extractor (TE), a Spatial-Temporal DiT, and a Motion-guidance Fuser (MGF). The TE encodes arbitrary trajectories into hierarchical spacetime motion patches with a 3D motion compression network. The MGF integrates the motion patches into the DiT blocks to generate consistent videos that accurately follow designated trajectories. Our design aligns seamlessly with DiT's scalability, allowing precise control of video content's dynamics with diverse durations, aspect ratios, and resolutions. Extensive experiments demonstrate that Tora excels in achieving high motion fidelity compared to the foundational DiT model, while also accurately simulating the complex movements of the physical world. Code is made available at https://github.com/alibaba/Tora .
Authors: An-Lan Wang, Bin Shan, Wei Shi, Kun-Yu Lin, Xiang Fei, Guozhi Tang, Lei Liao, Can Huang, Jingqun Tang, Wei-Shi Zheng
Abstract: This work presents ParGo, a novel Partial-Global projector designed to connect the vision and language modalities for Multimodal Large Language Models (MLLMs). Unlike previous works that rely on global attention-based projectors, our ParGo bridges the representation gap between the separately pre-trained vision encoders and the LLMs by integrating global and partial views, which alleviates the overemphasis on prominent regions. To facilitate the effective training of ParGo, we collect a large-scale detail-captioned image-text dataset named ParGoCap-1M-PT, consisting of 1 million images paired with high-quality captions. Extensive experiments on several MLLM benchmarks demonstrate the effectiveness of our ParGo, highlighting its superiority in aligning vision and language modalities. Compared to conventional Q-Former projector, our ParGo achieves an improvement of 259.96 in MME benchmark. Furthermore, our experiments reveal that ParGo significantly outperforms other projectors, particularly in tasks that emphasize detail perception ability.
Authors: Lin Geng Foo, Yixuan He, Ajmal Saeed Mian, Hossein Rahmani, Jun Liu, Christian Theobalt
Abstract: Text-based editing of 3D human avatars to precisely match user requirements is challenging due to the inherent ambiguity and limited expressiveness of natural language. To overcome this, we propose the Avatar Concept Slider (ACS), a 3D avatar editing method that allows precise editing of semantic concepts in human avatars towards a specified intermediate point between two extremes of concepts, akin to moving a knob along a slider track. To achieve this, our ACS has three designs: Firstly, a Concept Sliding Loss based on linear discriminant analysis to pinpoint the concept-specific axes for precise editing. Secondly, an Attribute Preserving Loss based on principal component analysis for improved preservation of avatar identity during editing. We further propose a 3D Gaussian Splatting primitive selection mechanism based on concept-sensitivity, which updates only the primitives that are the most sensitive to our target concept, to improve efficiency. Results demonstrate that our ACS enables controllable 3D avatar editing, without compromising the avatar quality or its identifying attributes.
Authors: Qifeng Chen, Sheng Yang, Sicong Du, Tao Tang, Peng Chen, Yuchi Huo
Abstract: We present LiDAR-GS, a Gaussian Splatting (GS) method for real-time, high-fidelity re-simulation of LiDAR scans in public urban road scenes. Recent GS methods proposed for cameras have achieved significant advancements in real-time rendering beyond Neural Radiance Fields (NeRF). However, applying GS representation to LiDAR, an active 3D sensor type, poses several challenges that must be addressed to preserve high accuracy and unique characteristics. Specifically, LiDAR-GS designs a differentiable laser beam splatting, using range-view representation for precise surface splatting by projecting lasers onto micro cross-sections, effectively eliminating artifacts associated with local affine approximations. Furthermore, LiDAR-GS leverages Neural Gaussian Representation, which further integrate view-dependent clues, to represent key LiDAR properties that are influenced by the incident direction and external factors. Combining these practices with some essential adaptations, e.g., dynamic instances decomposition, LiDAR-GS succeeds in simultaneously re-simulating depth, intensity, and ray-drop channels, achieving state-of-the-art results in both rendering frame rate and quality on publically available large scene datasets when compared with the methods using explicit mesh or implicit NeRF. Our source code is publicly available at https://www.github.com/cqf7419/LiDAR-GS.
Authors: Qianli Ma, Xuefei Ning, Dongrui Liu, Li Niu, Linfeng Zhang
Abstract: Diffusion models are trained by learning a sequence of models that reverse each step of noise corruption. Typically, the model parameters are fully shared across multiple timesteps to enhance training efficiency. However, since the denoising tasks differ at each timestep, the gradients computed at different timesteps may conflict, potentially degrading the overall performance of image generation. To solve this issue, this work proposes a \textbf{De}couple-then-\textbf{Me}rge (\textbf{DeMe}) framework, which begins with a pretrained model and finetunes separate models tailored to specific timesteps. We introduce several improved techniques during the finetuning stage to promote effective knowledge sharing while minimizing training interference across timesteps. Finally, after finetuning, these separate models can be merged into a single model in the parameter space, ensuring efficient and practical inference. Experimental results show significant generation quality improvements upon 6 benchmarks including Stable Diffusion on COCO30K, ImageNet1K, PartiPrompts, and DDPM on LSUN Church, LSUN Bedroom, and CIFAR10. Code is available at \href{https://github.com/MqLeet/DeMe}{GitHub}.
Authors: Shaohao Rui, Lingzhi Chen, Zhenyu Tang, Lilong Wang, Mianxin Liu, Shaoting Zhang, Xiaosong Wang
Abstract: Self-supervised learning has greatly facilitated medical image analysis by suppressing the training data requirement for real-world applications. Current paradigms predominantly rely on self-supervision within uni-modal image data, thereby neglecting the inter-modal correlations essential for effective learning of cross-modal image representations. This limitation is particularly significant for naturally grouped multi-modal data, e.g., multi-parametric MRI scans for a patient undergoing various functional imaging protocols in the same study. To bridge this gap, we conduct a novel multi-modal image pre-training with three proxy tasks to facilitate the learning of cross-modality representations and correlations using multi-modal brain MRI scans (over 2.4 million images in 16,022 scans of 3,755 patients), i.e., cross-modal image reconstruction, modality-aware contrastive learning, and modality template distillation. To demonstrate the generalizability of our pre-trained model, we conduct extensive experiments on various benchmarks with ten downstream tasks. The superior performance of our method is reported in comparison to state-of-the-art pre-training methods, with Dice Score improvement of 0.28\%-14.47\% across six segmentation benchmarks and a consistent accuracy boost of 0.65\%-18.07\% in four individual image classification tasks.
Authors: Jaehong Yoon, Shoubin Yu, Vaidehi Patil, Huaxiu Yao, Mohit Bansal
Abstract: Recent advances in diffusion models have significantly enhanced their ability to generate high-quality images and videos, but they have also increased the risk of producing unsafe content. Existing unlearning/editing-based methods for safe generation remove harmful concepts from models but face several challenges: (1) They cannot instantly remove harmful concepts without training. (2) Their safe generation capabilities depend on collected training data. (3) They alter model weights, risking degradation in quality for content unrelated to toxic concepts. To address these, we propose SAFREE, a novel, training-free approach for safe T2I and T2V, that does not alter the model's weights. Specifically, we detect a subspace corresponding to a set of toxic concepts in the text embedding space and steer prompt embeddings away from this subspace, thereby filtering out harmful content while preserving intended semantics. To balance the trade-off between filtering toxicity and preserving safe concepts, SAFREE incorporates a novel self-validating filtering mechanism that dynamically adjusts the denoising steps when applying the filtered embeddings. Additionally, we incorporate adaptive re-attention mechanisms within the diffusion latent space to selectively diminish the influence of features related to toxic concepts at the pixel level. In the end, SAFREE ensures coherent safety checking, preserving the fidelity, quality, and safety of the output. SAFREE achieves SOTA performance in suppressing unsafe content in T2I generation compared to training-free baselines and effectively filters targeted concepts while maintaining high-quality images. It also shows competitive results against training-based methods. We extend SAFREE to various T2I backbones and T2V tasks, showcasing its flexibility and generalization. SAFREE provides a robust and adaptable safeguard for ensuring safe visual generation.
Authors: Yuxun Qu, Yongqiang Tang, Chenyang Zhang, Wensheng Zhang
Abstract: Different from the traditional semi-supervised learning paradigm that is constrained by the close-world assumption, Generalized Category Discovery (GCD) presumes that the unlabeled dataset contains new categories not appearing in the labeled set, and aims to not only classify old categories but also discover new categories in the unlabeled data. Existing studies on GCD typically devote to transferring the general knowledge from the self-supervised pretrained model to the target GCD task via some fine-tuning strategies, such as partial tuning and prompt learning. Nevertheless, these fine-tuning methods fail to make a sound balance between the generalization capacity of pretrained backbone and the adaptability to the GCD task. To fill this gap, in this paper, we propose a novel adapter-tuning-based method named AdaptGCD, which is the first work to introduce the adapter tuning into the GCD task and provides some key insights expected to enlighten future research. Furthermore, considering the discrepancy of supervision information between the old and new classes, a multi-expert adapter structure equipped with a route assignment constraint is elaborately devised, such that the data from old and new classes are separated into different expert groups. Extensive experiments are conducted on 7 widely-used datasets. The remarkable improvements in performance highlight the effectiveness of our proposals.
Authors: Cristiano Patr\'icio, Lu\'is F. Teixeira, Jo\~ao C. Neves
Abstract: The main challenges hindering the adoption of deep learning-based systems in clinical settings are the scarcity of annotated data and the lack of interpretability and trust in these systems. Concept Bottleneck Models (CBMs) offer inherent interpretability by constraining the final disease prediction on a set of human-understandable concepts. However, this inherent interpretability comes at the cost of greater annotation burden. Additionally, adding new concepts requires retraining the entire system. In this work, we introduce a novel two-step methodology that addresses both of these challenges. By simulating the two stages of a CBM, we utilize a pretrained Vision Language Model (VLM) to automatically predict clinical concepts, and an off-the-shelf Large Language Model (LLM) to generate disease diagnoses based on the predicted concepts. Furthermore, our approach supports test-time human intervention, enabling corrections to predicted concepts, which improves final diagnoses and enhances transparency in decision-making. We validate our approach on three skin lesion datasets, demonstrating that it outperforms traditional CBMs and state-of-the-art explainable methods, all without requiring any training and utilizing only a few annotated examples. The code is available at https://github.com/CristianoPatricio/2-step-concept-based-skin-diagnosis.
URLs: https://github.com/CristianoPatricio/2-step-concept-based-skin-diagnosis.
Authors: Andr\'es Mu\~noz, Nancy Thomas, Annita Vapsi, Daniel Borrajo
Abstract: Many industrial and service sectors require tools to extract vehicle characteristics from images. This is a complex task not only by the variety of noise, and large number of classes, but also by the constant introduction of new vehicle models to the market. In this paper, we present Veri-Car, an information retrieval integrated approach designed to help on this task. It leverages supervised learning techniques to accurately identify the make, type, model, year, color, and license plate of cars. The approach also addresses the challenge of handling open-world problems, where new car models and variations frequently emerge, by employing a sophisticated combination of pre-trained models, and a hierarchical multi-similarity loss. Veri-Car demonstrates robust performance, achieving high precision and accuracy in classifying both seen and unseen data. Additionally, it integrates an ensemble license plate detection, and an OCR model to extract license plate numbers with impressive accuracy.
