Authors: Hongbin Liang, Hezhe Qiao, Wei Huang, Qizhou Wang, Mingsheng Shang, Lin Chen
Abstract: Ensuring the safety of vulnerable road users through accurate prediction of pedestrian crossing intention (PCI) plays a crucial role in the context of autonomous and assisted driving. Analyzing the set of observation video frames in ego-view has been widely used in most PCI prediction methods to forecast the cross intent. However, they struggle to capture the critical events related to pedestrian behaviour along the temporal dimension due to the high redundancy of the video frames, which results in the sub-optimal performance of PCI prediction. Our research addresses the challenge by introducing a novel approach called \underline{T}emporal-\underline{c}ontextual Event \underline{L}earning (TCL). The TCL is composed of the Temporal Merging Module (TMM), which aims to manage the redundancy by clustering the observed video frames into multiple key temporal events. Then, the Contextual Attention Block (CAB) is employed to adaptively aggregate multiple event features along with visual and non-visual data. By synthesizing the temporal feature extraction and contextual attention on the key information across the critical events, TCL can learn expressive representation for the PCI prediction. Extensive experiments are carried out on three widely adopted datasets, including PIE, JAAD-beh, and JAAD-all. The results show that TCL substantially surpasses the state-of-the-art methods. Our code can be accessed at https://github.com/dadaguailhb/TCL.
Authors: Ben Crulis, Cyril De Runz, Barthelemy Serres, Gilles Venturini
Abstract: We propose a process to compress a pre-trained Vision Language Model into a ternary version of itself instead of training a ternary model from scratch. A new initialization scheme from pre-trained weights based on the k-means algorithm is proposed to reduce the ternarization time. We implement different custom operators for executing the ternary model on the TensorFlow Lite Engine. We compare the original model with its ternary and binary versions in terms of memory consumption, inference speed and perplexity. We find that the ternary model using our custom ternary matrix multiplication operator provides a good compromise in term of memory usage and perplexity, while having the fastest token generation speed.
Authors: Abdulkareem Alsudais
Abstract: The primary objective of this paper is to investigate how a popular Text-to-Image (T2I) model represents people from 208 different nationalities when prompted to generate images of individuals performing typical everyday tasks. Two scenarios were developed, and images were generated based on input prompts that specified nationalities. The results show that in one scenario, the majority of images, and in the other, a substantial portion, depict individuals wearing traditional attire. This suggests that the model emphasizes such characteristics even when they are impractical for the given task. A statistically significant relationship was observed between this representation pattern and the regions associated with the specified countries. This indicates that the issue disproportionately affects certain areas, particularly the Middle East & North Africa and Sub-Saharan Africa. A notable association with income groups was also found. CLIP was used to measure alignment scores between generated images and various prompts and captions. The findings indicate statistically significant higher scores for images featuring individuals in traditional attire in one scenario. The study also examined revised prompts (additional contextual information automatically added to the original input prompts) to assess their potential influence on how individuals are represented in the generated images, finding that the word "traditional" was commonly added to revised prompts. These findings provide valuable insights into how T2I models represent individuals from various countries and highlight potential areas for improvement in future models.
Authors: Hicham Talaoubrid, Anissa Mokraoui, Ismail Ben Ayed, Axel Prouvost, Sonimith Hang, Monit Korn, R\'emi Harvey
Abstract: This paper investigates the application of Low-Rank Adaptation (LoRA) to small models for cross-domain few-shot object detection in aerial images. Originally designed for large-scale models, LoRA helps mitigate overfitting, making it a promising approach for resource-constrained settings. We integrate LoRA into DiffusionDet, and evaluate its performance on the DOTA and DIOR datasets. Our results show that LoRA applied after an initial fine-tuning slightly improves performance in low-shot settings (e.g., 1-shot and 5-shot), while full fine-tuning remains more effective in higher-shot configurations. These findings highlight LoRA's potential for efficient adaptation in aerial object detection, encouraging further research into parameter-efficient fine-tuning strategies for few-shot learning. Our code is available here: https://github.com/HichTala/LoRA-DiffusionDet.
Authors: Vladimir Golovkin, Nikolay Nemtsev, Vasyl Shandyba, Oleg Udin, Nikita Kasatkin, Pavel Kononov, Anton Afanasiev, Sergey Ulasen, Andrei Boiarov
Abstract: Game State Reconstruction (GSR), a critical task in Sports Video Understanding, involves precise tracking and localization of all individuals on the football field-players, goalkeepers, referees, and others - in real-world coordinates. This capability enables coaches and analysts to derive actionable insights into player movements, team formations, and game dynamics, ultimately optimizing training strategies and enhancing competitive advantage. Achieving accurate GSR using a single-camera setup is highly challenging due to frequent camera movements, occlusions, and dynamic scene content. In this work, we present a robust end-to-end pipeline for tracking players across an entire match using a single-camera setup. Our solution integrates a fine-tuned YOLOv5m for object detection, a SegFormer-based camera parameter estimator, and a DeepSORT-based tracking framework enhanced with re-identification, orientation prediction, and jersey number recognition. By ensuring both spatial accuracy and temporal consistency, our method delivers state-of-the-art game state reconstruction, securing first place in the SoccerNet Game State Reconstruction Challenge 2024 and significantly outperforming competing methods.
Authors: Yupeng Cheng, Zi Pong Lim, Sarthak Ketanbhai Modi, Yon Shin Teo, Yushi Cao, Shang-Wei Lin
Abstract: Test automation has become increasingly important as the complexity of both design and content in Human Machine Interface (HMI) software continues to grow. Current standard practice uses Optical Character Recognition (OCR) techniques to automatically extract textual information from HMI screens for validation. At present, one of the key challenges faced during the automation of HMI screen validation is the noise handling for the OCR models. In this paper, we propose to utilize adversarial training techniques to enhance OCR models in HMI testing scenarios. More specifically, we design a new adversarial attack objective for OCR models to discover the decision boundaries in the context of HMI testing. We then adopt adversarial training to optimize the decision boundaries towards a more robust and accurate OCR model. In addition, we also built an HMI screen dataset based on real-world requirements and applied multiple types of perturbation onto the clean HMI dataset to provide a more complete coverage for the potential scenarios. We conduct experiments to demonstrate how using adversarial training techniques yields more robust OCR models against various kinds of noises, while still maintaining high OCR model accuracy. Further experiments even demonstrate that the adversarial training models exhibit a certain degree of robustness against perturbations from other patterns.
Authors: Hritam Basak, Zhaozheng Yin
Abstract: Domain Adaptation (DA) and Semi-supervised Learning (SSL) converge in Semi-supervised Domain Adaptation (SSDA), where the objective is to transfer knowledge from a source domain to a target domain using a combination of limited labeled target samples and abundant unlabeled target data. Although intuitive, a simple amalgamation of DA and SSL is suboptimal in semantic segmentation due to two major reasons: (1) previous methods, while able to learn good segmentation boundaries, are prone to confuse classes with similar visual appearance due to limited supervision; and (2) skewed and imbalanced training data distribution preferring source representation learning whereas impeding from exploring limited information about tailed classes. Language guidance can serve as a pivotal semantic bridge, facilitating robust class discrimination and mitigating visual ambiguities by leveraging the rich semantic relationships encoded in pre-trained language models to enhance feature representations across domains. Therefore, we propose the first language-guided SSDA setting for semantic segmentation in this work. Specifically, we harness the semantic generalization capabilities inherent in vision-language models (VLMs) to establish a synergistic framework within the SSDA paradigm. To address the inherent class-imbalance challenges in long-tailed distributions, we introduce class-balanced segmentation loss formulations that effectively regularize the learning process. Through extensive experimentation across diverse domain adaptation scenarios, our approach demonstrates substantial performance improvements over contemporary state-of-the-art (SoTA) methodologies. Code is available: \href{https://github.com/hritam-98/SemiDAViL}{GitHub}.
Authors: Yufu Wang, Yu Sun, Priyanka Patel, Kostas Daniilidis, Michael J. Black, Muhammed Kocabas
Abstract: Human pose and shape (HPS) estimation presents challenges in diverse scenarios such as crowded scenes, person-person interactions, and single-view reconstruction. Existing approaches lack mechanisms to incorporate auxiliary "side information" that could enhance reconstruction accuracy in such challenging scenarios. Furthermore, the most accurate methods rely on cropped person detections and cannot exploit scene context while methods that process the whole image often fail to detect people and are less accurate than methods that use crops. While recent language-based methods explore HPS reasoning through large language or vision-language models, their metric accuracy is well below the state of the art. In contrast, we present PromptHMR, a transformer-based promptable method that reformulates HPS estimation through spatial and semantic prompts. Our method processes full images to maintain scene context and accepts multiple input modalities: spatial prompts like bounding boxes and masks, and semantic prompts like language descriptions or interaction labels. PromptHMR demonstrates robust performance across challenging scenarios: estimating people from bounding boxes as small as faces in crowded scenes, improving body shape estimation through language descriptions, modeling person-person interactions, and producing temporally coherent motions in videos. Experiments on benchmarks show that PromptHMR achieves state-of-the-art performance while offering flexible prompt-based control over the HPS estimation process.
Authors: Rupayan Mallick, Sibo Dong, Nataniel Ruiz, Sarah Adel Bargal
Abstract: Applications of diffusion models for visual tasks have been quite noteworthy. This paper targets making classification models more robust to occlusions for the task of object recognition by proposing a pipeline that utilizes a frozen diffusion model. Diffusion features have demonstrated success in image generation and image completion while understanding image context. Occlusion can be posed as an image completion problem by deeming the pixels of the occluder to be `missing.' We hypothesize that such features can help hallucinate object visual features behind occluding objects, and hence we propose using them to enable models to become more occlusion robust. We design experiments to include input-based augmentations as well as feature-based augmentations. Input-based augmentations involve finetuning on images where the occluder pixels are inpainted, and feature-based augmentations involve augmenting classification features with intermediate diffusion features. We demonstrate that our proposed use of diffusion-based features results in models that are more robust to partial object occlusions for both Transformers and ConvNets on ImageNet with simulated occlusions. We also propose a dataset that encompasses real-world occlusions and demonstrate that our method is more robust to partial object occlusions.
Authors: Jos\'e A. Pilartes-Congo, Matthew Kastl, Michael J. Starek, Marina Vicens-Miquel, Philippe Tissot
Abstract: The increasing population, thus financial interests, in coastal areas have increased the need to monitor coastal elevation and shoreline change. Though several resources exist to obtain this information, they often lack the required temporal resolution for short-term monitoring (e.g., every hour). To address this issue, this study implements a low-cost ZED 2i stereo camera system and close-range photogrammetry to collect images for generating 3D point clouds, digital surface models (DSMs) of beach elevation, and georectified imagery at a localized scale and high temporal resolution. The main contributions of this study are (i) intrinsic camera calibration, (ii) georectification and registration of acquired imagery and point cloud, (iii) generation of the DSM of the beach elevation, and (iv) a comparison of derived products against those from uncrewed aircraft system structure-from-motion photogrammetry. Preliminary results show that despite its limitations, the ZED 2i can provide the desired mapping products at localized and high temporal scales. The system achieved a mean reprojection error of 0.20 px, a point cloud registration of 27 cm, a vertical error of 37.56 cm relative to ground truth, and georectification root mean square errors of 2.67 cm and 2.81 cm for x and y.
Authors: Samuel Stevens, S M Rayeed, Jenna Kline
Abstract: The practical application of AI tools for specific computer vision tasks relies on the "small-data regime" of hundreds to thousands of labeled samples. This small-data regime is vital for applications requiring expensive expert annotations, such as ecological monitoring, medical diagnostics or industrial quality control. We find, however, that computer vision research has ignored the small data regime as evaluations increasingly focus on zero- and few-shot learning. We use the Natural World Tasks (NeWT) benchmark to compare multi-modal large language models (MLLMs) and vision-only methods across varying training set sizes. MLLMs exhibit early performance plateaus, while vision-only methods improve throughout the small-data regime, with performance gaps widening beyond 10 training examples. We provide the first comprehensive comparison between these approaches in small-data contexts and advocate for explicit small-data evaluations in AI research to better bridge theoretical advances with practical deployments.
Authors: Xiaohang Yang, Qing Wang, Jiahao Yang, Gregory Slabaugh, Shanxin Yuan
Abstract: Motion retargeting seeks to faithfully replicate the spatio-temporal motion characteristics of a source character onto a target character with a different body shape. Apart from motion semantics preservation, ensuring geometric plausibility and maintaining temporal consistency are also crucial for effective motion retargeting. However, many existing methods prioritize either geometric plausibility or temporal consistency. Neglecting geometric plausibility results in interpenetration while neglecting temporal consistency leads to motion jitter. In this paper, we propose a novel sequence-to-sequence model for seamless Spatial-Temporal aware motion Retargeting (STaR), with penetration and consistency constraints. STaR consists of two modules: (1) a spatial module that incorporates dense shape representation and a novel limb penetration constraint to ensure geometric plausibility while preserving motion semantics, and (2) a temporal module that utilizes a temporal transformer and a novel temporal consistency constraint to predict the entire motion sequence at once while enforcing multi-level trajectory smoothness. The seamless combination of the two modules helps us achieve a good balance between the semantic, geometric, and temporal targets. Extensive experiments on the Mixamo and ScanRet datasets demonstrate that our method produces plausible and coherent motions while significantly reducing interpenetration rates compared with other approaches.
Authors: Songze Li, Tonghua Su, Xu-Yao Zhang, Qixing Xu, Zhongjie Wang
Abstract: Pre-trained model-based continual learning (PTMCL) has garnered growing attention, as it enables more rapid acquisition of new knowledge by leveraging the extensive foundational understanding inherent in pre-trained model (PTM). Most existing PTMCL methods use Parameter-Efficient Fine-Tuning (PEFT) to learn new knowledge while consolidating existing memory. However, they often face some challenges. A major challenge lies in the misalignment of classification heads, as the classification head of each task is trained within a distinct feature space, leading to inconsistent decision boundaries across tasks and, consequently, increased forgetting. Another critical limitation stems from the restricted feature-level knowledge accumulation, with feature learning typically restricted to the initial task only, which constrains the model's representation capabilities. To address these issues, we propose a method named DUal-level Knowledge Accumulation and Ensemble (DUKAE) that leverages both feature-level and decision-level knowledge accumulation by aligning classification heads into a unified feature space through Gaussian distribution sampling and introducing an adaptive expertise ensemble to fuse knowledge across feature subspaces.Extensive experiments on CIFAR-100, ImageNet-R, CUB-200, and Cars-196 datasets demonstrate the superior performance of our approach.
Authors: Xinyu Liu, Xiaoguang Lin, Xiang Liu, Yong Yang, Hongqian Wang, Qilong Sun
Abstract: Recording the open surgery process is essential for educational and medical evaluation purposes; however, traditional single-camera methods often face challenges such as occlusions caused by the surgeon's head and body, as well as limitations due to fixed camera angles, which reduce comprehensibility of the video content. This study addresses these limitations by employing a multi-viewpoint camera recording system, capturing the surgical procedure from six different angles to mitigate occlusions. We propose a fully supervised learning-based time series prediction method to choose the best shot sequences from multiple simultaneously recorded video streams, ensuring optimal viewpoints at each moment. Our time series prediction model forecasts future camera selections by extracting and fusing visual and semantic features from surgical videos using pre-trained models. These features are processed by a temporal prediction network with TimeBlocks to capture sequential dependencies. A linear embedding layer reduces dimensionality, and a Softmax classifier selects the optimal camera view based on the highest probability. In our experiments, we created five groups of open thyroidectomy videos, each with simultaneous recordings from six different angles. The results demonstrate that our method achieves competitive accuracy compared to traditional supervised methods, even when predicting over longer time horizons. Furthermore, our approach outperforms state-of-the-art time series prediction techniques on our dataset. This manuscript makes a unique contribution by presenting an innovative framework that advances surgical video analysis techniques, with significant implications for improving surgical education and patient safety.
Authors: Weiwei Xing, Yue Cheng, Hongzhu Yi, Xiaohui Gao, Xiang Wei, Xiaoyu Guo, Yuming Zhang, Xinyu Pang
Abstract: Classifiers often learn to be biased corresponding to the class-imbalanced dataset, especially under the semi-supervised learning (SSL) set. While previous work tries to appropriately re-balance the classifiers by subtracting a class-irrelevant image's logit, but lacks a firm theoretical basis. We theoretically analyze why exploiting a baseline image can refine pseudo-labels and prove that the black image is the best choice. We also indicated that as the training process deepens, the pseudo-labels before and after refinement become closer. Based on this observation, we propose a debiasing scheme dubbed LCGC, which Learning from Consistency Gradient Conflicting, by encouraging biased class predictions during training. We intentionally update the pseudo-labels whose gradient conflicts with the debiased logits, representing the optimization direction offered by the over-imbalanced classifier predictions. Then, we debiased the predictions by subtracting the baseline image logits during testing. Extensive experiments demonstrate that LCGC can significantly improve the prediction accuracy of existing CISSL models on public benchmarks.
Authors: Shaocong Long, Qianyu Zhou, Xikun Jiang, Chenhao Ying, Lizhuang Ma, Yuan Luo
Abstract: Domain generalization (DG) strives to address distribution shifts across diverse environments to enhance model's generalizability. Current DG approaches are confined to acquiring robust representations with continuous features, specifically training at the pixel level. However, this DG paradigm may struggle to mitigate distribution gaps in dealing with a large space of continuous features, rendering it susceptible to pixel details that exhibit spurious correlations or noise. In this paper, we first theoretically demonstrate that the domain gaps in continuous representation learning can be reduced by the discretization process. Based on this inspiring finding, we introduce a novel learning paradigm for DG, termed Discrete Domain Generalization (DDG). DDG proposes to use a codebook to quantize the feature map into discrete codewords, aligning semantic-equivalent information in a shared discrete representation space that prioritizes semantic-level information over pixel-level intricacies. By learning at the semantic level, DDG diminishes the number of latent features, optimizing the utilization of the representation space and alleviating the risks associated with the wide-ranging space of continuous features. Extensive experiments across widely employed benchmarks in DG demonstrate DDG's superior performance compared to state-of-the-art approaches, underscoring its potential to reduce the distribution gaps and enhance the model's generalizability.
Authors: Rahul Singh Maharjan, Marta Romeo, Angelo Cangelosi
Abstract: Visual emotion analysis or recognition has gained considerable attention due to the growing interest in understanding how images can convey rich semantics and evoke emotions in human perception. However, visual emotion analysis poses distinctive challenges compared to traditional vision tasks, especially due to the intricate relationship between general visual features and the different affective states they evoke, known as the affective gap. Researchers have used deep representation learning methods to address this challenge of extracting generalized features from entire images. However, most existing methods overlook the importance of specific emotional attributes such as brightness, colorfulness, scene understanding, and facial expressions. Through this paper, we introduce A4Net, a deep representation network to bridge the affective gap by leveraging four key attributes: brightness (Attribute 1), colorfulness (Attribute 2), scene context (Attribute 3), and facial expressions (Attribute 4). By fusing and jointly training all aspects of attribute recognition and visual emotion analysis, A4Net aims to provide a better insight into emotional content in images. Experimental results show the effectiveness of A4Net, showcasing competitive performance compared to state-of-the-art methods across diverse visual emotion datasets. Furthermore, visualizations of activation maps generated by A4Net offer insights into its ability to generalize across different visual emotion datasets.
Authors: Joochan Kim, Minjoon Jung, Byoung-Tak Zhang
Abstract: Action recognition models have achieved promising results in understanding instructional videos. However, they often rely on dominant, dataset-specific action sequences rather than true video comprehension, a problem that we define as ordinal bias. To address this issue, we propose two effective video manipulation methods: Action Masking, which masks frames of frequently co-occurring actions, and Sequence Shuffling, which randomizes the order of action segments. Through comprehensive experiments, we demonstrate that current models exhibit significant performance drops when confronted with nonstandard action sequences, underscoring their vulnerability to ordinal bias. Our findings emphasize the importance of rethinking evaluation strategies and developing models capable of generalizing beyond fixed action patterns in diverse instructional videos.
Authors: Minghe Gao, Xuqi Liu, Zhongqi Yue, Yang Wu, Shuang Chen, Juncheng Li, Siliang Tang, Fei Wu, Tat-Seng Chua, Yueting Zhuang
Abstract: Recent advancements in reward signal usage for Large Language Models (LLMs) are remarkable. However, significant challenges exist when transitioning reward signal to the multimodal domain, including labor-intensive annotations, over-reliance on one-step rewards, and inadequate evaluation. To address these issues, we propose SVIP, a novel approach to train a step-level multi-dimensional Chain-of-Thought~(CoT) reward model automatically. It generates code for solving visual tasks and transforms the analysis of code blocks into the evaluation of CoT step as training samples. Then, we train SVIP-Reward model using a multi-head attention mechanism called TriAtt-CoT. The advantages of SVIP-Reward are evident throughout the entire process of MLLM. We also introduce a benchmark for CoT reward model training and testing. Experimental results demonstrate that SVIP-Reward improves MLLM performance across training and inference-time scaling, yielding better results on benchmarks while reducing hallucinations and enhancing reasoning ability.
Authors: Onkar Krishna, Hiroki Ohashi
Abstract: Domain gaps between training data (source) and real-world environments (target) often degrade the performance of object detection models. Most existing methods aim to bridge this gap by aligning features across source and target domains but often fail to account for visual differences, such as color or orientation, in alignment pairs. This limitation leads to less effective domain adaptation, as the model struggles to manage both domain-specific shifts (e.g., fog) and visual variations simultaneously. In this work, we demonstrate for the first time, using a custom-built dataset, that aligning visually similar pairs significantly improves domain adaptation. Based on this insight, we propose a novel memory-based system to enhance domain alignment. This system stores precomputed features of foreground objects and background areas from the source domain, which are periodically updated during training. By retrieving visually similar source features for alignment with target foreground and background features, the model effectively addresses domain-specific differences while reducing the impact of visual variations. Extensive experiments across diverse domain shift scenarios validate our method's effectiveness, achieving 53.1 mAP on Foggy Cityscapes and 62.3 on Sim10k, surpassing prior state-of-the-art methods by 1.2 and 4.1 mAP, respectively.
Authors: Jiajun Chen, Hongpeng Yin, Yifu Yang
Abstract: In task-based few-shot learning paradigms, it is commonly assumed that different tasks are independently and identically distributed (i.i.d.). However, in real-world scenarios, the distribution encountered in few-shot learning can significantly differ from the distribution of existing data. Thus, how to effectively leverage existing data knowledge to enable models to quickly adapt to class variations under non-i.i.d. assumptions has emerged as a key research challenge. To address this challenge, this paper proposes a new cross-domain few-shot learning approach based on domain knowledge mapping, applied consistently throughout the pre-training, training, and testing phases. In the pre-training phase, our method integrates self-supervised and supervised losses by maximizing mutual information, thereby mitigating mode collapse. During the training phase, the domain knowledge mapping layer collaborates with a domain classifier to learn both domain mapping capabilities and the ability to assess domain adaptation difficulty. Finally, this approach is applied during the testing phase, rapidly adapting to domain variations through meta-training tasks on support sets, consequently enhancing the model's capability to transfer domain knowledge effectively. Experimental validation conducted across six datasets from diverse domains demonstrates the effectiveness of the proposed method.
Authors: Zijun Lin, M Ganesh Kumar, Cheston Tan
Abstract: The compositional structure of language enables humans to decompose complex phrases and map them to novel visual concepts, showcasing flexible intelligence. While several algorithms exhibit compositionality, they fail to elucidate how humans learn to compose concept classes and ground visual cues through trial and error. To investigate this multi-modal learning challenge, we designed a 3D synthetic environment in which an agent learns, via reinforcement, to navigate to a target specified by a natural language instruction. These instructions comprise nouns, attributes, and critically, determiners, prepositions, or both. The vast array of word combinations heightens the compositional complexity of the visual grounding task, as navigating to a blue cube above red spheres is not rewarded when the instruction specifies navigating to "some blue cubes below the red sphere". We first demonstrate that reinforcement learning agents can ground determiner concepts to visual targets but struggle with more complex prepositional concepts. Second, we show that curriculum learning, a strategy humans employ, enhances concept learning efficiency, reducing the required training episodes by 15% in determiner environments and enabling agents to easily learn prepositional concepts. Finally, we establish that agents trained on determiner or prepositional concepts can decompose held-out test instructions and rapidly adapt their navigation policies to unseen visual object combinations. Leveraging synthetic environments, our findings demonstrate that multi-modal reinforcement learning agents can achieve compositional understanding of complex concept classes and highlight the efficacy of human-like learning strategies in improving artificial systems' learning efficiency.
