Authors: Junlin Guo, Siqi Lu, Can Cui, Ruining Deng, Tianyuan Yao, Zhewen Tao, Yizhe Lin, Marilyn Lionts, Quan Liu, Juming Xiong, Yu Wang, Shilin Zhao, Catie Chang, Mitchell Wilkes, Mengmeng Yin, Haichun Yang, Yuankai Huo
Abstract: Training AI foundation models has emerged as a promising large-scale learning approach for addressing real-world healthcare challenges, including digital pathology. While many of these models have been developed for tasks like disease diagnosis and tissue quantification using extensive and diverse training datasets, their readiness for deployment on some arguably simplest tasks, such as nuclei segmentation within a single organ (e.g., the kidney), remains uncertain. This paper seeks to answer this key question, "How good are we?", by thoroughly evaluating the performance of recent cell foundation models on a curated multi-center, multi-disease, and multi-species external testing dataset. Additionally, we tackle a more challenging question, "How can we improve?", by developing and assessing human-in-the-loop data enrichment strategies aimed at enhancing model performance while minimizing the reliance on pixel-level human annotation. To address the first question, we curated a multicenter, multidisease, and multispecies dataset consisting of 2,542 kidney whole slide images (WSIs). Three state-of-the-art (SOTA) cell foundation models-Cellpose, StarDist, and CellViT-were selected for evaluation. To tackle the second question, we explored data enrichment algorithms by distilling predictions from the different foundation models with a human-in-the-loop framework, aiming to further enhance foundation model performance with minimal human efforts. Our experimental results showed that all three foundation models improved over their baselines with model fine-tuning with enriched data. Interestingly, the baseline model with the highest F1 score does not yield the best segmentation outcomes after fine-tuning. This study establishes a benchmark for the development and deployment of cell vision foundation models tailored for real-world data applications.
Authors: David Schneider, Simon Rei{\ss}, Marco Kugler, Alexander Jaus, Kunyu Peng, Susanne Sutschet, M. Saquib Sarfraz, Sven Matthiesen, Rainer Stiefelhagen
Abstract: Exploring the intricate dynamics between muscular and skeletal structures is pivotal for understanding human motion. This domain presents substantial challenges, primarily attributed to the intensive resources required for acquiring ground truth muscle activation data, resulting in a scarcity of datasets. In this work, we address this issue by establishing Muscles in Time (MinT), a large-scale synthetic muscle activation dataset. For the creation of MinT, we enriched existing motion capture datasets by incorporating muscle activation simulations derived from biomechanical human body models using the OpenSim platform, a common approach in biomechanics and human motion research. Starting from simple pose sequences, our pipeline enables us to extract detailed information about the timing of muscle activations within the human musculoskeletal system. Muscles in Time contains over nine hours of simulation data covering 227 subjects and 402 simulated muscle strands. We demonstrate the utility of this dataset by presenting results on neural network-based muscle activation estimation from human pose sequences with two different sequence-to-sequence architectures. Data and code are provided under https://simplexsigil.github.io/mint.
Authors: Chen Zhao, Xuan Wang, Tong Zhang, Saqib Javed, Mathieu Salzmann
Abstract: 3D Gaussian Splatting (3DGS) has demonstrated remarkable effectiveness for novel view synthesis (NVS). However, the 3DGS model tends to overfit when trained with sparse posed views, limiting its generalization capacity for broader pose variations. In this paper, we alleviate the overfitting problem by introducing a self-ensembling Gaussian Splatting (SE-GS) approach. We present two Gaussian Splatting models named the $\mathbf{\Sigma}$-model and the $\mathbf{\Delta}$-model. The $\mathbf{\Sigma}$-model serves as the primary model that generates novel-view images during inference. At the training stage, the $\mathbf{\Sigma}$-model is guided away from specific local optima by an uncertainty-aware perturbing strategy. We dynamically perturb the $\mathbf{\Delta}$-model based on the uncertainties of novel-view renderings across different training steps, resulting in diverse temporal models sampled from the Gaussian parameter space without additional training costs. The geometry of the $\mathbf{\Sigma}$-model is regularized by penalizing discrepancies between the $\mathbf{\Sigma}$-model and the temporal samples. Therefore, our SE-GS conducts an effective and efficient regularization across a large number of Gaussian Splatting models, resulting in a robust ensemble, the $\mathbf{\Sigma}$-model. Experimental results on the LLFF, Mip-NeRF360, DTU, and MVImgNet datasets show that our approach improves NVS quality with few-shot training views, outperforming existing state-of-the-art methods. The code is released at https://github.com/sailor-z/SE-GS.
Authors: Nursena K\"opr\"uc\"u, Destiny Okpekpe, Antonio Orvieto
Abstract: Transformers have become dominant in large-scale deep learning tasks across various domains, including text, 2D and 3D vision. However, the quadratic complexity of their attention mechanism limits their efficiency as the sequence length increases, particularly in high-resolution 3D data such as point clouds. Recently, state space models (SSMs) like Mamba have emerged as promising alternatives, offering linear complexity, scalability, and high performance in long-sequence tasks. The key challenge in the application of SSMs in this domain lies in reconciling the non-sequential structure of point clouds with the inherently directional (or bi-directional) order-dependent processing of recurrent models like Mamba. To achieve this, previous research proposed reorganizing point clouds along multiple directions or predetermined paths in 3D space, concatenating the results to produce a single 1D sequence capturing different views. In our work, we introduce a method to convert point clouds into 1D sequences that maintain 3D spatial structure with no need for data replication, allowing Mamba sequential processing to be applied effectively in an almost permutation-invariant manner. In contrast to other works, we found that our method does not require positional embeddings and allows for shorter sequence lengths while still achieving state-of-the-art results in ModelNet40 and ScanObjectNN datasets and surpassing Transformer-based models in both accuracy and efficiency.
Authors: Marcelo Eduardo Marques Ribas (Department of Informatics - Federal University of Paran\'a), Heloisa Benedet Mendes (Department of Informatics - Federal University of Paran\'a), Luiz Eduardo Soares de Oliveira (Department of Informatics - Federal University of Paran\'a), Luiz Antonio Zanlorensi (DeepNeuronic), Paulo Ricardo Lisboa de Almeida (Department of Informatics - Federal University of Paran\'a)
Abstract: In smart cities, it is common practice to define a maximum length of stay for a given parking space to increase the space's rotativity and discourage the usage of individual transportation solutions. However, automatically determining individual car dwell times from images faces challenges, such as images collected from low-resolution cameras, lighting variations, and weather effects. In this work, we propose a method that combines two deep neural networks to compute the dwell time of each car in a parking lot. The proposed method first defines the parking space status between occupied and empty using a deep classification network. Then, it uses a Siamese network to check if the parked car is the same as the previous image. Using an experimental protocol that focuses on a cross-dataset scenario, we show that if a perfect classifier is used, the proposed system generates 75% of perfect dwell time predictions, where the predicted value matched exactly the time the car stayed parked. Nevertheless, our experiments show a drop in prediction quality when a real-world classifier is used to predict the parking space statuses, reaching 49% of perfect predictions, showing that the proposed Siamese network is promising but impacted by the quality of the classifier used at the beginning of the pipeline.
Authors: Mohammad Farazi, Yalin Wang
Abstract: Utilizing patch-based transformers for unstructured geometric data such as polygon meshes presents significant challenges, primarily due to the absence of a canonical ordering and variations in input sizes. Prior approaches to handling 3D meshes and point clouds have either relied on computationally intensive node-level tokens for large objects or resorted to resampling to standardize patch size. Moreover, these methods generally lack a geometry-aware, stable Structural Embedding (SE), often depending on simplistic absolute SEs such as 3D coordinates, which compromise isometry invariance essential for tasks like semantic segmentation. In our study, we meticulously examine the various components of a geometry-aware 3D mesh transformer, from tokenization to structural encoding, assessing the contribution of each. Initially, we introduce a spectral-preserving tokenization rooted in algebraic multigrid methods. Subsequently, we detail an approach for embedding features at the patch level, accommodating patches with variable node counts. Through comparative analyses against a baseline model employing simple point-wise Multi-Layer Perceptrons (MLP), our research highlights critical insights: 1) the importance of structural and positional embeddings facilitated by heat diffusion in general 3D mesh transformers; 2) the effectiveness of novel components such as geodesic masking and feature interaction via cross-attention in enhancing learning; and 3) the superior performance and efficiency of our proposed methods in challenging segmentation and classification tasks.