Authors: Neale Ratzlaff, Matthew Lyle Olson, Musashi Hinck, Estelle Aflalo, Shao-Yen Tseng, Vasudev Lal, Phillip Howard
Abstract: Large Multi-Modal Models (LMMs) have demonstrated impressive capabilities as general-purpose chatbots able to engage in conversations about visual inputs. However, their responses are influenced by societal biases present in their training datasets, leading to undesirable differences in how the model responds when presented with images depicting people of different demographics. In this work, we propose a training-free debiasing framework for LMMs that intervenes on the model's representations during text generation by constructing a steering vector that reduces reference on protected attributes. Our framework introduces two complementary methods: (1) a dataset-based approach that constructs a steering vector by contrasting model activations on biased and neutral inputs, and (2) a novel optimization-based approach designed for low-resource settings, which constructs the steering vector using a single step of gradient-based perturbation without requiring additional data. Our experiments show that these interventions effectively reduce the propensity of LMMs to generate text related to protected attributes while maintaining sentiment and fluency. Furthermore, we demonstrate that debiased LMMs achieve comparable accuracy to their unmodified counterparts on downstream tasks, indicating that bias mitigation can be achieved without sacrificing model performance.
Authors: Yiyang Cai, Zhengkai Jiang, Yulong Liu, Chunyang Jiang, Wei Xue, Wenhan Luo, Yike Guo
Abstract: Facial personalization faces challenges to maintain identity fidelity without disrupting the foundation model's prompt consistency. The mainstream personalization models employ identity embedding to integrate identity information within the cross-attention mechanisms of UNet. However, our preliminary experimental findings reveal that identity embeddings compromise the effectiveness of other tokens in the prompt, thereby limiting high prompt consistency and controllability. Moreover, by deactivating identity embedding, personalization models still demonstrate the underlying foundation models' ability to control facial attributes precisely. It suggests that such foundation models' knowledge can be leveraged to \textbf{cure} the ill-aligned prompt consistency of personalization models. Building upon these insights, we propose \textbf{FreeCure}, a framework that improves the prompt consistency of personalization models with their latent foundation models' knowledge. First, by setting a dual inference paradigm with/without identity embedding, we identify attributes (\textit{e.g.}, hair, accessories, etc.) for enhancements. Second, we introduce a novel foundation-aware self-attention module, coupled with an inversion-based process to bring well-aligned attribute information to the personalization process. Our approach is \textbf{training-free}, and can effectively enhance a wide array of facial attributes in a non-intrusive manner; and it can be seamlessly integrated into existing popular personalization models, without harming their well-trained modules. FreeCure has demonstrated significant improvements in prompt consistency across a diverse set of state-of-the-art facial personalization models while maintaining the integrity of original identity fidelity. The project page is available \href{https://github.com/YIYANGCAI/freecure-project-page}{here}.
Authors: Nazia Tasnim, Bryan A. Plummer
Abstract: Incremental learning aims to adapt to new sets of categories over time with minimal computational overhead. Prior work often addresses this task by training efficient task-specific adaptors that modify frozen layer weights or features to capture relevant information without affecting predictions on previously learned categories. While these adaptors are generally more efficient than finetuning the entire network, they still require tens to hundreds of thousands of task-specific trainable parameters even for relatively small networks, making it challenging to operate on resource-constrained environments with high communication costs like edge devices or mobile phones. Thus, we propose Reparameterized, Compact weight Adaptation for Sequential Tasks (RECAST), a novel method that dramatically reduces task-specific trainable parameters to fewer than 50 - several orders of magnitude less than competing methods like LoRA. RECAST accomplishes this efficiency by learning to decompose layer weights into a soft parameter-sharing framework consisting of shared weight templates and very few module-specific scaling factors or coefficients. This soft parameter-sharing framework allows for effective task-wise reparameterization by tuning only these coefficients while keeping templates frozen.A key innovation of RECAST is the novel weight reconstruction pipeline called Neural Mimicry, which eliminates the need for pretraining from scratch. This allows for high-fidelity emulation of existing pretrained weights within our framework and provides quick adaptability to any model scale and architecture. Extensive experiments across six datasets demonstrate RECAST outperforms the state-of-the-art by up to 3% across various scales, architectures, and parameter spaces Moreover, we show that RECAST's architecture-agnostic nature allows for seamless integration with existing methods, further boosting performance.
Authors: Yuhang Han, Xuyang Liu, Zihan Zhang, Pengxiang Ding, Donglin Wang, Honggang Chen, Qingsen Yan, Siteng Huang
Abstract: The quadratic complexity of Multimodal Large Language Models (MLLMs) with respect to sequence length poses significant computational and memory challenges, hindering their real-world deployment. While existing training-free token reduction methods aim to address these inefficiencies, how to precisely identify redundant visual tokens and recover the essential information from the discarded tokens remain unclear. In this paper, we propose a ''filter-correlate-compress'' framework that decomposes the token reduction into three stages: filtering redundant tokens, correlating discarded information to preserved tokens, and compressing tokens to minimize redundancy. Following the framework, we propose a solution FiCoCo to identify limitations in single redundancy assessment, propose adaptive strategies to retain critical information from discarded tokens, and mitigate semantic dilution during token fusion. Two specialized variants, FiCoCo-V (for vision encoders) and FiCoCo-L (for LLM decoders), further optimize efficiency across MLLM architectures. Extensive experiments demonstrate that FiCoCo achieves up to 5.7x/14.7x FLOPs reduction with 92.8%/93.6% performance retention on LLaVA-1.5-7B/LLaVA-NeXT-7B. Our methods consistently outperform state-of-the-art training-free approaches, showcasing effectiveness and generalizability across model architectures, sizes, and tasks without requiring retraining. Our project page is at https://ficoco-accelerate.github.io/.
Authors: Gwanghyun Kim, Suh Yoon Jeon, Seunggyu Lee, Se Young Chun
Abstract: We present PersonaCraft, a framework for controllable and occlusion-robust full-body personalized image synthesis of multiple individuals in complex scenes. Current methods struggle with occlusion-heavy scenarios and complete body personalization, as 2D pose conditioning lacks 3D geometry, often leading to ambiguous occlusions and anatomical distortions, and many approaches focus solely on facial identity. In contrast, our PersonaCraft integrates diffusion models with 3D human modeling, employing SMPLx-ControlNet, to utilize 3D geometry like depth and normal maps for robust 3D-aware pose conditioning and enhanced anatomical coherence. To handle fine-grained occlusions, we propose Occlusion Boundary Enhancer Network that exploits depth edge signals with occlusion-focused training, and Occlusion-Aware Classifier-Free Guidance strategy that selectively reinforces conditioning in occluded regions without affecting unoccluded areas. PersonaCraft can seamlessly be combined with Face Identity ControlNet, achieving full-body multi-human personalization and thus marking a significant advancement beyond prior approaches that concentrate only on facial identity. Our dual-pathway body shape representation with SMPLx-based shape parameters and textual refinement, enables precise full-body personalization and flexible user-defined body shape adjustments. Extensive quantitative experiments and user studies demonstrate that PersonaCraft significantly outperforms existing methods in generating high-quality, multi-person images with accurate personalization and robust occlusion handling.
Authors: Mingxing Rao, Bohan Jiang, Daniel Moyer
Abstract: In the present work we present Training Noise Token (TNT) Pruning for vision transformers. Our method relaxes the discrete token dropping condition to continuous additive noise, providing smooth optimization in training, while retaining discrete dropping computational gains in deployment settings. We provide theoretical connections to Rate-Distortion literature, and empirical evaluations on the ImageNet dataset using ViT and DeiT architectures demonstrating TNT's advantages over previous pruning methods.
Authors: Zixuan Chen, Guangcong Wang, Jiahao Zhu, Jianhuang Lai, Xiaohua Xie
Abstract: 3D Gaussian Splatting (3DGS) has recently created impressive 3D assets for various applications. However, considering security, capacity, invisibility, and training efficiency, the copyright of 3DGS assets is not well protected as existing watermarking methods are unsuited for its rendering pipeline. In this paper, we propose GuardSplat, an innovative and efficient framework for watermarking 3DGS assets. Specifically, 1) We propose a CLIP-guided pipeline for optimizing the message decoder with minimal costs. The key objective is to achieve high-accuracy extraction by leveraging CLIP's aligning capability and rich representations, demonstrating exceptional capacity and efficiency. 2) We tailor a Spherical-Harmonic-aware (SH-aware) Message Embedding module for 3DGS, seamlessly embedding messages into the SH features of each 3D Gaussian while preserving the original 3D structure. This enables watermarking 3DGS assets with minimal fidelity trade-offs and prevents malicious users from removing the watermarks from the model files, meeting the demands for invisibility and security. 3) We present an Anti-distortion Message Extraction module to improve robustness against various distortions. Experiments demonstrate that GuardSplat outperforms state-of-the-art and achieves fast optimization speed. Project page is at https://narcissusex.github.io/GuardSplat, and Code is at https://github.com/NarcissusEx/GuardSplat.
URLs: https://narcissusex.github.io/GuardSplat,, https://github.com/NarcissusEx/GuardSplat.
Authors: Yunpeng Bai, Qixing Huang
Abstract: Monocular Depth Estimation (MDE) is a fundamental 3D vision problem with numerous applications such as 3D scene reconstruction, autonomous navigation, and AI content creation. However, robust and generalizable MDE remains challenging due to limited real-world labeled data and distribution gaps between synthetic datasets and real data. Existing methods often struggle with real-world test data with low efficiency, reduced accuracy, and lack of detail. To address these issues, we propose an efficient MDE approach named FiffDepth. The key feature of FiffDepth is its use of diffusion priors. It transforms diffusion-based image generators into a feed-forward architecture for detailed depth estimation. FiffDepth preserves key generative features and integrates the strong generalization capabilities of models like DINOv2. Through benchmark evaluations, we demonstrate that FiffDepth achieves exceptional accuracy, stability, and fine-grained detail, offering significant improvements in MDE performance against state-of-the-art MDE approaches. The paper's source code is available here: https://yunpeng1998.github.io/FiffDepth/
Authors: Zewei Zhou, Hao Xiang, Zhaoliang Zheng, Seth Z. Zhao, Mingyue Lei, Yun Zhang, Tianhui Cai, Xinyi Liu, Johnson Liu, Maheswari Bajji, Xin Xia, Zhiyu Huang, Bolei Zhou, Jiaqi Ma
Abstract: Vehicle-to-everything (V2X) technologies offer a promising paradigm to mitigate the limitations of constrained observability in single-vehicle systems. Prior work primarily focuses on single-frame cooperative perception, which fuses agents' information across different spatial locations but ignores temporal cues and temporal tasks (e.g., temporal perception and prediction). In this paper, we focus on the spatio-temporal fusion in V2X scenarios and design one-step and multi-step communication strategies (when to transmit) as well as examine their integration with three fusion strategies - early, late, and intermediate (what to transmit), providing comprehensive benchmarks with 11 fusion models (how to fuse). Furthermore, we propose V2XPnP, a novel intermediate fusion framework within one-step communication for end-to-end perception and prediction. Our framework employs a unified Transformer-based architecture to effectively model complex spatio-temporal relationships across multiple agents, frames, and high-definition map. Moreover, we introduce the V2XPnP Sequential Dataset that supports all V2X collaboration modes and addresses the limitations of existing real-world datasets, which are restricted to single-frame or single-mode cooperation. Extensive experiments demonstrate our framework outperforms state-of-the-art methods in both perception and prediction tasks. The codebase and dataset will be released to facilitate future V2X research.