Authors: Yi Zhang, Xiaoyang Huang, Yishun Dou, Yue Shi, Rui Shi, Ye Chen, Bingbing Ni, Wenjun Zhang
Abstract: We present InstantSticker, a disentangled reconstruction pipeline based on Image-Based Lighting (IBL), which focuses on highly realistic decal blending, simulates stickers attached to the reconstructed surface, and allows for instant editing and real-time rendering. To achieve stereoscopic impression of the decal, we introduce shadow factor into IBL, which can be adaptively optimized during training. This allows the shadow brightness of surfaces to be accurately decomposed rather than baked into the diffuse color, ensuring that the edited texture exhibits authentic shading. To address the issues of warping and blurriness in previous methods, we apply As-Rigid-As-Possible (ARAP) parameterization to pre-unfold a specified area of the mesh and use the local UV mapping combined with a neural texture map to enhance the ability to express high-frequency details in that area. For instant editing, we utilize the Disney BRDF model, explicitly defining material colors with 3-channel diffuse albedo. This enables instant replacement of albedo RGB values during the editing process, avoiding the prolonged optimization required in previous approaches. In our experiment, we introduce the Ratio Variance Warping (RVW) metric to evaluate the local geometric warping of the decal area. Extensive experimental results demonstrate that our method surpasses previous decal blending methods in terms of editing quality, editing speed and rendering speed, achieving the state-of-the-art.
Authors: Ludvig Dill\'en, Per-Erik Forss\'en, Johan Edstedt
Abstract: We present FACT, a method for predicting alignment quality (i.e., registration error) of registered lidar point cloud pairs. This is useful e.g. for quality assurance of large, automatically registered 3D models. FACT extracts local features from a registered pair and processes them with a point transformer-based network to predict a misalignment class. We generalize prior work that study binary alignment classification of registration errors, by recasting it as multinomial misalignment classification. To achieve this, we introduce a custom regression-by-classification loss function that combines the cross-entropy and Wasserstein losses, and demonstrate that it outperforms both direct regression and prior binary classification. FACT successfully classifies point-cloud pairs registered with both the classical ICP and GeoTransformer, while other choices, such as standard point-cloud-quality metrics and registration residuals are shown to be poor choices for predicting misalignment. On a synthetically perturbed point-cloud task introduced by the CorAl method, we show that FACT achieves substantially better performance than CorAl. Finally, we demonstrate how FACT can assist experts in correcting misaligned point-cloud maps. Our code is available at https://github.com/LudvigDillen/FACT_for_PCMC.
Authors: MinKyu Lee, Sangeek Hyun, Woojin Jun, Hyunjun Kim, Jiwoo Chung, Jae-Pil Heo
Abstract: This work investigates abnormal feature behaviors observed in image restoration (IR) Transformers. Specifically, we identify two critical issues: feature entropy becoming excessively small and feature magnitudes diverging up to a million-fold scale. We pinpoint the root cause to the per-token normalization aspect of conventional LayerNorm, which disrupts essential spatial correlations and internal feature statistics. To address this, we propose a simple normalization strategy tailored for IR Transformers. Our approach applies normalization across the entire spatio-channel dimension, effectively preserving spatial correlations. Additionally, we introduce an input-adaptive rescaling method that aligns feature statistics to the unique statistical requirements of each input. Experimental results verify that this combined strategy effectively resolves feature divergence, significantly enhancing both the stability and performance of IR Transformers across various IR tasks.
Authors: Yifan Gao, Zihang Lin, Chuanbin Liu, Min Zhou, Tiezheng Ge, Bo Zheng, Hongtao Xie
Abstract: Product posters, which integrate subject, scene, and text, are crucial promotional tools for attracting customers. Creating such posters using modern image generation methods is valuable, while the main challenge lies in accurately rendering text, especially for complex writing systems like Chinese, which contains over 10,000 individual characters. In this work, we identify the key to precise text rendering as constructing a character-discriminative visual feature as a control signal. Based on this insight, we propose a robust character-wise representation as control and we develop TextRenderNet, which achieves a high text rendering accuracy of over 90%. Another challenge in poster generation is maintaining the fidelity of user-specific products. We address this by introducing SceneGenNet, an inpainting-based model, and propose subject fidelity feedback learning to further enhance fidelity. Based on TextRenderNet and SceneGenNet, we present PosterMaker, an end-to-end generation framework. To optimize PosterMaker efficiently, we implement a two-stage training strategy that decouples text rendering and background generation learning. Experimental results show that PosterMaker outperforms existing baselines by a remarkable margin, which demonstrates its effectiveness.
Authors: Junyoung Kim, Youngrok Kim, Siyeol Jung, Donghyun Min
Abstract: Single Image Super-Resolution (SISR) reconstructs high-resolution images from low-resolution inputs, enhancing image details. While Vision Transformer (ViT)-based models improve SISR by capturing long-range dependencies, they suffer from quadratic computational costs or employ selective attention mechanisms that do not explicitly focus on query-relevant regions. Despite these advancements, prior work has overlooked how selective attention mechanisms should be effectively designed for SISR. We propose SSCAN, which dynamically selects the most relevant key-value windows based on query similarity, ensuring focused feature extraction while maintaining efficiency. In contrast to prior approaches that apply attention globally or heuristically, our method introduces a query-aware window selection strategy that better aligns attention computation with important image regions. By incorporating fixed-sized windows, SSCAN reduces memory usage and enforces linear token-to-token complexity, making it scalable for large images. Our experiments demonstrate that SSCAN outperforms existing attention-based SISR methods, achieving up to 0.14 dB PSNR improvement on urban datasets, guaranteeing both computational efficiency and reconstruction quality in SISR.
Authors: Hu Cui, Tessai Hayama
Abstract: 3D human pose lifting is a promising research area that leverages estimated and ground-truth 2D human pose data for training. While existing approaches primarily aim to enhance the performance of estimated 2D poses, they often struggle when applied to ground-truth 2D pose data. We observe that achieving accurate 3D pose reconstruction from ground-truth 2D poses requires precise modeling of local pose structures, alongside the ability to extract robust global spatio-temporal features. To address these challenges, we propose a novel Hyper-GCN and Shuffle Mamba (HGMamba) block, which processes input data through two parallel streams: Hyper-GCN and Shuffle-Mamba. The Hyper-GCN stream models the human body structure as hypergraphs with varying levels of granularity to effectively capture local joint dependencies. Meanwhile, the Shuffle Mamba stream leverages a state space model to perform spatio-temporal scanning across all joints, enabling the establishment of global dependencies. By adaptively fusing these two representations, HGMamba achieves strong global feature modeling while excelling at local structure modeling. We stack multiple HGMamba blocks to create three variants of our model, allowing users to select the most suitable configuration based on the desired speed-accuracy trade-off. Extensive evaluations on the Human3.6M and MPI-INF-3DHP benchmark datasets demonstrate the effectiveness of our approach. HGMamba-B achieves state-of-the-art results, with P1 errors of 38.65 mm and 14.33 mm on the respective datasets. Code and models are available: https://github.com/HuCui2022/HGMamba
Authors: Nan Peng, Xun Zhou, Mingming Wang, Guisong Chen, Songming Chen
Abstract: Safety constitutes a foundational imperative for autonomous driving systems, necessitating the maximal incorporation of accessible external prior information. This study establishes that temporal perception buffers and cost-efficient maps inherently form complementary prior sources for online vectorized high-definition (HD) map construction. We present Uni-PrevPredMap, a unified prior-informed framework that systematically integrates two synergistic information sources: previous predictions and simulated outdated HD maps. The framework introduces two core innovations: a tile-indexed 3D vectorized global map processor enabling efficient refreshment, storage, and retrieval of 3D vectorized priors; a tri-mode operational optimization paradigm ensuring consistency across prior-free, map-absent, and map-prior scenarios while mitigating reliance on idealized map fidelity assumptions. Uni-PrevPredMap achieves state-of-the-art performance in map-free scenarios across established online vectorized HD map construction benchmarks. When provided with simulated outdated HD maps, the framework exhibits robust capabilities in error-resilient prior fusion, empirically confirming the synergistic complementarity between previous predictions and simulated outdated HD maps. Code will be available at https://github.com/pnnnnnnn/Uni-PrevPredMap.
Authors: Ruotian Peng, Haiying He, Yake Wei, Yandong Wen, Di Hu
Abstract: High-quality image captions play a crucial role in improving the performance of cross-modal applications such as text-to-image generation, text-to-video generation, and text-image retrieval. To generate long-form, high-quality captions, many recent studies have employed multimodal large language models (MLLMs). However, current MLLMs often produce captions that lack fine-grained details or suffer from hallucinations, a challenge that persists in both open-source and closed-source models. Inspired by Feature-Integration theory, which suggests that attention must focus on specific regions to integrate visual information effectively, we propose a \textbf{divide-then-aggregate} strategy. Our method first divides the image into semantic and spatial patches to extract fine-grained details, enhancing the model's local perception of the image. These local details are then hierarchically aggregated to generate a comprehensive global description. To address hallucinations and inconsistencies in the generated captions, we apply a semantic-level filtering process during hierarchical aggregation. This training-free pipeline can be applied to both open-source models (LLaVA-1.5, LLaVA-1.6, Mini-Gemini) and closed-source models (Claude-3.5-Sonnet, GPT-4o, GLM-4V-Plus). Extensive experiments demonstrate that our method generates more detailed, reliable captions, advancing multimodal description generation without requiring model retraining. The source code are available at https://github.com/GeWu-Lab/Patch-Matters
Authors: Elia Peruzzo, Dejia Xu, Xingqian Xu, Humphrey Shi, Nicu Sebe
Abstract: Video generation is experiencing rapid growth, driven by advances in diffusion models and the development of better and larger datasets. However, producing high-quality videos remains challenging due to the high-dimensional data and the complexity of the task. Recent efforts have primarily focused on enhancing visual quality and addressing temporal inconsistencies, such as flickering. Despite progress in these areas, the generated videos often fall short in terms of motion complexity and physical plausibility, with many outputs either appearing static or exhibiting unrealistic motion. In this work, we propose a framework to improve the realism of motion in generated videos, exploring a complementary direction to much of the existing literature. Specifically, we advocate for the incorporation of a retrieval mechanism during the generation phase. The retrieved videos act as grounding signals, providing the model with demonstrations of how the objects move. Our pipeline is designed to apply to any text-to-video diffusion model, conditioning a pretrained model on the retrieved samples with minimal fine-tuning. We demonstrate the superiority of our approach through established metrics, recently proposed benchmarks, and qualitative results, and we highlight additional applications of the framework.
Authors: Qingtao Yu, Jaskirat Singh, Zhaoyuan Yang, Peter Henry Tu, Jing Zhang, Hongdong Li, Richard Hartley, Dylan Campbell
Abstract: Diffusion models indirectly estimate the probability density over a data space, which can be used to study its structure. In this work, we show that geodesics can be computed in diffusion latent space, where the norm induced by the spatially-varying inner product is inversely proportional to the probability density. In this formulation, a path that traverses a high density (that is, probable) region of image latent space is shorter than the equivalent path through a low density region. We present algorithms for solving the associated initial and boundary value problems and show how to compute the probability density along the path and the geodesic distance between two points. Using these techniques, we analyze how closely video clips approximate geodesics in a pre-trained image diffusion space. Finally, we demonstrate how these techniques can be applied to training-free image sequence interpolation and extrapolation, given a pre-trained image diffusion model.
Authors: Christoph Balada, Aida Romano-Martinez, Vincent ten Cate, Katharina Geschke, Jonas Tesarz, Paul Cla{\ss}en, Alexander K. Schuster, Dativa Tibyampansha, Karl-Patrik Kresoja, Philipp S. Wild, Sheraz Ahmed, Andreas Dengel
Abstract: In this study, hypertension is utilized as an indicator of individual vascular damage. This damage can be identified through machine learning techniques, providing an early risk marker for potential major cardiovascular events and offering valuable insights into the overall arterial condition of individual patients. To this end, the VideoMAE deep learning model, originally developed for video classification, was adapted by finetuning for application in the domain of ultrasound imaging. The model was trained and tested using a dataset comprising over 31,000 carotid sonography videos sourced from the Gutenberg Health Study (15,010 participants), one of the largest prospective population health studies. This adaptation facilitates the classification of individuals as hypertensive or non-hypertensive (75.7% validation accuracy), functioning as a proxy for detecting visual arterial damage. We demonstrate that our machine learning model effectively captures visual features that provide valuable insights into an individual's overall cardiovascular health.
Authors: Anil Armagan, Albert Sa\`a-Garriga, Bruno Manganelli, Kyuwon Kim, M. Kerim Yucel
Abstract: Gaussian Splatting (GS) is a popular approach for 3D reconstruction, mostly due to its ability to converge reasonably fast, faithfully represent the scene and render (novel) views in a fast fashion. However, it suffers from large storage and memory requirements, and its training speed still lags behind the hash-grid based radiance field approaches (e.g. Instant-NGP), which makes it especially difficult to deploy them in robotics scenarios, where 3D reconstruction is crucial for accurate operation. In this paper, we propose GSta that dynamically identifies Gaussians that have converged well during training, based on their positional and color gradient norms. By forcing such Gaussians into a siesta and stopping their updates (freezing) during training, we improve training speed with competitive accuracy compared to state of the art. We also propose an early stopping mechanism based on the PSNR values computed on a subset of training images. Combined with other improvements, such as integrating a learning rate scheduler, GSta achieves an improved Pareto front in convergence speed, memory and storage requirements, while preserving quality. We also show that GSta can improve other methods and complement orthogonal approaches in efficiency improvement; once combined with Trick-GS, GSta achieves up to 5x faster training, 16x smaller disk size compared to vanilla GS, while having comparable accuracy and consuming only half the peak memory. More visualisations are available at https://anilarmagan.github.io/SRUK-GSta.
Authors: Pedro Hermosilla, Christian Stippel, Leon Sick
Abstract: Self-supervised learning has transformed 2D computer vision by enabling models trained on large, unannotated datasets to provide versatile off-the-shelf features that perform similarly to models trained with labels. However, in 3D scene understanding, self-supervised methods are typically only used as a weight initialization step for task-specific fine-tuning, limiting their utility for general-purpose feature extraction. This paper addresses this shortcoming by proposing a robust evaluation protocol specifically designed to assess the quality of self-supervised features for 3D scene understanding. Our protocol uses multi-resolution feature sampling of hierarchical models to create rich point-level representations that capture the semantic capabilities of the model and, hence, are suitable for evaluation with linear probing and nearest-neighbor methods. Furthermore, we introduce the first self-supervised model that performs similarly to supervised models when only off-the-shelf features are used in a linear probing setup. In particular, our model is trained natively in 3D with a novel self-supervised approach based on a Masked Scene Modeling objective, which reconstructs deep features of masked patches in a bottom-up manner and is specifically tailored to hierarchical 3D models. Our experiments not only demonstrate that our method achieves competitive performance to supervised models, but also surpasses existing self-supervised approaches by a large margin. The model and training code can be found at our Github repository (https://github.com/phermosilla/msm).
Authors: Wenfeng Feng, Guoying Sun
Abstract: In this paper, we propose EDIT (Encoder-Decoder Image Transformer), a novel architecture designed to mitigate the attention sink phenomenon observed in Vision Transformer models. Attention sink occurs when an excessive amount of attention is allocated to the [CLS] token, distorting the model's ability to effectively process image patches. To address this, we introduce a layer-aligned encoder-decoder architecture, where the encoder utilizes self-attention to process image patches, while the decoder uses cross-attention to focus on the [CLS] token. Unlike traditional encoder-decoder framework, where the decoder depends solely on high-level encoder representations, EDIT allows the decoder to extract information starting from low-level features, progressively refining the representation layer by layer. EDIT is naturally interpretable demonstrated through sequential attention maps, illustrating the refined, layer-by-layer focus on key image features. Experiments on ImageNet-1k and ImageNet-21k, along with transfer learning tasks, show that EDIT achieves consistent performance improvements over DeiT3 models. These results highlight the effectiveness of EDIT's design in addressing attention sink and improving visual feature extraction.
Authors: Ylli Sadikaj, Hongkuan Zhou, Lavdim Halilaj, Stefan Schmid, Steffen Staab, Claudia Plant
Abstract: Precise optical inspection in industrial applications is crucial for minimizing scrap rates and reducing the associated costs. Besides merely detecting if a product is anomalous or not, it is crucial to know the distinct type of defect, such as a bent, cut, or scratch. The ability to recognize the "exact" defect type enables automated treatments of the anomalies in modern production lines. Current methods are limited to solely detecting whether a product is defective or not without providing any insights on the defect type, nevertheless detecting and identifying multiple defects. We propose MultiADS, a zero-shot learning approach, able to perform Multi-type Anomaly Detection and Segmentation. The architecture of MultiADS comprises CLIP and extra linear layers to align the visual- and textual representation in a joint feature space. To the best of our knowledge, our proposal, is the first approach to perform a multi-type anomaly segmentation task in zero-shot learning. Contrary to the other baselines, our approach i) generates specific anomaly masks for each distinct defect type, ii) learns to distinguish defect types, and iii) simultaneously identifies multiple defect types present in an anomalous product. Additionally, our approach outperforms zero/few-shot learning SoTA methods on image-level and pixel-level anomaly detection and segmentation tasks on five commonly used datasets: MVTec-AD, Visa, MPDD, MAD and Real-IAD.
Authors: Constantin Ulrich, Tassilo Wald, Fabian Isensee, Klaus H. Maier-Hein
Abstract: The segmentation of lesions in Moderate to Severe Traumatic Brain Injury (msTBI) presents a significant challenge in neuroimaging due to the diverse characteristics of these lesions, which vary in size, shape, and distribution across brain regions and tissue types. This heterogeneity complicates traditional image processing techniques, resulting in critical errors in tasks such as image registration and brain parcellation. To address these challenges, the AIMS-TBI Segmentation Challenge 2024 aims to advance innovative segmentation algorithms specifically designed for T1-weighted MRI data, the most widely utilized imaging modality in clinical practice. Our proposed solution leverages a large-scale multi-dataset supervised pretraining approach inspired by the MultiTalent method. We train a Resenc L network on a comprehensive collection of datasets covering various anatomical and pathological structures, which equips the model with a robust understanding of brain anatomy and pathology. Following this, the model is fine-tuned on msTBI-specific data to optimize its performance for the unique characteristics of T1-weighted MRI scans and outperforms the baseline without pretraining up to 2 Dice points.
Authors: Alexandra Ertl, Shuhan Xiao, Stefan Denner, Robin Peretzke, David Zimmerer, Peter Neher, Fabian Isensee, Klaus Maier-Hein
Abstract: Landmark detection plays a crucial role in medical imaging tasks that rely on precise spatial localization, including specific applications in diagnosis, treatment planning, image registration, and surgical navigation. However, manual annotation is labor-intensive and requires expert knowledge. While deep learning shows promise in automating this task, progress is hindered by limited public datasets, inconsistent benchmarks, and non-standardized baselines, restricting reproducibility, fair comparisons, and model generalizability.This work introduces nnLandmark, a self-configuring deep learning framework for 3D medical landmark detection, adapting nnU-Net to perform heatmap-based regression. By leveraging nnU-Net's automated configuration, nnLandmark eliminates the need for manual parameter tuning, offering out-of-the-box usability. It achieves state-of-the-art accuracy across two public datasets, with a mean radial error (MRE) of 1.5 mm on the Mandibular Molar Landmark (MML) dental CT dataset and 1.2 mm for anatomical fiducials on a brain MRI dataset (AFIDs), where nnLandmark aligns with the inter-rater variability of 1.5 mm. With its strong generalization, reproducibility, and ease of deployment, nnLandmark establishes a reliable baseline for 3D landmark detection, supporting research in anatomical localization and clinical workflows that depend on precise landmark identification. The code will be available soon.
Authors: Leszek Luchowski, Dariusz Pojda
Abstract: The article presents an innovative approach to the visualisation of multidimensional data, using icons inspired by Chernoff faces. The approach merges classical projection techniques with the assignment of particular data dimensions to mimic features, capitalizing on the natural ability of the human brain to interpret facial expressions. The technique is implemented as a plugin to the dpVision open-source image handling platform. The plugin allows the data to be interactively explored in the form of a swarm of "totems" whose position in hyperspace as well as facial features represent various aspects of the data. Sample visualisations, based on synthetic test data as well as the vinhoverde 15-dimensional database on Portuguese wines, confirm the usefulness of our approach to the analysis of complex data structures.
Authors: Rishbuh Parihar, Vaibhav Agrawal, Sachidanand VS, R. Venkatesh Babu
Abstract: Existing approaches for controlling text-to-image diffusion models, while powerful, do not allow for explicit 3D object-centric control, such as precise control of object orientation. In this work, we address the problem of multi-object orientation control in text-to-image diffusion models. This enables the generation of diverse multi-object scenes with precise orientation control for each object. The key idea is to condition the diffusion model with a set of orientation-aware \textbf{compass} tokens, one for each object, along with text tokens. A light-weight encoder network predicts these compass tokens taking object orientation as the input. The model is trained on a synthetic dataset of procedurally generated scenes, each containing one or two 3D assets on a plain background. However, direct training this framework results in poor orientation control as well as leads to entanglement among objects. To mitigate this, we intervene in the generation process and constrain the cross-attention maps of each compass token to its corresponding object regions. The trained model is able to achieve precise orientation control for a) complex objects not seen during training and b) multi-object scenes with more than two objects, indicating strong generalization capabilities. Further, when combined with personalization methods, our method precisely controls the orientation of the new object in diverse contexts. Our method achieves state-of-the-art orientation control and text alignment, quantified with extensive evaluations and a user study.
Authors: Li Yu, Zhihui Li, Jimin Xiao, Moncef Gabbouj
Abstract: Neural representations for video (NeRV) have gained considerable attention for their strong performance across various video tasks. However, existing NeRV methods often struggle to capture fine spatial details, resulting in vague reconstructions. In this paper, we present a Frequency Separation and Augmentation based Neural Representation for video (FANeRV), which addresses these limitations with its core Wavelet Frequency Upgrade Block.This block explicitly separates input frames into high and low-frequency components using discrete wavelet transform, followed by targeted enhancement using specialized modules. Finally, a specially designed gated network effectively fuses these frequency components for optimal reconstruction. Additionally, convolutional residual enhancement blocks are integrated into the later stages of the network to balance parameter distribution and improve the restoration of high-frequency details. Experimental results demonstrate that FANeRV significantly improves reconstruction performance and excels in multiple tasks, including video compression, inpainting, and interpolation, outperforming existing NeRV methods.
Authors: Xi Tao, Liyan Lin
Abstract: Computed tomography (CT) examination poses radiation injury to patient. A consensus performing CT imaging is to make the radiation dose as low as reasonably achievable, i.e. the ALARA law. In this paper, we propose an end-to-end learning framework, named End2end-ALARA, that jointly optimizes dose modulation and image reconstruction to meet the goal of ALARA in CT imaging. End2end-ALARA works by building a dose modulation module and an image reconstruction module, connecting these modules with a differentiable simulation function, and optimizing the them with a constrained hinge loss function. The objective is to minimize radiation dose subject to a prescribed image quality (IQ) index. The results show that End2end-ALARA is able to preset personalized dose levels to gain a stable IQ level across patients, which may facilitate image-based diagnosis and downstream model training. Moreover, compared to fixed-dose and conventional dose modulation strategies, End2end-ALARA consumes lower dose to reach the same IQ level. Our study sheds light on a way of realizing the ALARA law in CT imaging.
Authors: Reiji Saito, Kazuhiro Hotta
Abstract: In this paper, we propose a new evaluation metric called Domain Independence (DI) and Attenuation of Domain-Specific Information (ADSI) which is specifically designed for domain-generalized semantic segmentation in automotive images. DI measures the presence of domain-specific information: a lower DI value indicates strong domain dependence, while a higher DI value suggests greater domain independence. This makes it roughly where domain-specific information exists and up to which frequency range it is present. As a result, it becomes possible to effectively suppress only the regions in the image that contain domain-specific information, enabling feature extraction independent of the domain. ADSI uses a Butterworth filter to remove the low-frequency components of images that contain inherent domain-specific information such as sensor characteristics and lighting conditions. However, since low-frequency components also contain important information such as color, we should not remove them completely. Thus, a scalar value (ranging from 0 to 1) is multiplied by the low-frequency components to retain essential information. This helps the model learn more domain-independent features. In experiments, GTA5 (synthetic dataset) was used as training images, and a real-world dataset was used for evaluation, and the proposed method outperformed conventional approaches. Similarly, in experiments that the Cityscapes (real-world dataset) was used for training and various environment datasets such as rain and nighttime were used for evaluation, the proposed method demonstrated its robustness under nighttime conditions.
Authors: Shuoshuo Xu, Kai Zhao, James Loney, Zili Li, Andrea Visentin
Abstract: Effective and rapid evaluation of pavement surface condition is critical for prioritizing maintenance, ensuring transportation safety, and minimizing vehicle wear and tear. While conventional manual inspections suffer from subjectivity, existing machine learning-based methods are constrained by their reliance on large and high-quality labeled datasets, which require significant resources and limit adaptability across varied road conditions. The revolutionary advancements in Large Language Models (LLMs) present significant potential for overcoming these challenges. In this study, we propose an innovative automated zero-shot learning approach that leverages the image recognition and natural language understanding capabilities of LLMs to assess road conditions effectively. Multiple LLM-based assessment models were developed, employing prompt engineering strategies aligned with the Pavement Surface Condition Index (PSCI) standards. These models' accuracy and reliability were evaluated against official PSCI results, with an optimized model ultimately selected. Extensive tests benchmarked the optimized model against evaluations from various levels experts using Google Street View road images. The results reveal that the LLM-based approach can effectively assess road conditions, with the optimized model -employing comprehensive and structured prompt engineering strategies -outperforming simpler configurations by achieving high accuracy and consistency, even surpassing expert evaluations. Moreover, successfully applying the optimized model to Google Street View images demonstrates its potential for future city-scale deployments. These findings highlight the transformative potential of LLMs in automating road damage evaluations and underscore the pivotal role of detailed prompt engineering in achieving reliable assessments.