Authors: Ibne Hassan, Aman Mujahid, Abdullah Al Hasib, Andalib Rahman Shagoto, Joyanta Jyoti Mondal, Meem Arafat Manab, Jannatun Noor
Abstract: Countries in South Asia experience many catastrophic flooding events regularly. Through image classification, it is possible to expedite search and rescue initiatives by classifying flood zones, including houses and humans. We create a new dataset collecting aerial imagery of flooding events across South Asian countries. For the classification, we propose a fine-tuned Compact Convolutional Transformer (CCT) based approach and some other cutting-edge transformer-based and Convolutional Neural Network-based architectures (CNN). We also implement the YOLOv8 object detection model and detect houses and humans within the imagery of our proposed dataset, and then compare the performance with our classification-based approach. Since the countries in South Asia have similar topography, housing structure, the color of flood water, and vegetation, this work can be more applicable to such a region as opposed to the rest of the world. The images are divided evenly into four classes: 'flood', 'flood with domicile', 'flood with humans', and 'no flood'. After experimenting with our proposed dataset on our fine-tuned CCT model, which has a comparatively lower number of weight parameters than many other transformer-based architectures designed for computer vision, it exhibits an accuracy and macro average precision of 98.62% and 98.50%. The other transformer-based architectures that we implement are the Vision Transformer (ViT), Swin Transformer, and External Attention Transformer (EANet), which give an accuracy of 88.66%, 84.74%, and 66.56% respectively. We also implement DCECNN (Deep Custom Ensembled Convolutional Neural Network), which is a custom ensemble model that we create by combining MobileNet, InceptionV3, and EfficientNetB0, and we obtain an accuracy of 98.78%. The architectures we implement are fine-tuned to achieve optimal performance on our dataset.
Authors: Kien X. Nguyen, Fengchun Qiao, Arthur Trembanis, Xi Peng
Abstract: A major obstacle to the advancements of machine learning models in marine science, particularly in sonar imagery analysis, is the scarcity of AI-ready datasets. While there have been efforts to make AI-ready sonar image dataset publicly available, they suffer from limitations in terms of environment setting and scale. To bridge this gap, we introduce SeafloorAI, the first extensive AI-ready datasets for seafloor mapping across 5 geological layers that is curated in collaboration with marine scientists. We further extend the dataset to SeafloorGenAI by incorporating the language component in order to facilitate the development of both vision- and language-capable machine learning models for sonar imagery. The dataset consists of 62 geo-distributed data surveys spanning 17,300 square kilometers, with 696K sonar images, 827K annotated segmentation masks, 696K detailed language descriptions and approximately 7M question-answer pairs. By making our data processing source code publicly available, we aim to engage the marine science community to enrich the data pool and inspire the machine learning community to develop more robust models. This collaborative approach will enhance the capabilities and applications of our datasets within both fields.
Authors: Pranav Singh Chib, Pravendra Singh
Abstract: Pedestrian trajectory prediction is crucial for several applications such as robotics and self-driving vehicles. Significant progress has been made in the past decade thanks to the availability of pedestrian trajectory datasets, which enable trajectory prediction methods to learn from pedestrians' past movements and predict future trajectories. However, these datasets and methods typically assume that the observed trajectory sequence is complete, ignoring real-world issues such as sensor failure, occlusion, and limited fields of view that can result in missing values in observed trajectories. To address this challenge, we present TrajImpute, a pedestrian trajectory prediction dataset that simulates missing coordinates in the observed trajectory, enhancing real-world applicability. TrajImpute maintains a uniform distribution of missing data within the observed trajectories. In this work, we comprehensively examine several imputation methods to reconstruct the missing coordinates and benchmark them for imputing pedestrian trajectories. Furthermore, we provide a thorough analysis of recent trajectory prediction methods and evaluate the performance of these models on the imputed trajectories. Our experimental evaluation of the imputation and trajectory prediction methods offers several valuable insights. Our dataset provides a foundational resource for future research on imputation-aware pedestrian trajectory prediction, potentially accelerating the deployment of these methods in real-world applications. Publicly accessible links to the datasets and code files are available at https://github.com/Pranav-chib/TrajImpute.
Authors: Panagiota Gatoula, Dimitrios E. Diamantis, Anastasios Koulaouzidis, Cristina Carretero, Stefania Chetcuti-Zammit, Pablo Cortegoso Valdivia, Bego\~na Gonz\'alez-Su\'arez, Alessandro Mussetto, John Plevris, Alexander Robertson, Bruno Rosa, Ervin Toth, Dimitris K. Iakovidis
Abstract: Sharing retrospectively acquired data is essential for both clinical research and training. Synthetic Data Generation (SDG), using Artificial Intelligence (AI) models, can overcome privacy barriers in sharing clinical data, enabling advancements in medical diagnostics. This study focuses on the clinical evaluation of medical SDG, with a proof-of-concept investigation on diagnosing Inflammatory Bowel Disease (IBD) using Wireless Capsule Endoscopy (WCE) images. The paper contributes by a) presenting a protocol for the systematic evaluation of synthetic images by medical experts and b) applying it to assess TIDE-II, a novel variational autoencoder-based model for high-resolution WCE image synthesis, with a comprehensive qualitative evaluation conducted by 10 international WCE specialists, focusing on image quality, diversity, realism, and clinical decision-making. The results show that TIDE-II generates clinically relevant WCE images, helping to address data scarcity and enhance diagnostic tools. The proposed protocol serves as a reference for future research on medical image-generation techniques.
Authors: Ce Zhou (Michigan State University), Qiben Yan (Michigan State University), Daniel Kent (Michigan State University), Guangjing Wang (University of South Florida), Weikang Ding (Michigan State University), Ziqi Zhang (Peking University), Hayder Radha (Michigan State University)
Abstract: Monocular Depth Estimation (MDE) is a pivotal component of vision-based Autonomous Driving (AD) systems, enabling vehicles to estimate the depth of surrounding objects using a single camera image. This estimation guides essential driving decisions, such as braking before an obstacle or changing lanes to avoid collisions. In this paper, we explore vulnerabilities of MDE algorithms in AD systems, presenting LensAttack, a novel physical attack that strategically places optical lenses on the camera of an autonomous vehicle to manipulate the perceived object depths. LensAttack encompasses two attack formats: concave lens attack and convex lens attack, each utilizing different optical lenses to induce false depth perception. We first develop a mathematical model that outlines the parameters of the attack, followed by simulations and real-world evaluations to assess its efficacy on state-of-the-art MDE models. Additionally, we adopt an attack optimization method to further enhance the attack success rate by optimizing the attack focal length. To better evaluate the implications of LensAttack on AD, we conduct comprehensive end-to-end system simulations using the CARLA platform. The results reveal that LensAttack can significantly disrupt the depth estimation processes in AD systems, posing a serious threat to their reliability and safety. Finally, we discuss some potential defense methods to mitigate the effects of the proposed attack.
Authors: Brody McNutt, Libby Zhang, Angus Carey-Douglas, Fritz Vollrath, Frank Pope, Leandra Brickson
Abstract: This research represents a pioneering application of automated pose estimation from drone data to study elephant behavior in the wild, utilizing video footage captured from Samburu National Reserve, Kenya. The study evaluates two pose estimation workflows: DeepLabCut, known for its application in laboratory settings and emerging wildlife fieldwork, and YOLO-NAS-Pose, a newly released pose estimation model not previously applied to wildlife behavioral studies. These models are trained to analyze elephant herd behavior, focusing on low-resolution ($\sim$50 pixels) subjects to detect key points such as the head, spine, and ears of multiple elephants within a frame. Both workflows demonstrated acceptable quality of pose estimation on the test set, facilitating the automated detection of basic behaviors crucial for studying elephant herd dynamics. For the metrics selected for pose estimation evaluation on the test set -- root mean square error (RMSE), percentage of correct keypoints (PCK), and object keypoint similarity (OKS) -- the YOLO-NAS-Pose workflow outperformed DeepLabCut. Additionally, YOLO-NAS-Pose exceeded DeepLabCut in object detection evaluation. This approach introduces a novel method for wildlife behavioral research, including the burgeoning field of wildlife drone monitoring, with significant implications for wildlife conservation.
Authors: Nidhal Jegham, Chan Young Koh, Marwan Abdelatti, Abdeltawab Hendawi
Abstract: This study presents a comprehensive benchmark analysis of various YOLO (You Only Look Once) algorithms, from YOLOv3 to the newest addition. It represents the first research to comprehensively evaluate the performance of YOLO11, the latest addition to the YOLO family. It evaluates their performance on three diverse datasets: Traffic Signs (with varying object sizes), African Wildlife (with diverse aspect ratios and at least one instance of the object per image), and Ships and Vessels (with small-sized objects of a single class), ensuring a comprehensive assessment across datasets with distinct challenges. To ensure a robust evaluation, we employ a comprehensive set of metrics, including Precision, Recall, Mean Average Precision (mAP), Processing Time, GFLOPs count, and Model Size. Our analysis highlights the distinctive strengths and limitations of each YOLO version. For example: YOLOv9 demonstrates substantial accuracy but struggles with detecting small objects and efficiency whereas YOLOv10 exhibits relatively lower accuracy due to architectural choices that affect its performance in overlapping object detection but excels in speed and efficiency. Additionally, the YOLO11 family consistently shows superior performance in terms of accuracy, speed, computational efficiency, and model size. YOLO11m achieved a remarkable balance of accuracy and efficiency, scoring mAP50-95 scores of 0.795, 0.81, and 0.325 on the Traffic Signs, African Wildlife, and Ships datasets, respectively, while maintaining an average inference time of 2.4ms, a model size of 38.8Mb, and around 67.6 GFLOPs on average. These results provide critical insights for both industry and academia, facilitating the selection of the most suitable YOLO algorithm for diverse applications and guiding future enhancements.