Authors: Tianyu Chang, Xiaohao Chen, Zhichao Wei, Xuanpu Zhang, Qing-Guo Chen, Weihua Luo, Peipei Song, Xun Yang
Abstract: Video Virtual Try-on aims to seamlessly transfer a reference garment onto a target person in a video while preserving both visual fidelity and temporal coherence. Existing methods typically rely on inpainting masks to define the try-on area, enabling accurate garment transfer for simple scenes (e.g., in-shop videos). However, these mask-based approaches struggle with complex real-world scenarios, as overly large and inconsistent masks often destroy spatial-temporal information, leading to distorted results. Mask-free methods alleviate this issue but face challenges in accurately determining the try-on area, especially for videos with dynamic body movements. To address these limitations, we propose PEMF-VTO, a novel Point-Enhanced Mask-Free Video Virtual Try-On framework that leverages sparse point alignments to explicitly guide garment transfer. Our key innovation is the introduction of point-enhanced guidance, which provides flexible and reliable control over both spatial-level garment transfer and temporal-level video coherence. Specifically, we design a Point-Enhanced Transformer (PET) with two core components: Point-Enhanced Spatial Attention (PSA), which uses frame-cloth point alignments to precisely guide garment transfer, and Point-Enhanced Temporal Attention (PTA), which leverages frame-frame point correspondences to enhance temporal coherence and ensure smooth transitions across frames. Extensive experiments demonstrate that our PEMF-VTO outperforms state-of-the-art methods, generating more natural, coherent, and visually appealing try-on videos, particularly for challenging in-the-wild scenarios. The link to our paper's homepage is https://pemf-vto.github.io/.
Authors: Hui Zhang, Dexiang Hong, Yitong Wang, Jie Shao, Xinglong Wu, Zuxuan Wu, Yu-Gang Jiang
Abstract: Diffusion models have been recognized for their ability to generate images that are not only visually appealing but also of high artistic quality. As a result, Layout-to-Image (L2I) generation has been proposed to leverage region-specific positions and descriptions to enable more precise and controllable generation. However, previous methods primarily focus on UNet-based models (e.g., SD1.5 and SDXL), and limited effort has explored Multimodal Diffusion Transformers (MM-DiTs), which have demonstrated powerful image generation capabilities. Enabling MM-DiT for layout-to-image generation seems straightforward but is challenging due to the complexity of how layout is introduced, integrated, and balanced among multiple modalities. To this end, we explore various network variants to efficiently incorporate layout guidance into MM-DiT, and ultimately present SiamLayout. To Inherit the advantages of MM-DiT, we use a separate set of network weights to process the layout, treating it as equally important as the image and text modalities. Meanwhile, to alleviate the competition among modalities, we decouple the image-layout interaction into a siamese branch alongside the image-text one and fuse them in the later stage. Moreover, we contribute a large-scale layout dataset, named LayoutSAM, which includes 2.7 million image-text pairs and 10.7 million entities. Each entity is annotated with a bounding box and a detailed description. We further construct the LayoutSAM-Eval benchmark as a comprehensive tool for evaluating the L2I generation quality. Finally, we introduce the Layout Designer, which taps into the potential of large language models in layout planning, transforming them into experts in layout generation and optimization. Our code, model, and dataset will be available at https://creatilayout.github.io.
Authors: Ruida Zhang, Chengxi Li, Chenyangguang Zhang, Xingyu Liu, Haili Yuan, Yanyan Li, Xiangyang Ji, Gim Hee Lee
Abstract: Realistic scene reconstruction in driving scenarios poses significant challenges due to fast-moving objects. Most existing methods rely on labor-intensive manual labeling of object poses to reconstruct dynamic objects in canonical space and move them based on these poses during rendering. While some approaches attempt to use 3D object trackers to replace manual annotations, the limited generalization of 3D trackers -- caused by the scarcity of large-scale 3D datasets -- results in inferior reconstructions in real-world settings. In contrast, 2D foundation models demonstrate strong generalization capabilities. To eliminate the reliance on 3D trackers and enhance robustness across diverse environments, we propose a stable object tracking module by leveraging associations from 2D deep trackers within a 3D object fusion strategy. We address inevitable tracking errors by further introducing a motion learning strategy in an implicit feature space that autonomously corrects trajectory errors and recovers missed detections. Experimental results on Waymo-NOTR and KITTI show that our method outperforms existing approaches. Our code will be made publicly available.
Authors: Xinpeng Liu, Junxuan Liang, Chenshuo Zhang, Zixuan Cai, Cewu Lu, Yong-Lu Li
Abstract: Analyses of human motion kinematics have achieved tremendous advances. However, the production mechanism, known as human dynamics, is still undercovered. In this paper, we aim to push data-driven human dynamics understanding forward. We identify a major obstacle to this as the heterogeneity of existing human motion understanding efforts. Specifically, heterogeneity exists in not only the diverse kinematics representations and hierarchical dynamics representations but also in the data from different domains, namely biomechanics and reinforcement learning. With an in-depth analysis of the existing heterogeneity, we propose to emphasize the beneath homogeneity: all of them represent the homogeneous fact of human motion, though from different perspectives. Given this, we propose Homogeneous Dynamics Space (HDyS) as a fundamental space for human dynamics by aggregating heterogeneous data and training a homogeneous latent space with inspiration from the inverse-forward dynamics procedure. Leveraging the heterogeneous representations and datasets, HDyS achieves decent mapping between human kinematics and dynamics. We demonstrate the feasibility of HDyS with extensive experiments and applications. The project page is https://foruck.github.io/HDyS.
Authors: Huaxin Zhang, Xiaohao Xu, Xiang Wang, Jialong Zuo, Xiaonan Huang, Changxin Gao, Shanjun Zhang, Li Yu, Nong Sang
Abstract: How can we enable models to comprehend video anomalies occurring over varying temporal scales and contexts? Traditional Video Anomaly Understanding (VAU) methods focus on frame-level anomaly prediction, often missing the interpretability of complex and diverse real-world anomalies. Recent multimodal approaches leverage visual and textual data but lack hierarchical annotations that capture both short-term and long-term anomalies. To address this challenge, we introduce HIVAU-70k, a large-scale benchmark for hierarchical video anomaly understanding across any granularity. We develop a semi-automated annotation engine that efficiently scales high-quality annotations by combining manual video segmentation with recursive free-text annotation using large language models (LLMs). This results in over 70,000 multi-granular annotations organized at clip-level, event-level, and video-level segments. For efficient anomaly detection in long videos, we propose the Anomaly-focused Temporal Sampler (ATS). ATS integrates an anomaly scorer with a density-aware sampler to adaptively select frames based on anomaly scores, ensuring that the multimodal LLM concentrates on anomaly-rich regions, which significantly enhances both efficiency and accuracy. Extensive experiments demonstrate that our hierarchical instruction data markedly improves anomaly comprehension. The integrated ATS and visual-language model outperform traditional methods in processing long videos. Our benchmark and model are publicly available at https://github.com/pipixin321/HolmesVAU.
Authors: Shihan Wu, Ji Zhang, Pengpeng Zeng, Lianli Gao, Jingkuan Song, Heng Tao Shen
Abstract: Prompt tuning (PT) has long been recognized as an effective and efficient paradigm for transferring large pre-trained vision-language models (VLMs) to downstream tasks by learning a tiny set of context vectors. Nevertheless, in this work, we reveal that freezing the parameters of VLMs during learning the context vectors neither facilitates the transferability of pre-trained knowledge nor improves the memory and time efficiency significantly. Upon further investigation, we find that reducing both the length and width of the feature-gradient propagation flows of the full fine-tuning (FT) baseline is key to achieving effective and efficient knowledge transfer. Motivated by this, we propose Skip Tuning, a novel paradigm for adapting VLMs to downstream tasks. Unlike existing PT or adapter-based methods, Skip Tuning applies Layer-wise Skipping (LSkip) and Class-wise Skipping (CSkip) upon the FT baseline without introducing extra context vectors or adapter modules. Extensive experiments across a wide spectrum of benchmarks demonstrate the superior effectiveness and efficiency of our Skip Tuning over both PT and adapter-based methods. Code: https://github.com/Koorye/SkipTuning.
Authors: Ashish Seth, Dinesh Manocha, Chirag Agarwal
Abstract: Large Vision-Language Models (LVLMs) have demonstrated remarkable performance in complex multimodal tasks. However, these models still suffer from hallucinations, particularly when required to implicitly recognize or infer diverse visual entities from images for complex vision-language tasks. To address this challenge, we propose HALLUCINOGEN, a novel visual question answering (VQA) benchmark that employs contextual reasoning prompts as hallucination attacks to evaluate the extent of hallucination in state-of-the-art LVLMs. Our benchmark provides a comprehensive study of the implicit reasoning capabilities of these models by first categorizing visual entities based on the ease of recognition in an image as either salient (prominent, visibly recognizable objects such as a car) or latent entities (such as identifying a disease from a chest X-ray), which are not readily visible and require domain knowledge or contextual reasoning for accurate inference. Next, we design hallucination attacks for both types of entities to assess hallucinations in LVLMs while performing various vision-language tasks, such as locating or reasoning about specific entities within an image, where models must perform implicit reasoning by verifying the existence of the queried entity within the image before generating responses. Finally, our extensive evaluations of eleven LVLMs, including powerful open-source models (like LLaMA-3.2 and DeepSeek-V2), commercial models like Gemini, and two hallucination mitigation strategies across multiple datasets, demonstrate that current LVLMs remain susceptible to hallucination attacks.
Authors: Jiajun Cao, Yuan Zhang, Tao Huang, Ming Lu, Qizhe Zhang, Ruichuan An, Ningning MA, Shanghang Zhang
Abstract: Visual encoders are fundamental components in vision-language models (VLMs), each showcasing unique strengths derived from various pre-trained visual foundation models. To leverage the various capabilities of these encoders, recent studies incorporate multiple encoders within a single VLM, leading to a considerable increase in computational cost. In this paper, we present Mixture-of-Visual-Encoder Knowledge Distillation (MoVE-KD), a novel framework that distills the unique proficiencies of multiple vision encoders into a single, efficient encoder model. Specifically, to mitigate conflicts and retain the unique characteristics of each teacher encoder, we employ low-rank adaptation (LoRA) and mixture-of-experts (MoEs) to selectively activate specialized knowledge based on input features, enhancing both adaptability and efficiency. To regularize the KD process and enhance performance, we propose an attention-based distillation strategy that adaptively weighs the different encoders and emphasizes valuable visual tokens, reducing the burden of replicating comprehensive but distinct features from multiple teachers. Comprehensive experiments on popular VLMs, such as LLaVA and LLaVA-NeXT, validate the effectiveness of our method. Our code is available at: https://github.com/hey-cjj/MoVE-KD.
Authors: Zongxia Li, Xiyang Wu, Hongyang Du, Huy Nghiem, Guangyao Shi
Abstract: Multimodal Vision Language Models (VLMs) have emerged as a transformative technology at the intersection of computer vision and natural language processing, enabling machines to perceive and reason about the world through both visual and textual modalities. For example, models such as CLIP, Claude, and GPT-4V demonstrate strong reasoning and understanding abilities on visual and textual data and beat classical single modality vision models on zero-shot classification. Despite their rapid advancements in research and growing popularity in applications, a comprehensive survey of existing studies on VLMs is notably lacking, particularly for researchers aiming to leverage VLMs in their specific domains. To this end, we provide a systematic overview of VLMs in the following aspects: model information of the major VLMs developed over the past five years (2019-2024); the main architectures and training methods of these VLMs; summary and categorization of the popular benchmarks and evaluation metrics of VLMs; the applications of VLMs including embodied agents, robotics, and video generation; the challenges and issues faced by current VLMs such as hallucination, fairness, and safety. Detailed collections including papers and model repository links are listed in https://github.com/zli12321/Awesome-VLM-Papers-And-Models.git.
URLs: https://github.com/zli12321/Awesome-VLM-Papers-And-Models.git.