Authors: Danielle Cohen, Hila Chefer, Lior Wolf
Abstract: Deep neural networks (DNNs) have demonstrated remarkable success, yet their wide adoption is often hindered by their opaque decision-making. To address this, attribution methods have been proposed to assign relevance values to each part of the input. However, different methods often produce entirely different relevance maps, necessitating the development of standardized metrics to evaluate them. Typically, such evaluation is performed through perturbation, wherein high- or low-relevance regions of the input image are manipulated to examine the change in prediction. In this work, we introduce a novel approach, which harnesses image generation models to perform targeted perturbation. Specifically, we focus on inpainting only the high-relevance pixels of an input image to modify the model's predictions while preserving image fidelity. This is in contrast to existing approaches, which often produce out-of-distribution modifications, leading to unreliable results. Through extensive experiments, we demonstrate the effectiveness of our approach in generating meaningful rankings across a wide range of models and attribution methods. Crucially, we establish that the ranking produced by our metric exhibits significantly higher correlation with human preferences compared to existing approaches, underscoring its potential for enhancing interpretability in DNNs.
Authors: Rishubh Parihar, Srinjay Sarkar, Sarthak Vora, Jogendra Kundu, R. Venkatesh Babu
Abstract: Current monocular 3D detectors are held back by the limited diversity and scale of real-world datasets. While data augmentation certainly helps, it's particularly difficult to generate realistic scene-aware augmented data for outdoor settings. Most current approaches to synthetic data generation focus on realistic object appearance through improved rendering techniques. However, we show that where and how objects are positioned is just as crucial for training effective 3D monocular detectors. The key obstacle lies in automatically determining realistic object placement parameters - including position, dimensions, and directional alignment when introducing synthetic objects into actual scenes. To address this, we introduce MonoPlace3D, a novel system that considers the 3D scene content to create realistic augmentations. Specifically, given a background scene, MonoPlace3D learns a distribution over plausible 3D bounding boxes. Subsequently, we render realistic objects and place them according to the locations sampled from the learned distribution. Our comprehensive evaluation on two standard datasets KITTI and NuScenes, demonstrates that MonoPlace3D significantly improves the accuracy of multiple existing monocular 3D detectors while being highly data efficient.
Authors: Wangbo Zhao, Yizeng Han, Jiasheng Tang, Kai Wang, Hao Luo, Yibing Song, Gao Huang, Fan Wang, Yang You
Abstract: Diffusion Transformer (DiT), an emerging diffusion model for visual generation, has demonstrated superior performance but suffers from substantial computational costs. Our investigations reveal that these costs primarily stem from the \emph{static} inference paradigm, which inevitably introduces redundant computation in certain \emph{diffusion timesteps} and \emph{spatial regions}. To overcome this inefficiency, we propose \textbf{Dy}namic \textbf{Di}ffusion \textbf{T}ransformer (DyDiT), an architecture that \emph{dynamically} adjusts its computation along both \emph{timestep} and \emph{spatial} dimensions. Specifically, we introduce a \emph{Timestep-wise Dynamic Width} (TDW) approach that adapts model width conditioned on the generation timesteps. In addition, we design a \emph{Spatial-wise Dynamic Token} (SDT) strategy to avoid redundant computation at unnecessary spatial locations. TDW and SDT can be seamlessly integrated into DiT and significantly accelerates the generation process. Building on these designs, we further enhance DyDiT in three key aspects. First, DyDiT is integrated seamlessly with flow matching-based generation, enhancing its versatility. Furthermore, we enhance DyDiT to tackle more complex visual generation tasks, including video generation and text-to-image generation, thereby broadening its real-world applications. Finally, to address the high cost of full fine-tuning and democratize technology access, we investigate the feasibility of training DyDiT in a parameter-efficient manner and introduce timestep-based dynamic LoRA (TD-LoRA). Extensive experiments on diverse visual generation models, including DiT, SiT, Latte, and FLUX, demonstrate the effectiveness of DyDiT.
Authors: Abhinav Roy, Bhavesh Gyanchandani, Aditya Oza
Abstract: Lung cancer remains one of the leading causes of cancer-related mortality worldwide, with early and accurate diagnosis playing a pivotal role in improving patient outcomes. Automated detection of pulmonary nodules in computed tomography (CT) scans is a challenging task due to variability in nodule size, shape, texture, and location. Traditional Convolutional Neural Networks (CNNs) have shown considerable promise in medical image analysis; however, their limited ability to capture fine-grained spatial-spectral variations restricts their performance in complex diagnostic scenarios. In this study, we propose a novel hybrid deep learning architecture that incorporates Chebyshev polynomial expansions into CNN layers to enhance expressive power and improve the representation of underlying anatomical structures. The proposed Chebyshev-CNN leverages the orthogonality and recursive properties of Chebyshev polynomials to extract high-frequency features and approximate complex nonlinear functions with greater fidelity. The model is trained and evaluated on benchmark lung cancer imaging datasets, including LUNA16 and LIDC-IDRI, achieving superior performance in classifying pulmonary nodules as benign or malignant. Quantitative results demonstrate significant improvements in accuracy, sensitivity, and specificity compared to traditional CNN-based approaches. This integration of polynomial-based spectral approximation within deep learning provides a robust framework for enhancing automated medical diagnostics and holds potential for broader applications in clinical decision support systems.
Authors: Hanxiao Sun, YuPeng Gao, Jin Xie, Jian Yang, Beibei Wang
Abstract: Reconstructing 3D assets from images, known as inverse rendering (IR), remains a challenging task due to its ill-posed nature. 3D Gaussian Splatting (3DGS) has demonstrated impressive capabilities for novel view synthesis (NVS) tasks. Methods apply it to relighting by separating radiance into BRDF parameters and lighting, yet produce inferior relighting quality with artifacts and unnatural indirect illumination due to the limited capability of each Gaussian, which has constant material parameters and normal, alongside the absence of physical constraints for indirect lighting. In this paper, we present a novel framework called Spatially-vayring Gaussian Inverse Rendering (SVG-IR), aimed at enhancing both NVS and relighting quality. To this end, we propose a new representation-Spatially-varying Gaussian (SVG)-that allows per-Gaussian spatially varying parameters. This enhanced representation is complemented by a SVG splatting scheme akin to vertex/fragment shading in traditional graphics pipelines. Furthermore, we integrate a physically-based indirect lighting model, enabling more realistic relighting. The proposed SVG-IR framework significantly improves rendering quality, outperforming state-of-the-art NeRF-based methods by 2.5 dB in peak signal-to-noise ratio (PSNR) and surpassing existing Gaussian-based techniques by 3.5 dB in relighting tasks, all while maintaining a real-time rendering speed.
Authors: Can Zhang, Gim Hee Lee
Abstract: This work presents IAAO, a novel framework that builds an explicit 3D model for intelligent agents to gain understanding of articulated objects in their environment through interaction. Unlike prior methods that rely on task-specific networks and assumptions about movable parts, our IAAO leverages large foundation models to estimate interactive affordances and part articulations in three stages. We first build hierarchical features and label fields for each object state using 3D Gaussian Splatting (3DGS) by distilling mask features and view-consistent labels from multi-view images. We then perform object- and part-level queries on the 3D Gaussian primitives to identify static and articulated elements, estimating global transformations and local articulation parameters along with affordances. Finally, scenes from different states are merged and refined based on the estimated transformations, enabling robust affordance-based interaction and manipulation of objects. Experimental results demonstrate the effectiveness of our method.
Authors: Ziyi Wang, Haoran Wu, Yiming Rong, Deyang Jiang, Yixin Zhang, Yunlong Zhao, Shuang Xu, Bo XU
Abstract: Long video understanding is a complex task that requires both spatial detail and temporal awareness. While Vision-Language Models (VLMs) obtain frame-level understanding capabilities through multi-frame input, they suffer from information loss due to the sparse sampling strategy. In contrast, Video Large Language Models (Video-LLMs) capture temporal relationships within visual features but are limited by the scarcity of high-quality video-text datasets. To transfer long video understanding capabilities to VLMs with minimal data and computational cost, we propose Lightweight Video Compression (LVC), a novel method featuring the Query-Attention Video Compression mechanism, which effectively tackles the sparse sampling problem in VLMs. By training only the alignment layer with 10k short video-text pairs, LVC significantly enhances the temporal reasoning abilities of VLMs. Extensive experiments show that LVC provides consistent performance improvements across various models, including the InternVL2 series and Phi-3.5-Vision. Notably, the InternVL2-40B-LVC achieves scores of 68.2 and 65.9 on the long video understanding benchmarks MLVU and Video-MME, respectively, with relative improvements of 14.6% and 7.7%. The enhanced models and code will be publicly available soon.
Authors: Jakub Maciej Wi\'sniewski, Anders Nymark Christensen, Mary Le Ngo, Martin Gr{\o}nneb{\ae}k Tolsgaard, Chun Kit Wong
Abstract: Cognitive demands of fetal ultrasound examinations pose unique challenges among clinicians. With the goal of providing an assistive tool, we developed an automated pipeline for predicting fetal orientation from ultrasound videos acquired following a simple blind sweep protocol. Leveraging on a pre-trained head detection and segmentation model, this is achieved by first determining the fetal presentation (cephalic or breech) with a template matching approach, followed by the fetal lie (facing left or right) by analyzing the spatial distribution of segmented brain anatomies. Evaluation on a dataset of third-trimester ultrasound scans demonstrated the promising accuracy of our pipeline. This work distinguishes itself by introducing automated fetal lie prediction and by proposing an assistive paradigm that augments sonographer expertise rather than replacing it. Future research will focus on enhancing acquisition efficiency, and exploring real-time clinical integration to improve workflow and support for obstetric clinicians.
Authors: Seonghwan Park, Jaehyeon Jeong, Yongjun Kim, Jaeho Lee, Namhoon Lee
Abstract: Recent studies have introduced various approaches for prompt-tuning black-box vision-language models, referred to as black-box prompt-tuning (BBPT). While BBPT has demonstrated considerable potential, it is often found that many existing methods require an excessive number of queries (i.e., function evaluations), which poses a significant challenge in real-world scenarios where the number of allowed queries is limited. To tackle this issue, we propose Zeroth-order Intrinsic-dimensional Prompt-tuning (ZIP), a novel approach that enables efficient and robust prompt optimization in a purely black-box setting. The key idea of ZIP is to reduce the problem dimensionality and the variance of zeroth-order gradient estimates, such that the training is done fast with far less queries. We achieve this by re-parameterizing prompts in low-rank representations and designing intrinsic-dimensional clipping of estimated gradients. We evaluate ZIP on 13+ vision-language tasks in standard benchmarks and show that it achieves an average improvement of approximately 6% in few-shot accuracy and 48% in query efficiency compared to the best-performing alternative BBPT methods, establishing a new state of the art. Our ablation analysis further shows that the proposed clipping mechanism is robust and nearly optimal, without the need to manually select the clipping threshold, matching the result of expensive hyperparameter search.
Authors: Tom Simon, William Mocaer, Pierrick Tranouez, Clement Chatelain, Thierry Paquet
Abstract: We introduce Rosetta, a multimodal model that leverages Multimodal In-Context Learning (MICL) to classify sequences of novel script patterns in documents by leveraging minimal examples, thus eliminating the need for explicit retraining. To enhance contextual learning, we designed a dataset generation process that ensures varying degrees of contextual informativeness, improving the model's adaptability in leveraging context across different scenarios. A key strength of our method is the use of a Context-Aware Tokenizer (CAT), which enables open-vocabulary classification. This allows the model to classify text and symbol patterns across an unlimited range of classes, extending its classification capabilities beyond the scope of its training alphabet of patterns. As a result, it unlocks applications such as the recognition of new alphabets and languages. Experiments on synthetic datasets demonstrate the potential of Rosetta to successfully classify Out-Of-Distribution visual patterns and diverse sets of alphabets and scripts, including but not limited to Chinese, Greek, Russian, French, Spanish, and Japanese.
Authors: Mishan Aliev, Dmitry Baranchuk, Kirill Struminsky
Abstract: This work investigates text-to-texture synthesis using diffusion models to generate physically-based texture maps. We aim to achieve realistic model appearances under varying lighting conditions. A prominent solution for the task is score distillation sampling. It allows recovering a complex texture using gradient guidance given a differentiable rasterization and shading pipeline. However, in practice, the aforementioned solution in conjunction with the widespread latent diffusion models produces severe visual artifacts and requires additional regularization such as implicit texture parameterization. As a more direct alternative, we propose an approach using cascaded diffusion models for texture synthesis (CasTex). In our setup, score distillation sampling yields high-quality textures out-of-the box. In particular, we were able to omit implicit texture parameterization in favor of an explicit parameterization to improve the procedure. In the experiments, we show that our approach significantly outperforms state-of-the-art optimization-based solutions on public texture synthesis benchmarks.
Authors: Diljeet Jagpal, Xi Chen, Vinay P. Namboodiri
Abstract: Zero-shot, training-free, image-based text-to-video generation is an emerging area that aims to generate videos using existing image-based diffusion models. Current methods in this space require specific architectural changes to image generation models, which limit their adaptability and scalability. In contrast to such methods, we provide a model-agnostic approach. We use intersections in diffusion trajectories, working only with the latent values. We could not obtain localized frame-wise coherence and diversity using only the intersection of trajectories. Thus, we instead use a grid-based approach. An in-context trained LLM is used to generate coherent frame-wise prompts; another is used to identify differences between frames. Based on these, we obtain a CLIP-based attention mask that controls the timing of switching the prompts for each grid cell. Earlier switching results in higher variance, while later switching results in more coherence. Therefore, our approach can ensure appropriate control between coherence and variance for the frames. Our approach results in state-of-the-art performance while being more flexible when working with diverse image-generation models. The empirical analysis using quantitative metrics and user studies confirms our model's superior temporal consistency, visual fidelity and user satisfaction, thus providing a novel way to obtain training-free, image-based text-to-video generation.
Authors: Chang Nie, Yiqing Xu, Guangming Wang, Zhe Liu, Yanzi Miao, Hesheng Wang
Abstract: Moving object segmentation plays a vital role in understanding dynamic visual environments. While existing methods rely on multi-frame image sequences to identify moving objects, single-image MOS is critical for applications like motion intention prediction and handling camera frame drops. However, segmenting moving objects from a single image remains challenging for existing methods due to the absence of temporal cues. To address this gap, we propose MovSAM, the first framework for single-image moving object segmentation. MovSAM leverages a Multimodal Large Language Model (MLLM) enhanced with Chain-of-Thought (CoT) prompting to search the moving object and generate text prompts based on deep thinking for segmentation. These prompts are cross-fused with visual features from the Segment Anything Model (SAM) and a Vision-Language Model (VLM), enabling logic-driven moving object segmentation. The segmentation results then undergo a deep thinking refinement loop, allowing MovSAM to iteratively improve its understanding of the scene context and inter-object relationships with logical reasoning. This innovative approach enables MovSAM to segment moving objects in single images by considering scene understanding. We implement MovSAM in the real world to validate its practical application and effectiveness for autonomous driving scenarios where the multi-frame methods fail. Furthermore, despite the inherent advantage of multi-frame methods in utilizing temporal information, MovSAM achieves state-of-the-art performance across public MOS benchmarks, reaching 92.5\% on J\&F. Our implementation will be available at https://github.com/IRMVLab/MovSAM.
Authors: Mingbo Li, Liying Liu, Ye Luo
Abstract: Convolutional neural networks have become increasingly deep and complex, leading to higher computational costs. While tropical convolutional neural networks (TCNNs) reduce multiplications, they underperform compared to standard CNNs. To address this, we propose two new variants - compound TCNN (cTCNN) and parallel TCNN (pTCNN)-that use combinations of tropical min-plus and max-plus kernels to replace traditional convolution kernels. This reduces multiplications and balances efficiency with performance. Experiments on various datasets show that cTCNN and pTCNN match or exceed the performance of other CNN methods. Combining these with conventional CNNs in deeper architectures also improves performance. We are further exploring simplified TCNN architectures that reduce parameters and multiplications with minimal accuracy loss, aiming for efficient and effective models.
Authors: Dingkun Yan, Xinrui Wang, Yusuke Iwasawa, Yutaka Matsuo, Suguru Saito, Jiaxian Guo
Abstract: Reference-based sketch colorization methods have garnered significant attention due to their potential applications in the animation production industry. However, most existing methods are trained with image triplets of sketch, reference, and ground truth that are semantically and spatially well-aligned, while real-world references and sketches often exhibit substantial misalignment. This mismatch in data distribution between training and inference leads to overfitting, consequently resulting in spatial artifacts and significant degradation in overall colorization quality, limiting potential applications of current methods for general purposes. To address this limitation, we conduct an in-depth analysis of the \textbf{carrier}, defined as the latent representation facilitating information transfer from reference to sketch. Based on this analysis, we propose a novel workflow that dynamically adapts the carrier to optimize distinct aspects of colorization. Specifically, for spatially misaligned artifacts, we introduce a split cross-attention mechanism with spatial masks, enabling region-specific reference injection within the diffusion process. To mitigate semantic neglect of sketches, we employ dedicated background and style encoders to transfer detailed reference information in the latent feature space, achieving enhanced spatial control and richer detail synthesis. Furthermore, we propose character-mask merging and background bleaching as preprocessing steps to improve foreground-background integration and background generation. Extensive qualitative and quantitative evaluations, including a user study, demonstrate the superior performance of our proposed method compared to existing approaches. An ablation study further validates the efficacy of each proposed component.
Authors: Jiawei Mao, Yuhan Wang, Yucheng Tang, Daguang Xu, Kang Wang, Yang Yang, Zongwei Zhou, Yuyin Zhou
Abstract: This paper presents MedSegFactory, a versatile medical synthesis framework that generates high-quality paired medical images and segmentation masks across modalities and tasks. It aims to serve as an unlimited data repository, supplying image-mask pairs to enhance existing segmentation tools. The core of MedSegFactory is a dual-stream diffusion model, where one stream synthesizes medical images and the other generates corresponding segmentation masks. To ensure precise alignment between image-mask pairs, we introduce Joint Cross-Attention (JCA), enabling a collaborative denoising paradigm by dynamic cross-conditioning between streams. This bidirectional interaction allows both representations to guide each other's generation, enhancing consistency between generated pairs. MedSegFactory unlocks on-demand generation of paired medical images and segmentation masks through user-defined prompts that specify the target labels, imaging modalities, anatomical regions, and pathological conditions, facilitating scalable and high-quality data generation. This new paradigm of medical image synthesis enables seamless integration into diverse medical imaging workflows, enhancing both efficiency and accuracy. Extensive experiments show that MedSegFactory generates data of superior quality and usability, achieving competitive or state-of-the-art performance in 2D and 3D segmentation tasks while addressing data scarcity and regulatory constraints.
Authors: Emmanuelle Bourigault, Amir Jamaludin, Abdullah Hamdi
Abstract: In medical imaging, the primary challenge is collecting large-scale labeled data due to privacy concerns, logistics, and high labeling costs. In this work, we present the UK Biobank Organs and Bones (UKBOB), the largest labeled dataset of body organs, comprising 51,761 MRI 3D samples (equivalent to 17.9 million 2D images) and more than 1.37 billion 2D segmentation masks of 72 organs, all based on the UK Biobank MRI dataset. We utilize automatic labeling, introduce an automated label cleaning pipeline with organ-specific filters, and manually annotate a subset of 300 MRIs with 11 abdominal classes to validate the quality (referred to as UKBOB-manual). This approach allows for scaling up the dataset collection while maintaining confidence in the labels. We further confirm the validity of the labels by demonstrating zero-shot generalization of trained models on the filtered UKBOB to other small labeled datasets from similar domains (e.g., abdominal MRI). To further mitigate the effect of noisy labels, we propose a novel method called Entropy Test-time Adaptation (ETTA) to refine the segmentation output. We use UKBOB to train a foundation model, Swin-BOB, for 3D medical image segmentation based on the Swin-UNetr architecture, achieving state-of-the-art results in several benchmarks in 3D medical imaging, including the BRATS brain MRI tumor challenge (with a 0.4% improvement) and the BTCV abdominal CT scan benchmark (with a 1.3% improvement). The pre-trained models and the code are available at https://emmanuelleb985.github.io/ukbob , and the filtered labels will be made available with the UK Biobank.
Authors: Masquil El\'ias, Mar\'i Roger, Ehret Thibaud, Meinhardt-Llopis Enric, Mus\'e Pablo, Facciolo Gabriele
Abstract: We introduce the S-EO dataset: a large-scale, high-resolution dataset, designed to advance geometry-aware shadow detection. Collected from diverse public-domain sources, including challenge datasets and government providers such as USGS, our dataset comprises 702 georeferenced tiles across the USA, each covering 500x500 m. Each tile includes multi-date, multi-angle WorldView-3 pansharpened RGB images, panchromatic images, and a ground-truth DSM of the area obtained from LiDAR scans. For each image, we provide a shadow mask derived from geometry and sun position, a vegetation mask based on the NDVI index, and a bundle-adjusted RPC model. With approximately 20,000 images, the S-EO dataset establishes a new public resource for shadow detection in remote sensing imagery and its applications to 3D reconstruction. To demonstrate the dataset's impact, we train and evaluate a shadow detector, showcasing its ability to generalize, even to aerial images. Finally, we extend EO-NeRF - a state-of-the-art NeRF approach for satellite imagery - to leverage our shadow predictions for improved 3D reconstructions.
Authors: Sergio Romero-Tapiador, Ruben Tolosana, Blanca Lacruz-Pleguezuelos, Laura Judith Marcos Zambrano, Guadalupe X. Baz\'an, Isabel Espinosa-Salinas, Julian Fierrez, Javier Ortega-Garcia, Enrique Carrillo de Santa Pau, Aythami Morales
Abstract: Automatic dietary assessment based on food images remains a challenge, requiring precise food detection, segmentation, and classification. Vision-Language Models (VLMs) offer new possibilities by integrating visual and textual reasoning. In this study, we evaluate six state-of-the-art VLMs (ChatGPT, Gemini, Claude, Moondream, DeepSeek, and LLaVA), analyzing their capabilities in food recognition at different levels. For the experimental framework, we introduce the FoodNExTDB, a unique food image database that contains 9,263 expert-labeled images across 10 categories (e.g., "protein source"), 62 subcategories (e.g., "poultry"), and 9 cooking styles (e.g., "grilled"). In total, FoodNExTDB includes 50k nutritional labels generated by seven experts who manually annotated all images in the database. Also, we propose a novel evaluation metric, Expert-Weighted Recall (EWR), that accounts for the inter-annotator variability. Results show that closed-source models outperform open-source ones, achieving over 90% EWR in recognizing food products in images containing a single product. Despite their potential, current VLMs face challenges in fine-grained food recognition, particularly in distinguishing subtle differences in cooking styles and visually similar food items, which limits their reliability for automatic dietary assessment. The FoodNExTDB database is publicly available at https://github.com/AI4Food/FoodNExtDB.
Authors: Sachin Kumar Danisetty, Alexandros Graikos, Srikar Yellapragada, Dimitris Samaras
Abstract: Image segmentation is crucial in many computational pathology pipelines, including accurate disease diagnosis, subtyping, outcome, and survivability prediction. The common approach for training a segmentation model relies on a pre-trained feature extractor and a dataset of paired image and mask annotations. These are used to train a lightweight prediction model that translates features into per-pixel classes. The choice of the feature extractor is central to the performance of the final segmentation model, and recent literature has focused on finding tasks to pre-train the feature extractor. In this paper, we propose PathSegDiff, a novel approach for histopathology image segmentation that leverages Latent Diffusion Models (LDMs) as pre-trained featured extractors. Our method utilizes a pathology-specific LDM, guided by a self-supervised encoder, to extract rich semantic information from H\&E stained histopathology images. We employ a simple, fully convolutional network to process the features extracted from the LDM and generate segmentation masks. Our experiments demonstrate significant improvements over traditional methods on the BCSS and GlaS datasets, highlighting the effectiveness of domain-specific diffusion pre-training in capturing intricate tissue structures and enhancing segmentation accuracy in histopathology images.
Authors: Marco Acerbis, Nata\v{s}a Sladoje, Joakim Lindblad
Abstract: Accurate and efficient cell detection is crucial in many biomedical image analysis tasks. We evaluate the performance of several Deep Learning (DL) methods for cell detection in Papanicolaou-stained cytological Whole Slide Images (WSIs), focusing on accuracy of predictions and computational efficiency. We examine recentoff-the-shelf algorithms as well as custom-designed detectors, applying them to two datasets: the CNSeg Dataset and the Oral Cancer (OC) Dataset. Our comparison includes well-established segmentation methods such as StarDist, Cellpose, and the Segment Anything Model 2 (SAM2), alongside centroid-based Fully Convolutional Regression Network (FCRN) approaches. We introduce a suitable evaluation metric to assess the accuracy of predictions based on the distance from ground truth positions. We also explore the impact of dataset size and data augmentation techniques on model performance. Results show that centroid-based methods, particularly the Improved Fully Convolutional Regression Network (IFCRN) method, outperform segmentation-based methods in terms of both detection accuracy and computational efficiency. This study highlights the potential of centroid-based detectors as a preferred option for cell detection in resource-limited environments, offering faster processing times and lower GPU memory usage without compromising accuracy.