Authors: Thanh-Dung Le, Vu Nguyen Ha, Ti Ti Nguyen, Geoffrey Eappen, Prabhu Thiruvasagam, Hong-fu Chou, Duc-Dung Tran, Luis M. Garces-Socarras, Jorge L. Gonzalez-Rios, Juan Carlos Merlano-Duncan, Symeon Chatzinotas
Abstract: This study presents an innovative dynamic weighting knowledge distillation (KD) framework tailored for efficient Earth observation (EO) image classification (IC) in resource-constrained settings. Utilizing EfficientViT and MobileViT as teacher models, this framework enables lightweight student models, particularly ResNet8 and ResNet16, to surpass 90% in accuracy, precision, and recall, adhering to the stringent confidence thresholds necessary for reliable classification tasks. Unlike conventional KD methods that rely on static weight distribution, our adaptive weighting mechanism responds to each teacher model's confidence, allowing student models to prioritize more credible sources of knowledge dynamically. Remarkably, ResNet8 delivers substantial efficiency gains, achieving a 97.5% reduction in parameters, a 96.7% decrease in FLOPs, an 86.2% cut in power consumption, and a 63.5% increase in inference speed over MobileViT. This significant optimization of complexity and resource demands establishes ResNet8 as an optimal candidate for EO tasks, combining robust performance with feasibility in deployment. The confidence-based, adaptable KD approach underscores the potential of dynamic distillation strategies to yield high-performing, resource-efficient models tailored for satellite-based EO applications. The reproducible code is accessible on our GitHub repository.
Authors: Shreelekha Revankar, Cheng Perng Phoo, Utkarsh Mall, Bharath Hariharan, Kavita Bala
Abstract: Recognition of features in satellite imagery (forests, swimming pools, etc.) depends strongly on the spatial scale of the concept and therefore the resolution of the images. This poses two challenges: Which resolution is best suited for recognizing a given concept, and where and when should the costlier higher-resolution (HR) imagery be acquired? We present a novel scheme to address these challenges by introducing three components: (1) A technique to distill knowledge from models trained on HR imagery to recognition models that operate on imagery of lower resolution (LR), (2) a sampling strategy for HR imagery based on model disagreement, and (3) an LLM-based approach for inferring concept "scale". With these components we present a system to efficiently perform scale-aware recognition in satellite imagery, improving accuracy over single-scale inference while following budget constraints. Our novel approach offers up to a 26.3% improvement over entirely HR baselines, using 76.3% fewer HR images.
Authors: Johanna Karras, Yingwei Li, Nan Liu, Luyang Zhu, Innfarn Yoo, Andreas Lugmayr, Chris Lee, Ira Kemelmacher-Shlizerman
Abstract: We present Fashion-VDM, a video diffusion model (VDM) for generating virtual try-on videos. Given an input garment image and person video, our method aims to generate a high-quality try-on video of the person wearing the given garment, while preserving the person's identity and motion. Image-based virtual try-on has shown impressive results; however, existing video virtual try-on (VVT) methods are still lacking garment details and temporal consistency. To address these issues, we propose a diffusion-based architecture for video virtual try-on, split classifier-free guidance for increased control over the conditioning inputs, and a progressive temporal training strategy for single-pass 64-frame, 512px video generation. We also demonstrate the effectiveness of joint image-video training for video try-on, especially when video data is limited. Our qualitative and quantitative experiments show that our approach sets the new state-of-the-art for video virtual try-on. For additional results, visit our project page: https://johannakarras.github.io/Fashion-VDM.
Authors: Shaohua Liu, Junzhe Lu, Zuoya Gu, Jiajun Li, Yue Deng
Abstract: Representing underwater 3D scenes is a valuable yet complex task, as attenuation and scattering effects during underwater imaging significantly couple the information of the objects and the water. This coupling presents a significant challenge for existing methods in effectively representing both the objects and the water medium simultaneously. To address this challenge, we propose Aquatic-GS, a hybrid 3D representation approach for underwater scenes that effectively represents both the objects and the water medium. Specifically, we construct a Neural Water Field (NWF) to implicitly model the water parameters, while extending the latest 3D Gaussian Splatting (3DGS) to model the objects explicitly. Both components are integrated through a physics-based underwater image formation model to represent complex underwater scenes. Moreover, to construct more precise scene geometry and details, we design a Depth-Guided Optimization (DGO) mechanism that uses a pseudo-depth map as auxiliary guidance. After optimization, Aquatic-GS enables the rendering of novel underwater viewpoints and supports restoring the true appearance of underwater scenes, as if the water medium were absent. Extensive experiments on both simulated and real-world datasets demonstrate that Aquatic-GS surpasses state-of-the-art underwater 3D representation methods, achieving better rendering quality and real-time rendering performance with a 410x increase in speed. Furthermore, regarding underwater image restoration, Aquatic-GS outperforms representative dewatering methods in color correction, detail recovery, and stability. Our models, code, and datasets can be accessed at https://aquaticgs.github.io.
Authors: Lorenzo Basile, Valentino Maiorca, Luca Bortolussi, Emanuele Rodol\`a, Francesco Locatello
Abstract: When examined through the lens of their residual streams, a puzzling property emerges in transformer networks: residual contributions (e.g., attention heads) sometimes specialize in specific tasks or input attributes. In this paper, we analyze this phenomenon in vision transformers, focusing on the spectral geometry of residuals, and explore its implications for modality alignment in vision-language models. First, we link it to the intrinsically low-dimensional structure of visual head representations, zooming into their principal components and showing that they encode specialized roles across a wide variety of input data distributions. Then, we analyze the effect of head specialization in multimodal models, focusing on how improved alignment between text and specialized heads impacts zero-shot classification performance. This specialization-performance link consistently holds across diverse pre-training data, network sizes, and objectives, demonstrating a powerful new mechanism for boosting zero-shot classification through targeted alignment. Ultimately, we translate these insights into actionable terms by introducing ResiDual, a technique for spectral alignment of the residual stream. Much like panning for gold, it lets the noise from irrelevant unit principal components (i.e., attributes) wash away to amplify task-relevant ones. Remarkably, this dual perspective on modality alignment yields fine-tuning level performances on different data distributions while modeling an extremely interpretable and parameter-efficient transformation, as we extensively show on more than 50 (pre-trained network, dataset) pairs.
Authors: Maxwell Meyer, Jack Spruyt
Abstract: Transformers and their derivatives have achieved state-of-the-art performance across text, vision, and speech recognition tasks. However, minimal effort has been made to train transformers capable of evaluating the output quality of other models. This paper examines SwinV2-based reward models, called the Input-Output Transformer (IO Transformer) and the Output Transformer. These reward models can be leveraged for tasks such as inference quality evaluation, data categorization, and policy optimization. Our experiments demonstrate highly accurate model output quality assessment across domains where the output is entirely dependent on the input, with the IO Transformer achieving perfect evaluation accuracy on the Change Dataset 25 (CD25). We also explore modified Swin V2 architectures. Ultimately Swin V2 remains on top with a score of 95.41 % on the IO Segmentation Dataset, outperforming the IO Transformer in scenarios where the output is not entirely dependent on the input. Our work expands the application of transformer architectures to reward modeling in computer vision and provides critical insights into optimizing these models for various tasks.
Authors: Kyung-hwan Lee, Kyung-tae Kim
Abstract: Recent advances in deep learning architectures have enabled efficient and accurate classification of pre-trained targets in Synthetic Aperture Radar (SAR) images. Nevertheless, the presence of unknown targets in real battlefield scenarios is unavoidable, resulting in misclassification and reducing the accuracy of the classifier. Over the past decades, various feature-based out-of-distribution (OOD) approaches have been developed to address this issue, yet defining the decision boundary between known and unknown targets remains challenging. Additionally, unlike optical images, detecting unknown targets in SAR imagery is further complicated by high speckle noise, the presence of clutter, and the inherent similarities in back-scattered microwave signals. In this work, we propose transforming feature-based OOD detection into a class-localized feature-residual-based approach, demonstrating that this method can improve stability across varying unknown targets' distribution conditions. Transforming feature-based OOD detection into a residual-based framework offers a more robust reference space for distinguishing between in-distribution (ID) and OOD data, particularly within the unique characteristics of SAR imagery. This adaptive residual transformation method standardizes feature-based inputs into distributional representations, enhancing OOD detection in noisy, low-information images. Our approach demonstrates promising performance in real-world SAR scenarios, effectively adapting to the high levels of noise and clutter inherent in these environments. These findings highlight the practical relevance of residual-based OOD detection for SAR applications and suggest a foundation for further advancements in unknown target detection in complex, operational settings.
Authors: Torin Gerhart, Justin Sunu, Ekaterina Merkurjev, Jen-Mei Chang, Jerome Gilles, Andrea L. Bertozzi
Abstract: Automated detection of chemical plumes presents a segmentation challenge. The segmentation problem for gas plumes is difficult due to the diffusive nature of the cloud. The advantage of considering hyperspectral images in the gas plume detection problem over the conventional RGB imagery is the presence of non-visual data, allowing for a richer representation of information. In this paper we present an effective method of visualizing hyperspectral video sequences containing chemical plumes and investigate the effectiveness of segmentation techniques on these post-processed videos. Our approach uses a combination of dimension reduction and histogram equalization to prepare the hyperspectral videos for segmentation. First, Principal Components Analysis (PCA) is used to reduce the dimension of the entire video sequence. This is done by projecting each pixel onto the first few Principal Components resulting in a type of spectral filter. Next, a Midway method for histogram equalization is used. These methods redistribute the intensity values in order to reduce flicker between frames. This properly prepares these high-dimensional video sequences for more traditional segmentation techniques. We compare the ability of various clustering techniques to properly segment the chemical plume. These include K-means, spectral clustering, and the Ginzburg-Landau functional.