Authors: Jiaxing Zhao, Boyuan Sun, Xiang Chen, Xihan Wei, Qibin Hou
Abstract: In this paper, we introduce LLaVA-Octopus, a novel video multimodal large language model. LLaVA-Octopus adaptively weights features from different visual projectors based on user instructions, enabling us to leverage the complementary strengths of each projector. We observe that different visual projectors exhibit distinct characteristics when handling specific tasks. For instance, some projectors excel at capturing static details, while others are more effective at processing temporal information, and some are better suited for tasks requiring temporal coherence. By dynamically adjusting feature weights according to user instructions, LLaVA-Octopus dynamically selects and combines the most suitable features, significantly enhancing the model's performance in multimodal tasks. Experimental results demonstrate that LLaVA-Octopus achieves excellent performance across multiple benchmarks, especially in tasks such as video question answering, long video understanding, and comprehensive multi-choices benchmarks, highlighting its broad application potential.
Authors: Xueyi Ke, Satoshi Tsutsui, Yayun Zhang, Bihan Wen
Abstract: Infants develop complex visual understanding rapidly, even preceding of the acquisition of linguistic inputs. As computer vision seeks to replicate the human vision system, understanding infant visual development may offer valuable insights. In this paper, we present an interdisciplinary study exploring this question: can a computational model that imitates the infant learning process develop broader visual concepts that extend beyond the vocabulary it has heard, similar to how infants naturally learn? To investigate this, we analyze a recently published model in Science by Vong et al.,which is trained on longitudinal, egocentric images of a single child paired with transcribed parental speech. We introduce a training-free framework that can discover visual concept neurons hidden in the model's internal representations. Our findings show that these neurons can classify objects outside its original vocabulary. Furthermore, we compare the visual representations in infant-like models with those in moder computer vision models, such as CLIP or ImageNet pre-trained model, highlighting key similarities and differences. Ultimately, our work bridges cognitive science and computer vision by analyzing the internal representations of a computational model trained on an infant's visual and linguistic inputs.
Authors: Arkaprava Sinha, Monish Soundar Raj, Pu Wang, Ahmed Helmy, Srijan Das
Abstract: Temporal Action Detection (TAD) in untrimmed videos requires models that can efficiently (1) process long-duration videos, (2) capture temporal variations within action classes, and (3) handle dense, overlapping actions, all while remaining suitable for resource-constrained edge deployment. While Transformer-based methods achieve high accuracy, their quadratic complexity hinders deployment in such scenarios. Given the recent popularity of linear complexity Mamba-based models, leveraging them for TAD is a natural choice. However, naively adapting Mamba from language or vision tasks fails to provide an optimal solution and does not address the challenges of long, untrimmed videos. Therefore, we propose Multi-Scale Temporal Mamba (MS-Temba), the first Mamba-based architecture specifically designed for densely labeled TAD tasks. MS-Temba features Temporal Mamba Blocks (Temba Blocks), consisting of Temporal Convolutional Module (TCM) and Dilated SSM (D-SSM). TCM captures short-term dependencies using dilated convolutions, while D-SSM introduces a novel dilated state-space mechanism to model long-range temporal relationships effectively at each temporal scale. These multi-scale representations are aggregated by Scale-Aware State Fuser, which learns a unified representation for detecting densely overlapping actions. Experiments show that MS-Temba achieves state-of-the-art performance on long-duration videos, remains competitive on shorter segments, and reduces model complexity by 88%. Its efficiency and effectiveness make MS-Temba well-suited for real-world edge deployment.
Authors: Wenping Jin, Li Zhu, Jing Sun
Abstract: Weakly supervised violence detection refers to the technique of training models to identify violent segments in videos using only video-level labels. Among these approaches, multimodal violence detection, which integrates modalities such as audio and optical flow, holds great potential. Existing methods in this domain primarily focus on designing multimodal fusion models to address modality discrepancies. In contrast, we take a different approach; leveraging the inherent discrepancies across modalities in violence event representation to propose a novel multimodal semantic feature alignment method. This method sparsely maps the semantic features of local, transient, and less informative modalities ( such as audio and optical flow ) into the more informative RGB semantic feature space. Through an iterative process, the method identifies the suitable no-zero feature matching subspace and aligns the modality-specific event representations based on this subspace, enabling the full exploitation of information from all modalities during the subsequent modality fusion stage. Building on this, we design a new weakly supervised violence detection framework that consists of unimodal multiple-instance learning for extracting unimodal semantic features, multimodal alignment, multimodal fusion, and final detection. Experimental results on benchmark datasets demonstrate the effectiveness of our method, achieving an average precision (AP) of 86.07% on the XD-Violence dataset. Our code is available at https://github.com/xjpp2016/MAVD.
Authors: Haozhe Xie, Zhaoxi Chen, Fangzhou Hong, Ziwei Liu
Abstract: 3D scene generation has garnered growing attention in recent years and has made significant progress. Generating 4D cities is more challenging than 3D scenes due to the presence of structurally complex, visually diverse objects like buildings and vehicles, and heightened human sensitivity to distortions in urban environments. To tackle these issues, we propose CityDreamer4D, a compositional generative model specifically tailored for generating unbounded 4D cities. Our main insights are 1) 4D city generation should separate dynamic objects (e.g., vehicles) from static scenes (e.g., buildings and roads), and 2) all objects in the 4D scene should be composed of different types of neural fields for buildings, vehicles, and background stuff. Specifically, we propose Traffic Scenario Generator and Unbounded Layout Generator to produce dynamic traffic scenarios and static city layouts using a highly compact BEV representation. Objects in 4D cities are generated by combining stuff-oriented and instance-oriented neural fields for background stuff, buildings, and vehicles. To suit the distinct characteristics of background stuff and instances, the neural fields employ customized generative hash grids and periodic positional embeddings as scene parameterizations. Furthermore, we offer a comprehensive suite of datasets for city generation, including OSM, GoogleEarth, and CityTopia. The OSM dataset provides a variety of real-world city layouts, while the Google Earth and CityTopia datasets deliver large-scale, high-quality city imagery complete with 3D instance annotations. Leveraging its compositional design, CityDreamer4D supports a range of downstream applications, such as instance editing, city stylization, and urban simulation, while delivering state-of-the-art performance in generating realistic 4D cities.
Authors: Jinrong Cui, Xiaohuang Wu, Haitao Zhang, Chongjie Dong, Jie Wen
Abstract: Deep multi-view clustering seeks to utilize the abundant information from multiple views to improve clustering performance. However, most of the existing clustering methods often neglect to fully mine multi-view structural information and fail to explore the distribution of multi-view data, limiting clustering performance. To address these limitations, we propose a structure-guided deep multi-view clustering model. Specifically, we introduce a positive sample selection strategy based on neighborhood relationships, coupled with a corresponding loss function. This strategy constructs multi-view nearest neighbor graphs to dynamically redefine positive sample pairs, enabling the mining of local structural information within multi-view data and enhancing the reliability of positive sample selection. Additionally, we introduce a Gaussian distribution model to uncover latent structural information and introduce a loss function to reduce discrepancies between view embeddings. These two strategies explore multi-view structural information and data distribution from different perspectives, enhancing consistency across views and increasing intra-cluster compactness. Experimental evaluations demonstrate the efficacy of our method, showing significant improvements in clustering performance on multiple benchmark datasets compared to state-of-the-art multi-view clustering approaches.
Authors: Chengwei Zheng, Lixin Xue, Juan Zarate, Jie Song
Abstract: 3D Gaussian Splatting techniques have enabled efficient photo-realistic rendering of static scenes. Recent works have extended these approaches to support surface reconstruction and tracking. However, tracking dynamic surfaces with 3D Gaussians remains challenging due to complex topology changes, such as surfaces appearing, disappearing, or splitting. To address these challenges, we propose GauSTAR, a novel method that achieves photo-realistic rendering, accurate surface reconstruction, and reliable 3D tracking for general dynamic scenes with changing topology. Given multi-view captures as input, GauSTAR binds Gaussians to mesh faces to represent dynamic objects. For surfaces with consistent topology, GauSTAR maintains the mesh topology and tracks the meshes using Gaussians. For regions where topology changes, GauSTAR adaptively unbinds Gaussians from the mesh, enabling accurate registration and generation of new surfaces based on these optimized Gaussians. Additionally, we introduce a surface-based scene flow method that provides robust initialization for tracking between frames. Experiments demonstrate that our method effectively tracks and reconstructs dynamic surfaces, enabling a range of applications. Our project page with the code release is available at https://eth-ait.github.io/GauSTAR/.
Authors: Ruojun Xu, Weijie Xi, Xiaodi Wang, Yongbo Mao, Zach Cheng
Abstract: Training-free diffusion-based methods have achieved remarkable success in style transfer, eliminating the need for extensive training or fine-tuning. However, due to the lack of targeted training for style information extraction and constraints on the content image layout, training-free methods often suffer from layout changes of original content and content leakage from style images. Through a series of experiments, we discovered that an effective startpoint in the sampling stage significantly enhances the style transfer process. Based on this discovery, we propose StyleSSP, which focuses on obtaining a better startpoint to address layout changes of original content and content leakage from style image. StyleSSP comprises two key components: (1) Frequency Manipulation: To improve content preservation, we reduce the low-frequency components of the DDIM latent, allowing the sampling stage to pay more attention to the layout of content images; and (2) Negative Guidance via Inversion: To mitigate the content leakage from style image, we employ negative guidance in the inversion stage to ensure that the startpoint of the sampling stage is distanced from the content of style image. Experiments show that StyleSSP surpasses previous training-free style transfer baselines, particularly in preserving original content and minimizing the content leakage from style image. Project page: https://github.com/bytedance/StyleSSP.
Authors: Kelsey Allen, Ishita Dasgupta, Eliza Kosoy, Andrew K. Lampinen
Abstract: Inductive biases are what allow learners to make guesses in the absence of conclusive evidence. These biases have often been studied in cognitive science using concepts or categories -- e.g. by testing how humans generalize a new category from a few examples that leave the category boundary ambiguous. We use these approaches to study generalization in foundation models during in-context learning. Modern foundation models can condition on both vision and text, and differences in how they interpret and learn from these different modalities is an emerging area of study. Here, we study how their generalizations vary by the modality in which stimuli are presented, and the way the stimuli are described in text. We study these biases with three different experimental paradigms, across three different vision-language models. We find that the models generally show some bias towards generalizing according to shape over color. This shape bias tends to be amplified when the examples are presented visually. By contrast, when examples are presented in text, the ordering of adjectives affects generalization. However, the extent of these effects vary across models and paradigms. These results help to reveal how vision-language models represent different types of inputs in context, and may have practical implications for the use of vision-language models.
Authors: Shilong Zhang, Wenbo Li, Shoufa Chen, Chongjian Ge, Peize Sun, Yida Zhang, Yi Jiang, Zehuan Yuan, Binyue Peng, Ping Luo
Abstract: DiT diffusion models have achieved great success in text-to-video generation, leveraging their scalability in model capacity and data scale. High content and motion fidelity aligned with text prompts, however, often require large model parameters and a substantial number of function evaluations (NFEs). Realistic and visually appealing details are typically reflected in high resolution outputs, further amplifying computational demands especially for single stage DiT models. To address these challenges, we propose a novel two stage framework, FlashVideo, which strategically allocates model capacity and NFEs across stages to balance generation fidelity and quality. In the first stage, prompt fidelity is prioritized through a low resolution generation process utilizing large parameters and sufficient NFEs to enhance computational efficiency. The second stage establishes flow matching between low and high resolutions, effectively generating fine details with minimal NFEs. Quantitative and visual results demonstrate that FlashVideo achieves state-of-the-art high resolution video generation with superior computational efficiency. Additionally, the two-stage design enables users to preview the initial output and accordingly adjust the prompt before committing to full-resolution generation, thereby significantly reducing computational costs and wait times as well as enhancing commercial viability.