Authors: Xinhao Li, Ziang Yan, Desen Meng, Lu Dong, Xiangyu Zeng, Yinan He, Yali Wang, Yu Qiao, Yi Wang, Limin Wang
Abstract: Recent advancements in reinforcement learning have significantly advanced the reasoning capabilities of multimodal large language models (MLLMs). While approaches such as Group Relative Policy Optimization (GRPO) and rule-based reward mechanisms demonstrate promise in text and image domains, their application to video understanding remains limited. This paper presents a systematic exploration of Reinforcement Fine-Tuning (RFT) with GRPO for video MLLMs, aiming to enhance spatio-temporal perception while maintaining general capabilities. Our experiments reveal that RFT is highly data-efficient for task-specific improvements. Through multi-task RFT on spatio-temporal perception objectives with limited samples, we develop VideoChat-R1, a powerful video MLLM that achieves state-of-the-art performance on spatio-temporal perception tasks without sacrificing chat ability, while exhibiting emerging spatio-temporal reasoning abilities. Compared to Qwen2.5-VL-7B, VideoChat-R1 boosts performance several-fold in tasks like temporal grounding (+31.8) and object tracking (+31.2). Additionally, it significantly improves on general QA benchmarks such as VideoMME (+0.9), MVBench (+1.0), and Perception Test (+0.9). Our findings underscore the potential of RFT for specialized task enhancement of Video MLLMs. We hope our work offers valuable insights for future RL research in video MLLMs.
Authors: Thomas Kerdreux, Alexandre Tuel, Quentin Febvre, Alexis Mouche, Bertrand Chapron
Abstract: Self-supervised learning (SSL) has enabled the development of vision foundation models for Earth Observation (EO), demonstrating strong transferability across diverse remote sensing tasks. While prior work has focused on network architectures and training strategies, the role of dataset curation, especially in balancing and diversifying pre-training datasets, remains underexplored. In EO, this challenge is amplified by the redundancy and heavy-tailed distributions common in satellite imagery, which can lead to biased representations and inefficient training. In this work, we propose a dynamic dataset pruning strategy designed to improve SSL pre-training by maximizing dataset diversity and balance. Our method iteratively refines the training set without requiring a pre-existing feature extractor, making it well-suited for domains where curated datasets are limited or unavailable. We demonstrate our approach on the Sentinel-1 Wave Mode (WV) Synthetic Aperture Radar (SAR) archive, a challenging dataset dominated by ocean observations. We train models from scratch on the entire Sentinel-1 WV archive spanning 10 years. Across three downstream tasks, our results show that dynamic pruning improves both computational efficiency and representation quality, leading to stronger transferability. We also release the weights of Nereus-SAR-1, the first model in the Nereus family, a series of foundation models for ocean observation and analysis using SAR imagery, at github.com/galeio-research/nereus-sar-models/.
Authors: Teng Xiao, Qi Hu, Qingsong Yan, Wei Liu, Zhiwei Ye, Fei Deng
Abstract: Pan-Tilt-Zoom (PTZ) cameras with wide-angle lenses are widely used in surveillance but often require image rectification due to their inherent nonlinear distortions. Current deep learning approaches typically struggle to maintain fine-grained geometric details, resulting in inaccurate rectification. This paper presents a Forward Distortion and Backward Warping Network (FDBW-Net), a novel framework for wide-angle image rectification. It begins by using a forward distortion model to synthesize barrel-distorted images, reducing pixel redundancy and preventing blur. The network employs a pyramid context encoder with attention mechanisms to generate backward warping flows containing geometric details. Then, a multi-scale decoder is used to restore distorted features and output rectified images. FDBW-Net's performance is validated on diverse datasets: public benchmarks, AirSim-rendered PTZ camera imagery, and real-scene PTZ camera datasets. It demonstrates that FDBW-Net achieves SOTA performance in distortion rectification, boosting the adaptability of PTZ cameras for practical visual applications.
Authors: Daiwei Zhang, Joaquin Gajardo, Tomislav Medic, Isinsu Katircioglu, Mike Boss, Norbert Kirchgessner, Achim Walter, Lukas Roth
Abstract: Automated extraction of plant morphological traits is crucial for supporting crop breeding and agricultural management through high-throughput field phenotyping (HTFP). Solutions based on multi-view RGB images are attractive due to their scalability and affordability, enabling volumetric measurements that 2D approaches cannot directly capture. While advanced methods like Neural Radiance Fields (NeRFs) have shown promise, their application has been limited to counting or extracting traits from only a few plants or organs. Furthermore, accurately measuring complex structures like individual wheat heads-essential for studying crop yields-remains particularly challenging due to occlusions and the dense arrangement of crop canopies in field conditions. The recent development of 3D Gaussian Splatting (3DGS) offers a promising alternative for HTFP due to its high-quality reconstructions and explicit point-based representation. In this paper, we present Wheat3DGS, a novel approach that leverages 3DGS and the Segment Anything Model (SAM) for precise 3D instance segmentation and morphological measurement of hundreds of wheat heads automatically, representing the first application of 3DGS to HTFP. We validate the accuracy of wheat head extraction against high-resolution laser scan data, obtaining per-instance mean absolute percentage errors of 15.1%, 18.3%, and 40.2% for length, width, and volume. We provide additional comparisons to NeRF-based approaches and traditional Muti-View Stereo (MVS), demonstrating superior results. Our approach enables rapid, non-destructive measurements of key yield-related traits at scale, with significant implications for accelerating crop breeding and improving our understanding of wheat development.
Authors: Yuhang Yang, Fengqi Liu, Yixing Lu, Qin Zhao, Pingyu Wu, Wei Zhai, Ran Yi, Yang Cao, Lizhuang Ma, Zheng-Jun Zha, Junting Dong
Abstract: 3D human digitization has long been a highly pursued yet challenging task. Existing methods aim to generate high-quality 3D digital humans from single or multiple views, but remain primarily constrained by current paradigms and the scarcity of 3D human assets. Specifically, recent approaches fall into several paradigms: optimization-based and feed-forward (both single-view regression and multi-view generation with reconstruction). However, they are limited by slow speed, low quality, cascade reasoning, and ambiguity in mapping low-dimensional planes to high-dimensional space due to occlusion and invisibility, respectively. Furthermore, existing 3D human assets remain small-scale, insufficient for large-scale training. To address these challenges, we propose a latent space generation paradigm for 3D human digitization, which involves compressing multi-view images into Gaussians via a UV-structured VAE, along with DiT-based conditional generation, we transform the ill-posed low-to-high-dimensional mapping problem into a learnable distribution shift, which also supports end-to-end inference. In addition, we employ the multi-view optimization approach combined with synthetic data to construct the HGS-1M dataset, which contains $1$ million 3D Gaussian assets to support the large-scale training. Experimental results demonstrate that our paradigm, powered by large-scale training, produces high-quality 3D human Gaussians with intricate textures, facial details, and loose clothing deformation.
Authors: Jonas Loos, Lorenz Linhardt
Abstract: Diffusion models have demonstrated remarkable capabilities in synthesizing realistic images, spurring interest in using their representations for various downstream tasks. To better understand the robustness of these representations, we analyze popular Stable Diffusion models using representational similarity and norms. Our findings reveal three phenomena: (1) the presence of a learned positional embedding in intermediate representations, (2) high-similarity corner artifacts, and (3) anomalous high-norm artifacts. These findings underscore the need to further investigate the properties of diffusion model representations before considering them for downstream tasks that require robust features. Project page: https://jonasloos.github.io/sd-representation-anomalies
URLs: https://jonasloos.github.io/sd-representation-anomalies
Authors: Bojian Wu, Yifan Peng, Ruizhen Hu, Xiaowei Zhou
Abstract: The challenge of image-based 3D reconstruction for glossy objects lies in separating diffuse and specular components on glossy surfaces from captured images, a task complicated by the ambiguity in discerning lighting conditions and material properties using RGB data alone. While state-of-the-art methods rely on tailored and/or high-end equipment for data acquisition, which can be cumbersome and time-consuming, this work introduces a scalable polarization-aided approach that employs cost-effective acquisition tools. By attaching a linear polarizer to readily available RGB cameras, multi-view polarization images can be captured without the need for advance calibration or precise measurements of the polarizer angle, substantially reducing system construction costs. The proposed approach represents polarimetric BRDF, Stokes vectors, and polarization states of object surfaces as neural implicit fields. These fields, combined with the polarizer angle, are retrieved by optimizing the rendering loss of input polarized images. By leveraging fundamental physical principles for the implicit representation of polarization rendering, our method demonstrates superiority over existing techniques through experiments in public datasets and real captured images on both reconstruction and novel view synthesis.
Authors: Ran Zhang, Xuanhua He, Ke Cao, Liu Liu, Li Zhang, Man Zhou, Jie Zhang
Abstract: Multi-modality image fusion aims to synthesize a single, comprehensive image from multiple source inputs. Traditional approaches, such as CNNs and GANs, offer efficiency but struggle to handle low-quality or complex inputs. Recent advances in text-guided methods leverage large model priors to overcome these limitations, but at the cost of significant computational overhead, both in memory and inference time. To address this challenge, we propose a novel framework for distilling large model priors, eliminating the need for text guidance during inference while dramatically reducing model size. Our framework utilizes a teacher-student architecture, where the teacher network incorporates large model priors and transfers this knowledge to a smaller student network via a tailored distillation process. Additionally, we introduce spatial-channel cross-fusion module to enhance the model's ability to leverage textual priors across both spatial and channel dimensions. Our method achieves a favorable trade-off between computational efficiency and fusion quality. The distilled network, requiring only 10\% of the parameters and inference time of the teacher network, retains 90\% of its performance and outperforms existing SOTA methods. Extensive experiments demonstrate the effectiveness of our approach. The implementation will be made publicly available as an open-source resource.
Authors: Jifang Wang, Xue Yang, Longyue Wang, Zhenran Xu, Yiyu Wang, Yaowei Wang, Weihua Luo, Kaifu Zhang, Baotian Hu, Min Zhang
Abstract: Conditional image generation has gained significant attention for its ability to personalize content. However, the field faces challenges in developing task-agnostic, reliable, and explainable evaluation metrics. This paper introduces CIGEval, a unified agentic framework for comprehensive evaluation of conditional image generation tasks. CIGEval utilizes large multimodal models (LMMs) as its core, integrating a multi-functional toolbox and establishing a fine-grained evaluation framework. Additionally, we synthesize evaluation trajectories for fine-tuning, empowering smaller LMMs to autonomously select appropriate tools and conduct nuanced analyses based on tool outputs. Experiments across seven prominent conditional image generation tasks demonstrate that CIGEval (GPT-4o version) achieves a high correlation of 0.4625 with human assessments, closely matching the inter-annotator correlation of 0.47. Moreover, when implemented with 7B open-source LMMs using only 2.3K training trajectories, CIGEval surpasses the previous GPT-4o-based state-of-the-art method. Case studies on GPT-4o image generation highlight CIGEval's capability in identifying subtle issues related to subject consistency and adherence to control guidance, indicating its great potential for automating evaluation of image generation tasks with human-level reliability.
Authors: Ruoyu Chen, Hua Zhang, Jingzhi Li, Li Liu, Zhen Huang, Xiaochun Cao
Abstract: The objective of few-shot object detection (FSOD) is to detect novel objects with few training samples. The core challenge of this task is how to construct a generalized feature space for novel categories with limited data on the basis of the base category space, which could adapt the learned detection model to unknown scenarios. However, limited by insufficient samples for novel categories, two issues still exist: (1) the features of the novel category are easily implicitly represented by the features of the base category, leading to inseparable classifier boundaries, (2) novel categories with fewer data are not enough to fully represent the distribution, where the model fine-tuning is prone to overfitting. To address these issues, we introduce the side information to alleviate the negative influences derived from the feature space and sample viewpoints and formulate a novel generalized feature representation learning method for FSOD. Specifically, we first utilize embedding side information to construct a knowledge matrix to quantify the semantic relationship between the base and novel categories. Then, to strengthen the discrimination between semantically similar categories, we further develop contextual semantic supervised contrastive learning which embeds side information. Furthermore, to prevent overfitting problems caused by sparse samples, a side-information guided region-aware masked module is introduced to augment the diversity of samples, which finds and abandons biased information that discriminates between similar categories via counterfactual explanation, and refines the discriminative representation space further. Extensive experiments using ResNet and ViT backbones on PASCAL VOC, MS COCO, LVIS V1, FSOD-1K, and FSVOD-500 benchmarks demonstrate that our model outperforms the previous state-of-the-art methods, significantly improving the ability of FSOD in most shots/splits.
Authors: Shi Pan, Jianan Chen, Maria Secrier
Abstract: Gene expression profiling provides critical insights into cellular heterogeneity, biological processes and disease mechanisms. There has been an increasing interest in computational approaches that can predict gene expression directly from digitalized histopathology images. While image foundation models have shown promise in a variety of pathology downstream analysis, their performances on gene-expression prediction are still limited. Explicitly incorporating information from the transcriptomic models can help image models to address domain shift, yet the fine-tuning and alignment of foundation models can be expensive. In the work, we propose Parameter Efficient Knowledge trAnsfer (PEKA), a novel framework that leverages Block-Affine Adaptation and integrates knowledge distillation and structure alignment losses for cross-modal knowledge transfer. We evaluated PEKA for gene expression prediction using multiple spatial transcriptomics datasets (comprising 206,123 image tiles with matched gene expression profiles) that encompassed various types of tissue. PEKA achieved at least 5\% performance improvement over baseline foundation models while also outperforming alternative parameter-efficient fine-tuning strategies. We will release the code, datasets and aligned models after peer-review to facilitate broader adoption and further development for parameter efficient model alignment.
Authors: Meien Li, Mark Stamp
Abstract: The high efficiency and quality of artwork generated by Artificial Intelligence (AI) has created new concerns and challenges for human artists. In particular, recent improvements in generative AI have made it difficult for people to distinguish between human-generated and AI-generated art. In this research, we consider the potential utility of various types of Machine Learning (ML) and Deep Learning (DL) models in distinguishing AI-generated artwork from human-generated artwork. We focus on three challenging artistic styles, namely, baroque, cubism, and expressionism. The learning models we test are Logistic Regression (LR), Support Vector Machine (SVM), Multilayer Perceptron (MLP), and Convolutional Neural Network (CNN). Our best experimental results yield a multiclass accuracy of 0.8208 over six classes, and an impressive accuracy of 0.9758 for the binary classification problem of distinguishing AI-generated from human-generated art.
Authors: Mengchen Zhang, Tong Wu, Jing Tan, Ziwei Liu, Gordon Wetzstein, Dahua Lin
Abstract: Camera trajectory design plays a crucial role in video production, serving as a fundamental tool for conveying directorial intent and enhancing visual storytelling. In cinematography, Directors of Photography meticulously craft camera movements to achieve expressive and intentional framing. However, existing methods for camera trajectory generation remain limited: Traditional approaches rely on geometric optimization or handcrafted procedural systems, while recent learning-based methods often inherit structural biases or lack textual alignment, constraining creative synthesis. In this work, we introduce an auto-regressive model inspired by the expertise of Directors of Photography to generate artistic and expressive camera trajectories. We first introduce DataDoP, a large-scale multi-modal dataset containing 29K real-world shots with free-moving camera trajectories, depth maps, and detailed captions in specific movements, interaction with the scene, and directorial intent. Thanks to the comprehensive and diverse database, we further train an auto-regressive, decoder-only Transformer for high-quality, context-aware camera movement generation based on text guidance and RGBD inputs, named GenDoP. Extensive experiments demonstrate that compared to existing methods, GenDoP offers better controllability, finer-grained trajectory adjustments, and higher motion stability. We believe our approach establishes a new standard for learning-based cinematography, paving the way for future advancements in camera control and filmmaking. Our project website: https://kszpxxzmc.github.io/GenDoP/.
Authors: Yiting Lu, Jiakang Yuan, Zhen Li, Shitian Zhao, Qi Qin, Xinyue Li, Le Zhuo, Licheng Wen, Dongyang Liu, Yuewen Cao, Xiangchao Yan, Xin Li, Botian Shi, Tao Chen, Zhibo Chen, Lei Bai, Bo Zhang, Peng Gao
Abstract: We propose OmniCaptioner, a versatile visual captioning framework for generating fine-grained textual descriptions across a wide variety of visual domains. Unlike prior methods limited to specific image types (e.g., natural images or geometric visuals), our framework provides a unified solution for captioning natural images, visual text (e.g., posters, UIs, textbooks), and structured visuals (e.g., documents, tables, charts). By converting low-level pixel information into semantically rich textual representations, our framework bridges the gap between visual and textual modalities. Our results highlight three key advantages: (i) Enhanced Visual Reasoning with LLMs, where long-context captions of visual modalities empower LLMs, particularly the DeepSeek-R1 series, to reason effectively in multimodal scenarios; (ii) Improved Image Generation, where detailed captions improve tasks like text-to-image generation and image transformation; and (iii) Efficient Supervised Fine-Tuning (SFT), which enables faster convergence with less data. We believe the versatility and adaptability of OmniCaptioner can offer a new perspective for bridging the gap between language and visual modalities.
Authors: Alexander Rubinstein, Ameya Prabhu, Matthias Bethge, Seong Joon Oh
Abstract: Object-centric learning (OCL) seeks to learn representations that only encode an object, isolated from other objects or background cues in a scene. This approach underpins various aims, including out-of-distribution (OOD) generalization, sample-efficient composition, and modeling of structured environments. Most research has focused on developing unsupervised mechanisms that separate objects into discrete slots in the representation space, evaluated using unsupervised object discovery. However, with recent sample-efficient segmentation models, we can separate objects in the pixel space and encode them independently. This achieves remarkable zero-shot performance on OOD object discovery benchmarks, is scalable to foundation models, and can handle a variable number of slots out-of-the-box. Hence, the goal of OCL methods to obtain object-centric representations has been largely achieved. Despite this progress, a key question remains: How does the ability to separate objects within a scene contribute to broader OCL objectives, such as OOD generalization? We address this by investigating the OOD generalization challenge caused by spurious background cues through the lens of OCL. We propose a novel, training-free probe called $\textbf{Object-Centric Classification with Applied Masks (OCCAM)}$, demonstrating that segmentation-based encoding of individual objects significantly outperforms slot-based OCL methods. However, challenges in real-world applications remain. We provide the toolbox for the OCL community to use scalable object-centric representations, and focus on practical applications and fundamental questions, such as understanding object perception in human cognition. Our code is available $\href{https://github.com/AlexanderRubinstein/OCCAM}{here}$.
Authors: Gene Chou, Wenqi Xian, Guandao Yang, Mohamed Abdelfattah, Bharath Hariharan, Noah Snavely, Ning Yu, Paul Debevec
Abstract: A versatile video depth estimation model should (1) be accurate and consistent across frames, (2) produce high-resolution depth maps, and (3) support real-time streaming. We propose FlashDepth, a method that satisfies all three requirements, performing depth estimation on a 2044x1148 streaming video at 24 FPS. We show that, with careful modifications to pretrained single-image depth models, these capabilities are enabled with relatively little data and training. We evaluate our approach across multiple unseen datasets against state-of-the-art depth models, and find that ours outperforms them in terms of boundary sharpness and speed by a significant margin, while maintaining competitive accuracy. We hope our model will enable various applications that require high-resolution depth, such as video editing, and online decision-making, such as robotics.
Authors: Jiali Cheng, Hadi Amiri
Abstract: Machine Unlearning removes specific knowledge about training data samples from an already trained model. It has significant practical benefits, such as purging private, inaccurate, or outdated information from trained models without the need for complete re-training. Unlearning within a multimodal setting presents unique challenges due to the complex dependencies between different data modalities and the expensive cost of training on large multimodal datasets and architectures. This paper presents the first machine unlearning approach for multimodal data and models, titled MultiDelete, which is designed to decouple associations between unimodal data points during unlearning without losing the overall representation strength of the trained model. MultiDelete advocates for three key properties for effective multimodal unlearning: (a): modality decoupling, which effectively decouples the association between individual unimodal data points marked for deletion, rendering them as unrelated data points, (b): multimodal knowledge retention, which retains the multimodal representation post-unlearning, and (c): unimodal knowledge retention, which retains the unimodal representation postunlearning. MultiDelete is efficient to train and is not constrained by using a strongly convex loss -- a common restriction among existing baselines. Experiments on two architectures and four datasets, including image-text and graph-text datasets, show that MultiDelete gains an average improvement of 17.6 points over best performing baseline in unlearning multimodal samples, can maintain the multimodal and unimodal knowledge of the original model post unlearning, and can provide better protection to unlearned data against adversarial attacks.
Authors: Jonas Br\"andli, Maurice Schneeberger, Lisa Herzog, Loran Avci, Nordin Dari, Martin H\"aansel, Hakim Baazaoui, Pascal B\"uhler, Susanne Wegener, Beate Sick
Abstract: Aim: This study aims to enhance interpretability and explainability of multi-modal prediction models integrating imaging and tabular patient data. Methods: We adapt the xAI methods Grad-CAM and Occlusion to multi-modal, partly interpretable deep transformation models (dTMs). DTMs combine statistical and deep learning approaches to simultaneously achieve state-of-the-art prediction performance and interpretable parameter estimates, such as odds ratios for tabular features. Based on brain imaging and tabular data from 407 stroke patients, we trained dTMs to predict functional outcome three months after stroke. We evaluated the models using different discriminatory metrics. The adapted xAI methods were used to generated explanation maps for identification of relevant image features and error analysis. Results: The dTMs achieve state-of-the-art prediction performance, with area under the curve (AUC) values close to 0.8. The most important tabular predictors of functional outcome are functional independence before stroke and NIHSS on admission, a neurological score indicating stroke severity. Explanation maps calculated from brain imaging dTMs for functional outcome highlighted critical brain regions such as the frontal lobe, which is known to be linked to age which in turn increases the risk for unfavorable outcomes. Similarity plots of the explanation maps revealed distinct patterns which give insight into stroke pathophysiology, support developing novel predictors of stroke outcome and enable to identify false predictions. Conclusion: By adapting methods for explanation maps to dTMs, we enhanced the explainability of multi-modal and partly interpretable prediction models. The resulting explanation maps facilitate error analysis and support hypothesis generation regarding the significance of specific image regions in outcome prediction.
Authors: Mohsen Jenadeleh, Jon Sneyers, Panqi Jia, Shima Mohammadi, Joao Ascenso, Dietmar Saupe
Abstract: Learning-based image compression methods have recently emerged as promising alternatives to traditional codecs, offering improved rate-distortion performance and perceptual quality. JPEG AI represents the latest standardized framework in this domain, leveraging deep neural networks for high-fidelity image reconstruction. In this study, we present a comprehensive subjective visual quality assessment of JPEG AI-compressed images using the JPEG AIC-3 methodology, which quantifies perceptual differences in terms of Just Noticeable Difference (JND) units. We generated a dataset of 50 compressed images with fine-grained distortion levels from five diverse sources. A large-scale crowdsourced experiment collected 96,200 triplet responses from 459 participants. We reconstructed JND-based quality scales using a unified model based on boosted and plain triplet comparisons. Additionally, we evaluated the alignment of multiple objective image quality metrics with human perception in the high-fidelity range. The CVVDP metric achieved the overall highest performance; however, most metrics including CVVDP were overly optimistic in predicting the quality of JPEG AI-compressed images. These findings emphasize the necessity for rigorous subjective evaluations in the development and benchmarking of modern image codecs, particularly in the high-fidelity range. Another technical contribution is the introduction of the well-known Meng-Rosenthal-Rubin statistical test to the field of Quality of Experience research. This test can reliably assess the significance of difference in performance of quality metrics in terms of correlation between metrics and ground truth. The complete dataset, including all subjective scores, is publicly available at https://github.com/jpeg-aic/dataset-JPEG-AI-SDR25.
Authors: Matvei Popov, Aymen Kallala, Anirudha Ramesh, Narimane Hennouni, Shivesh Khaitan, Rick Gentry, Alain-Sam Cohen
Abstract: Long-range dependencies are critical for understanding genomic structure and function, yet most conventional methods struggle with them. Widely adopted transformer-based models, while excelling at short-context tasks, are limited by the attention module's quadratic computational complexity and inability to extrapolate to sequences longer than those seen in training. In this work, we explore State Space Models (SSMs) as a promising alternative by benchmarking two SSM-inspired architectures, Caduceus and Hawk, on long-range genomics modeling tasks under conditions parallel to a 50M parameter transformer baseline. We discover that SSMs match transformer performance and exhibit impressive zero-shot extrapolation across multiple tasks, handling contexts 10 to 100 times longer than those seen during training, indicating more generalizable representations better suited for modeling the long and complex human genome. Moreover, we demonstrate that these models can efficiently process sequences of 1M tokens on a single GPU, allowing for modeling entire genomic regions at once, even in labs with limited compute. Our findings establish SSMs as efficient and scalable for long-context genomic analysis.