Authors: Sraavya Sambara (Harvard University), Serena Zhang (Stanford University), Oishi Banerjee (Harvard University), Julian Acosta (Harvard University), John Fahrner (Harvard University), Pranav Rajpurkar (Harvard University)
Abstract: Generating accurate radiology reports from medical images is a clinically important but challenging task. While current Vision Language Models (VLMs) show promise, they are prone to generating hallucinations, potentially compromising patient care. We introduce RadFlag, a black-box method to enhance the accuracy of radiology report generation. Our method uses a sampling-based flagging technique to find hallucinatory generations that should be removed. We first sample multiple reports at varying temperatures and then use a Large Language Model (LLM) to identify claims that are not consistently supported across samples, indicating that the model has low confidence in those claims. Using a calibrated threshold, we flag a fraction of these claims as likely hallucinations, which should undergo extra review or be automatically rejected. Our method achieves high precision when identifying both individual hallucinatory sentences and reports that contain hallucinations. As an easy-to-use, black-box system that only requires access to a model's temperature parameter, RadFlag is compatible with a wide range of radiology report generation models and has the potential to broadly improve the quality of automated radiology reporting.
Authors: Wei Chow, Juncheng Li, Qifan Yu, Kaihang Pan, Hao Fei, Zhiqi Ge, Shuai Yang, Siliang Tang, Hanwang Zhang, Qianru Sun
Abstract: In recent times, Vision-Language Models (VLMs) have been trained under two predominant paradigms. Generative training has enabled Multimodal Large Language Models (MLLMs) to tackle various complex tasks, yet issues such as hallucinations and weak object discrimination persist. Discriminative training, exemplified by models like CLIP, excels in zero-shot image-text classification and retrieval, yet struggles with complex scenarios requiring fine-grained semantic differentiation. This paper addresses these challenges by proposing a unified approach that integrates the strengths of both paradigms. Considering interleaved image-text sequences as the general format of input samples, we introduce a structure-induced training strategy that imposes semantic relationships between input samples and the MLLM's hidden state. This approach enhances the MLLM's ability to capture global semantics and distinguish fine-grained semantics. By leveraging dynamic sequence alignment within the Dynamic Time Warping framework and integrating a novel kernel for fine-grained semantic differentiation, our method effectively balances generative and discriminative tasks. Extensive experiments demonstrate the effectiveness of our approach, achieving state-of-the-art results in multiple generative tasks, especially those requiring cognitive and discrimination abilities. Additionally, our method surpasses discriminative benchmarks in interleaved and fine-grained retrieval tasks. By employing a retrieval-augmented generation strategy, our approach further enhances performance in some generative tasks within one model, offering a promising direction for future research in vision-language modeling.
Authors: Shengxun Wei, Zan Gao, Yibo Zhao, Weili Guan
Abstract: Cloth-changing person re-identification is a subject closer to the real world, which focuses on solving the problem of person re-identification after pedestrians change clothes. The primary challenge in this field is to overcome the complex interplay between intra-class and inter-class variations and to identify features that remain unaffected by changes in appearance. Sufficient data collection for model training would significantly aid in addressing this problem. However, it is challenging to gather diverse datasets in practice. Current methods focus on implicitly learning identity information from the original image or introducing additional auxiliary models, which are largely limited by the quality of the image and the performance of the additional model. To address these issues, inspired by prompt learning, we propose a novel multiple information prompt learning (MIPL) scheme for cloth-changing person ReID, which learns identity robust features through the common prompt guidance of multiple messages. Specifically, the clothing information stripping (CIS) module is designed to decouple the clothing information from the original RGB image features to counteract the influence of clothing appearance. The Bio-guided attention (BGA) module is proposed to increase the learning intensity of the model for key information. A dual-length hybrid patch (DHP) module is employed to make the features have diverse coverage to minimize the impact of feature bias. Extensive experiments demonstrate that the proposed method outperforms all state-of-the-art methods on the LTCC, Celeb-reID, Celeb-reID-light, and CSCC datasets, achieving rank-1 scores of 74.8%, 73.3%, 66.0%, and 88.1%, respectively. When compared to AIM (CVPR23), ACID (TIP23), and SCNet (MM23), MIPL achieves rank-1 improvements of 11.3%, 13.8%, and 7.9%, respectively, on the PRCC dataset.
Authors: Xintao Jiang, Yaosen Chen, Siqin Zhang, Wei Wang, Xuming Wen
Abstract: Video color style transfer aims to transform the color style of an original video by using a reference style image. Most existing methods employ neural networks, which come with challenges like opaque transfer processes and limited user control over the outcomes. Typically, users cannot fine-tune the resulting images or videos. To tackle this issue, we introduce a method that predicts specific parameters for color style transfer using two images. Initially, we train a neural network to learn the corresponding color adjustment parameters. When applying style transfer to a video, we fine-tune the network with key frames from the video and the chosen style image, generating precise transformation parameters. These are then applied to convert the color style of both images and videos. Our experimental results demonstrate that our algorithm surpasses current methods in color style transfer quality. Moreover, each parameter in our method has a specific, interpretable meaning, enabling users to understand the color style transfer process and allowing them to perform manual fine-tuning if desired.
Authors: Xiaotian Li, Baojie Fan, Jiandong Tian, Huijie Fan
Abstract: Recent years have witnessed the remarkable progress of 3D multi-modality object detection methods based on the Bird's-Eye-View (BEV) perspective. However, most of them overlook the complementary interaction and guidance between LiDAR and camera. In this work, we propose a novel multi-modality 3D objection detection method, named GAFusion, with LiDAR-guided global interaction and adaptive fusion. Specifically, we introduce sparse depth guidance (SDG) and LiDAR occupancy guidance (LOG) to generate 3D features with sufficient depth information. In the following, LiDAR-guided adaptive fusion transformer (LGAFT) is developed to adaptively enhance the interaction of different modal BEV features from a global perspective. Meanwhile, additional downsampling with sparse height compression and multi-scale dual-path transformer (MSDPT) are designed to enlarge the receptive fields of different modal features. Finally, a temporal fusion module is introduced to aggregate features from previous frames. GAFusion achieves state-of-the-art 3D object detection results with 73.6$\%$ mAP and 74.9$\%$ NDS on the nuScenes test set.
Authors: Mengcheng Li, Mingbao Lin, Fei Chao, Chia-Wen Lin, Rongrong Ji
Abstract: In this paper, we propose TextDestroyer, the first training- and annotation-free method for scene text destruction using a pre-trained diffusion model. Existing scene text removal models require complex annotation and retraining, and may leave faint yet recognizable text information, compromising privacy protection and content concealment. TextDestroyer addresses these issues by employing a three-stage hierarchical process to obtain accurate text masks. Our method scrambles text areas in the latent start code using a Gaussian distribution before reconstruction. During the diffusion denoising process, self-attention key and value are referenced from the original latent to restore the compromised background. Latent codes saved at each inversion step are used for replacement during reconstruction, ensuring perfect background restoration. The advantages of TextDestroyer include: (1) it eliminates labor-intensive data annotation and resource-intensive training; (2) it achieves more thorough text destruction, preventing recognizable traces; and (3) it demonstrates better generalization capabilities, performing well on both real-world scenes and generated images.
Authors: Li Liu, Diji Yang, Sijia Zhong, Kalyana Suma Sree Tholeti, Lei Ding, Yi Zhang, Leilani H. Gilpin
Abstract: In question-answering scenarios, humans can assess whether the available information is sufficient and seek additional information if necessary, rather than providing a forced answer. In contrast, Vision Language Models (VLMs) typically generate direct, one-shot responses without evaluating the sufficiency of the information. To investigate this gap, we identify a critical and challenging task in the Visual Question Answering (VQA) scenario: can VLMs indicate how to adjust an image when the visual information is insufficient to answer a question? This capability is especially valuable for assisting visually impaired individuals who often need guidance to capture images correctly. To evaluate this capability of current VLMs, we introduce a human-labeled dataset as a benchmark for this task. Additionally, we present an automated framework that generates synthetic training data by simulating ``where to know'' scenarios. Our empirical results show significant performance improvements in mainstream VLMs when fine-tuned with this synthetic data. This study demonstrates the potential to narrow the gap between information assessment and acquisition in VLMs, bringing their performance closer to humans.