Authors: Aditya Vora, Sauradip Nag, Hao Zhang
Abstract: We present ATOP (Articulate That Object Part), a novel few-shot method based on motion personalization to articulate a static 3D object with respect to a part and its motion as prescribed in a text prompt. Given the scarcity of available datasets with motion attribute annotations, existing methods struggle to generalize well in this task. In our work, the text input allows us to tap into the power of modern-day diffusion models to generate plausible motion samples for the right object category and part. In turn, the input 3D object provides image prompting to personalize the generated video to that very object we wish to articulate. Our method starts with a few-shot finetuning for category-specific motion generation, a key first step to compensate for the lack of articulation awareness by current diffusion models. For this, we finetune a pre-trained multi-view image generation model for controllable multi-view video generation, using a small collection of video samples obtained for the target object category. This is followed by motion video personalization that is realized by multi-view rendered images of the target 3D object. At last, we transfer the personalized video motion to the target 3D object via differentiable rendering to optimize part motion parameters by a score distillation sampling loss. Experimental results on PartNet-Sapien and ACD datasets show that our method is capable of generating realistic motion videos and predicting 3D motion parameters in a more accurate and generalizable way, compared to prior works in the few-shot setting.
Authors: Ai Chen, Yuxu Lu, Dong Yang, Junlin Zhou, Yan Fu, Duanbing Chen
Abstract: Salient object detection (SOD) plays a critical role in intelligent transportation systems (ITS), facilitating the detection and segmentation of key visual elements in an image. However, adverse imaging conditions such as haze during the day, low light, and haze at night severely degrade image quality and hinder reliable object detection in real-world scenarios. To address these challenges, we propose a multi-knowledge-oriented nighttime haze imaging enhancer (MKoIE), which integrates three tasks: daytime dehazing, low-light enhancement, and nighttime dehazing. The MKoIE incorporates two key innovative components: First, the network employs a task-oriented node learning mechanism to handle three specific degradation types: day-time haze, low light, and night-time haze conditions, with an embedded self-attention module enhancing its performance in nighttime imaging. In addition, multi-receptive field enhancement module that efficiently extracts multi-scale features through three parallel depthwise separable convolution branches with different dilation rates, capturing comprehensive spatial information with minimal computational overhead to meet the requirements of real-time ITS deployment. To ensure optimal image reconstruction quality and visual characteristics, we suggest a hybrid loss function. Extensive experiments on different types of weather/imaging conditions illustrate that MKoIE surpasses existing methods, enhancing the reliability, accuracy, and operational efficiency of ITS. The code is available at https://github.com/Ai-Chen-Lab/MKoIE.
Authors: Nie Lin, Takehiko Ohkawa, Yifei Huang, Mingfang Zhang, Minjie Cai, Ming Li, Ryosuke Furuta, Yoichi Sato
Abstract: We present a framework for pre-training of 3D hand pose estimation from in-the-wild hand images sharing with similar hand characteristics, dubbed SimHand. Pre-training with large-scale images achieves promising results in various tasks, but prior methods for 3D hand pose pre-training have not fully utilized the potential of diverse hand images accessible from in-the-wild videos. To facilitate scalable pre-training, we first prepare an extensive pool of hand images from in-the-wild videos and design our pre-training method with contrastive learning. Specifically, we collect over 2.0M hand images from recent human-centric videos, such as 100DOH and Ego4D. To extract discriminative information from these images, we focus on the similarity of hands: pairs of non-identical samples with similar hand poses. We then propose a novel contrastive learning method that embeds similar hand pairs closer in the feature space. Our method not only learns from similar samples but also adaptively weights the contrastive learning loss based on inter-sample distance, leading to additional performance gains. Our experiments demonstrate that our method outperforms conventional contrastive learning approaches that produce positive pairs sorely from a single image with data augmentation. We achieve significant improvements over the state-of-the-art method (PeCLR) in various datasets, with gains of 15% on FreiHand, 10% on DexYCB, and 4% on AssemblyHands. Our code is available at https://github.com/ut-vision/SiMHand.
Authors: Chen-Chen Zong, Sheng-Jun Huang
Abstract: Active learning (AL), which iteratively queries the most informative examples from a large pool of unlabeled candidates for model training, faces significant challenges in the presence of open-set classes. Existing methods either prioritize query examples likely to belong to known classes, indicating low epistemic uncertainty (EU), or focus on querying those with highly uncertain predictions, reflecting high aleatoric uncertainty (AU). However, they both yield suboptimal performance, as low EU corresponds to limited useful information, and closed-set AU metrics for unknown class examples are less meaningful. In this paper, we propose an Energy-based Active Open-set Annotation (EAOA) framework, which effectively integrates EU and AU to achieve superior performance. EAOA features a $(C+1)$-class detector and a target classifier, incorporating an energy-based EU measure and a margin-based energy loss designed for the detector, alongside an energy-based AU measure for the target classifier. Another crucial component is the target-driven adaptive sampling strategy. It first forms a smaller candidate set with low EU scores to ensure closed-set properties, making AU metrics meaningful. Subsequently, examples with high AU scores are queried to form the final query set, with the candidate set size adjusted adaptively. Extensive experiments show that EAOA achieves state-of-the-art performance while maintaining high query precision and low training overhead. The code is available at https://github.com/chenchenzong/EAOA.
Authors: Kyle Stein, Arash Mahyari, Guillermo Francia, Eman El-Sheikh
Abstract: Vision-Language Models (VLMs) have demonstrated impressive capabilities in learning joint representations of visual and textual data, making them powerful tools for tasks such as Compositional Zero-Shot Learning (CZSL). CZSL requires models to generalize to novel combinations of visual primitives-such as attributes and objects-that were not explicitly encountered during training. Recent works in prompting for CZSL have focused on modifying inputs for the text encoder, often using static prompts that do not change across varying visual contexts. However, these approaches struggle to fully capture varying visual contexts, as they focus on text adaptation rather than leveraging visual features for compositional reasoning. To address this, we propose Visual Adaptive Prompting System (VAPS) that leverages a learnable visual prompt repository and similarity-based retrieval mechanism within the framework of VLMs to bridge the gap between semantic and visual features. Our method introduces a dynamic visual prompt repository mechanism that selects the most relevant attribute and object prompts based on the visual features of the image. Our proposed system includes a visual prompt adapter that encourages the model to learn a more generalizable embedding space. Experiments on three CZSL benchmarks, across both closed and open-world scenarios, demonstrate state-of-the-art results.
Authors: Zhiwei Ling, Yachen Chang, Hailiang Zhao, Xinkui Zhao, Kingsum Chow, Shuiguang Deng
Abstract: Deep neural networks (DNNs) have been widely criticized for their overconfidence when dealing with out-of-distribution (OOD) samples, highlighting the critical need for effective OOD detection to ensure the safe deployment of DNNs in real-world settings. Existing post-hoc OOD detection methods primarily enhance the discriminative power of logit-based approaches by reshaping sample features, yet they often neglect critical information inherent in the features themselves. In this paper, we propose the Class-Aware Relative Feature-based method (CARef), which utilizes the error between a sample's feature and its class-aware average feature as a discriminative criterion. To further refine this approach, we introduce the Class-Aware Decoupled Relative Feature-based method (CADRef), which decouples sample features based on the alignment of signs between the relative feature and corresponding model weights, enhancing the discriminative capabilities of CARef. Extensive experimental results across multiple datasets and models demonstrate that both proposed methods exhibit effectiveness and robustness in OOD detection compared to state-of-the-art methods. Specifically, our two methods outperform the best baseline by 2.82% and 3.27% in AUROC, with improvements of 4.03% and 6.32% in FPR95, respectively.
Authors: Adnen Abdessaied, Anna Rohrbach, Marcus Rohrbach, Andreas Bulling
Abstract: We present V$^2$Dial - a novel expert-based model specifically geared towards simultaneously handling image and video input data for multimodal conversational tasks. Current multimodal models primarily focus on simpler tasks (e.g., VQA, VideoQA, video-text retrieval) and often neglect the more challenging conversational counterparts, such as video and visual/image dialog. Moreover, works on both conversational tasks evolved separately from each other despite their apparent similarities limiting their applicability potential. To this end, we propose to unify both tasks using a single model that for the first time jointly learns the spatial and temporal features of images and videos by routing them through dedicated experts and aligns them using matching and contrastive learning techniques. Furthermore, we systemically study the domain shift between the two tasks by investigating whether and to what extent these seemingly related tasks can mutually benefit from their respective training data. Extensive evaluations on the widely used video and visual dialog datasets of AVSD and VisDial show that our model achieves new state-of-the-art results across four benchmarks both in zero-shot and fine-tuning settings.
Authors: Zhen Yang, Guibao Shen, Liang Hou, Mushui Liu, Luozhou Wang, Xin Tao, Pengfei Wan, Di Zhang, Ying-Cong Chen
Abstract: Diffusion models have achieved remarkable advances in various image generation tasks. However, their performance notably declines when generating images at resolutions higher than those used during the training period. Despite the existence of numerous methods for producing high-resolution images, they either suffer from inefficiency or are hindered by complex operations. In this paper, we propose RectifiedHR, an straightforward and efficient solution for training-free high-resolution image generation. Specifically, we introduce the noise refresh strategy, which theoretically only requires a few lines of code to unlock the model's high-resolution generation ability and improve efficiency. Additionally, we first observe the phenomenon of energy decay that may cause image blurriness during the high-resolution image generation process. To address this issue, we introduce average latent energy analysis and discover that an improved classifier-free guidance hyperparameter can significantly enhance generation performance. Our method is entirely training-free and boasts a simple implementation logic and efficient performance. Through extensive comparisons with numerous baseline methods, our RectifiedHR demonstrates superior effectiveness and efficiency.
Authors: Alaa Mazouz, Sumanta Chaudhuri, Marco Cagnanzzo, Mihai Mitrea, Enzo Tartaglione, Attilio Fiandrotti
Abstract: Learnable Image Compression (LIC) has shown the potential to outperform standardized video codecs in RD efficiency, prompting the research for hardware-friendly implementations. Most existing LIC hardware implementations prioritize latency to RD-efficiency and through an extensive exploration of the hardware design space. We present a novel design paradigm where the burden of tuning the design for a specific hardware platform is shifted towards model dimensioning and without compromising on RD-efficiency. First, we design a framework for distilling a leaner student LIC model from a reference teacher: by tuning a single model hyperparameters, we can meet the constraints of different hardware platforms without a complex hardware design exploration. Second, we propose a hardware-friendly implementation of the Generalized Divisive Normalization - GDN activation that preserves RD efficiency even post parameter quantization. Third, we design a pipelined FPGA configuration which takes full advantage of available FPGA resources by leveraging parallel processing and optimizing resource allocation. Our experiments with a state of the art LIC model show that we outperform all existing FPGA implementations while performing very close to the original model.
Authors: Xiang Gao, Shuai Yang, Jiaying Liu
Abstract: Optical illusion hidden picture is an interesting visual perceptual phenomenon where an image is cleverly integrated into another picture in a way that is not immediately obvious to the viewer. Established on the off-the-shelf text-to-image (T2I) diffusion model, we propose a novel training-free text-guided image-to-image (I2I) translation framework dubbed as \textbf{P}hase-\textbf{T}ransferred \textbf{Diffusion} Model (PTDiffusion) for hidden art syntheses. PTDiffusion embeds an input reference image into arbitrary scenes as described by the text prompts, while exhibiting hidden visual cues of the reference image. At the heart of our method is a plug-and-play phase transfer mechanism that dynamically and progressively transplants diffusion features' phase spectrum from the denoising process to reconstruct the reference image into the one to sample the generated illusion image, realizing harmonious fusion of the reference structural information and the textual semantic information. Furthermore, we propose asynchronous phase transfer to enable flexible control to the degree of hidden content discernability. Our method bypasses any model training and fine-tuning, all while substantially outperforming related methods in image quality, text fidelity, visual discernibility, and contextual naturalness for illusion picture synthesis, as demonstrated by extensive qualitative and quantitative experiments.