Authors: Paul Roetzer, Florian Bernard
Abstract: Geometric consistency, i.e. the preservation of neighbourhoods, is a natural and strong prior in 3D shape matching. Geometrically consistent matchings are crucial for many downstream applications, such as texture transfer or statistical shape modelling. Yet, in practice, geometric consistency is often overlooked, or only achieved under severely limiting assumptions (e.g. a good initialisation). In this work, we propose a novel formalism for computing globally optimal and geometrically consistent matchings between 3D shapes which is scalable in practice. Our key idea is to represent the surface of the source shape as a collection of cyclic paths, which are then consistently matched to the target shape. Mathematically, we construct a hyper product graph (between source and target shape), and then cast 3D shape matching as a minimum-cost circulation flow problem in this hyper graph, which yields global geometrically consistent matchings between both shapes. We empirically show that our formalism is efficiently solvable and that it leads to high-quality results.
Authors: Jiali Cheng, Hadi Amiri
Abstract: Machine Unlearning aims to remove undesired information from trained models without requiring full retraining from scratch. Despite recent advancements, their underlying loss landscapes and optimization dynamics received less attention. In this paper, we investigate and analyze machine unlearning through the lens of mode connectivity - the phenomenon where independently trained models can be connected by smooth low-loss paths in the parameter space. We define and study mode connectivity in unlearning across a range of overlooked conditions, including connections between different unlearning methods, models trained with and without curriculum learning, and models optimized with first-order and secondorder techniques. Our findings show distinct patterns of fluctuation of different evaluation metrics along the curve, as well as the mechanistic (dis)similarity between unlearning methods. To the best of our knowledge, this is the first study on mode connectivity in the context of machine unlearning.
Authors: Huzaifa Arif, Keerthiram Murugesan, Payel Das, Alex Gittens, Pin-Yu Chen
Abstract: This paper explores inference-time data leakage risks of deep neural networks (NNs), where a curious and honest model service provider is interested in retrieving users' private data inputs solely based on the model inference results. Particularly, we revisit residual NNs due to their popularity in computer vision and our hypothesis that residual blocks are a primary cause of data leakage owing to the use of skip connections. By formulating inference-time data leakage as a constrained optimization problem, we propose a novel backward feature inversion method, \textbf{PEEL}, which can effectively recover block-wise input features from the intermediate output of residual NNs. The surprising results in high-quality input data recovery can be explained by the intuition that the output from these residual blocks can be considered as a noisy version of the input and thus the output retains sufficient information for input recovery. We demonstrate the effectiveness of our layer-by-layer feature inversion method on facial image datasets and pre-trained classifiers. Our results show that PEEL outperforms the state-of-the-art recovery methods by an order of magnitude when evaluated by mean squared error (MSE). The code is available at \href{https://github.com/Huzaifa-Arif/PEEL}{https://github.com/Huzaifa-Arif/PEEL}
URLs: https://github.com/Huzaifa-Arif/PEEL, https://github.com/Huzaifa-Arif/PEEL
Authors: Adam McArthur, Stephanie Wichuk, Stephen Burnside, Andrew Kirby, Alexander Scammon, Damian Sol, Abhilash Hareendranathan, Jacob L. Jaremko
Abstract: Developmental dysplasia of the hip (DDH) poses significant diagnostic challenges, hindering timely intervention. Current screening methodologies lack standardization, and AI-driven studies suffer from reproducibility issues due to limited data and code availability. To address these limitations, we introduce Retuve, an open-source framework for multi-modality DDH analysis, encompassing both ultrasound (US) and X-ray imaging. Retuve provides a complete and reproducible workflow, offering open datasets comprising expert-annotated US and X-ray images, pre-trained models with training code and weights, and a user-friendly Python Application Programming Interface (API). The framework integrates segmentation and landmark detection models, enabling automated measurement of key diagnostic parameters such as the alpha angle and acetabular index. By adhering to open-source principles, Retuve promotes transparency, collaboration, and accessibility in DDH research. This initiative has the potential to democratize DDH screening, facilitate early diagnosis, and ultimately improve patient outcomes by enabling widespread screening and early intervention. The GitHub repository/code can be found here: https://github.com/radoss-org/retuve
Authors: Yashil Sukurdeep, Fausto Navarro, Tam\'as Budav\'ari
Abstract: Recovering high-fidelity images of the night sky from blurred observations is a fundamental problem in astronomy, where traditional methods typically fall short. In ground-based astronomy, combining multiple exposures to enhance signal-to-noise ratios is further complicated by variations in the point-spread function caused by atmospheric turbulence. In this work, we present a self-supervised multi-frame method, based on deep image priors, for denoising, deblurring, and coadding ground-based exposures. Central to our approach is a carefully designed convolutional neural network that integrates information across multiple observations and enforces physically motivated constraints. We demonstrate the method's potential by processing Hyper Suprime-Cam exposures, yielding promising preliminary results with sharper restored images.
Authors: Julian Nubert, Turcan Tuna, Jonas Frey, Cesar Cadena, Katherine J. Kuchenbecker, Shehryar Khattak, Marco Hutter
Abstract: Seamless operation of mobile robots in challenging environments requires low-latency local motion estimation (e.g., dynamic maneuvers) and accurate global localization (e.g., wayfinding). While most existing sensor-fusion approaches are designed for specific scenarios, this work introduces a flexible open-source solution for task- and setup-agnostic multimodal sensor fusion that is distinguished by its generality and usability. Holistic Fusion formulates sensor fusion as a combined estimation problem of i) the local and global robot state and ii) a (theoretically unlimited) number of dynamic context variables, including automatic alignment of reference frames; this formulation fits countless real-world applications without any conceptual modifications. The proposed factor-graph solution enables the direct fusion of an arbitrary number of absolute, local, and landmark measurements expressed with respect to different reference frames by explicitly including them as states in the optimization and modeling their evolution as random walks. Moreover, local smoothness and consistency receive particular attention to prevent jumps in the robot state belief. HF enables low-latency and smooth online state estimation on typical robot hardware while simultaneously providing low-drift global localization at the IMU measurement rate. The efficacy of this released framework is demonstrated in five real-world scenarios on three robotic platforms, each with distinct task requirements.
Authors: Yun Chang, Leonor Fermoselle, Duy Ta, Bernadette Bucher, Luca Carlone, Jiuguang Wang
Abstract: While recent work in scene reconstruction and understanding has made strides in grounding natural language to physical 3D environments, it is still challenging to ground abstract, high-level instructions to a 3D scene. High-level instructions might not explicitly invoke semantic elements in the scene, and even the process of breaking a high-level task into a set of more concrete subtasks, a process called hierarchical task analysis, is environment-dependent. In this work, we propose ASHiTA, the first framework that generates a task hierarchy grounded to a 3D scene graph by breaking down high-level tasks into grounded subtasks. ASHiTA alternates LLM-assisted hierarchical task analysis, to generate the task breakdown, with task-driven 3D scene graph construction to generate a suitable representation of the environment. Our experiments show that ASHiTA performs significantly better than LLM baselines in breaking down high-level tasks into environment-dependent subtasks and is additionally able to achieve grounding performance comparable to state-of-the-art methods.
Authors: Jaehong Chung, Wei Cai, Tapan Mukerji
Abstract: This study proposes a systematic image registration approach to align 2D optical thin-section images within a 3D digital rock volume. Using template image matching with differential evolution optimization, we identify the most similar 2D plane in 3D. The method is validated on a synthetic porous medium, achieving exact registration, and applied to Berea sandstone, where it achieves a structural similarity index (SSIM) of 0.990. With the registered images, we explore upscaling properties based on paired multimodal images, focusing on pore characteristics and effective elastic moduli. The thin-section image reveals 50 % more porosity and submicron pores than the registered CT plane. In addition, bulk and shear moduli from thin sections are 25 % and 30 % lower, respectively, than those derived from CT images. Beyond numerical comparisons, thin sections provide additional geological insights, including cementation, mineral phases, and weathering effects, which are not clear in CT images. This study demonstrates the potential of multimodal image registration to improve computed rock properties in digital rock physics by integrating complementary imaging modalities.
Authors: Sumeyye Meryem Tasyurek, Tugce Kiziltepe, Hacer Yalim Keles
Abstract: In this work, we propose a simple gloss-free, transformer-based sign language production (SLP) framework that directly maps spoken-language text to sign pose sequences. We first train a pose autoencoder that encodes sign poses into a compact latent space using an articulator-based disentanglement strategy, where features corresponding to the face, right hand, left hand, and body are modeled separately to promote structured and interpretable representation learning. Next, a non-autoregressive transformer decoder is trained to predict these latent representations from sentence-level text embeddings. To guide this process, we apply channel-aware regularization by aligning predicted latent distributions with priors extracted from the ground-truth encodings using a KL-divergence loss. The contribution of each channel to the loss is weighted according to its associated articulator region, enabling the model to account for the relative importance of different articulators during training. Our approach does not rely on gloss supervision or pretrained models, and achieves state-of-the-art results on the PHOENIX14T dataset using only a modest training set.
Authors: Alexandre Banks, Richard Cook, Septimiu E. Salcudean
Abstract: Augmented reality (AR) is an effective tool in robotic surgery education as it combines exploratory learning with three-dimensional guidance. However, existing AR systems require expert supervision and do not account for differences in the mentor and mentee robot configurations. To enable novices to train outside the operating room while receiving expert-informed guidance, we present dV-STEAR: an open-source system that plays back task-aligned expert demonstrations without assuming identical setup joint positions between expert and novice. Pose estimation was rigorously quantified, showing a registration error of 3.86 (SD=2.01)mm. In a user study (N=24), dV-STEAR significantly improved novice performance on tasks from the Fundamentals of Laparoscopic Surgery. In a single-handed ring-over-wire task, dV-STEAR increased completion speed (p=0.03) and reduced collision time (p=0.01) compared to dry-lab training alone. During a pick-and-place task, it improved success rates (p=0.004). Across both tasks, participants using dV-STEAR exhibited significantly more balanced hand use and reported lower frustration levels. This work presents a novel educational tool implemented on the da Vinci Research Kit, demonstrates its effectiveness in teaching novices, and builds the foundation for further AR integration into robot-assisted surgery.
Authors: Yoshihiro Yamada
Abstract: Transformers have driven remarkable breakthroughs in natural language processing and computer vision, yet their standard attention mechanism still imposes O(N^2) complexity, hindering scalability to longer sequences. We introduce Circular-convolutional ATtention (CAT), a Fourier-based approach that efficiently applies circular convolutions to reduce complexity without sacrificing representational power. CAT achieves O(NlogN) computations, requires fewer learnable parameters by streamlining fully-connected layers, and introduces no heavier operations, resulting in consistent accuracy improvements and about a 10% speedup in naive PyTorch implementations on large-scale benchmarks such as ImageNet-1k and WikiText-103. Grounded in an engineering-isomorphism framework, CAT's design not only offers practical efficiency and ease of implementation but also provides insights to guide the development of next-generation, high-performance Transformer architectures. Finally, our ablation studies highlight the key conditions underlying CAT's success, shedding light on broader principles for scalable attention mechanisms.
Authors: Paolo Angella, Luca Balbi, Fabrizio Ferrando, Paolo Traverso, Rosario Varriale, Vito Paolo Pastore, Matteo Santacesaria
Abstract: Motion artifacts remain a significant challenge in Magnetic Resonance Imaging (MRI), compromising diagnostic quality and potentially leading to misdiagnosis or repeated scans. Existing deep learning approaches for motion artifact correction typically require paired motion-free and motion-affected images for training, which are rarely available in clinical settings. To overcome this requirement, we present DIMA (DIffusing Motion Artifacts), a novel framework that leverages diffusion models to enable unsupervised motion artifact correction in brain MRI. Our two-phase approach first trains a diffusion model on unpaired motion-affected images to learn the distribution of motion artifacts. This model then generates realistic motion artifacts on clean images, creating paired datasets suitable for supervised training of correction networks. Unlike existing methods, DIMA operates without requiring k-space manipulation or detailed knowledge of MRI sequence parameters, making it adaptable across different scanning protocols and hardware. Comprehensive evaluations across multiple datasets and anatomical planes demonstrate that our method achieves comparable performance to state-of-the-art supervised approaches while offering superior generalizability to real clinical data. DIMA represents a significant advancement in making motion artifact correction more accessible for routine clinical use, potentially reducing the need for repeat scans and improving diagnostic accuracy.
Authors: Seunghyeok Back, Joosoon Lee, Kangmin Kim, Heeseon Rho, Geonhyup Lee, Raeyoung Kang, Sangbeom Lee, Sangjun Noh, Youngjin Lee, Taeyeop Lee, Kyoobin Lee
Abstract: Robust grasping in cluttered environments remains an open challenge in robotics. While benchmark datasets have significantly advanced deep learning methods, they mainly focus on simplistic scenes with light occlusion and insufficient diversity, limiting their applicability to practical scenarios. We present GraspClutter6D, a large-scale real-world grasping dataset featuring: (1) 1,000 highly cluttered scenes with dense arrangements (14.1 objects/scene, 62.6\% occlusion), (2) comprehensive coverage across 200 objects in 75 environment configurations (bins, shelves, and tables) captured using four RGB-D cameras from multiple viewpoints, and (3) rich annotations including 736K 6D object poses and 9.3B feasible robotic grasps for 52K RGB-D images. We benchmark state-of-the-art segmentation, object pose estimation, and grasping detection methods to provide key insights into challenges in cluttered environments. Additionally, we validate the dataset's effectiveness as a training resource, demonstrating that grasping networks trained on GraspClutter6D significantly outperform those trained on existing datasets in both simulation and real-world experiments. The dataset, toolkit, and annotation tools are publicly available on our project website: https://sites.google.com/view/graspclutter6d.
Authors: Wuyang Liu, Yi Chai, Yongpeng Yan, Yanzhen Ren
Abstract: Audio-visual event localization (AVEL) plays a critical role in multimodal scene understanding. While existing datasets for AVEL predominantly comprise landscape-oriented long videos with clean and simple audio context, short videos have become the primary format of online video content due to the the proliferation of smartphones. Short videos are characterized by portrait-oriented framing and layered audio compositions (e.g., overlapping sound effects, voiceovers, and music), which brings unique challenges unaddressed by conventional methods. To this end, we introduce AVE-PM, the first AVEL dataset specifically designed for portrait mode short videos, comprising 25,335 clips that span 86 fine-grained categories with frame-level annotations. Beyond dataset creation, our empirical analysis shows that state-of-the-art AVEL methods suffer an average 18.66% performance drop during cross-mode evaluation. Further analysis reveals two key challenges of different video formats: 1) spatial bias from portrait-oriented framing introduces distinct domain priors, and 2) noisy audio composition compromise the reliability of audio modality. To address these issues, we investigate optimal preprocessing recipes and the impact of background music for AVEL on portrait mode videos. Experiments show that these methods can still benefit from tailored preprocessing and specialized model design, thus achieving improved performance. This work provides both a foundational benchmark and actionable insights for advancing AVEL research in the era of mobile-centric video content. Dataset and code will be released.
Authors: Miro Miranda, Francisco Mena, Andreas Dengel
Abstract: Missing instances in time series data impose a significant challenge to deep learning models, particularly in regression tasks. In the Earth Observation field, satellite failure or cloud occlusion frequently results in missing time-steps, introducing uncertainties in the predicted output and causing a decline in predictive performance. While many studies address missing time-steps through data augmentation to improve model robustness, the uncertainty arising at the input level is commonly overlooked. To address this gap, we introduce Monte Carlo Temporal Dropout (MC-TD), a method that explicitly accounts for input-level uncertainty by randomly dropping time-steps during inference using a predefined dropout ratio, thereby simulating the effect of missing data. To bypass the need for costly searches for the optimal dropout ratio, we extend this approach with Monte Carlo Concrete Temporal Dropout (MC-ConcTD), a method that learns the optimal dropout distribution directly. Both MC-TD and MC-ConcTD are applied during inference, leveraging Monte Carlo sampling for uncertainty quantification. Experiments on three EO time-series datasets demonstrate that MC-ConcTD improves predictive performance and uncertainty calibration compared to existing approaches. Additionally, we highlight the advantages of adaptive dropout tuning over manual selection, making uncertainty quantification more robust and accessible for EO applications.
Authors: Anisa V. Prasad, Tejas Sudharshan Mathai, Pritam Mukherjee, Jianfei Liu, Ronald M. Summers
Abstract: An accurate segmentation of the pancreas on CT is crucial to identify pancreatic pathologies and extract imaging-based biomarkers. However, prior research on pancreas segmentation has primarily focused on modifying the segmentation model architecture or utilizing pre- and post-processing techniques. In this article, we investigate the utility of anatomical priors to enhance the segmentation performance of the pancreas. Two 3D full-resolution nnU-Net models were trained, one with 8 refined labels from the public PANORAMA dataset, and another that combined them with labels derived from the public TotalSegmentator (TS) tool. The addition of anatomical priors resulted in a 6\% increase in Dice score ($p < .001$) and a 36.5 mm decrease in Hausdorff distance for pancreas segmentation ($p < .001$). Moreover, the pancreas was always detected when anatomy priors were used, whereas there were 8 instances of failed detections without their use. The use of anatomy priors shows promise for pancreas segmentation and subsequent derivation of imaging biomarkers.
Authors: Tejas Sudharshan Mathai, Benjamin Hou, Ronald M. Summers
Abstract: In the U.S., lung cancer is the second major cause of death. Early detection of suspicious lung nodules is crucial for patient treatment planning, management, and improving outcomes. Many approaches for lung nodule segmentation and volumetric analysis have been proposed, but few have looked at longitudinal changes in total lung tumor burden. In this work, we trained two 3D models (nnUNet) with and without anatomical priors to automatically segment lung lesions and quantified total lesion burden for each patient. The 3D model without priors significantly outperformed ($p < .001$) the model trained with anatomy priors. For detecting clinically significant lesions $>$ 1cm, a precision of 71.3\%, sensitivity of 68.4\%, and F1-score of 69.8\% was achieved. For segmentation, a Dice score of 77.1 $\pm$ 20.3 and Hausdorff distance error of 11.7 $\pm$ 24.1 mm was obtained. The median lesion burden was 6.4 cc (IQR: 2.1, 18.1) and the median volume difference between manual and automated measurements was 0.02 cc (IQR: -2.8, 1.2). Agreements were also evaluated with linear regression and Bland-Altman plots. The proposed approach can produce a personalized evaluation of the total tumor burden for a patient and facilitate interval change tracking over time.
Authors: Yu Qi, Yuanchen Ju, Tianming Wei, Chi Chu, Lawson L. S. Wong, Huazhe Xu
Abstract: 3D assembly tasks, such as furniture assembly and component fitting, play a crucial role in daily life and represent essential capabilities for future home robots. Existing benchmarks and datasets predominantly focus on assembling geometric fragments or factory parts, which fall short in addressing the complexities of everyday object interactions and assemblies. To bridge this gap, we present 2BY2, a large-scale annotated dataset for daily pairwise objects assembly, covering 18 fine-grained tasks that reflect real-life scenarios, such as plugging into sockets, arranging flowers in vases, and inserting bread into toasters. 2BY2 dataset includes 1,034 instances and 517 pairwise objects with pose and symmetry annotations, requiring approaches that align geometric shapes while accounting for functional and spatial relationships between objects. Leveraging the 2BY2 dataset, we propose a two-step SE(3) pose estimation method with equivariant features for assembly constraints. Compared to previous shape assembly methods, our approach achieves state-of-the-art performance across all 18 tasks in the 2BY2 dataset. Additionally, robot experiments further validate the reliability and generalization ability of our method for complex 3D assembly tasks.
Authors: Omar Alama, Avigyan Bhattacharya, Haoyang He, Seungchan Kim, Yuheng Qiu, Wenshan Wang, Cherie Ho, Nikhil Keetha, Sebastian Scherer
Abstract: Open-set semantic mapping is crucial for open-world robots. Current mapping approaches either are limited by the depth range or only map beyond-range entities in constrained settings, where overall they fail to combine within-range and beyond-range observations. Furthermore, these methods make a trade-off between fine-grained semantics and efficiency. We introduce RayFronts, a unified representation that enables both dense and beyond-range efficient semantic mapping. RayFronts encodes task-agnostic open-set semantics to both in-range voxels and beyond-range rays encoded at map boundaries, empowering the robot to reduce search volumes significantly and make informed decisions both within & beyond sensory range, while running at 8.84 Hz on an Orin AGX. Benchmarking the within-range semantics shows that RayFronts's fine-grained image encoding provides 1.34x zero-shot 3D semantic segmentation performance while improving throughput by 16.5x. Traditionally, online mapping performance is entangled with other system components, complicating evaluation. We propose a planner-agnostic evaluation framework that captures the utility for online beyond-range search and exploration, and show RayFronts reduces search volume 2.2x more efficiently than the closest online baselines.
Authors: Israfel Salazar, Manuel Fern\'andez Burda, Shayekh Bin Islam, Arshia Soltani Moakhar, Shivalika Singh, Fabian Farestam, Angelika Romanou, Danylo Boiko, Dipika Khullar, Mike Zhang, Dominik Krzemi\'nski, Jekaterina Novikova, Lu\'isa Shimabucoro, Joseph Marvin Imperial, Rishabh Maheshwary, Sharad Duwal, Alfonso Amayuelas, Swati Rajwal, Jebish Purbey, Ahmed Ruby, Nicholas Popovi\v{c}, Marek Suppa, Azmine Toushik Wasi, Ram Mohan Rao Kadiyala, Olga Tsymboi, Maksim Kostritsya, Bardia Soltani Moakhar, Gabriel da Costa Merlin, Ot\'avio Ferracioli Coletti, Maral Jabbari Shiviari, MohammadAmin farahani fard, Silvia Fernandez, Mar\'ia Grandury, Dmitry Abulkhanov, Drishti Sharma, Andre Guarnier De Mitri, Leticia Bossatto Marchezi, Johan Obando-Ceron, Nazar Kohut, Beyza Ermis, Desmond Elliott, Enzo Ferrante, Sara Hooker, Marzieh Fadaee
Abstract: The evaluation of vision-language models (VLMs) has mainly relied on English-language benchmarks, leaving significant gaps in both multilingual and multicultural coverage. While multilingual benchmarks have expanded, both in size and languages, many rely on translations of English datasets, failing to capture cultural nuances. In this work, we propose Kaleidoscope, as the most comprehensive exam benchmark to date for the multilingual evaluation of vision-language models. Kaleidoscope is a large-scale, in-language multimodal benchmark designed to evaluate VLMs across diverse languages and visual inputs. Kaleidoscope covers 18 languages and 14 different subjects, amounting to a total of 20,911 multiple-choice questions. Built through an open science collaboration with a diverse group of researchers worldwide, Kaleidoscope ensures linguistic and cultural authenticity. We evaluate top-performing multilingual vision-language models and find that they perform poorly on low-resource languages and in complex multimodal scenarios. Our results highlight the need for progress on culturally inclusive multimodal evaluation frameworks.
Authors: Boyuan Zheng, Michael Y. Fatemi, Xiaolong Jin, Zora Zhiruo Wang, Apurva Gandhi, Yueqi Song, Yu Gu, Jayanth Srinivasa, Gaowen Liu, Graham Neubig, Yu Su
Abstract: To survive and thrive in complex environments, humans have evolved sophisticated self-improvement mechanisms through environment exploration, hierarchical abstraction of experiences into reuseable skills, and collaborative construction of an ever-growing skill repertoire. Despite recent advancements, autonomous web agents still lack crucial self-improvement capabilities, struggling with procedural knowledge abstraction, refining skills, and skill composition. In this work, we introduce SkillWeaver, a skill-centric framework enabling agents to self-improve by autonomously synthesizing reusable skills as APIs. Given a new website, the agent autonomously discovers skills, executes them for practice, and distills practice experiences into robust APIs. Iterative exploration continually expands a library of lightweight, plug-and-play APIs, significantly enhancing the agent's capabilities. Experiments on WebArena and real-world websites demonstrate the efficacy of SkillWeaver, achieving relative success rate improvements of 31.8% and 39.8%, respectively. Additionally, APIs synthesized by strong agents substantially enhance weaker agents through transferable skills, yielding improvements of up to 54.3% on WebArena. These results demonstrate the effectiveness of honing diverse website interactions into APIs, which can be seamlessly shared among various web agents.
Authors: Jie Zhang, Minghui Nie, Changqing Zou, Jian Liu, Ligang Liu, Junjie Cao
Abstract: Although supervised deep normal estimators have recently shown impressive results on synthetic benchmarks, their performance deteriorates significantly in real-world scenarios due to the domain gap between synthetic and real data. Building high-quality real training data to boost those supervised methods is not trivial because point-wise annotation of normals for varying-scale real-world 3D scenes is a tedious and expensive task. This paper introduces PointNorm-Net, the first self-supervised deep learning framework to tackle this challenge. The key novelty of PointNorm-Net is a three-stage multi-modal normal distribution estimation paradigm that can be integrated into either deep or traditional optimization-based normal estimation frameworks. Extensive experiments show that our method achieves superior generalization and outperforms state-of-the-art conventional and deep learning approaches across three real-world datasets that exhibit distinct characteristics compared to the synthetic training data.