Authors: Zhiyu Xie, Yuqing Zhang, Xiangjun Tang, Yiqian Wu, Dehan Chen, Gongsheng Li, Xaogang Jin
Abstract: Style-guided texture generation aims to generate a texture that is harmonious with both the style of the reference image and the geometry of the input mesh, given a reference style image and a 3D mesh with its text description. Although diffusion-based 3D texture generation methods, such as distillation sampling, have numerous promising applications in stylized games and films, it requires addressing two challenges: 1) decouple style and content completely from the reference image for 3D models, and 2) align the generated texture with the color tone, style of the reference image, and the given text prompt. To this end, we introduce StyleTex, an innovative diffusion-model-based framework for creating stylized textures for 3D models. Our key insight is to decouple style information from the reference image while disregarding content in diffusion-based distillation sampling. Specifically, given a reference image, we first decompose its style feature from the image CLIP embedding by subtracting the embedding's orthogonal projection in the direction of the content feature, which is represented by a text CLIP embedding. Our novel approach to disentangling the reference image's style and content information allows us to generate distinct style and content features. We then inject the style feature into the cross-attention mechanism to incorporate it into the generation process, while utilizing the content feature as a negative prompt to further dissociate content information. Finally, we incorporate these strategies into StyleTex to obtain stylized textures. The resulting textures generated by StyleTex retain the style of the reference image, while also aligning with the text prompts and intrinsic details of the given 3D mesh. Quantitative and qualitative experiments show that our method outperforms existing baseline methods by a significant margin.
Authors: Yinxuan Huang, Chengmin Gao, Bin Li, Xiangyang Xue
Abstract: Given the complexities inherent in visual scenes, such as object occlusion, a comprehensive understanding often requires observation from multiple viewpoints. Existing multi-viewpoint object-centric learning methods typically employ random or sequential viewpoint selection strategies. While applicable across various scenes, these strategies may not always be ideal, as certain scenes could benefit more from specific viewpoints. To address this limitation, we propose a novel active viewpoint selection strategy. This strategy predicts images from unknown viewpoints based on information from observation images for each scene. It then compares the object-centric representations extracted from both viewpoints and selects the unknown viewpoint with the largest disparity, indicating the greatest gain in information, as the next observation viewpoint. Through experiments on various datasets, we demonstrate the effectiveness of our active viewpoint selection strategy, significantly enhancing segmentation and reconstruction performance compared to random viewpoint selection. Moreover, our method can accurately predict images from unknown viewpoints.
Authors: Naufal Suryanto, Andro Aprila Adiputra, Ahmada Yusril Kadiptya, Thi-Thu-Huong Le, Derry Pratama, Yongsu Kim, Howon Kim
Abstract: Recent advancements in generative AI, particularly diffusion-based image editing, have enabled the transformation of images into highly realistic scenes using only text instructions. This technology offers significant potential for generating diverse synthetic datasets to evaluate model robustness. In this paper, we introduce Cityscape-Adverse, a benchmark that employs diffusion-based image editing to simulate eight adverse conditions, including variations in weather, lighting, and seasons, while preserving the original semantic labels. We evaluate the reliability of diffusion-based models in generating realistic scene modifications and assess the performance of state-of-the-art CNN and Transformer-based semantic segmentation models under these challenging conditions. Additionally, we analyze which modifications have the greatest impact on model performance and explore how training on synthetic datasets can improve robustness in real-world adverse scenarios. Our results demonstrate that all tested models, particularly CNN-based architectures, experienced significant performance degradation under extreme conditions, while Transformer-based models exhibited greater resilience. We verify that models trained on Cityscape-Adverse show significantly enhanced resilience when applied to unseen domains. Code and datasets will be released at https://github.com/naufalso/cityscape-adverse.
Authors: Donghyun Kim, Hyeonkyeong Kwon, Yumin Kim, Seong Jae Hwang
Abstract: 3D point clouds are increasingly vital for applications like autonomous driving and robotics, yet the raw data captured by sensors often suffer from noise and sparsity, creating challenges for downstream tasks. Consequently, point cloud upsampling becomes essential for improving density and uniformity, with recent approaches showing promise by projecting randomly generated query points onto the underlying surface of sparse point clouds. However, these methods often result in outliers, non-uniformity, and difficulties in handling regions with high curvature and intricate structures. In this work, we address these challenges by introducing the Progressive Local Surface Estimator (PLSE), which more effectively captures local features in complex regions through a curvature-based sampling technique that selectively targets high-curvature areas. Additionally, we incorporate a curriculum learning strategy that leverages the curvature distribution within the point cloud to naturally assess the sample difficulty, enabling curriculum learning on point cloud data for the first time. The experimental results demonstrate that our approach significantly outperforms existing methods, achieving high-quality, dense point clouds with superior accuracy and detail.
Authors: Jianhua Sun, Yuxuan Li, Longfei Xu, Nange Wang, Jiude Wei, Yining Zhang, Cewu Lu
Abstract: We present ConceptFactory, a novel scope to facilitate more efficient annotation of 3D object knowledge by recognizing 3D objects through generalized concepts (i.e. object conceptualization), aiming at promoting machine intelligence to learn comprehensive object knowledge from both vision and robotics aspects. This idea originates from the findings in human cognition research that the perceptual recognition of objects can be explained as a process of arranging generalized geometric components (e.g. cuboids and cylinders). ConceptFactory consists of two critical parts: i) ConceptFactory Suite, a unified toolbox that adopts Standard Concept Template Library (STL-C) to drive a web-based platform for object conceptualization, and ii) ConceptFactory Asset, a large collection of conceptualized objects acquired using ConceptFactory suite. Our approach enables researchers to effortlessly acquire or customize extensive varieties of object knowledge to comprehensively study different object understanding tasks. We validate our idea on a wide range of benchmark tasks from both vision and robotics aspects with state-of-the-art algorithms, demonstrating the high quality and versatility of annotations provided by our approach. Our website is available at https://apeirony.github.io/ConceptFactory.
Authors: Jie Wang, Tingfa Xu, Lihe Ding, Jianan Li
Abstract: Achieving robust 3D perception in the face of corrupted data presents an challenging hurdle within 3D vision research. Contemporary transformer-based point cloud recognition models, albeit advanced, tend to overfit to specific patterns, consequently undermining their robustness against corruption. In this work, we introduce the Target-Guided Adversarial Point Cloud Transformer, termed APCT, a novel architecture designed to augment global structure capture through an adversarial feature erasing mechanism predicated on patterns discerned at each step during training. Specifically, APCT integrates an Adversarial Significance Identifier and a Target-guided Promptor. The Adversarial Significance Identifier, is tasked with discerning token significance by integrating global contextual analysis, utilizing a structural salience index algorithm alongside an auxiliary supervisory mechanism. The Target-guided Promptor, is responsible for accentuating the propensity for token discard within the self-attention mechanism, utilizing the value derived above, consequently directing the model attention towards alternative segments in subsequent stages. By iteratively applying this strategy in multiple steps during training, the network progressively identifies and integrates an expanded array of object-associated patterns. Extensive experiments demonstrate that our method achieves state-of-the-art results on multiple corruption benchmarks.
Authors: Lianjun Liu
Abstract: Underwater instance segmentation is a fundamental and critical step in various underwater vision tasks. However, the decline in image quality caused by complex underwater environments presents significant challenges to existing segmentation models. While the state-of-the-art USIS-SAM model has demonstrated impressive performance, it struggles to effectively adapt to feature variations across different channels in addressing issues such as light attenuation, color distortion, and complex backgrounds. This limitation hampers its segmentation performance in challenging underwater scenarios. To address these issues, we propose the MarineVision Adapter (MV-Adapter). This module introduces an adaptive channel attention mechanism that enables the model to dynamically adjust the feature weights of each channel based on the characteristics of underwater images. By adaptively weighting features, the model can effectively handle challenges such as light attenuation, color shifts, and complex backgrounds. Experimental results show that integrating the MV-Adapter module into the USIS-SAM network architecture further improves the model's overall performance, especially in high-precision segmentation tasks. On the USIS10K dataset, the module achieves improvements in key metrics such as mAP, AP50, and AP75 compared to competitive baseline models.
Authors: Yuchen Zheng, Yuxin Jing, Jufeng Zhao, Guangmang Cui
Abstract: Drone-based target detection presents inherent challenges, such as the high density and overlap of targets in drone-based images, as well as the blurriness of targets under varying lighting conditions, which complicates identification. Traditional methods often struggle to recognize numerous densely packed small targets under complex background. To address these challenges, we propose LAM-YOLO, an object detection model specifically designed for drone-based. First, we introduce a light-occlusion attention mechanism to enhance the visibility of small targets under different lighting conditions. Meanwhile, we incroporate incorporate Involution modules to improve interaction among feature layers. Second, we utilize an improved SIB-IoU as the regression loss function to accelerate model convergence and enhance localization accuracy. Finally, we implement a novel detection strategy that introduces two auxiliary detection heads for identifying smaller-scale targets.Our quantitative results demonstrate that LAM-YOLO outperforms methods such as Faster R-CNN, YOLOv9, and YOLOv10 in terms of mAP@0.5 and mAP@0.5:0.95 on the VisDrone2019 public dataset. Compared to the original YOLOv8, the average precision increases by 7.1\%. Additionally, the proposed SIB-IoU loss function shows improved faster convergence speed during training and improved average precision over the traditional loss function.
Authors: Hairuo Hu, Haiyong Cong, Zhuyu Shao, Yubo Bi, Jinghao Liu
Abstract: In the context of firefighting and rescue operations, a cross-modal semantic segmentation model based on a single-chip millimeter-wave (mmWave) radar for indoor environmental perception is proposed and discussed. To efficiently obtain high-quality labels, an automatic label generation method utilizing LiDAR point clouds and occupancy grid maps is introduced. The proposed segmentation model is based on U-Net. A spatial attention module is incorporated, which enhanced the performance of the mode. The results demonstrate that cross-modal semantic segmentation provides a more intuitive and accurate representation of indoor environments. Unlike traditional methods, the model's segmentation performance is minimally affected by azimuth. Although performance declines with increasing distance, this can be mitigated by a well-designed model. Additionally, it was found that using raw ADC data as input is ineffective; compared to RA tensors, RD tensors are more suitable for the proposed model.