Authors: Chengxuan Qian, Kai Han, Jingchao Wang, Zhenlong Yuan, Chongwen Lyu, Jun Chen, Zhe Liu
Abstract: Multimodal learning integrates complementary information from diverse modalities to enhance the decision-making process. However, the potential of multimodal collaboration remains under-exploited due to disparities in data quality and modality representation capabilities. To address this, we introduce DynCIM, a novel dynamic curriculum learning framework designed to quantify the inherent imbalances from both sample and modality perspectives. DynCIM employs a sample-level curriculum to dynamically assess each sample's difficulty according to prediction deviation, consistency, and stability, while a modality-level curriculum measures modality contributions from global and local. Furthermore, a gating-based dynamic fusion mechanism is introduced to adaptively adjust modality contributions, minimizing redundancy and optimizing fusion effectiveness. Extensive experiments on six multimodal benchmarking datasets, spanning both bimodal and trimodal scenarios, demonstrate that DynCIM consistently outperforms state-of-the-art methods. Our approach effectively mitigates modality and sample imbalances while enhancing adaptability and robustness in multimodal learning tasks. Our code is available at https://github.com/Raymond-Qiancx/DynCIM.
Authors: Shiu-hong Kao, Yu-Wing Tai, Chi-Keung Tang
Abstract: Reasoning segmentation is a challenging vision-language task that aims to output the segmentation mask with respect to a complex, implicit, and even non-visual query text. Previous works incorporated multimodal Large Language Models (MLLMs) with segmentation models to approach the difficult problem. However, their segmentation quality often falls short in complex cases, particularly when dealing with out-of-domain objects with intricate structures, blurry boundaries, occlusions, or high similarity with surroundings. In this paper, we introduce ThinkFirst, a training-free reasoning segmentation framework that leverages GPT's chain of thought to address these challenging cases. Our approach allows GPT-4o or other powerful MLLMs to generate a detailed, chain-of-thought description of an image. This summarized description is then passed to a language-instructed segmentation assistant to aid the segmentation process. Our framework allows users to easily interact with the segmentation agent using multimodal inputs, such as easy text and image scribbles, for successive refinement or communication. We evaluate the performance of ThinkFirst on diverse objects. Extensive experiments show that, this zero-shot-CoT approach significantly improves the vanilla reasoning segmentation agent, both qualitatively and quantitatively, while being less sensitive or critical to user-supplied prompts after Thinking First.
Authors: Beitao Chen, Xinyu Lyu, Lianli Gao, Jingkuan Song, Heng Tao Shen
Abstract: Despite their success, Large Vision-Language Models (LVLMs) remain vulnerable to hallucinations. While existing studies attribute the cause of hallucinations to insufficient visual attention to image tokens, our findings indicate that hallucinations also arise from interference from instruction tokens during decoding. Intuitively, certain instruction tokens continuously distort LVLMs' visual perception during decoding, hijacking their visual attention toward less discriminative visual regions. This distortion prevents them integrating broader contextual information from images, ultimately leading to hallucinations. We term this phenomenon 'Attention Hijacking', where disruptive instruction tokens act as 'Attention Hijackers'. To address this, we propose a novel, training-free strategy namely Attention HIjackers Detection and Disentanglement (AID), designed to isolate the influence of Hijackers, enabling LVLMs to rely on their context-aware intrinsic attention map. Specifically, AID consists of three components: First, Attention Hijackers Detection identifies Attention Hijackers by calculating instruction-driven visual salience. Next, Attention Disentanglement mechanism is proposed to mask the visual attention of these identified Hijackers, and thereby mitigate their disruptive influence on subsequent tokens. Finally, Re-Disentanglement recalculates the balance between instruction-driven and image-driven visual salience to avoid over-masking effects. Extensive experiments demonstrate that AID significantly reduces hallucination across various LVLMs on several benchmarks.
Authors: Qingsong Xie, Zhao Zhang, Zhe Huang, Yanhao Zhang, Haonan Lu, Zhenyu Yang
Abstract: Image tokenization has significantly advanced visual generation and multimodal modeling, particularly when paired with autoregressive models. However, current methods face challenges in balancing efficiency and fidelity: high-resolution image reconstruction either requires an excessive number of tokens or compromises critical details through token reduction. To resolve this, we propose Latent Consistency Tokenizer (Layton) that bridges discrete visual tokens with the compact latent space of pre-trained Latent Diffusion Models (LDMs), enabling efficient representation of 1024x1024 images using only 256 tokens-a 16 times compression over VQGAN. Layton integrates a transformer encoder, a quantized codebook, and a latent consistency decoder. Direct application of LDM as the decoder results in color and brightness discrepancies. Thus, we convert it to latent consistency decoder, reducing multi-step sampling to 1-2 steps for direct pixel-level supervision. Experiments demonstrate Layton's superiority in high-fidelity reconstruction, with 10.8 reconstruction Frechet Inception Distance on MSCOCO-2017 5K benchmark for 1024x1024 image reconstruction. We also extend Layton to a text-to-image generation model, LaytonGen, working in autoregression. It achieves 0.73 score on GenEval benchmark, surpassing current state-of-the-art methods. Project homepage: https://github.com/OPPO-Mente-Lab/Layton
Authors: Ahmad Mustafa Anis, Hasnain Ali, Saquib Sarfraz
Abstract: Vision Language Models (VLMs) have demonstrated significant potential in various downstream tasks, including Image/Video Generation, Visual Question Answering, Multimodal Chatbots, and Video Understanding. However, these models often struggle with basic image transformations. This paper investigates the image-level understanding of VLMs, specifically CLIP by OpenAI and SigLIP by Google. Our findings reveal that these models lack comprehension of multiple image-level augmentations. To facilitate this study, we created an augmented version of the Flickr8k dataset, pairing each image with a detailed description of the applied transformation. We further explore how this deficiency impacts downstream tasks, particularly in image editing, and evaluate the performance of state-of-the-art Image2Image models on simple transformations.
Authors: Shunqi Mao, Chaoyi Zhang, Weidong Cai
Abstract: Existing vision-language models (VLMs) often suffer from visual hallucination, where the generated responses contain inaccuracies that are not grounded in the visual input. Efforts to address this issue without model finetuning primarily mitigate hallucination by reducing biases contrastively or amplifying the weights of visual embedding during decoding. However, these approaches improve visual perception at the cost of impairing the language reasoning capability. In this work, we propose the Perception Magnifier (PM), a novel visual decoding method that iteratively isolates relevant visual tokens based on attention and magnifies the corresponding regions, spurring the model to concentrate on fine-grained visual details during decoding. Specifically, by magnifying critical regions while preserving the structural and contextual information at each decoding step, PM allows the VLM to enhance its scrutiny of the visual input, hence producing more accurate and faithful responses. Extensive experimental results demonstrate that PM not only achieves superior hallucination mitigation but also enhances language generation while preserving strong reasoning capabilities. Code is available at https://github.com/ShunqiM/PM .
Authors: Ye Zhang, Zijie Fang, Yifeng Wang, Lingbo Zhang, Xianchao Guan, Yongbing Zhang
Abstract: Nuclei segmentation and classification provide an essential basis for tumor immune microenvironment analysis. The previous nuclei segmentation and classification models require splitting large images into smaller patches for training, leading to two significant issues. First, nuclei at the borders of adjacent patches often misalign during inference. Second, this patch-based approach significantly increases the model's training and inference time. Recently, Mamba has garnered attention for its ability to model large-scale images with linear time complexity and low memory consumption. It offers a promising solution for training nuclei segmentation and classification models on full-sized images. However, the Mamba orientation-based scanning method lacks account for category-specific features, resulting in sub-optimal performance in scenarios with imbalanced class distributions. To address these challenges, this paper introduces a novel scanning strategy based on category probability sorting, which independently ranks and scans features for each category according to confidence from high to low. This approach enhances the feature representation of uncertain samples and mitigates the issues caused by imbalanced distributions. Extensive experiments conducted on four public datasets demonstrate that our method outperforms state-of-the-art approaches, delivering superior performance in nuclei segmentation and classification tasks.
Authors: Jun Yu, Lingsi Zhu, Yanjun Chi, Yunxiang Zhang, Yang Zheng, Yongqi Wang, Xilong Lu
Abstract: Emotional Mimicry Intensity (EMI) estimation serves as a critical technology for understanding human social behavior and enhancing human-computer interaction experiences, where the core challenge lies in dynamic correlation modeling and robust fusion of multimodal temporal signals. To address the limitations of existing methods in insufficient exploitation of modal synergistic effects, noise sensitivity, and limited fine-grained alignment capabilities, this paper proposes a dual-stage cross-modal alignment framework. First, we construct vision-text and audio-text contrastive learning networks based on an improved CLIP architecture, achieving preliminary alignment in the feature space through modality-decoupled pre-training. Subsequently, we design a temporal-aware dynamic fusion module that combines Temporal Convolutional Networks (TCN) and gated bidirectional LSTM to respectively capture the macro-evolution patterns of facial expressions and local dynamics of acoustic features. Innovatively, we introduce a quality-guided modality fusion strategy that enables modality compensation under occlusion and noisy scenarios through differentiable weight allocation. Experimental results on the Hume-Vidmimic2 dataset demonstrate that our method achieves an average Pearson correlation coefficient of 0.35 across six emotion dimensions, outperforming the best baseline by 40\%. Ablation studies further validate the effectiveness of the dual-stage training strategy and dynamic fusion mechanism, providing a novel technical pathway for fine-grained emotion analysis in open environments.
Authors: Hsuan-Kung Yang, Tsung-Chih Chiang, Jou-Min Liu, Ting-Ru Liu, Chun-Wei Huang, Tsu-Ching Hsiao, Chun-Yi Lee
Abstract: In the context of autonomous navigation, effectively conveying abstract navigational cues to agents in dynamic environments presents significant challenges, particularly when navigation information is derived from diverse modalities such as both vision and high-level language descriptions. To address this issue, we introduce a novel technique termed `Virtual Guidance,' which is designed to visually represent non-visual instructional signals. These visual cues are overlaid onto the agent's camera view and served as comprehensible navigational guidance signals. To validate the concept of virtual guidance, we propose a sim-to-real framework that enables the transfer of the trained policy from simulated environments to real world, ensuring the adaptability of virtual guidance in practical scenarios. We evaluate and compare the proposed method against a non-visual guidance baseline through detailed experiments in simulation. The experimental results demonstrate that the proposed virtual guidance approach outperforms the baseline methods across multiple scenarios and offers clear evidence of its effectiveness in autonomous navigation tasks.
Authors: Juyoung Yun, Sol Choi, Francois Rameau, Byungkon Kang, Zhoulai Fu
Abstract: With the increasing complexity of machine learning models, managing computational resources like memory and processing power has become a critical concern. Mixed precision techniques, which leverage different numerical precisions during model training and inference to optimize resource usage, have been widely adopted. However, access to hardware that supports lower precision formats (e.g., FP8 or FP4) remains limited, especially for practitioners with hardware constraints. For many with limited resources, the available options are restricted to using 32-bit, 16-bit, or a combination of the two. While it is commonly believed that 16-bit precision can achieve results comparable to full (32-bit) precision, this study is the first to systematically validate this assumption through both rigorous theoretical analysis and extensive empirical evaluation. Our theoretical formalization of floating-point errors and classification tolerance provides new insights into the conditions under which 16-bit precision can approximate 32-bit results. This study fills a critical gap, proving for the first time that standalone 16-bit precision neural networks match 32-bit and mixed-precision in accuracy while boosting computational speed. Given the widespread availability of 16-bit across GPUs, these findings are especially valuable for machine learning practitioners with limited hardware resources to make informed decisions.