Authors: Andrei Voinea, Robin Kock, Maruf A. Dhali
Abstract: Losing pets can be highly distressing for pet owners, and finding a lost pet is often challenging and time-consuming. An artificial intelligence-based application can significantly improve the speed and accuracy of finding lost pets. To facilitate such an application, this study introduces a contrastive neural network model capable of accurately distinguishing between images of pets. The model was trained on a large dataset of dog images and evaluated through 3-fold cross-validation. Following 350 epochs of training, the model achieved a test accuracy of 90%. Furthermore, overfitting was avoided, as the test accuracy closely matched the training accuracy. Our findings suggest that contrastive neural network models hold promise as a tool for locating lost pets. This paper presents the foundational framework for a potential web application designed to assist users in locating their missing pets. The application will allow users to upload images of their lost pets and provide notifications when matching images are identified within its image database. This functionality aims to enhance the efficiency and accuracy with which pet owners can search for and reunite with their beloved animals.
Authors: Abhishek Ramanathapura Satyanarayana, Maruf A. Dhali
Abstract: Crude oil is an integral component of the world economy and transportation sectors. With the growing demand for crude oil due to its widespread applications, accidental oil spills are unfortunate yet unavoidable. Even though oil spills are difficult to clean up, the first and foremost challenge is to detect them. In this research, the authors test the feasibility of deep encoder-decoder models that can be trained effectively to detect oil spills remotely. The work examines and compares the results from several segmentation models on high dimensional satellite Synthetic Aperture Radar (SAR) image data to pave the way for further in-depth research. Multiple combinations of models are used to run the experiments. The best-performing model is the one with the ResNet-50 encoder and DeepLabV3+ decoder. It achieves a mean Intersection over Union (IoU) of 64.868% and an improved class IoU of 61.549% for the ``oil spill" class when compared with the previous benchmark model, which achieved a mean IoU of 65.05% and a class IoU of 53.38% for the ``oil spill" class.
Authors: Siyu Ren, Junhui Hou, Xiaodong Chen, Hongkai Xiong, Wenping Wang
Abstract: Qualifying the discrepancy between 3D geometric models, which could be represented with either point clouds or triangle meshes, is a pivotal issue with board applications. Existing methods mainly focus on directly establishing the correspondence between two models and then aggregating point-wise distance between corresponding points, resulting in them being either inefficient or ineffective. In this paper, we propose DirDist, an efficient, effective, robust, and differentiable distance metric for 3D geometry data. Specifically, we construct DirDist based on the proposed implicit representation of 3D models, namely directional distance field (DDF), which defines the directional distances of 3D points to a model to capture its local surface geometry. We then transfer the discrepancy between two 3D geometric models as the discrepancy between their DDFs defined on an identical domain, naturally establishing model correspondence. To demonstrate the advantage of our DirDist, we explore various distance metric-driven 3D geometric modeling tasks, including template surface fitting, rigid registration, non-rigid registration, scene flow estimation and human pose optimization. Extensive experiments show that our DirDist achieves significantly higher accuracy under all tasks. As a generic distance metric, DirDist has the potential to advance the field of 3D geometric modeling. The source code is available at https://github.com/rsy6318/DirDist.
Authors: Ant\'onio Filgueiras, Eduardo R. B. Marques, Lu\'is M. B. Lopes, Miguel Marques, Hugo Silva
Abstract: Machine-learning techniques, especially deep convolutional neural networks, are pivotal for image-based identification of biological species in many Citizen Science platforms. In this paper, we describe the construction of a dataset for the Portuguese native flora based on publicly available research-grade datasets, and the derivation of a high-accuracy model from it using off-the-shelf deep convolutional neural networks. We anchored the dataset in high-quality data provided by Sociedade Portuguesa de Bot\^anica and added further sampled data from research-grade datasets available from GBIF. We find that with a careful dataset design, off-the-shelf machine-learning cloud services such as Google's AutoML Vision produce accurate models, with results comparable to those of Pl@ntNet, a state-of-the-art citizen science platform. The best model we derived, dubbed Floralens, has been integrated into the public website of Project Biolens, where we gather models for other taxa as well. The dataset used to train the model is also publicly available on Zenodo.
Authors: Atsuyuki Miyai, Jingkang Yang, Jingyang Zhang, Yifei Ming, Qing Yu, Go Irie, Yixuan Li, Hai Li, Ziwei Liu, Kiyoharu Aizawa
Abstract: This paper introduces a novel task to evaluate the robust understanding capability of Large Multimodal Models (LMMs), termed $\textbf{Unsolvable Problem Detection (UPD)}$. Multiple-choice question answering (MCQA) is widely used to assess the understanding capability of LMMs, but it does not guarantee that LMMs truly comprehend the answer. UPD assesses the LMM's ability to withhold answers when encountering unsolvable problems of MCQA, verifying whether the model truly understands the answer. UPD encompasses three problems: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD), covering unsolvable cases like answer-lacking or incompatible choices and image-question mismatches. For the evaluation, we introduce the MM-UPD Bench, a benchmark for assessing performance across various ability dimensions. Our experiments reveal that even most LMMs, which demonstrate adequate performance on existing benchmarks, struggle significantly with MM-UPD, underscoring a novel aspect of trustworthiness that current benchmarks have overlooked. A detailed analysis shows that LMMs have different bottlenecks and chain-of-thought and self-reflection improved performance for LMMs with the bottleneck in their LLM capability. We hope our insights will enhance the broader understanding and development of more reliable LMMs.
Authors: Genjia Liu, Yue Hu, Chenxin Xu, Weibo Mao, Junhao Ge, Zhengxiang Huang, Yifan Lu, Yinda Xu, Junkai Xia, Yafei Wang, Siheng Chen
Abstract: Vehicle-to-everything-aided autonomous driving (V2X-AD) has a huge potential to provide a safer driving solution. Despite extensive researches in transportation and communication to support V2X-AD, the actual utilization of these infrastructures and communication resources in enhancing driving performances remains largely unexplored. This highlights the necessity of collaborative autonomous driving: a machine learning approach that optimizes the information sharing strategy to improve the driving performance of each vehicle. This effort necessitates two key foundations: a platform capable of generating data to facilitate the training and testing of V2X-AD, and a comprehensive system that integrates full driving-related functionalities with mechanisms for information sharing. From the platform perspective, we present V2Xverse, a comprehensive simulation platform for collaborative autonomous driving. This platform provides a complete pipeline for collaborative driving. From the system perspective, we introduce CoDriving, a novel end-to-end collaborative driving system that properly integrates V2X communication over the entire autonomous pipeline, promoting driving with shared perceptual information. The core idea is a novel driving-oriented communication strategy. Leveraging this strategy, CoDriving improves driving performance while optimizing communication efficiency. We make comprehensive benchmarks with V2Xverse, analyzing both modular performance and closed-loop driving performance. Experimental results show that CoDriving: i) significantly improves the driving score by 62.49% and drastically reduces the pedestrian collision rate by 53.50% compared to the SOTA end-to-end driving method, and ii) achieves sustaining driving performance superiority over dynamic constraint communication conditions.
Authors: Jiamin Wu, Kenkun Liu, Han Gao, Xiaoke Jiang, Yao Yuan, Lei Zhang
Abstract: Recently, Gaussian splatting has demonstrated significant success in novel view synthesis. Current methods often regress Gaussians with pixel or point cloud correspondence, linking each Gaussian with a pixel or a 3D point. This leads to the redundancy of Gaussians being used to overfit the correspondence rather than the objects represented by the 3D Gaussians themselves, consequently wasting resources and lacking accurate geometries or textures. In this paper, we introduce LeanGaussian, a novel approach that treats each query in deformable Transformer as one 3D Gaussian ellipsoid, breaking the pixel or point cloud correspondence constraints. We leverage deformable decoder to iteratively refine the Gaussians layer-by-layer with the image features as keys and values. Notably, the center of each 3D Gaussian is defined as 3D reference points, which are then projected onto the image for deformable attention in 2D space. On both the ShapeNet SRN dataset (category level) and the Google Scanned Objects dataset (open-category level, trained with the Objaverse dataset), our approach, outperforms prior methods by approximately 6.1%, achieving a PSNR of 25.44 and 22.36, respectively. Additionally, our method achieves a 3D reconstruction speed of 7.2 FPS and rendering speed 500 FPS. Codes are available at https://github.com/jwubz123/LeanGaussian.
Authors: Sanghyeon Na, Yonggyu Kim, Hyunjoon Lee
Abstract: Human image generation is a key focus in image synthesis due to its broad applications, but even slight inaccuracies in anatomy, pose, or details can compromise realism. To address these challenges, we explore Direct Preference Optimization (DPO), which trains models to generate preferred (winning) images while diverging from non-preferred (losing) ones. However, conventional DPO methods use generated images as winning images, limiting realism. To overcome this limitation, we propose an enhanced DPO approach that incorporates high-quality real images as winning images, encouraging outputs to resemble real images rather than generated ones. However, implementing this concept is not a trivial task. Therefore, our approach, HG-DPO (Human image Generation through DPO), employs a novel curriculum learning framework that gradually improves the output of the model toward greater realism, making training more feasible. Furthermore, HG-DPO effectively adapts to personalized text-to-image tasks, generating high-quality and identity-specific images, which highlights the practical value of our approach.
Authors: Dazhou Yu, Yuntong Hu, Yun Li, Liang Zhao
Abstract: Polygon representation learning is essential for diverse applications, encompassing tasks such as shape coding, building pattern classification, and geographic question answering. While recent years have seen considerable advancements in this field, much of the focus has been on single polygons, overlooking the intricate inner- and inter-polygonal relationships inherent in multipolygons. To address this gap, our study introduces a comprehensive framework specifically designed for learning representations of polygonal geometries, particularly multipolygons. Central to our approach is the incorporation of a heterogeneous visibility graph, which seamlessly integrates both inner- and inter-polygonal relationships. To enhance computational efficiency and minimize graph redundancy, we implement a heterogeneous spanning tree sampling method. Additionally, we devise a rotation-translation invariant geometric representation, ensuring broader applicability across diverse scenarios. Finally, we introduce Multipolygon-GNN, a novel model tailored to leverage the spatial and semantic heterogeneity inherent in the visibility graph. Experiments on five real-world and synthetic datasets demonstrate its ability to capture informative representations for polygonal geometries. Code and data are available at \href{https://github.com/dyu62/PolyGNN}{$github.com/dyu62/PolyGNN$}.
Authors: Xingyue Lin, Xingjian Hu, Shuai Peng, Jianhua Zhu, Liangcai Gao
Abstract: Sketching is a powerful artistic technique for capturing essential visual information about real-world objects and has increasingly attracted attention in image synthesis research. However, the field lacks a unified benchmark to evaluate the performance of various synthesis methods. To address this, we propose SketchRef, the first comprehensive multi-task evaluation benchmark for sketch synthesis. SketchRef fully leverages the shared characteristics between sketches and reference photos. It introduces two primary tasks: category prediction and structural consistency estimation, the latter being largely overlooked in previous studies. These tasks are further divided into five sub-tasks across four domains: animals, common things, human body, and faces. Recognizing the inherent trade-off between recognizability and simplicity in sketches, we are the first to quantify this balance by introducing a recognizability calculation method constrained by simplicity, mRS, ensuring fair and meaningful evaluations. To validate our approach, we collected 7,920 responses from art enthusiasts, confirming the effectiveness of our proposed evaluation metrics. Additionally, we evaluate the performance of existing sketch synthesis methods on our benchmark, highlighting their strengths and weaknesses. We hope this study establishes a standardized benchmark and offers valuable insights for advancing sketch synthesis algorithms.
Authors: Yang Nan, Huichi Zhou, Xiaodan Xing, Guang Yang
Abstract: Recent advancements in Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across diverse tasks, garnering significant attention in AI communities. However, their performance and reliability in specialized domains such as medicine remain insufficiently assessed. In particular, most assessments over-concentrate on evaluating VLMs based on simple Visual Question Answering (VQA) on multi-modality data, while ignoring the in-depth characteristics of LVLMs. In this study, we introduce RadVUQA, a novel Radiological Visual Understanding and Question Answering benchmark, to comprehensively evaluate existing LVLMs. RadVUQA mainly validates LVLMs across five dimensions: 1) Anatomical understanding, assessing the models' ability to visually identify biological structures; 2) Multimodal comprehension, which involves the capability of interpreting linguistic and visual instructions to produce desired outcomes; 3) Quantitative and spatial reasoning, evaluating the models' spatial awareness and proficiency in combining quantitative analysis with visual and linguistic information; 4) Physiological knowledge, measuring the models' capability to comprehend functions and mechanisms of organs and systems; and 5) Robustness, which assesses the models' capabilities against unharmonized and synthetic data. The results indicate that both generalized LVLMs and medical-specific LVLMs have critical deficiencies with weak multimodal comprehension and quantitative reasoning capabilities. Our findings reveal the large gap between existing LVLMs and clinicians, highlighting the urgent need for more robust and intelligent LVLMs. The code is available at https://github.com/Nandayang/RadVUQA
Authors: Haitao Yang, Yuan Dong, Hanwen Jiang, Dejia Xu, Georgios Pavlakos, Qixing Huang
Abstract: Using the latent diffusion model has proven effective in developing novel 3D generation techniques. To harness the latent diffusion model, a key challenge is designing a high-fidelity and efficient representation that links the latent space and the 3D space. In this paper, we introduce Atlas Gaussians, a novel representation for feed-forward native 3D generation. Atlas Gaussians represent a shape as the union of local patches, and each patch can decode 3D Gaussians. We parameterize a patch as a sequence of feature vectors and design a learnable function to decode 3D Gaussians from the feature vectors. In this process, we incorporate UV-based sampling, enabling the generation of a sufficiently large, and theoretically infinite, number of 3D Gaussian points. The large amount of 3D Gaussians enables the generation of high-quality details. Moreover, due to local awareness of the representation, the transformer-based decoding procedure operates on a patch level, ensuring efficiency. We train a variational autoencoder to learn the Atlas Gaussians representation, and then apply a latent diffusion model on its latent space for learning 3D Generation. Experiments show that our approach outperforms the prior arts of feed-forward native 3D generation. Project page: https://yanghtr.github.io/projects/atlas_gaussians.
Authors: Manu Gaur, Darshan Singh, Makarand Tapaswi
Abstract: Image captioning systems are unable to generate fine-grained captions as they are trained on data that is either noisy (alt-text) or generic (human annotations). This is further exacerbated by maximum likelihood training that encourages generation of frequently occurring phrases. Previous works have tried to address this limitation by fine-tuning captioners with a self-retrieval (SR) reward. However, we find that SR fine-tuning has a tendency to reduce caption faithfulness and even hallucinate. In this work, we circumvent this bottleneck by improving the MLE initialization of the captioning system and designing a curriculum for the SR fine-tuning process. To this extent, we present (1) Visual Caption Boosting, a novel framework to instill fine-grainedness in generic image captioning datasets while remaining anchored in human annotations; and (2) BagCurri, a carefully designed training curriculum that more optimally leverages the contrastive nature of the self-retrieval reward. Jointly, they enable the captioner to describe fine-grained aspects in the image while preserving faithfulness to ground-truth captions. Our approach outperforms previous work by +8.9% on SR against 99 random distractors (RD100) (Dessi et al., 2023); and +7.6% on ImageCoDe. Additionally, existing metrics to evaluate captioning systems fail to reward diversity or evaluate a model's fine-grained understanding ability. Our third contribution addresses this by proposing self-retrieval from the lens of evaluation. We introduce TrueMatch, a benchmark comprising bags of highly similar images that uses SR to assess the captioner's ability to capture subtle visual distinctions. We evaluate and compare several state-of-the-art open-source MLLMs on TrueMatch, and find that our SR approach outperforms them all by a significant margin (e.g. +4.8% - 7.1% over Cambrian) while having 1-2 orders of magnitude fewer parameters.
Authors: Md Ferdous Alam, Faez Ahmed
Abstract: The creation of manufacturable and editable 3D shapes through Computer-Aided Design (CAD) remains a highly manual and time-consuming task, hampered by the complex topology of boundary representations of 3D solids and unintuitive design tools. While most work in the 3D shape generation literature focuses on representations like meshes, voxels, or point clouds, practical engineering applications demand the modifiability and manufacturability of CAD models and the ability for multi-modal conditional CAD model generation. This paper introduces GenCAD, a generative model that employs autoregressive transformers with a contrastive learning framework and latent diffusion models to transform image inputs into parametric CAD command sequences, resulting in editable 3D shape representations. Extensive evaluations demonstrate that GenCAD significantly outperforms existing state-of-the-art methods in terms of the unconditional and conditional generations of CAD models. Additionally, the contrastive learning framework of GenCAD facilitates the retrieval of CAD models using image queries from large CAD databases, which is a critical challenge within the CAD community. Our results provide a significant step forward in highlighting the potential of generative models to expedite the entire design-to-production pipeline and seamlessly integrate different design modalities.
Authors: Wenhao Chai, Enxin Song, Yilun Du, Chenlin Meng, Vashisht Madhavan, Omer Bar-Tal, Jenq-Neng Hwang, Saining Xie, Christopher D. Manning
Abstract: Video detailed captioning is a key task which aims to generate comprehensive and coherent textual descriptions of video content, benefiting both video understanding and generation. In this paper, we propose AuroraCap, a video captioner based on a large multimodal model. We follow the simplest architecture design without additional parameters for temporal modeling. To address the overhead caused by lengthy video sequences, we implement the token merging strategy, reducing the number of input visual tokens. Surprisingly, we found that this strategy results in little performance loss. AuroraCap shows superior performance on various video and image captioning benchmarks, for example, obtaining a CIDEr of 88.9 on Flickr30k, beating GPT-4V (55.3) and Gemini-1.5 Pro (82.2). However, existing video caption benchmarks only include simple descriptions, consisting of a few dozen words, which limits research in this field. Therefore, we develop VDC, a video detailed captioning benchmark with over one thousand carefully annotated structured captions. In addition, we propose a new LLM-assisted metric VDCscore for bettering evaluation, which adopts a divide-and-conquer strategy to transform long caption evaluation into multiple short question-answer pairs. With the help of human Elo ranking, our experiments show that this benchmark better correlates with human judgments of video detailed captioning quality.
Authors: Ziyue Wang, Chi Chen, Fuwen Luo, Yurui Dong, Yuanchi Zhang, Yuzhuang Xu, Xiaolong Wang, Peng Li, Yang Liu
Abstract: Active perception, a crucial human capability, involves setting a goal based on the current understanding of the environment and performing actions to achieve that goal. Despite significant efforts in evaluating Multimodal Large Language Models (MLLMs), active perception has been largely overlooked. To address this gap, we propose a novel benchmark named ActiView to evaluate active perception in MLLMs. We focus on a specialized form of Visual Question Answering (VQA) that eases and quantifies the evaluation yet challenging for existing MLLMs. Meanwhile, intermediate reasoning behaviors of models are also discussed. Given an image, we restrict the perceptual field of a model, requiring it to actively zoom or shift its perceptual field based on reasoning to answer the question successfully. We conduct extensive evaluation over 30 models, including proprietary and open-source models, and observe that restricted perceptual fields play a significant role in enabling active perception. Results reveal a significant gap in the active perception capability of MLLMs, indicating that this area deserves more attention. We hope that ActiView could help develop methods for MLLMs to understand multimodal inputs in more natural and holistic ways.
Authors: Phoenix Yu, Tilo Burghardt, Andrew W Dowsey, Neill W Campbell
Abstract: We present MultiCamCows2024, a farm-scale image dataset filmed across multiple cameras for the biometric identification of individual Holstein-Friesian cattle exploiting their unique black and white coat-patterns. Captured by three ceiling-mounted visual sensors covering adjacent barn areas over seven days on a working dairy farm, the dataset comprises 101,329 images of 90 cows, plus underlying original CCTV footage. The dataset is provided with full computer vision recognition baselines, that is both a supervised and self-supervised learning framework for individual cow identification trained on cattle tracklets. We report a performance above 96% single image identification accuracy from the dataset and demonstrate that combining data from multiple cameras during learning enhances self-supervised identification. We show that our framework enables automatic cattle identification, barring only the simple human verification of tracklet integrity during data collection. Crucially, our study highlights that multi-camera, supervised and self-supervised components in tandem not only deliver highly accurate individual cow identification, but also achieve this efficiently with no labelling of cattle identities by humans. We argue that this improvement in efficacy has practical implications for livestock management, behaviour analysis, and agricultural monitoring. For reproducibility and practical ease of use, we publish all key software and code including re-identification components and the species detector with this paper, available at https://tinyurl.com/MultiCamCows2024.
Authors: Bo Han, Yuheng Jia, Hui Liu, Junhui Hou
Abstract: Spectral variations pose a common challenge in analyzing hyperspectral images (HSI). To address this, low-rank tensor representation has emerged as a robust strategy, leveraging inherent correlations within HSI data. However, the spatial distribution of ground objects in HSIs is inherently irregular, existing naturally in tensor format, with numerous class-specific regions manifesting as irregular tensors. Current low-rank representation techniques are designed for regular tensor structures and overlook this fundamental irregularity in real-world HSIs, leading to performance limitations. To tackle this issue, we propose a novel model for irregular tensor low-rank representation tailored to efficiently model irregular 3D cubes. By incorporating a non-convex nuclear norm to promote low-rankness and integrating a global negative low-rank term to enhance the discriminative ability, our proposed model is formulated as a constrained optimization problem and solved using an alternating augmented Lagrangian method. Experimental validation conducted on four public datasets demonstrates the superior performance of our method compared to existing state-of-the-art approaches. The code is publicly available at https://github.com/hb-studying/ITLRR.
Authors: Qiqi Hou, Randall Rauwendaal, Zifeng Li, Hoang Le, Farzad Farhadzadeh, Fatih Porikli, Alexei Bourd, Amir Said
Abstract: Recently, 3D Gaussian Splatting (3DGS) has emerged as a significant advancement in 3D scene reconstruction, attracting considerable attention due to its ability to recover high-fidelity details while maintaining low complexity. Despite the promising results achieved by 3DGS, its rendering performance is constrained by its dependence on costly non-commutative alpha-blending operations. These operations mandate complex view dependent sorting operations that introduce computational overhead, especially on the resource-constrained platforms such as mobile phones. In this paper, we propose Weighted Sum Rendering, which approximates alpha blending with weighted sums, thereby removing the need for sorting. This simplifies implementation, delivers superior performance, and eliminates the "popping" artifacts caused by sorting. Experimental results show that optimizing a generalized Gaussian splatting formulation to the new differentiable rendering yields competitive image quality. The method was implemented and tested in a mobile device GPU, achieving on average $1.23\times$ faster rendering.
Authors: Joseph Geo Benjamin, Mothilal Asokan, Mohammad Yaqub, Karthik Nandakumar
Abstract: One of the most common defense strategies against Byzantine clients in federated learning (FL) is to employ a robust aggregator mechanism that makes the training more resilient. While many existing Byzantine robust aggregators provide theoretical convergence guarantees and are empirically effective against certain categories of attacks, we observe that certain high-strength attacks can subvert the robust aggregator and collapse the training. To overcome this limitation, we propose a method called FedSECA for robust Sign Election and Coordinate-wise Aggregation of gradients in FL that is less susceptible to malicious updates by an omniscient attacker. The proposed method has two main components. The Concordance Ratio Induced Sign Election(CRISE) module determines the consensus direction (elected sign) for each individual parameter gradient through a weighted voting strategy. The client weights are assigned based on a novel metric called concordance ratio, which quantifies the degree of sign agreement between the client gradient updates. Based on the elected sign, a Robust Coordinate-wise Aggregation(RoCA) strategy is employed, where variance-reduced sparse gradients are aggregated only if they are in alignment with the corresponding elected sign. We compare our proposed FedSECA method against 10 robust aggregators under 7 Byzantine attacks on 3 datasets and architectures. The results show that existing robust aggregators fail for at least some attacks, while FedSECA exhibits better robustness. Code - https://github.com/JosephGeoBenjamin/FedSECA-ByzantineTolerance
URLs: https://github.com/JosephGeoBenjamin/FedSECA-ByzantineTolerance
Authors: Hang Ye, Xiaoxuan Ma, Hai Ci, Wentao Zhu, Yizhou Wang
Abstract: Achieving realistic animated human avatars requires accurate modeling of pose-dependent clothing deformations. Existing learning-based methods heavily rely on the Linear Blend Skinning (LBS) of minimally-clothed human models like SMPL to model deformation. However, they struggle to handle loose clothing, such as long dresses, where the canonicalization process becomes ill-defined when the clothing is far from the body, leading to disjointed and fragmented results. To overcome this limitation, we propose FreeCloth, a novel hybrid framework to model challenging clothed humans. Our core idea is to use dedicated strategies to model different regions, depending on whether they are close to or distant from the body. Specifically, we segment the human body into three categories: unclothed, deformed, and generated. We simply replicate unclothed regions that require no deformation. For deformed regions close to the body, we leverage LBS to handle the deformation. As for the generated regions, which correspond to loose clothing areas, we introduce a novel free-form, part-aware generator to model them, as they are less affected by movements. This free-form generation paradigm brings enhanced flexibility and expressiveness to our hybrid framework, enabling it to capture the intricate geometric details of challenging loose clothing, such as skirts and dresses. Experimental results on the benchmark dataset featuring loose clothing demonstrate that FreeCloth achieves state-of-the-art performance with superior visual fidelity and realism, particularly in the most challenging cases.