Authors: Jongmin Lee, Minsu Cho
Abstract: Determining the 3D orientations of an object in an image, known as single-image pose estimation, is a crucial task in 3D vision applications. Existing methods typically learn 3D rotations parametrized in the spatial domain using Euler angles or quaternions, but these representations often introduce discontinuities and singularities. SO(3)-equivariant networks enable the structured capture of pose patterns with data-efficient learning, but the parametrizations in spatial domain are incompatible with their architecture, particularly spherical CNNs, which operate in the frequency domain to enhance computational efficiency. To overcome these issues, we propose a frequency-domain approach that directly predicts Wigner-D coefficients for 3D rotation regression, aligning with the operations of spherical CNNs. Our SO(3)-equivariant pose harmonics predictor overcomes the limitations of spatial parameterizations, ensuring consistent pose estimation under arbitrary rotations. Trained with a frequency-domain regression loss, our method achieves state-of-the-art results on benchmarks such as ModelNet10-SO(3) and PASCAL3D+, with significant improvements in accuracy, robustness, and data efficiency.
Authors: Sourav Modak, Anthony Stein
Abstract: In automated crop protection tasks such as weed control, disease diagnosis, and pest monitoring, deep learning has demonstrated significant potential. However, these advanced models rely heavily on high-quality, diverse datasets, often limited and costly in agricultural settings. Traditional data augmentation can increase dataset volume but usually lacks the real-world variability needed for robust training. This study presents a new approach for generating synthetic images to improve deep learning-based object detection models for intelligent weed control. Our GenAI-based image generation pipeline integrates the Segment Anything Model (SAM) for zero-shot domain adaptation with a text-to-image Stable Diffusion Model, enabling the creation of synthetic images that capture diverse real-world conditions. We evaluate these synthetic datasets using lightweight YOLO models, measuring data efficiency with mAP50 and mAP50-95 scores across varying proportions of real and synthetic data. Notably, YOLO models trained on datasets with 10% synthetic and 90% real images generally demonstrate superior mAP50 and mAP50-95 scores compared to those trained solely on real images. This approach not only reduces dependence on extensive real-world datasets but also enhances predictive performance. The integration of this approach opens opportunities for achieving continual self-improvement of perception modules in intelligent technical systems.
Authors: J. Seiffarth, L. Bl\"obaum, R. D. Paul, N. Friederich, A. J. Yamachui Sitcheu, R. Mikut, H. Scharr, A. Gr\"unberger, K. N\"oh
Abstract: Tracking the development of living cells in live-cell time-lapses reveals crucial insights into single-cell behavior and presents tremendous potential for biomedical and biotechnological applications. In microbial live-cell imaging (MLCI), a few to thousands of cells have to be detected and tracked within dozens of growing cell colonies. The challenge of tracking cells is heavily influenced by the experiment parameters, namely the imaging interval and maximal cell number. For now, tracking benchmarks are not widely available in MLCI and the effect of these parameters on the tracking performance are not yet known. Therefore, we present the largest publicly available and annotated dataset for MLCI, containing more than 1.4 million cell instances, 29k cell tracks, and 14k cell divisions. With this dataset at hand, we generalize existing tracking metrics to incorporate relevant imaging and experiment parameters into experiment-aware metrics. These metrics reveal that current cell tracking methods crucially depend on the choice of the experiment parameters, where their performance deteriorates at high imaging intervals and large cell colonies. Thus, our new benchmark quantifies the influence of experiment parameters on the tracking quality, and gives the opportunity to develop new data-driven methods that generalize across imaging and experiment parameters. The benchmark dataset is publicly available at https://zenodo.org/doi/10.5281/zenodo.7260136.
Authors: Gianluca Mancusi, Mattia Bernardi, Aniello Panariello, Angelo Porrello, Rita Cucchiara, Simone Calderara
Abstract: End-to-end transformer-based trackers have achieved remarkable performance on most human-related datasets. However, training these trackers in heterogeneous scenarios poses significant challenges, including negative interference - where the model learns conflicting scene-specific parameters - and limited domain generalization, which often necessitates expensive fine-tuning to adapt the models to new domains. In response to these challenges, we introduce Parameter-efficient Scenario-specific Tracking Architecture (PASTA), a novel framework that combines Parameter-Efficient Fine-Tuning (PEFT) and Modular Deep Learning (MDL). Specifically, we define key scenario attributes (e.g, camera-viewpoint, lighting condition) and train specialized PEFT modules for each attribute. These expert modules are combined in parameter space, enabling systematic generalization to new domains without increasing inference time. Extensive experiments on MOTSynth, along with zero-shot evaluations on MOT17 and PersonPath22 demonstrate that a neural tracker built from carefully selected modules surpasses its monolithic counterpart. We release models and code.
Authors: Ruiyu Xia, Jianqiang Li, Xi Xu, Guanghui Fu
Abstract: Ocular Myasthenia Gravis (OMG) is a rare and challenging disease to detect in its early stages, but symptoms often first appear in the eye muscles, such as drooping eyelids and double vision. Ocular images can be used for early diagnosis by segmenting different regions, such as the sclera, iris, and pupil, which allows for the calculation of area ratios to support accurate medical assessments. However, no publicly available dataset and tools currently exist for this purpose. To address this, we propose a new topology and intersection-union constrained loss function (TIU loss) that improves performance using small training datasets. We conducted experiments on a public dataset consisting of 55 subjects and 2,197 images. Our proposed method outperformed two widely used loss functions across three deep learning networks, achieving a mean Dice score of 83.12% [82.47%, 83.81%] with a 95% bootstrap confidence interval. In a low-percentage training scenario (10% of the training data), our approach showed an 8.32% improvement in Dice score compared to the baseline. Additionally, we evaluated the method in a clinical setting with 47 subjects and 501 images, achieving a Dice score of 64.44% [63.22%, 65.62%]. We did observe some bias when applying the model in clinical settings. These results demonstrate that the proposed method is accurate, and our code along with the trained model is publicly available.
Authors: Valentina Vadori, Antonella Peruffo, Jean-Marie Gra\"ic, Livio Finos, Enrico Grisan
Abstract: This study addresses the challenge of classifying cell shapes from noisy contours, such as those obtained through cell instance segmentation of histological images. We assess the performance of various features for shape classification, including Elliptical Fourier Descriptors, curvature features, and lower dimensional representations. Using an annotated synthetic dataset of noisy contours, we identify the most suitable shape descriptors and apply them to a set of real images for qualitative analysis. Our aim is to provide a comprehensive evaluation of descriptors for classifying cell shapes, which can support cell type identification and tissue characterization-critical tasks in both biological research and histopathological assessments.
Authors: Curtis Wigington
Abstract: Document capture applications on smartphones have emerged as popular tools for digitizing documents. For many individuals, capturing documents with their smartphones is more convenient than using dedicated photocopiers or scanners, even if the quality of digitization is lower. However, using a smartphone for digitization can become excessively time-consuming and tedious when a user needs to digitize a document with multiple pages. In this work, we propose a novel approach to automatically scan multi-page documents from a video stream as the user turns through the pages of the document. Unlike previous methods that required constrained settings such as mounting the phone on a tripod, our technique is designed to allow the user to hold the phone in their hand. Our technique is trained to be robust to the motion and instability inherent in handheld scanning. Our primary contributions in this work include: (1) an efficient, on-device deep learning model that is accurate and robust for handheld scanning, (2) a novel data collection and annotation technique for video document scanning, and (3) state-of-the-art results on the PUCIT page turn dataset.
Authors: Antoine P. Sanner, Jonathan Stieber, Nils F. Grauhan, Suam Kim, Marc A. Brockmann, Ahmed E. Othman, Anirban Mukhopadhyay
Abstract: Intracranial Hemorrhage is a potentially lethal condition whose manifestation is vastly diverse and shifts across clinical centers worldwide. Deep-learning-based solutions are starting to model complex relations between brain structures, but still struggle to generalize. While gathering more diverse data is the most natural approach, privacy regulations often limit the sharing of medical data. We propose the first application of Federated Scene Graph Generation. We show that our models can leverage the increased training data diversity. For Scene Graph Generation, they can recall up to 20% more clinically relevant relations across datasets compared to models trained on a single centralized dataset. Learning structured data representation in a federated setting can open the way to the development of new methods that can leverage this finer information to regularize across clients more effectively.
Authors: Li Li
Abstract: 3D LiDAR point cloud data is crucial for scene perception in computer vision, robotics, and autonomous driving. Geometric and semantic scene understanding, involving 3D point clouds, is essential for advancing autonomous driving technologies. However, significant challenges remain, particularly in improving the overall accuracy (e.g., segmentation accuracy, depth estimation accuracy, etc.) and efficiency of these systems. To address the challenge in terms of accuracy related to LiDAR-based tasks, we present DurLAR, the first high-fidelity 128-channel 3D LiDAR dataset featuring panoramic ambient (near infrared) and reflectivity imagery. To improve efficiency in 3D segmentation while ensuring the accuracy, we propose a novel pipeline that employs a smaller architecture, requiring fewer ground-truth annotations while achieving superior segmentation accuracy compared to contemporary approaches. To improve the segmentation accuracy, we introduce Range-Aware Pointwise Distance Distribution (RAPiD) features and the associated RAPiD-Seg architecture. All contributions have been accepted by peer-reviewed conferences, underscoring the advancements in both accuracy and efficiency in 3D LiDAR applications for autonomous driving. Full abstract: https://etheses.dur.ac.uk/15738/.