Authors: Mingyan Zhu, Yiming Li, Junfeng Guo, Tao Wei, Shu-Tao Xia, Zhan Qin
Abstract: Currently, sample-specific backdoor attacks (SSBAs) are the most advanced and malicious methods since they can easily circumvent most of the current backdoor defenses. In this paper, we reveal that SSBAs are not sufficiently stealthy due to their poisoned-label nature, where users can discover anomalies if they check the image-label relationship. In particular, we demonstrate that it is ineffective to directly generalize existing SSBAs to their clean-label variants by poisoning samples solely from the target class. We reveal that it is primarily due to two reasons, including \textbf{(1)} the `antagonistic effects' of ground-truth features and \textbf{(2)} the learning difficulty of sample-specific features. Accordingly, trigger-related features of existing SSBAs cannot be effectively learned under the clean-label setting due to their mild trigger intensity required for ensuring stealthiness. We argue that the intensity constraint of existing SSBAs is mostly because their trigger patterns are `content-irrelevant' and therefore act as `noises' for both humans and DNNs. Motivated by this understanding, we propose to exploit content-relevant features, $a.k.a.$ (human-relied) attributes, as the trigger patterns to design clean-label SSBAs. This new attack paradigm is dubbed backdoor attack with attribute trigger (BAAT). Extensive experiments are conducted on benchmark datasets, which verify the effectiveness of our BAAT and its resistance to existing defenses.
Authors: Ribana Roscher, Marc Ru{\ss}wurm, Caroline Gevaert, Michael Kampffmeyer, Jefersson A. dos Santos, Maria Vakalopoulou, Ronny H\"ansch, Stine Hansen, Keiller Nogueira, Jonathan Prexl, Devis Tuia
Abstract: Recent developments and research in modern machine learning have led to substantial improvements in the geospatial field. Although numerous deep learning architectures and models have been proposed, the majority of them have been solely developed on benchmark datasets that lack strong real-world relevance. Furthermore, the performance of many methods has already saturated on these datasets. We argue that a shift from a model-centric view to a complementary data-centric perspective is necessary for further improvements in accuracy, generalization ability, and real impact on end-user applications. Furthermore, considering the entire machine learning cycle-from problem definition to model deployment with feedback-is crucial for enhancing machine learning models that can be reliable in unforeseen situations. This work presents a definition as well as a precise categorization and overview of automated data-centric learning approaches for geospatial data. It highlights the complementary role of data-centric learning with respect to model-centric in the larger machine learning deployment cycle. We review papers across the entire geospatial field and categorize them into different groups. A set of representative experiments shows concrete implementation examples. These examples provide concrete steps to act on geospatial data with data-centric machine learning approaches.
Authors: Tim Selig, Thomas M\"arz, Martin Storath, Andreas Weinmann
Abstract: Computed tomography from a low radiation dose (LDCT) is challenging due to high noise in the projection data. Popular approaches for LDCT image reconstruction are two-stage methods, typically consisting of the filtered backprojection (FBP) algorithm followed by a neural network for LDCT image enhancement. Two-stage methods are attractive for their simplicity and potential for computational efficiency, typically requiring only a single FBP and a neural network forward pass for inference. However, the best reconstruction quality is currently achieved by unrolled iterative methods (Learned Primal-Dual and ItNet), which are more complex and thus have a higher computational cost for training and inference. We propose a method combining the simplicity and efficiency of two-stage methods with state-of-the-art reconstruction quality. Our strategy utilizes a neural network pretrained for Gaussian noise removal from natural grayscale images, fine-tuned for LDCT image enhancement. We call this method FBP-DTSGD (Domain and Task Shifted Gaussian Denoisers) as the fine-tuning is a task shift from Gaussian denoising to enhancing LDCT images and a domain shift from natural grayscale to LDCT images. An ablation study with three different pretrained Gaussian denoisers indicates that the performance of FBP-DTSGD does not depend on a specific denoising architecture, suggesting future advancements in Gaussian denoising could benefit the method. The study also shows that pretraining on natural images enhances LDCT reconstruction quality, especially with limited training data. Notably, pretraining involves no additional cost, as existing pretrained models are used. The proposed method currently holds the top mean position in the LoDoPaB-CT challenge.
Authors: Anna Breger, Ander Biguri, Malena Sabat\'e Landman, Ian Selby, Nicole Amberg, Elisabeth Brunner, Janek Gr\"ohl, Sepideh Hatamikia, Clemens Karner, Lipeng Ning, S\"oren Dittmer, Michael Roberts, AIX-COVNET Collaboration, Carola-Bibiane Sch\"onlieb
Abstract: Image quality assessment (IQA) is indispensable in clinical practice to ensure high standards, as well as in the development stage of machine learning algorithms that operate on medical images. The popular full reference (FR) IQA measures PSNR and SSIM are known and tested for working successfully in many natural imaging tasks, but discrepancies in medical scenarios have been reported in the literature, highlighting the gap between development and actual clinical application. Such inconsistencies are not surprising, as medical images have very different properties than natural images, and PSNR and SSIM have neither been targeted nor properly tested for medical images. This may cause unforeseen problems in clinical applications due to wrong judgment of novel methods. This paper provides a structured and comprehensive overview of examples where PSNR and SSIM prove to be unsuitable for the assessment of novel algorithms using different kinds of medical images, including real-world MRI, CT, OCT, X-Ray, digital pathology and photoacoustic imaging data. Therefore, improvement is urgently needed in particular in this era of AI to increase reliability and explainability in machine learning for medical imaging and beyond. Lastly, we will provide ideas for future research as well as suggesting guidelines for the usage of FR-IQA measures applied to medical images.
Authors: Zhengzhuo Xu, Bowen Qu, Yiyan Qi, Sinan Du, Chengjin Xu, Chun Yuan, Jian Guo
Abstract: Automatic chart understanding is crucial for content comprehension and document parsing. Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in chart understanding through domain-specific alignment and fine-tuning. However, current MLLMs still struggle to provide faithful data and reliable analysis only based on charts. To address it, we propose ChartMoE, which employs the Mixture of Expert (MoE) architecture to replace the traditional linear projector to bridge the modality gap. Specifically, we train several linear connectors through distinct alignment tasks, which are utilized as the foundational initialization parameters for different experts. Additionally, we introduce ChartMoE-Align, a dataset with nearly 1 million chart-table-JSON-code quadruples to conduct three alignment tasks (chart-table/JSON/code). Combined with the vanilla connector, we initialize different experts diversely and adopt high-quality knowledge learning to further refine the MoE connector and LLM parameters. Extensive experiments demonstrate the effectiveness of the MoE connector and our initialization strategy, e.g., ChartMoE improves the accuracy of the previous state-of-the-art from 80.48\% to 84.64\% on the ChartQA benchmark.
Authors: Angela Lopez-Cardona, Carlos Segura, Alexandros Karatzoglou, Sergi Abadal, Ioannis Arapakis
Abstract: Advancements in Natural Language Processing (NLP), have led to the emergence of Large Language Models (LLMs) such as GPT, Llama, Claude, and Gemini, which excel across a range of tasks but require extensive fine-tuning to align their outputs with human expectations. A widely used method for achieving this alignment is Reinforcement Learning from Human Feedback (RLHF), which, despite its success, faces challenges in accurately modelling human preferences. In this paper, we introduce GazeReward, a novel framework that integrates implicit feedback -- and specifically eye-tracking (ET) data -- into the Reward Model (RM). In addition, we explore how ET-based features can provide insights into user preferences. Through ablation studies we test our framework with different integration methods, LLMs, and ET generator models, demonstrating that our approach significantly improves the accuracy of the RM on established human preference datasets. This work advances the ongoing discussion on optimizing AI alignment with human values, exploring the potential of cognitive data for shaping future NLP research.
Authors: Minoh Jeong, Min Namgung, Zae Myung Kim, Dongyeop Kang, Yao-Yi Chiang, Alfred Hero
Abstract: A unified representation space in multi-modal learning is essential for effectively integrating diverse data sources, such as text, images, and audio, to enhance efficiency and performance across various downstream tasks. Recent binding methods, such as ImageBind (Girdhar et al., 2023), typically rely on a single, fixed anchor modality for aligning multi-modal data. We mathematically analyze these fixed anchor binding method and uncover significant limitations: (1) over-reliance on the choice of the anchor modality, (2) inadequate capture of intra-modal information, and (3) failure to account for cross-modal correlation among non-anchored modalities. To address these issues, we propose the need for adaptive anchor binding methods, exemplified by our framework CentroBind. The proposed method uses adaptively adjustable centroid-based anchors generated from all available modalities, leading to a balanced and rich representation space. We theoretically demonstrate that our approach captures three critical properties of multi-modal learning -- intra-modal learning, inter-modal learning, and multi-modal alignment -- while constructing a unified representation that spans all modalities. Experiments on both synthetic and real-world datasets show that adaptive anchor methods such as CentroBind consistently outperform fixed anchor binding methods, verifying our analysis.
Authors: Changdae Oh, Yixuan Li, Kyungwoo Song, Sangdoo Yun, Dongyoon Han
Abstract: Adapting a pre-trained foundation model on downstream tasks should ensure robustness against distribution shifts without the need to retrain the whole model. Although existing weight interpolation methods are simple yet effective, we argue their static nature limits downstream performance while achieving efficiency. In this work, we propose DaWin, a training-free dynamic weight interpolation method that leverages the entropy of individual models over each unlabeled test sample to assess model expertise, and compute per-sample interpolation coefficients dynamically. Unlike previous works that typically rely on additional training to learn such coefficients, our approach requires no training. Then, we propose a mixture modeling approach that greatly reduces inference overhead raised by dynamic interpolation. We validate DaWin on the large-scale visual recognition benchmarks, spanning 14 tasks across robust fine-tuning -- ImageNet and derived five distribution shift benchmarks -- and multi-task learning with eight classification tasks. Results demonstrate that DaWin achieves significant performance gain in considered settings, with minimal computational overhead. We further discuss DaWin's analytic behavior to explain its empirical success.
Authors: Po-han Li, Sandeep P. Chinchali, Ufuk Topcu
Abstract: Multimodal encoders like CLIP excel in tasks such as zero-shot image classification and cross-modal retrieval. However, they require excessive training data. We propose canonical similarity analysis (CSA), which uses two unimodal encoders to replicate multimodal encoders using limited data. CSA maps unimodal features into a multimodal space, using a new similarity score to retain only the multimodal information. CSA only involves the inference of unimodal encoders and a cubic-complexity matrix decomposition, eliminating the need for extensive GPU-based model training. Experiments show that CSA outperforms CLIP while requiring $50,000\times$ fewer multimodal data pairs to bridge the modalities given pre-trained unimodal encoders on ImageNet classification and misinformative news caption detection. CSA surpasses the state-of-the-art method to map unimodal features to multimodal features. We also demonstrate the ability of CSA with modalities beyond image and text, paving the way for future modality pairs with limited paired multimodal data but abundant unpaired unimodal data, such as lidar and text.