Authors: Jiongtong Hu, Wei Zhuo, Jun Cheng, Yingying Liu, Wufeng Xue, Dong Ni
Abstract: In clinical practice of echocardiography examinations, multiple planes containing the heart structures of different view are usually required in screening, diagnosis and treatment of cardiac disease. AI models for echocardiography have to be tailored for each specific plane due to the dramatic structure differences, thus resulting in repetition development and extra complexity. Effective solution for such a multi-plane segmentation (MPS) problem is highly demanded for medical images, yet has not been well investigated. In this paper, we propose a novel solution, EchoONE, for this problem with a SAM-based segmentation architecture, a prior-composable mask learning (PC-Mask) module for semantic-aware dense prompt generation, and a learnable CNN-branch with a simple yet effective local feature fusion and adaption (LFFA) module for SAM adapting. We extensively evaluated our method on multiple internal and external echocardiography datasets, and achieved consistently state-of-the-art performance for multi-source datasets with different heart planes. This is the first time that the MPS problem is solved in one model for echocardiography data. The code will be available at https://github.com/a2502503/EchoONE.
Authors: Rao Fu, Dingxi Zhang, Alex Jiang, Wanjia Fu, Austin Funk, Daniel Ritchie, Srinath Sridhar
Abstract: Understanding bimanual human hand activities is a critical problem in AI and robotics. We cannot build large models of bimanual activities because existing datasets lack the scale, coverage of diverse hand activities, and detailed annotations. We introduce GigaHands, a massive annotated dataset capturing 34 hours of bimanual hand activities from 56 subjects and 417 objects, totaling 14k motion clips derived from 183 million frames paired with 84k text annotations. Our markerless capture setup and data acquisition protocol enable fully automatic 3D hand and object estimation while minimizing the effort required for text annotation. The scale and diversity of GigaHands enable broad applications, including text-driven action synthesis, hand motion captioning, and dynamic radiance field reconstruction. Our website are avaliable at https://ivl.cs.brown.edu/research/gigahands.html .
Authors: Luis Wiedmann, Luca Wiehe, David Rozenberszki
Abstract: Open-set 3D segmentation represents a major point of interest for multiple downstream robotics and augmented/virtual reality applications. We present a decoupled 3D segmentation pipeline to ensure modularity and adaptability to novel 3D representations as well as semantic segmentation foundation models. We first reconstruct a scene with 3D Gaussians and learn class-agnostic features through contrastive supervision from a 2D instance proposal network. These 3D features are then clustered to form coarse object- or part-level masks. Finally, we match each 3D cluster to class-aware masks predicted by a 2D open-vocabulary segmentation model, assigning semantic labels without retraining the 3D representation. Our decoupled design (1) provides a plug-and-play interface for swapping different 2D or 3D modules, (2) ensures multi-object instance segmentation at no extra cost, and (3) leverages rich 3D geometry for robust scene understanding. We evaluate on synthetic and real-world indoor datasets, demonstrating improved performance over comparable NeRF-based pipelines on mIoU and mAcc, particularly for challenging or long-tail classes. We also show how varying the 2D backbone affects the final segmentation, highlighting the modularity of our framework. These results confirm that decoupling 3D mask proposal and semantic classification can deliver flexible, efficient, and open-vocabulary 3D segmentation.
Authors: Burak Ekim, Girmaw Abebe Tadesse, Caleb Robinson, Gilles Hacheme, Michael Schmitt, Rahul Dodhia, Juan M. Lavista Ferres
Abstract: Training robust deep learning models is crucial in Earth Observation, where globally deployed models often face distribution shifts that degrade performance, especially in low-data regions. Out-of-distribution (OOD) detection addresses this by identifying inputs that deviate from in-distribution (ID) data. However, existing methods either assume access to OOD data or compromise primary task performance, limiting real-world use. We introduce TARDIS, a post-hoc OOD detection method designed for scalable geospatial deployment. Our core innovation lies in generating surrogate distribution labels by leveraging ID data within the feature space. TARDIS takes a pre-trained model, ID data, and data from an unknown distribution (WILD), separates WILD into surrogate ID and OOD labels based on internal activations, and trains a binary classifier to detect distribution shifts. We validate on EuroSAT and xBD across 17 setups covering covariate and semantic shifts, showing near-upper-bound surrogate labeling performance in 13 cases and matching the performance of top post-hoc activation- and scoring-based methods. Finally, deploying TARDIS on Fields of the World reveals actionable insights into pre-trained model behavior at scale. The code is available at \href{https://github.com/microsoft/geospatial-ood-detection}{https://github.com/microsoft/geospatial-ood-detection}
URLs: https://github.com/microsoft/geospatial-ood-detection, https://github.com/microsoft/geospatial-ood-detection
Authors: Yung-Hong Sun, Gefei Shen, Jiangang Chen, Jayer Fernandes, Amber L. Shada, Charles P. Heise, Hongrui Jiang, Yu Hen Hu
Abstract: EasyVis2 is a system designed to provide hands-free, real-time 3D visualization for laparoscopic surgery. It incorporates a surgical trocar equipped with an array of micro-cameras, which can be inserted into the body cavity to offer an enhanced field of view and a 3D perspective of the surgical procedure. A specialized deep neural network algorithm, YOLOv8-Pose, is utilized to estimate the position and orientation of surgical instruments in each individual camera view. These multi-view estimates enable the calculation of 3D poses of surgical tools, facilitating the rendering of a 3D surface model of the instruments, overlaid on the background scene, for real-time visualization. This study presents methods for adapting YOLOv8-Pose to the EasyVis2 system, including the development of a tailored training dataset. Experimental results demonstrate that, with an identical number of cameras, the new system improves 3D reconstruction accuracy and reduces computation time. Additionally, the adapted YOLOv8-Pose system shows high accuracy in 2D pose estimation.
Authors: Yuhui Zhang, Yuchang Su, Yiming Liu, Xiaohan Wang, James Burgess, Elaine Sui, Chenyu Wang, Josiah Aklilu, Alejandro Lozano, Anjiang Wei, Ludwig Schmidt, Serena Yeung-Levy
Abstract: The rapid development of vision language models (VLMs) demands rigorous and reliable evaluation. However, current visual question answering (VQA) benchmarks often depend on open-ended questions, making accurate evaluation difficult due to the variability in natural language responses. To address this, we introduce AutoConverter, an agentic framework that automatically converts these open-ended questions into multiple-choice format, enabling objective evaluation while reducing the costly multiple-choice question creation process. Our experiments demonstrate that AutoConverter can generate correct and challenging multiple-choice questions, with VLMs demonstrating consistently similar or lower accuracy on these questions compared to human-created ones. Using AutoConverter, we construct VMCBench, a benchmark created by transforming 20 existing VQA datasets into a unified multiple-choice format, totaling 9,018 questions. We comprehensively evaluate 33 state-of-the-art VLMs on VMCBench, setting a new standard for scalable, consistent, and reproducible VLM evaluation.
Authors: Nannan Li, Kevin J. Shih, Bryan A. Plummer
Abstract: Given an isolated garment image in a canonical product view and a separate image of a person, the virtual try-on task aims to generate a new image of the person wearing the target garment. Prior virtual try-on works face two major challenges in achieving this goal: a) the paired (human, garment) training data has limited availability; b) generating textures on the human that perfectly match that of the prompted garment is difficult, often resulting in distorted text and faded textures. Our work explores ways to tackle these issues through both synthetic data as well as model refinement. We introduce a garment extraction model that generates (human, synthetic garment) pairs from a single image of a clothed individual. The synthetic pairs can then be used to augment the training of virtual try-on. We also propose an Error-Aware Refinement-based Schr\"odinger Bridge (EARSB) that surgically targets localized generation errors for correcting the output of a base virtual try-on model. To identify likely errors, we propose a weakly-supervised error classifier that localizes regions for refinement, subsequently augmenting the Schr\"odinger Bridge's noise schedule with its confidence heatmap. Experiments on VITON-HD and DressCode-Upper demonstrate that our synthetic data augmentation enhances the performance of prior work, while EARSB improves the overall image quality. In user studies, our model is preferred by the users in an average of 59% of cases.
Authors: Yu Qiao, Apurba Adhikary, Huy Q. Le, Eui-Nam Huh, Zhu Han, Choong Seon Hong
Abstract: Federated learning (FL) has gained significant attention for enabling decentralized training on edge networks without exposing raw data. However, FL models remain susceptible to adversarial attacks and performance degradation in non-IID data settings, thus posing challenges to both robustness and accuracy. This paper aims to achieve communication-efficient adversarial federated learning (AFL) by leveraging a pre-trained model to enhance both robustness and accuracy under adversarial attacks and non-IID challenges in AFL. By leveraging the knowledge from a pre-trained model for both clean and adversarial images, we propose a pre-trained model-guided adversarial federated learning (PM-AFL) framework. This framework integrates vanilla and adversarial mixture knowledge distillation to effectively balance accuracy and robustness while promoting local models to learn from diverse data. Specifically, for clean accuracy, we adopt a dual distillation strategy where the class probabilities of randomly paired images, and their blended versions are aligned between the teacher model and the local models. For adversarial robustness, we employ a similar distillation approach but replace clean samples on the local side with adversarial examples. Moreover, by considering the bias between local and global models, we also incorporate a consistency regularization term to ensure that local adversarial predictions stay aligned with their corresponding global clean ones. These strategies collectively enable local models to absorb diverse knowledge from the teacher model while maintaining close alignment with the global model, thereby mitigating overfitting to local optima and enhancing the generalization of the global model. Experiments demonstrate that the PM-AFL-based framework not only significantly outperforms other methods but also maintains communication efficiency.
Authors: Antoni Kowalczuk, Jan Dubi\'nski, Franziska Boenisch, Adam Dziedzic
Abstract: Image autoregressive generation has emerged as a powerful new paradigm, with image autoregressive models (IARs) matching state-of-the-art diffusion models (DMs) in image quality (FID: 1.48 vs. 1.58) while allowing for higher generation speed. However, the privacy risks associated with IARs remain unexplored, raising concerns about their responsible deployment. To address this gap, we conduct a comprehensive privacy analysis of IARs, comparing their privacy risks to those of DMs as a reference point. Specifically, we develop a novel membership inference attack (MIA) that achieves a remarkably high success rate in detecting training images, with a True Positive Rate at False Positive Rate = 1% (TPR@FPR=1%) of 86.38%, compared to just 6.38% for DMs using comparable attacks. We leverage our novel MIA to perform dataset inference (DI) for IARs and show that it requires as few as 6 samples to detect dataset membership, compared to 200 samples for DI in DMs. This confirms a higher level of information leakage in IARs. Finally, we are able to extract hundreds of training data points from an IAR (e.g., 698 from VAR-d30). Our results suggest a fundamental privacy-utility trade-off: while IARs excel in image generation quality and speed, they are empirically significantly more vulnerable to privacy attacks compared to DMs that achieve similar performance. This trend suggests that incorporating techniques from DMs into IARs, such as modeling the per-token probability distribution using a diffusion procedure, could help mitigate IARs' vulnerability to privacy attacks. We make our code available at: https://github.com/sprintml/privacy_attacks_against_iars
URLs: https://github.com/sprintml/privacy_attacks_against_iars
Authors: Ebenezer Tarubinga, Jenifer Kalafatovich, Seong-Whan Lee
Abstract: Semi-supervised semantic segmentation (SSSS) aims to improve segmentation performance by utilizing large amounts of unlabeled data with limited labeled samples. Existing methods often suffer from coupling, where over-reliance on initial labeled data leads to suboptimal learning; confirmation bias, where incorrect predictions reinforce themselves repeatedly; and boundary blur caused by limited boundary-awareness and ambiguous edge cues. To address these issues, we propose CW-BASS, a novel framework for SSSS. In order to mitigate the impact of incorrect predictions, we assign confidence weights to pseudo-labels. Additionally, we leverage boundary-delineation techniques, which, despite being extensively explored in weakly-supervised semantic segmentation (WSSS), remain underutilized in SSSS. Specifically, our method: (1) reduces coupling via a confidence-weighted loss that adjusts pseudo-label influence based on their predicted confidence scores, (2) mitigates confirmation bias with a dynamic thresholding mechanism that learns to filter out pseudo-labels based on model performance, (3) tackles boundary blur using a boundary-aware module to refine segmentation near object edges, and (4) reduces label noise through a confidence decay strategy that progressively refines pseudo-labels during training. Extensive experiments on Pascal VOC 2012 and Cityscapes demonstrate that CW-BASS achieves state-of-the-art performance. Notably, CW-BASS achieves a 65.9% mIoU on Cityscapes under a challenging and underexplored 1/30 (3.3%) split (100 images), highlighting its effectiveness in limited-label settings. Our code is available at https://github.com/psychofict/CW-BASS.
Authors: Nikita Shvetsov, Thomas K. Kilvaer, Masoud Tafavvoghi, Anders Sildnes, Kajsa M{\o}llersen, Lill-Tove Rasmussen Busund, Lars Ailo Bongo
Abstract: Developing clinically useful cell-level analysis tools in digital pathology remains challenging due to limitations in dataset granularity, inconsistent annotations, high computational demands, and difficulties integrating new technologies into workflows. To address these issues, we propose a solution that enhances data quality, model performance, and usability by creating a lightweight, extensible cell segmentation and classification model. First, we update data labels through cross-relabeling to refine annotations of PanNuke and MoNuSAC, producing a unified dataset with seven distinct cell types. Second, we leverage the H-Optimus foundation model as a fixed encoder to improve feature representation for simultaneous segmentation and classification tasks. Third, to address foundation models' computational demands, we distill knowledge to reduce model size and complexity while maintaining comparable performance. Finally, we integrate the distilled model into QuPath, a widely used open-source digital pathology platform. Results demonstrate improved segmentation and classification performance using the H-Optimus-based model compared to a CNN-based model. Specifically, average $R^2$ improved from 0.575 to 0.871, and average $PQ$ score improved from 0.450 to 0.492, indicating better alignment with actual cell counts and enhanced segmentation quality. The distilled model maintains comparable performance while reducing parameter count by a factor of 48. By reducing computational complexity and integrating into workflows, this approach may significantly impact diagnostics, reduce pathologist workload, and improve outcomes. Although the method shows promise, extensive validation is necessary prior to clinical deployment.
Authors: Itay Benou, Tammy Riklin-Raviv
Abstract: Modern deep neural networks have now reached human-level performance across a variety of tasks. However, unlike humans they lack the ability to explain their decisions by showing where and telling what concepts guided them. In this work, we present a unified framework for transforming any vision neural network into a spatially and conceptually interpretable model. We introduce a spatially-aware concept bottleneck layer that projects "black-box" features of pre-trained backbone models into interpretable concept maps, without requiring human labels. By training a classification layer over this bottleneck, we obtain a self-explaining model that articulates which concepts most influenced its prediction, along with heatmaps that ground them in the input image. Accordingly, we name this method "Spatially-Aware and Label-Free Concept Bottleneck Model" (SALF-CBM). Our results show that the proposed SALF-CBM: (1) Outperforms non-spatial CBM methods, as well as the original backbone, on a variety of classification tasks; (2) Produces high-quality spatial explanations, outperforming widely used heatmap-based methods on a zero-shot segmentation task; (3) Facilitates model exploration and debugging, enabling users to query specific image regions and refine the model's decisions by locally editing its concept maps.
Authors: Yuxuan Bian, Zhaoyang Zhang, Xuan Ju, Mingdeng Cao, Liangbin Xie, Ying Shan, Qiang Xu
Abstract: Video inpainting, which aims to restore corrupted video content, has experienced substantial progress. Despite these advances, existing methods, whether propagating unmasked region pixels through optical flow and receptive field priors, or extending image-inpainting models temporally, face challenges in generating fully masked objects or balancing the competing objectives of background context preservation and foreground generation in one model, respectively. To address these limitations, we propose a novel dual-stream paradigm VideoPainter that incorporates an efficient context encoder (comprising only 6% of the backbone parameters) to process masked videos and inject backbone-aware background contextual cues to any pre-trained video DiT, producing semantically consistent content in a plug-and-play manner. This architectural separation significantly reduces the model's learning complexity while enabling nuanced integration of crucial background context. We also introduce a novel target region ID resampling technique that enables any-length video inpainting, greatly enhancing our practical applicability. Additionally, we establish a scalable dataset pipeline leveraging current vision understanding models, contributing VPData and VPBench to facilitate segmentation-based inpainting training and assessment, the largest video inpainting dataset and benchmark to date with over 390K diverse clips. Using inpainting as a pipeline basis, we also explore downstream applications including video editing and video editing pair data generation, demonstrating competitive performance and significant practical potential. Extensive experiments demonstrate VideoPainter's superior performance in both any-length video inpainting and editing, across eight key metrics, including video quality, mask region preservation, and textual coherence.
Authors: Yifan Wang, Yifei Liu, Yingdong Shi, Changming Li, Anqi Pang, Sibei Yang, Jingyi Yu, Kan Ren
Abstract: Vision Transformer models exhibit immense power yet remain opaque to human understanding, posing challenges and risks for practical applications. While prior research has attempted to demystify these models through input attribution and neuron role analysis, there's been a notable gap in considering layer-level information and the holistic path of information flow across layers. In this paper, we investigate the significance of influential neuron paths within vision Transformers, which is a path of neurons from the model input to output that impacts the model inference most significantly. We first propose a joint influence measure to assess the contribution of a set of neurons to the model outcome. And we further provide a layer-progressive neuron locating approach that efficiently selects the most influential neuron at each layer trying to discover the crucial neuron path from input to output within the target model. Our experiments demonstrate the superiority of our method finding the most influential neuron path along which the information flows, over the existing baseline solutions. Additionally, the neuron paths have illustrated that vision Transformers exhibit some specific inner working mechanism for processing the visual information within the same image category. We further analyze the key effects of these neurons on the image classification task, showcasing that the found neuron paths have already preserved the model capability on downstream tasks, which may also shed some lights on real-world applications like model pruning. The project website including implementation code is available at https://foundation-model-research.github.io/NeuronPath/.
URLs: https://foundation-model-research.github.io/NeuronPath/.
Authors: SaiKiran Tedla, Junyong Lee, Beixuan Yang, Mahmoud Afifi, Michael S. Brown
Abstract: Multispectral (MS) images capture detailed scene information across a wide range of spectral bands, making them invaluable for applications requiring rich spectral data. Integrating MS imaging into multi camera devices, such as smartphones, has the potential to enhance both spectral applications and RGB image quality. A critical step in processing MS data is demosaicing, which reconstructs color information from the mosaic MS images captured by the camera. This paper proposes a method for MS image demosaicing specifically designed for dual-camera setups where both RGB and MS cameras capture the same scene. Our approach leverages co-captured RGB images, which typically have higher spatial fidelity, to guide the demosaicing of lower-fidelity MS images. We introduce the Dual-camera RGB-MS Dataset - a large collection of paired RGB and MS mosaiced images with ground-truth demosaiced outputs - that enables training and evaluation of our method. Experimental results demonstrate that our method achieves state-of-the-art accuracy compared to existing techniques.
Authors: Bar{\i}\c{s} Batuhan Topal, Umut \"Ozyurt, Zafer Do\u{g}an Budak, Ramazan Gokberk Cinbis
Abstract: Recent advancements in text-to-image generative models, particularly latent diffusion models (LDMs), have demonstrated remarkable capabilities in synthesizing high-quality images from textual prompts. However, achieving identity personalization-ensuring that a model consistently generates subject-specific outputs from limited reference images-remains a fundamental challenge. To address this, we introduce Meta-Low-Rank Adaptation (Meta-LoRA), a novel framework that leverages meta-learning to encode domain-specific priors into LoRA-based identity personalization. Our method introduces a structured three-layer LoRA architecture that separates identity-agnostic knowledge from identity-specific adaptation. In the first stage, the LoRA Meta-Down layers are meta-trained across multiple subjects, learning a shared manifold that captures general identity-related features. In the second stage, only the LoRA-Mid and LoRA-Up layers are optimized to specialize on a given subject, significantly reducing adaptation time while improving identity fidelity. To evaluate our approach, we introduce Meta-PHD, a new benchmark dataset for identity personalization, and compare Meta-LoRA against state-of-the-art methods. Our results demonstrate that Meta-LoRA achieves superior identity retention, computational efficiency, and adaptability across diverse identity conditions. Our code, model weights, and dataset are released on barisbatuhan.github.io/Meta-LoRA.
Authors: Mahan Rafidashti, Ji Lan, Maryam Fatemi, Junsheng Fu, Lars Hammarstrand, Lennart Svensson
Abstract: Radar is an important sensor for autonomous driving (AD) systems due to its robustness to adverse weather and different lighting conditions. Novel view synthesis using neural radiance fields (NeRFs) has recently received considerable attention in AD due to its potential to enable efficient testing and validation but remains unexplored for radar point clouds. In this paper, we present NeuRadar, a NeRF-based model that jointly generates radar point clouds, camera images, and lidar point clouds. We explore set-based object detection methods such as DETR, and propose an encoder-based solution grounded in the NeRF geometry for improved generalizability. We propose both a deterministic and a probabilistic point cloud representation to accurately model the radar behavior, with the latter being able to capture radar's stochastic behavior. We achieve realistic reconstruction results for two automotive datasets, establishing a baseline for NeRF-based radar point cloud simulation models. In addition, we release radar data for ZOD's Sequences and Drives to enable further research in this field. To encourage further development of radar NeRFs, we release the source code for NeuRadar.
Authors: Wenjun Zeng, Dana Kurniawan, Ryan Mullins, Yuchi Liu, Tamoghna Saha, Dirichi Ike-Njoku, Jindong Gu, Yiwen Song, Cai Xu, Jingjing Zhou, Aparna Joshi, Shravan Dheep, Mani Malek, Hamid Palangi, Joon Baek, Rick Pereira, Karthik Narasimhan
Abstract: We introduce ShieldGemma 2, a 4B parameter image content moderation model built on Gemma 3. This model provides robust safety risk predictions across the following key harm categories: Sexually Explicit, Violence \& Gore, and Dangerous Content for synthetic images (e.g. output of any image generation model) and natural images (e.g. any image input to a Vision-Language Model). We evaluated on both internal and external benchmarks to demonstrate state-of-the-art performance compared to LlavaGuard \citep{helff2024llavaguard}, GPT-4o mini \citep{hurst2024gpt}, and the base Gemma 3 model \citep{gemma_2025} based on our policies. Additionally, we present a novel adversarial data generation pipeline which enables a controlled, diverse, and robust image generation. ShieldGemma 2 provides an open image moderation tool to advance multimodal safety and responsible AI development.
Authors: Kaiwei Zhang, Dandan Zhu, Xiongkuo Min, Guangtao Zhai
Abstract: Mesh saliency enhances the adaptability of 3D vision by identifying and emphasizing regions that naturally attract visual attention. To investigate the interaction between geometric structure and texture in shaping visual attention, we establish a comprehensive mesh saliency dataset, which is the first to systematically capture the differences in saliency distribution under both textured and non-textured visual conditions. Furthermore, we introduce mesh Mamba, a unified saliency prediction model based on a state space model (SSM), designed to adapt across various mesh types. Mesh Mamba effectively analyzes the geometric structure of the mesh while seamlessly incorporating texture features into the topological framework, ensuring coherence throughout appearance-enhanced modeling. More importantly, by subgraph embedding and a bidirectional SSM, the model enables global context modeling for both local geometry and texture, preserving the topological structure and improving the understanding of visual details and structural complexity. Through extensive theoretical and empirical validation, our model not only improves performance across various mesh types but also demonstrates high scalability and versatility, particularly through cross validations of various visual features.
Authors: Ulas Gunes, Matias Turkulainen, Xuqian Ren, Arno Solin, Juho Kannala, Esa Rahtu
Abstract: The development of large-scale 3D scene reconstruction and novel view synthesis methods mostly rely on datasets comprising perspective images with narrow fields of view (FoV). While effective for small-scale scenes, these datasets require large image sets and extensive structure-from-motion (SfM) processing, limiting scalability. To address this, we introduce a fisheye image dataset tailored for scene reconstruction tasks. Using dual 200-degree fisheye lenses, our dataset provides full 360-degree coverage of 5 indoor and 5 outdoor scenes. Each scene has sparse SfM point clouds and precise LIDAR-derived dense point clouds that can be used as geometric ground-truth, enabling robust benchmarking under challenging conditions such as occlusions and reflections. While the baseline experiments focus on vanilla Gaussian Splatting and NeRF based Nerfacto methods, the dataset supports diverse approaches for scene reconstruction, novel view synthesis, and image-based rendering.