Authors: Xiang Li, Cheng Chen, Yuan-yao Lou, Mustafa Abdallah, Kwang Taik Kim, Saurabh Bagchi
Abstract: Multi-Object Tracking (MOT) poses significant challenges in computer vision. Despite its wide application in robotics, autonomous driving, and smart manufacturing, there is limited literature addressing the specific challenges of running MOT on embedded devices. State-of-the-art MOT trackers designed for high-end GPUs often experience low processing rates (<11fps) when deployed on embedded devices. Existing MOT frameworks for embedded devices proposed strategies such as fusing the detector model with the feature embedding model to reduce inference latency or combining different trackers to improve tracking accuracy, but tend to compromise one for the other. This paper introduces HopTrack, a real-time multi-object tracking system tailored for embedded devices. Our system employs a novel discretized static and dynamic matching approach along with an innovative content-aware dynamic sampling technique to enhance tracking accuracy while meeting the real-time requirement. Compared with the best high-end GPU modified baseline Byte (Embed) and the best existing baseline on embedded devices MobileNet-JDE, HopTrack achieves a processing speed of up to 39.29 fps on NVIDIA AGX Xavier with a multi-object tracking accuracy (MOTA) of up to 63.12% on the MOT16 benchmark, outperforming both counterparts by 2.15% and 4.82%, respectively. Additionally, the accuracy improvement is coupled with the reduction in energy consumption (20.8%), power (5%), and memory usage (8%), which are crucial resources on embedded devices. HopTrack is also detector agnostic allowing the flexibility of plug-and-play.
Authors: Huancheng Chen, Jingtao Li, Nidham Gazagnadou, Weiming Zhuang, Chen Chen, Lingjuan Lyu
Abstract: In the era of foundation models, we revisit continual learning~(CL), which aims to enable vision transformers (ViTs) to learn new tasks over time. However, as the scale of these models increases, catastrophic forgetting remains a persistent challenge, particularly in the presence of significant domain shifts across tasks. Recent studies highlight a crossover between CL techniques and parameter-efficient fine-tuning (PEFT), which focuses on fine-tuning only a small set of trainable parameters to adapt to downstream tasks, such as low-rank adaptation (LoRA). While LoRA achieves faster convergence and requires fewer trainable parameters, it has seldom been explored in the context of continual learning. To address this gap, we propose a novel PEFT-CL method called Dual Low-Rank Adaptation (DualLoRA), which introduces both an orthogonal LoRA adapter and a residual LoRA adapter parallel to pre-trained weights in each layer. These components are orchestrated by a dynamic memory mechanism to strike a balance between stability and plasticity. The orthogonal LoRA adapter's parameters are updated in an orthogonal subspace of previous tasks to mitigate catastrophic forgetting, while the residual LoRA adapter's parameters are updated in the residual subspace spanned by task-specific bases without interaction across tasks, offering complementary capabilities for fine-tuning new tasks. On ViT-based models, we demonstrate that DualLoRA offers significant advantages in accuracy, inference speed, and memory efficiency over existing CL methods across multiple benchmarks.
Authors: Beomyoung Kim, Chanyong Shin, Joonhyun Jeong, Hyungsik Jung, Se-Yun Lee, Sewhan Chun, Dong-Hyun Hwang, Joonsang Yu
Abstract: The recent segmentation foundation model, Segment Anything Model (SAM), exhibits strong zero-shot segmentation capabilities, but it falls short in generating fine-grained precise masks. To address this limitation, we propose a novel zero-shot image matting model, called ZIM, with two key contributions: First, we develop a label converter that transforms segmentation labels into detailed matte labels, constructing the new SA1B-Matte dataset without costly manual annotations. Training SAM with this dataset enables it to generate precise matte masks while maintaining its zero-shot capability. Second, we design the zero-shot matting model equipped with a hierarchical pixel decoder to enhance mask representation, along with a prompt-aware masked attention mechanism to improve performance by enabling the model to focus on regions specified by visual prompts. We evaluate ZIM using the newly introduced MicroMat-3K test set, which contains high-quality micro-level matte labels. Experimental results show that ZIM outperforms existing methods in fine-grained mask generation and zero-shot generalization. Furthermore, we demonstrate the versatility of ZIM in various downstream tasks requiring precise masks, such as image inpainting and 3D NeRF. Our contributions provide a robust foundation for advancing zero-shot matting and its downstream applications across a wide range of computer vision tasks. The code is available at \url{https://github.com/naver-ai/ZIM}.
Authors: Yuyan Zhang, Derya Soydaner, Fatemeh Behrad, Lisa Ko{\ss}mann, Johan Wagemans
Abstract: Deep neural networks perform well in object recognition, but do they perceive objects like humans? This study investigates the Gestalt principle of closure in convolutional neural networks. We propose a protocol to identify closure and conduct experiments using simple visual stimuli with progressively removed edge sections. We evaluate well-known networks on their ability to classify incomplete polygons. Our findings reveal a performance degradation as the edge removal percentage increases, indicating that current models heavily rely on complete edge information for accurate classification. The data used in our study is available on Github.
Authors: Zerui Wang, Yan Liu
Abstract: Transformer-based models have achieved state-of-the-art performance in various computer vision tasks, including image and video analysis. However, Transformer's complex architecture and black-box nature pose challenges for explainability, a crucial aspect for real-world applications and scientific inquiry. Current Explainable AI (XAI) methods can only provide one-dimensional feature importance, either spatial or temporal explanation, with significant computational complexity. This paper introduces STAA (Spatio-Temporal Attention Attribution), an XAI method for interpreting video Transformer models. Differ from traditional methods that separately apply image XAI techniques for spatial features or segment contribution analysis for temporal aspects, STAA offers both spatial and temporal information simultaneously from attention values in Transformers. The study utilizes the Kinetics-400 dataset, a benchmark collection of 400 human action classes used for action recognition research. We introduce metrics to quantify explanations. We also apply optimization to enhance STAA's raw output. By implementing dynamic thresholding and attention focusing mechanisms, we improve the signal-to-noise ratio in our explanations, resulting in more precise visualizations and better evaluation results. In terms of computational overhead, our method requires less than 3\% of the computational resources of traditional XAI methods, making it suitable for real-time video XAI analysis applications. STAA contributes to the growing field of XAI by offering a method for researchers and practitioners to analyze Transformer models.
Authors: Jincen Jiang, Qianyu Zhou, Yuhang Li, Xinkui Zhao, Meili Wang, Lizhuang Ma, Jian Chang, Jian Jun Zhang, Xuequan Lu
Abstract: In this paper, we present PCoTTA, an innovative, pioneering framework for Continual Test-Time Adaptation (CoTTA) in multi-task point cloud understanding, enhancing the model's transferability towards the continually changing target domain. We introduce a multi-task setting for PCoTTA, which is practical and realistic, handling multiple tasks within one unified model during the continual adaptation. Our PCoTTA involves three key components: automatic prototype mixture (APM), Gaussian Splatted feature shifting (GSFS), and contrastive prototype repulsion (CPR). Firstly, APM is designed to automatically mix the source prototypes with the learnable prototypes with a similarity balancing factor, avoiding catastrophic forgetting. Then, GSFS dynamically shifts the testing sample toward the source domain, mitigating error accumulation in an online manner. In addition, CPR is proposed to pull the nearest learnable prototype close to the testing feature and push it away from other prototypes, making each prototype distinguishable during the adaptation. Experimental comparisons lead to a new benchmark, demonstrating PCoTTA's superiority in boosting the model's transferability towards the continually changing target domain.
Authors: Zhen Yao, Mooi Choo Chuah
Abstract: Recent video semantic segmentation (VSS) methods have demonstrated promising results in well-lit environments. However, their performance significantly drops in low-light scenarios due to limited visibility and reduced contextual details. In addition, unfavorable low-light conditions make it harder to incorporate temporal consistency across video frames and thus, lead to video flickering effects. Compared with conventional cameras, event cameras can capture motion dynamics, filter out temporal-redundant information, and are robust to lighting conditions. To this end, we propose EVSNet, a lightweight framework that leverages event modality to guide the learning of a unified illumination-invariant representation. Specifically, we leverage a Motion Extraction Module to extract short-term and long-term temporal motions from event modality and a Motion Fusion Module to integrate image features and motion features adaptively. Furthermore, we use a Temporal Decoder to exploit video contexts and generate segmentation predictions. Such designs in EVSNet result in a lightweight architecture while achieving SOTA performance. Experimental results on 3 large-scale datasets demonstrate our proposed EVSNet outperforms SOTA methods with up to 11x higher parameter efficiency.