Authors: Zhi Hou, Tianyi Zhang, Yuwen Xiong, Haonan Duan, Hengjun Pu, Ronglei Tong, Chengyang Zhao, Xizhou Zhu, Yu Qiao, Jifeng Dai, Yuntao Chen
Abstract: While recent vision-language-action models trained on diverse robot datasets exhibit promising generalization capabilities with limited in-domain data, their reliance on compact action heads to predict discretized or continuous actions constrains adaptability to heterogeneous action spaces. We present Dita, a scalable framework that leverages Transformer architectures to directly denoise continuous action sequences through a unified multimodal diffusion process. Departing from prior methods that condition denoising on fused embeddings via shallow networks, Dita employs in-context conditioning -- enabling fine-grained alignment between denoised actions and raw visual tokens from historical observations. This design explicitly models action deltas and environmental nuances. By scaling the diffusion action denoiser alongside the Transformer's scalability, Dita effectively integrates cross-embodiment datasets across diverse camera perspectives, observation scenes, tasks, and action spaces. Such synergy enhances robustness against various variances and facilitates the successful execution of long-horizon tasks. Evaluations across extensive benchmarks demonstrate state-of-the-art or comparative performance in simulation. Notably, Dita achieves robust real-world adaptation to environmental variances and complex long-horizon tasks through 10-shot finetuning, using only third-person camera inputs. The architecture establishes a versatile, lightweight and open-source baseline for generalist robot policy learning. Project Page: https://robodita.github.io/
Authors: Rania Briq, Jiangtao Wang, Stefan Kesselheim
Abstract: Data pruning is the problem of identifying a core subset that is most beneficial to training and discarding the remainder. While pruning strategies are well studied for discriminative models like those used in classification, little research has gone into their application to generative models. Generative models aim to estimate the underlying distribution of the data, so presumably they should benefit from larger datasets. In this work we aim to shed light on the accuracy of this statement, specifically answer the question of whether data pruning for generative diffusion models could have a positive impact. Contrary to intuition, we show that eliminating redundant or noisy data in large datasets is beneficial particularly when done strategically. We experiment with several pruning methods including recent-state-of-art methods, and evaluate over CelebA-HQ and ImageNet datasets. We demonstrate that a simple clustering method outperforms other sophisticated and computationally demanding methods. We further exhibit how we can leverage clustering to balance skewed datasets in an unsupervised manner to allow fair sampling for underrepresented populations in the data distribution, which is a crucial problem in generative models.
Authors: Jialin Lu, Junjie Shan, Ziqi Zhao, Ka-Ho Chow
Abstract: As object detection becomes integral to many safety-critical applications, understanding its vulnerabilities is essential. Backdoor attacks, in particular, pose a serious threat by implanting hidden triggers in victim models, which adversaries can later exploit to induce malicious behaviors during inference. However, current understanding is limited to single-target attacks, where adversaries must define a fixed malicious behavior (target) before training, making inference-time adaptability impossible. Given the large output space of object detection (including object existence prediction, bounding box estimation, and classification), the feasibility of flexible, inference-time model control remains unexplored. This paper introduces AnywhereDoor, a multi-target backdoor attack for object detection. Once implanted, AnywhereDoor allows adversaries to make objects disappear, fabricate new ones, or mislabel them, either across all object classes or specific ones, offering an unprecedented degree of control. This flexibility is enabled by three key innovations: (i) objective disentanglement to scale the number of supported targets; (ii) trigger mosaicking to ensure robustness even against region-based detectors; and (iii) strategic batching to address object-level data imbalances that hinder manipulation. Extensive experiments demonstrate that AnywhereDoor grants attackers a high degree of control, improving attack success rates by 26% compared to adaptations of existing methods for such flexible control.
Authors: Qizhou Chen, Chengyu Wang, Dakan Wang, Taolin Zhang, Wangyue Li, Xiaofeng He
Abstract: Model editing aims to correct inaccurate knowledge, update outdated information, and incorporate new data into Large Language Models (LLMs) without the need for retraining. This task poses challenges in lifelong scenarios where edits must be continuously applied for real-world applications. While some editors demonstrate strong robustness for lifelong editing in pure LLMs, Vision LLMs (VLLMs), which incorporate an additional vision modality, are not directly adaptable to existing LLM editors. In this paper, we propose LiveEdit, a LIfelong Vision language modEl Edit to bridge the gap between lifelong LLM editing and VLLMs. We begin by training an editing expert generator to independently produce low-rank experts for each editing instance, with the goal of correcting the relevant responses of the VLLM. A hard filtering mechanism is developed to utilize visual semantic knowledge, thereby coarsely eliminating visually irrelevant experts for input queries during the inference stage of the post-edited model. Finally, to integrate visually relevant experts, we introduce a soft routing mechanism based on textual semantic relevance to achieve multi-expert fusion. For evaluation, we establish a benchmark for lifelong VLLM editing. Extensive experiments demonstrate that LiveEdit offers significant advantages in lifelong VLLM editing scenarios. Further experiments validate the rationality and effectiveness of each module design in LiveEdit.
Authors: Deep Chakraborty, Yann LeCun, Tim G. J. Rudner, Erik Learned-Miller
Abstract: A number of different architectures and loss functions have been applied to the problem of self-supervised learning (SSL), with the goal of developing embeddings that provide the best possible pre-training for as-yet-unknown, lightly supervised downstream tasks. One of these SSL criteria is to maximize the entropy of a set of embeddings in some compact space. But the goal of maximizing the embedding entropy often depends -- whether explicitly or implicitly -- upon high dimensional entropy estimates, which typically perform poorly in more than a few dimensions. In this paper, we motivate an effective entropy maximization criterion (E2MC), defined in terms of easy-to-estimate, low-dimensional constraints. We demonstrate that using it to continue training an already-trained SSL model for only a handful of epochs leads to a consistent and, in some cases, significant improvement in downstream performance. We perform careful ablation studies to show that the improved performance is due to the proposed add-on criterion. We also show that continued pre-training with alternative criteria does not lead to notable improvements, and in some cases, even degrades performance.
Authors: Alvaro Becerra, Roberto Daza, Ruth Cobos, Aythami Morales, Julian Fierrez
Abstract: We present a demonstration of a web-based system called M2LADS ("System for Generating Multimodal Learning Analytics Dashboards"), designed to integrate, synchronize, visualize, and analyze multimodal data recorded during computer-based learning sessions with biosensors. This system presents a range of biometric and behavioral data on web-based dashboards, providing detailed insights into various physiological and activity-based metrics. The multimodal data visualized include electroencephalogram (EEG) data for assessing attention and brain activity, heart rate metrics, eye-tracking data to measure visual attention, webcam video recordings, and activity logs of the monitored tasks. M2LADS aims to assist data scientists in two key ways: (1) by providing a comprehensive view of participants' experiences, displaying all data categorized by the activities in which participants are engaged, and (2) by synchronizing all biosignals and videos, facilitating easier data relabeling if any activity information contains errors.
Authors: Gen Shi, Hui Zhang, Jie Tian
Abstract: Accurate segmentation of 3D vascular structures is essential for various medical imaging applications. The dispersed nature of vascular structures leads to inherent spatial uncertainty and necessitates location awareness, yet most current 3D medical segmentation models rely on the patch-wise training strategy that usually loses this spatial context. In this study, we introduce the Coordinate-aware Modulated Mamba Network (COMMA) and contribute a manually labeled dataset of 570 cases, the largest publicly available 3D vessel dataset to date. COMMA leverages both entire and cropped patch data through global and local branches, ensuring robust and efficient spatial location awareness. Specifically, COMMA employs a channel-compressed Mamba (ccMamba) block to encode entire image data, capturing long-range dependencies while optimizing computational costs. Additionally, we propose a coordinate-aware modulated (CaM) block to enhance interactions between the global and local branches, allowing the local branch to better perceive spatial information. We evaluate COMMA on six datasets, covering two imaging modalities and five types of vascular tissues. The results demonstrate COMMA's superior performance compared to state-of-the-art methods with computational efficiency, especially in segmenting small vessels. Ablation studies further highlight the importance of our proposed modules and spatial information. The code and data will be open source at https://github.com/shigen-StoneRoot/COMMA.
Authors: Huanhuan Zhao, Ruben Millan-Solsona, Marti Checa, Spenser R. Brown, Jennifer L. Morrell-Falvey, Liam Collins, Arpan Biswas
Abstract: Microscopy is an essential tool in scientific research, enabling the visualization of structures at micro- and nanoscale resolutions. However, the field of microscopy often encounters limitations in field-of-view (FOV), restricting the amount of sample that can be imaged in a single capture. To overcome this limitation, image stitching techniques have been developed to seamlessly merge multiple overlapping images into a single, high-resolution composite. The images collected from microscope need to be optimally stitched before accurate physical information can be extracted from post analysis. However, the existing stitching tools either struggle to stitch images together when the microscopy images are feature sparse or cannot address all the transformations of images. To address these issues, we propose a bi-channel aided feature-based image stitching method and demonstrate its use on AFM generated biofilm images. The topographical channel image of AFM data captures the morphological details of the sample, and a stitched topographical image is desired for researchers. We utilize the amplitude channel of AFM data to maximize the matching features and to estimate the position of the original topographical images and show that the proposed bi-channel aided stitching method outperforms the traditional stitching approach. Furthermore, we found that the differentiation of the topographical images along the x-axis provides similar feature information to the amplitude channel image, which generalizes our approach when the amplitude images are not available. Here we demonstrated the application on AFM, but similar approaches could be employed of optical microscopy with brightfield and fluorescence channels. We believe this proposed workflow will benefit the experimentalist to avoid erroneous analysis and discovery due to incorrect stitching.
Authors: Theo Di Piazza, Loic Boussel
Abstract: Early detection of cervical cancer is crucial for improving patient outcomes and reducing mortality by identifying precancerous lesions as soon as possible. As a result, the use of pap smear screening has significantly increased, leading to a growing demand for automated tools that can assist cytologists managing their rising workload. To address this, the Pap Smear Cell Classification Challenge (PS3C) has been organized in association with ISBI in 2025. This project aims to promote the development of automated tools for pap smear images classification. The analyzed images are grouped into four categories: healthy, unhealthy, both, and rubbish images which are considered as unsuitable for diagnosis. In this work, we propose a two-stage ensemble approach: first, a neural network determines whether an image is rubbish or not. If not, a second neural network classifies the image as containing a healthy cell, an unhealthy cell, or both.
Authors: Ho Kei Cheng, Alexander Schwing
Abstract: Minibatch optimal transport coupling straightens paths in unconditional flow matching. This leads to computationally less demanding inference as fewer integration steps and less complex numerical solvers can be employed when numerically solving an ordinary differential equation at test time. However, in the conditional setting, minibatch optimal transport falls short. This is because the default optimal transport mapping disregards conditions, resulting in a conditionally skewed prior distribution during training. In contrast, at test time, we have no access to the skewed prior, and instead sample from the full, unbiased prior distribution. This gap between training and testing leads to a subpar performance. To bridge this gap, we propose conditional optimal transport C^2OT that adds a conditional weighting term in the cost matrix when computing the optimal transport assignment. Experiments demonstrate that this simple fix works with both discrete and continuous conditions in 8gaussians-to-moons, CIFAR-10, ImageNet-32x32, and ImageNet-256x256. Our method performs better overall compared to the existing baselines across different function evaluation budgets. Code is available at https://hkchengrex.github.io/C2OT
Authors: Rohit Gandikota, David Bau
Abstract: Distilled diffusion models suffer from a critical limitation: reduced sample diversity compared to their base counterparts. In this work, we uncover that despite this diversity loss, distilled models retain the fundamental concept representations of base models. We demonstrate control distillation - where control mechanisms like Concept Sliders and LoRAs trained on base models can be seamlessly transferred to distilled models and vice-versa, effectively distilling control without any retraining. This preservation of representational structure prompted our investigation into the mechanisms of diversity collapse during distillation. To understand how distillation affects diversity, we introduce Diffusion Target (DT) Visualization, an analysis and debugging tool that reveals how models predict final outputs at intermediate steps. Through DT-Visualization, we identify generation artifacts, inconsistencies, and demonstrate that initial diffusion timesteps disproportionately determine output diversity, while later steps primarily refine details. Based on these insights, we introduce diversity distillation - a hybrid inference approach that strategically employs the base model for only the first critical timestep before transitioning to the efficient distilled model. Our experiments demonstrate that this simple modification not only restores the diversity capabilities from base to distilled models but surprisingly exceeds it, while maintaining nearly the computational efficiency of distilled inference, all without requiring additional training or model modifications. Our code and data are available at https://distillation.baulab.info