Authors: Yingyan Li, Yuqi Wang, Yang Liu, Jiawei He, Lue Fan, Zhaoxiang Zhang
Abstract: End-to-end autonomous driving has achieved remarkable progress by integrating perception, prediction, and planning into a fully differentiable framework. Yet, to fully realize its potential, an effective online trajectory evaluation is indispensable to ensure safety. By forecasting the future outcomes of a given trajectory, trajectory evaluation becomes much more effective. This goal can be achieved by employing a world model to capture environmental dynamics and predict future states. Therefore, we propose an end-to-end driving framework WoTE, which leverages a BEV World model to predict future BEV states for Trajectory Evaluation. The proposed BEV world model is latency-efficient compared to image-level world models and can be seamlessly supervised using off-the-shelf BEV-space traffic simulators. We validate our framework on both the NAVSIM benchmark and the closed-loop Bench2Drive benchmark based on the CARLA simulator, achieving state-of-the-art performance. Code is released at https://github.com/liyingyanUCAS/WoTE.
Authors: Hao Wang, Shuo Zhang, Biao Leng
Abstract: The computer vision community has witnessed an extensive exploration of vision transformers in the past two years. Drawing inspiration from traditional schemes, numerous works focus on introducing vision-specific inductive biases. However, the implicit modeling of permutation invariance and fully-connected interaction with individual tokens disrupts the regional context and spatial topology, further hindering higher-order modeling. This deviates from the principle of perceptual organization that emphasizes the local groups and overall topology of visual elements. Thus, we introduce the concept of hypergraph for perceptual exploration. Specifically, we propose a topology-aware vision transformer called HyperGraph Transformer (HGFormer). Firstly, we present a Center Sampling K-Nearest Neighbors (CS-KNN) algorithm for semantic guidance during hypergraph construction. Secondly, we present a topology-aware HyperGraph Attention (HGA) mechanism that integrates hypergraph topology as perceptual indications to guide the aggregation of global and unbiased information during hypergraph messaging. Using HGFormer as visual backbone, we develop an effective and unitive representation, achieving distinct and detailed scene depictions. Empirical experiments show that the proposed HGFormer achieves competitive performance compared to the recent SoTA counterparts on various visual benchmarks. Extensive ablation and visualization studies provide comprehensive explanations of our ideas and contributions.
Authors: Joel Sol, Shadi Alijani, Homayoun Najjaran
Abstract: This paper introduces DRANet-SWD as a novel complete pipeline for disentangling content and style representations of images for unsupervised domain adaptation (UDA). The approach builds upon DRANet by incorporating the sliced Wasserstein discrepancy (SWD) as a style loss instead of the traditional Gram matrix loss. The potential advantages of SWD over the Gram matrix loss for capturing style variations in domain adaptation are investigated. Experiments using digit classification datasets and driving scenario segmentation validate the method, demonstrating that DRANet-SWD enhances performance. Results indicate that SWD provides a more robust statistical comparison of feature distributions, leading to better style adaptation. These findings highlight the effectiveness of SWD in refining feature alignment and improving domain adaptation tasks across these benchmarks. Our code can be found here.
Authors: Haishan Wang, Mohammad Hassan Vali, Arno Solin
Abstract: 3D Gaussian Splatting (3DGS) has demonstrated remarkable effectiveness in 3D reconstruction, achieving high-quality results with real-time radiance field rendering. However, a key challenge is the substantial storage cost: reconstructing a single scene typically requires millions of Gaussian splats, each represented by 59 floating-point parameters, resulting in approximately 1 GB of memory. To address this challenge, we propose a compression method by building separate attribute codebooks and storing only discrete code indices. Specifically, we employ noise-substituted vector quantization technique to jointly train the codebooks and model features, ensuring consistency between gradient descent optimization and parameter discretization. Our method reduces the memory consumption efficiently (around $45\times$) while maintaining competitive reconstruction quality on standard 3D benchmark scenes. Experiments on different codebook sizes show the trade-off between compression ratio and image quality. Furthermore, the trained compressed model remains fully compatible with popular 3DGS viewers and enables faster rendering speed, making it well-suited for practical applications.
Authors: Zahid Hassan Tushar, Adeleke Ademakinwa, Jianwu Wang, Zhibo Zhang, Sanjay Purushotham
Abstract: Accurate cloud property retrieval is vital for understanding cloud behavior and its impact on climate, including applications in weather forecasting, climate modeling, and estimating Earth's radiation balance. The Independent Pixel Approximation (IPA), a widely used physics-based approach, simplifies radiative transfer calculations by assuming each pixel is independent of its neighbors. While computationally efficient, IPA has significant limitations, such as inaccuracies from 3D radiative effects, errors at cloud edges, and ineffectiveness for overlapping or heterogeneous cloud fields. Recent AI/ML-based deep learning models have improved retrieval accuracy by leveraging spatial relationships across pixels. However, these models are often memory-intensive, retrieve only a single cloud property, or struggle with joint property retrievals. To overcome these challenges, we introduce CloudUNet with Attention Module (CAM), a compact UNet-based model that employs attention mechanisms to reduce errors in thick, overlapping cloud regions and a specialized loss function for joint retrieval of Cloud Optical Thickness (COT) and Cloud Effective Radius (CER). Experiments on a Large Eddy Simulation (LES) dataset show that our CAM model outperforms state-of-the-art deep learning methods, reducing mean absolute errors (MAE) by 34% for COT and 42% for CER, and achieving 76% and 86% lower MAE for COT and CER retrievals compared to the IPA method.
Authors: Bingyang Wang, Kaer Huang, Bin Li, Yiqiang Yan, Lihe Zhang, Huchuan Lu, You He
Abstract: Open-World Tracking (OWT) aims to track every object of any category, which requires the model to have strong generalization capabilities. Trackers can improve their generalization ability by leveraging Visual Language Models (VLMs). However, challenges arise with the fine-tuning strategies when VLMs are transferred to OWT: full fine-tuning results in excessive parameter and memory costs, while the zero-shot strategy leads to sub-optimal performance. To solve the problem, EffOWT is proposed for efficiently transferring VLMs to OWT. Specifically, we build a small and independent learnable side network outside the VLM backbone. By freezing the backbone and only executing backpropagation on the side network, the model's efficiency requirements can be met. In addition, EffOWT enhances the side network by proposing a hybrid structure of Transformer and CNN to improve the model's performance in the OWT field. Finally, we implement sparse interactions on the MLP, thus reducing parameter updates and memory costs significantly. Thanks to the proposed methods, EffOWT achieves an absolute gain of 5.5% on the tracking metric OWTA for unknown categories, while only updating 1.3% of the parameters compared to full fine-tuning, with a 36.4% memory saving. Other metrics also demonstrate obvious improvement.
Authors: Yunlong Tang, Jing Bi, Chao Huang, Susan Liang, Daiki Shimada, Hang Hua, Yunzhong Xiao, Yizhi Song, Pinxin Liu, Mingqian Feng, Junjia Guo, Zhuo Liu, Luchuan Song, Ali Vosoughi, Jinxi He, Liu He, Zeliang Zhang, Jiebo Luo, Chenliang Xu
Abstract: We present CAT-V (Caption AnyThing in Video), a training-free framework for fine-grained object-centric video captioning that enables detailed descriptions of user-selected objects through time. CAT-V integrates three key components: a Segmenter based on SAMURAI for precise object segmentation across frames, a Temporal Analyzer powered by TRACE-Uni for accurate event boundary detection and temporal analysis, and a Captioner using InternVL-2.5 for generating detailed object-centric descriptions. Through spatiotemporal visual prompts and chain-of-thought reasoning, our framework generates detailed, temporally-aware descriptions of objects' attributes, actions, statuses, interactions, and environmental contexts without requiring additional training data. CAT-V supports flexible user interactions through various visual prompts (points, bounding boxes, and irregular regions) and maintains temporal sensitivity by tracking object states and interactions across different time segments. Our approach addresses limitations of existing video captioning methods, which either produce overly abstract descriptions or lack object-level precision, enabling fine-grained, object-specific descriptions while maintaining temporal coherence and spatial accuracy. The GitHub repository for this project is available at https://github.com/yunlong10/CAT-V
Authors: Shuai Wang, Zhi Tian, Weilin Huang, Limin Wang
Abstract: Diffusion transformers have demonstrated remarkable generation quality, albeit requiring longer training iterations and numerous inference steps. In each denoising step, diffusion transformers encode the noisy inputs to extract the lower-frequency semantic component and then decode the higher frequency with identical modules. This scheme creates an inherent optimization dilemma: encoding low-frequency semantics necessitates reducing high-frequency components, creating tension between semantic encoding and high-frequency decoding. To resolve this challenge, we propose a new \textbf{\color{ddt}D}ecoupled \textbf{\color{ddt}D}iffusion \textbf{\color{ddt}T}ransformer~(\textbf{\color{ddt}DDT}), with a decoupled design of a dedicated condition encoder for semantic extraction alongside a specialized velocity decoder. Our experiments reveal that a more substantial encoder yields performance improvements as model size increases. For ImageNet $256\times256$, Our DDT-XL/2 achieves a new state-of-the-art performance of {1.31 FID}~(nearly $4\times$ faster training convergence compared to previous diffusion transformers). For ImageNet $512\times512$, Our DDT-XL/2 achieves a new state-of-the-art FID of 1.28. Additionally, as a beneficial by-product, our decoupled architecture enhances inference speed by enabling the sharing self-condition between adjacent denoising steps. To minimize performance degradation, we propose a novel statistical dynamic programming approach to identify optimal sharing strategies.
Authors: Hao Zhang, Yanping Zha, Qingwei Zhuang, Zhenfeng Shao, Jiayi Ma
Abstract: Current image fusion methods struggle to adapt to real-world environments encompassing diverse degradations with spatially varying characteristics. To address this challenge, we propose a robust fusion controller (RFC) capable of achieving degradation-aware image fusion through fine-grained language instructions, ensuring its reliable application in adverse environments. Specifically, RFC first parses language instructions to innovatively derive the functional condition and the spatial condition, where the former specifies the degradation type to remove, while the latter defines its spatial coverage. Then, a composite control priori is generated through a multi-condition coupling network, achieving a seamless transition from abstract language instructions to latent control variables. Subsequently, we design a hybrid attention-based fusion network to aggregate multi-modal information, in which the obtained composite control priori is deeply embedded to linearly modulate the intermediate fused features. To ensure the alignment between language instructions and control outcomes, we introduce a novel language-feature alignment loss, which constrains the consistency between feature-level gains and the composite control priori. Extensive experiments on publicly available datasets demonstrate that our RFC is robust against various composite degradations, particularly in highly challenging flare scenarios.
Authors: Xiaoxing Hu, Ziyang Gong, Yupei Wang, Yuru Jia, Gen Luo, Xue Yang
Abstract: Parameter-Efficient Fine-Tuning (PEFT) is a technique that allows us to adapt powerful Foundation Models (FMs) to diverse downstream tasks while preserving and unleashing their inherent capabilities. However, we have observed that existing PEFT methods, which are often designed with natural imagery in mind, struggle when applied to Remote Sensing (RS) scenarios. This is primarily due to their inability to handle artifact influences, a problem particularly severe in RS image features. To tackle this challenge, we introduce Earth-Adapter, the first PEFT method specifically designed for RS artifacts conquering. Earth-Adapter introduces a novel Mixture of Frequency Adaptation process that combines a Mixture of Adapter (MoA) with Discrete Fourier Transformation (DFT). By utilizing DFT, Earth-Adapter can decompose features into different frequency components, precisely separating artifacts from original features. The MoA then dynamically assigns weights to each adapter expert, allowing for the combination of features across various frequency domains. These simple-yet-effective approaches enable Earth-Adapter to more efficiently overcome the disturbances caused by artifacts than previous PEFT methods, significantly enhancing the FMs' performance on RS scenarios. Experiments on Domain Adaptation (DA), and Domain Generalization (DG) semantic segmentation benchmarks showcase the Earth-Adapter's effectiveness. Compared with baseline Rein, Earth-Adapter significantly improves 9.0% mIoU in DA and 3.1% mIoU in DG benchmarks. Our code will be released at https://github.com/VisionXLab/Earth-Adapter.
Authors: Geongyu Lee, Joonho Lee, Tae-Yeong Kwak, Sun Woo Kim, Youngmee Kwon, Chungyeul Kim, Hyeyoon Chang
Abstract: Accurate prediction of the likelihood of recurrence is important in the selection of postoperative treatment for patients with early-stage breast cancer. In this study, we investigated whether deep learning algorithms can predict patients' risk of recurrence by analyzing the pathology images of their cancer histology.We analyzed 125 hematoxylin and eosin-stained whole slide images (WSIs) from 125 patients across two institutions (National Cancer Center and Korea University Medical Center Guro Hospital) to predict breast cancer recurrence risk using deep learning. Sensitivity reached 0.857, 0.746, and 0.529 for low, intermediate, and high-risk categories, respectively, with specificity of 0.816, 0.803, and 0.972, and a Pearson correlation of 0.61 with histological grade. Class activation maps highlighted features like tubule formation and mitotic rate, suggesting a cost-effective approach to risk stratification, pending broader validation. These findings suggest that deep learning models trained exclusively on hematoxylin and eosin stained whole slide images can approximate genomic assay results, offering a cost-effective and scalable tool for breast cancer recurrence risk assessment. However, further validation using larger and more balanced datasets is needed to confirm the clinical applicability of our approach.
Authors: Dung Thuy Nguyen, Taylor T. Johnson, Kevin Leach
Abstract: Federated Learning (FL) shows promise in preserving privacy and enabling collaborative learning. However, most current solutions focus on private data collected from a single domain. A significant challenge arises when client data comes from diverse domains (i.e., domain shift), leading to poor performance on unseen domains. Existing Federated Domain Generalization approaches address this problem but assume each client holds data for an entire domain, limiting their practicality in real-world scenarios with domain-based heterogeneity and client sampling. In addition, certain methods enable information sharing among clients, raising privacy concerns as this information could be used to reconstruct sensitive private data. To overcome this, we introduce FISC, a novel FedDG paradigm designed to robustly handle more complicated domain distributions between clients while ensuring security. FISC enables learning across domains by extracting an interpolative style from local styles and employing contrastive learning. This strategy gives clients multi-domain representations and unbiased convergent targets. Empirical results on multiple datasets, including PACS, Office-Home, and IWildCam, show FISC outperforms state-of-the-art (SOTA) methods. Our method achieves accuracy on unseen domains, with improvements ranging from 3.64% to 57.22% on unseen domains. Our code is available at https://github.com/judydnguyen/PARDON-FedDG.
Authors: Pit Henrich, Franziska Mathis-Ullrich, Paul Maria Scheikl
Abstract: Accurately determining the shape of objects and the location of their internal structures within deformable objects is crucial for medical tasks that require precise targeting, such as robotic biopsies. We introduce LUDO, a method for accurate low-latency understanding of deformable objects. LUDO reconstructs objects in their deformed state, including their internal structures, from a single-view point cloud observation in under 30 ms using occupancy networks. LUDO provides uncertainty estimates for its predictions. Additionally, it provides explainability by highlighting key features in its input observations. Both uncertainty and explainability are important for safety-critical applications such as surgical interventions. We demonstrate LUDO's abilities for autonomous targeting of internal regions of interest (ROIs) in deformable objects. %Additionally, LUDO provides uncertainty estimates and explainability for its predictions, both of which are important in safety-critical applications such as surgical interventions. We evaluate LUDO in real-world robotic experiments, achieving a success rate of 98.9% for puncturing various ROIs inside deformable objects. LUDO demonstrates the potential to interact with deformable objects without the need for deformable registration methods.
Authors: Feng Zhou, Ruiyang Liu, Chen Liu, Gaofeng He, Yong-Lu Li, Xiaogang Jin, Huamin Wang
Abstract: Sewing patterns, the essential blueprints for fabric cutting and tailoring, act as a crucial bridge between design concepts and producible garments. However, existing uni-modal sewing pattern generation models struggle to effectively encode complex design concepts with a multi-modal nature and correlate them with vectorized sewing patterns that possess precise geometric structures and intricate sewing relations. In this work, we propose a novel sewing pattern generation approach \textbf{Design2GarmentCode} based on Large Multimodal Models (LMMs), to generate parametric pattern-making programs from multi-modal design concepts. LMM offers an intuitive interface for interpreting diverse design inputs, while pattern-making programs could serve as well-structured and semantically meaningful representations of sewing patterns, and act as a robust bridge connecting the cross-domain pattern-making knowledge embedded in LMMs with vectorized sewing patterns. Experimental results demonstrate that our method can flexibly handle various complex design expressions such as images, textual descriptions, designer sketches, or their combinations, and convert them into size-precise sewing patterns with correct stitches. Compared to previous methods, our approach significantly enhances training efficiency, generation quality, and authoring flexibility.
Authors: Jiakang Yuan, Xiangchao Yan, Shiyang Feng, Bo Zhang, Tao Chen, Botian Shi, Wanli Ouyang, Yu Qiao, Lei Bai, Bowen Zhou
Abstract: The scientific research paradigm is undergoing a profound transformation owing to the development of Artificial Intelligence (AI). Recent works demonstrate that various AI-assisted research methods can largely improve research efficiency by improving data analysis, accelerating computation, and fostering novel idea generation. To further move towards the ultimate goal (i.e., automatic scientific research), in this paper, we introduce Dolphin, a closed-loop LLM-driven framework to enhance the automation level of scientific research. Dolphin first generates novel ideas based on feedback from previous experiments and relevant papers ranked by the topic and task attributes. Then, the generated ideas can be implemented using a code template refined and debugged with the designed exception-traceback-guided local code structure. Finally, Dolphin automatically analyzes the results of each idea and feeds the results back to the next round of idea generation. Experiments are conducted on the benchmark datasets of different topics and a subset of MLE-bench. Results show that Dolphin can continuously improve the performance of the input topic in a loop. We highlight that Dolphin can automatically propose methods that are comparable to the state-of-the-art in some tasks such as 3D point classification.
Authors: Ella Koresh, Ronit D. Gross, Yuval Meir, Yarden Tzach, Tal Halevi, Ido Kanter
Abstract: Convolutional neural networks (CNNs) evaluate short-range correlations in input images which progress along the layers, whereas vision transformer (ViT) architectures evaluate long-range correlations, using repeated transformer encoders composed of fully connected layers. Both are designed to solve complex classification tasks but from different perspectives. This study demonstrates that CNNs and ViT architectures stem from a unified underlying learning mechanism, which quantitatively measures the single-nodal performance (SNP) of each node in feedforward (FF) and multi-head attention (MHA) sub-blocks. Each node identifies small clusters of possible output labels, with additional noise represented as labels outside these clusters. These features are progressively sharpened along the transformer encoders, enhancing the signal-to-noise ratio. This unified underlying learning mechanism leads to two main findings. First, it enables an efficient applied nodal diagonal connection (ANDC) pruning technique without affecting the accuracy. Second, based on the SNP, spontaneous symmetry breaking occurs among the MHA heads, such that each head focuses its attention on a subset of labels through cooperation among its SNPs. Consequently, each head becomes an expert in recognizing its designated labels, representing a quantitative MHA modus vivendi mechanism. This statistical mechanics inspired viewpoint enables to reveal macroscopic behavior of the entire network from the microscopic performance of each node. These results are based on a compact convolutional transformer architecture trained on the CIFAR-100 and Flowers-102 datasets and call for their extension to other architectures and applications, such as natural language processing.
Authors: Peiyan Yue, Die Cai, Chu Guo, Mengxing Liu, Jun Xia, Yi Wang
Abstract: Accurate automated segmentation of tibial plateau fractures (TPF) from computed tomography (CT) requires large amounts of annotated data to train deep learning models, but obtaining such annotations presents unique challenges. The process demands expert knowledge to identify diverse fracture patterns, assess severity, and account for individual anatomical variations, making the annotation process highly time-consuming and expensive. Although semi-supervised learning methods can utilize unlabeled data, existing approaches often struggle with the complexity and variability of fracture morphologies, as well as limited generalizability across datasets. To tackle these issues, we propose an effective training strategy based on masked autoencoder (MAE) for the accurate TPF segmentation in CT. Our method leverages MAE pretraining to capture global skeletal structures and fine-grained fracture details from unlabeled data, followed by fine-tuning with a small set of labeled data. This strategy reduces the dependence on extensive annotations while enhancing the model's ability to learn generalizable and transferable features. The proposed method is evaluated on an in-house dataset containing 180 CT scans with TPF. Experimental results demonstrate that our method consistently outperforms semi-supervised methods, achieving an average Dice similarity coefficient (DSC) of 95.81%, average symmetric surface distance (ASSD) of 1.91mm, and Hausdorff distance (95HD) of 9.42mm with only 20 annotated cases. Moreover, our method exhibits strong transferability when applying to another public pelvic CT dataset with hip fractures, highlighting its potential for broader applications in fracture segmentation tasks.
Authors: Kasra Arabi, R. Teal Witter, Chinmay Hegde, Niv Cohen
Abstract: Generative models have rapidly evolved to generate realistic outputs. However, their synthetic outputs increasingly challenge the clear distinction between natural and AI-generated content, necessitating robust watermarking techniques. Watermarks are typically expected to preserve the integrity of the target image, withstand removal attempts, and prevent unauthorized replication onto unrelated images. To address this need, recent methods embed persistent watermarks into images produced by diffusion models using the initial noise. Yet, to do so, they either distort the distribution of generated images or rely on searching through a long dictionary of used keys for detection. In this paper, we propose a novel watermarking method that embeds semantic information about the generated image directly into the watermark, enabling a distortion-free watermark that can be verified without requiring a database of key patterns. Instead, the key pattern can be inferred from the semantic embedding of the image using locality-sensitive hashing. Furthermore, conditioning the watermark detection on the original image content improves robustness against forgery attacks. To demonstrate that, we consider two largely overlooked attack strategies: (i) an attacker extracting the initial noise and generating a novel image with the same pattern; (ii) an attacker inserting an unrelated (potentially harmful) object into a watermarked image, possibly while preserving the watermark. We empirically validate our method's increased robustness to these attacks. Taken together, our results suggest that content-aware watermarks can mitigate risks arising from image-generative models.
Authors: Xuechen Liang, Meiling Tao, Yinghui Xia, Jianhui Wang, Kun Li, Yijin Wang, Jingsong Yang, Tianyu Shi, Yuantao Wang, Miao Zhang, Xueqian Wang
Abstract: Large language models (LLMs) have made significant advances in the field of natural language processing, but they still face challenges such as continuous decision-making, lack of long-term memory, and limited context windows in dynamic environments. To address these issues, this paper proposes an innovative framework Memory-Enhanced Agents with Reflective Self-improvement. The MARS framework comprises three agents: the User, the Assistant, and the Checker. By integrating iterative feedback, reflective mechanisms, and a memory optimization mechanism based on the Ebbinghaus forgetting curve, it significantly enhances the agents capabilities in handling multi-tasking and long-span information.
Authors: Rana Muhammad Shahroz Khan, Dongwen Tang, Pingzhi Li, Kai Wang, Tianlong Chen
Abstract: Parameter generation has emerged as a novel paradigm for neural network development, offering an alternative to traditional neural network training by synthesizing high-quality model weights directly. In the context of Low-Rank Adaptation (LoRA) for evolving ($\textit{i.e.}$, constantly updated) large language models (LLMs), this approach promises efficient adaptation without costly retraining. However, existing methods face critical limitations in simultaneously achieving scalability and controllability. In this paper, we introduce $\texttt{ORAL}$, a novel $\textbf{conditional recurrent diffusion}$ framework that addresses these challenges. $\texttt{ORAL}$ incorporates a novel conditioning mechanism that integrates model architecture and textual task specifications, enabling the generation of task-specific LoRA parameters that can seamlessly transfer across evolving foundation models. Our approach successfully scales to billions-of-parameter LLMs and maintains controllability. Through extensive experiments across seven language tasks, four vision tasks, and three multimodal tasks using five pre-trained LLMs, we demonstrate that $\texttt{ORAL}$ generates high-quality LoRA parameters that achieve comparable or superior performance to vanilla trained counterparts.
Authors: Jianhua Sun, Cewu Lu
Abstract: Reviewing the progress in artificial intelligence over the past decade, various significant advances (e.g. object detection, image generation, large language models) have enabled AI systems to produce more semantically meaningful outputs and achieve widespread adoption in internet scenarios. Nevertheless, AI systems still struggle when it comes to understanding and interacting with the physical world. This reveals an important issue: relying solely on semantic-level concepts learned from internet data (e.g. texts, images) to understand the physical world is far from sufficient -- machine intelligence currently lacks an effective way to learn about the physical world. This research introduces the idea of analytic concept -- representing the concepts related to the physical world through programs of mathematical procedures, providing machine intelligence a portal to perceive, reason about, and interact with the physical world. Except for detailing the design philosophy and providing guidelines for the application of analytic concepts, this research also introduce about the infrastructure that has been built around analytic concepts. I aim for my research to contribute to addressing these questions: What is a proper abstraction of general concepts in the physical world for machine intelligence? How to systematically integrate structured priors with neural networks to constrain AI systems to comply with physical laws?