Authors: Hah Min Lew, Sahng-Min Yoo, Hyunwoo Kang, Gyeong-Moon Park
Abstract: We introduce an industrial Head Blending pipeline for the task of seamlessly integrating an actor's head onto a target body in digital content creation. The key challenge stems from discrepancies in head shape and hair structure, which lead to unnatural boundaries and blending artifacts. Existing methods treat foreground and background as a single task, resulting in suboptimal blending quality. To address this problem, we propose CHANGER, a novel pipeline that decouples background integration from foreground blending. By utilizing chroma keying for artifact-free background generation and introducing Head shape and long Hair augmentation ($H^2$ augmentation) to simulate a wide range of head shapes and hair styles, CHANGER improves generalization on innumerable various real-world cases. Furthermore, our Foreground Predictive Attention Transformer (FPAT) module enhances foreground blending by predicting and focusing on key head and body regions. Quantitative and qualitative evaluations on benchmark datasets demonstrate that our CHANGER outperforms state-of-the-art methods, delivering high-fidelity, industrial-grade results.
Authors: Srikumar Sastry, Subash Khanal, Aayush Dhakal, Adeel Ahmad, Nathan Jacobs
Abstract: We present TaxaBind, a unified embedding space for characterizing any species of interest. TaxaBind is a multimodal embedding space across six modalities: ground-level images of species, geographic location, satellite image, text, audio, and environmental features, useful for solving ecological problems. To learn this joint embedding space, we leverage ground-level images of species as a binding modality. We propose multimodal patching, a technique for effectively distilling the knowledge from various modalities into the binding modality. We construct two large datasets for pretraining: iSatNat with species images and satellite images, and iSoundNat with species images and audio. Additionally, we introduce TaxaBench-8k, a diverse multimodal dataset with six paired modalities for evaluating deep learning models on ecological tasks. Experiments with TaxaBind demonstrate its strong zero-shot and emergent capabilities on a range of tasks including species classification, cross-model retrieval, and audio classification. The datasets and models are made available at https://github.com/mvrl/TaxaBind.
Authors: Sara Oblak, Despoina Paschalidou, Sanja Fidler, Matan Atzmon
Abstract: Reconstructing dynamic scenes from image inputs is a fundamental computer vision task with many downstream applications. Despite recent advancements, existing approaches still struggle to achieve high-quality reconstructions from unseen viewpoints and timestamps. This work introduces the ReMatching framework, designed to improve generalization quality by incorporating deformation priors into dynamic reconstruction models. Our approach advocates for velocity-field-based priors, for which we suggest a matching procedure that can seamlessly supplement existing dynamic reconstruction pipelines. The framework is highly adaptable and can be applied to various dynamic representations. Moreover, it supports integrating multiple types of model priors and enables combining simpler ones to create more complex classes. Our evaluations on popular benchmarks involving both synthetic and real-world dynamic scenes demonstrate a clear improvement in reconstruction accuracy of current state-of-the-art models.
Authors: Nourhan Bayasi, Jamil Fayyad, Ghassan Hamarneh, Rafeef Garbi, Homayoun Najjaran
Abstract: Simplicity bias poses a significant challenge in neural networks, often leading models to favor simpler solutions and inadvertently learn decision rules influenced by spurious correlations. This results in biased models with diminished generalizability. While many current approaches depend on human supervision, obtaining annotations for various bias attributes is often impractical. To address this, we introduce Debiasify, a novel self-distillation approach that requires no prior knowledge about the nature of biases. Our method leverages a new distillation loss to transfer knowledge within the network, from deeper layers containing complex, highly-predictive features to shallower layers with simpler, attribute-conditioned features in an unsupervised manner. This enables Debiasify to learn robust, debiased representations that generalize effectively across diverse biases and datasets, improving both worst-group performance and overall accuracy. Extensive experiments on computer vision and medical imaging benchmarks demonstrate the effectiveness of our approach, significantly outperforming previous unsupervised debiasing methods (e.g., a 10.13% improvement in worst-group accuracy for Wavy Hair classification in CelebA) and achieving comparable or superior performance to supervised approaches. Our code is publicly available at the following link: Debiasify.
Authors: Shreyash Arya, Sukrut Rao, Moritz B\"ohle, Bernt Schiele
Abstract: B-cos Networks have been shown to be effective for obtaining highly human interpretable explanations of model decisions by architecturally enforcing stronger alignment between inputs and weight. B-cos variants of convolutional networks (CNNs) and vision transformers (ViTs), which primarily replace linear layers with B-cos transformations, perform competitively to their respective standard variants while also yielding explanations that are faithful by design. However, it has so far been necessary to train these models from scratch, which is increasingly infeasible in the era of large, pre-trained foundation models. In this work, inspired by the architectural similarities in standard DNNs and B-cos networks, we propose 'B-cosification', a novel approach to transform existing pre-trained models to become inherently interpretable. We perform a thorough study of design choices to perform this conversion, both for convolutional neural networks and vision transformers. We find that B-cosification can yield models that are on par with B-cos models trained from scratch in terms of interpretability, while often outperforming them in terms of classification performance at a fraction of the training cost. Subsequently, we apply B-cosification to a pretrained CLIP model, and show that, even with limited data and compute cost, we obtain a B-cosified version that is highly interpretable and competitive on zero shot performance across a variety of datasets. We release our code and pre-trained model weights at https://github.com/shrebox/B-cosification.
Authors: Mehdi Sefidgar Dilmaghani, Waseem Shariff, Cian Ryan, Joseph Lemley, Peter Corcoran
Abstract: This paper presents an autonomous method to address challenges arising from severe lighting conditions in machine vision applications that use event cameras. To manage these conditions, the research explores the built in potential of these cameras to adjust pixel functionality, named bias settings. As cars are driven at various times and locations, shifts in lighting conditions are unavoidable. Consequently, this paper utilizes the neuromorphic YOLO-based face tracking module of a driver monitoring system as the event-based application to study. The proposed method uses numerical metrics to continuously monitor the performance of the event-based application in real-time. When the application malfunctions, the system detects this through a drop in the metrics and automatically adjusts the event cameras bias values. The Nelder-Mead simplex algorithm is employed to optimize this adjustment, with finetuning continuing until performance returns to a satisfactory level. The advantage of bias optimization lies in its ability to handle conditions such as flickering or darkness without requiring additional hardware or software. To demonstrate the capabilities of the proposed system, it was tested under conditions where detecting human faces with default bias values was impossible. These severe conditions were simulated using dim ambient light and various flickering frequencies. Following the automatic and dynamic process of bias modification, the metrics for face detection significantly improved under all conditions. Autobiasing resulted in an increase in the YOLO confidence indicators by more than 33 percent for object detection and 37 percent for face detection highlighting the effectiveness of the proposed method.
Authors: Han-Wei Kung, Tuomas Varanka, Sanjay Saha, Terence Sim, Nicu Sebe
Abstract: Current face anonymization techniques often depend on identity loss calculated by face recognition models, which can be inaccurate and unreliable. Additionally, many methods require supplementary data such as facial landmarks and masks to guide the synthesis process. In contrast, our approach uses diffusion models with only a reconstruction loss, eliminating the need for facial landmarks or masks while still producing images with intricate, fine-grained details. We validated our results on two public benchmarks through both quantitative and qualitative evaluations. Our model achieves state-of-the-art performance in three key areas: identity anonymization, facial attribute preservation, and image quality. Beyond its primary function of anonymization, our model can also perform face swapping tasks by incorporating an additional facial image as input, demonstrating its versatility and potential for diverse applications. Our code and models are available at https://github.com/hanweikung/face_anon_simple .
Authors: Haoxuan Che, Xuanhua He, Quande Liu, Cheng Jin, Hao Chen
Abstract: We introduce GameGen-X, the first diffusion transformer model specifically designed for both generating and interactively controlling open-world game videos. This model facilitates high-quality, open-domain generation by simulating an extensive array of game engine features, such as innovative characters, dynamic environments, complex actions, and diverse events. Additionally, it provides interactive controllability, predicting and altering future content based on the current clip, thus allowing for gameplay simulation. To realize this vision, we first collected and built an Open-World Video Game Dataset from scratch. It is the first and largest dataset for open-world game video generation and control, which comprises over a million diverse gameplay video clips sampling from over 150 games with informative captions from GPT-4o. GameGen-X undergoes a two-stage training process, consisting of foundation model pre-training and instruction tuning. Firstly, the model was pre-trained via text-to-video generation and video continuation, endowing it with the capability for long-sequence, high-quality open-domain game video generation. Further, to achieve interactive controllability, we designed InstructNet to incorporate game-related multi-modal control signal experts. This allows the model to adjust latent representations based on user inputs, unifying character interaction and scene content control for the first time in video generation. During instruction tuning, only the InstructNet is updated while the pre-trained foundation model is frozen, enabling the integration of interactive controllability without loss of diversity and quality of generated video content.
Authors: Yang Liu, Chuanchen Luo, Zhongkai Mao, Junran Peng, Zhaoxiang Zhang
Abstract: Recently, 3D Gaussian Splatting (3DGS) has revolutionized radiance field reconstruction, manifesting efficient and high-fidelity novel view synthesis. However, accurately representing surfaces, especially in large and complex scenarios, remains a significant challenge due to the unstructured nature of 3DGS. In this paper, we present CityGaussianV2, a novel approach for large-scale scene reconstruction that addresses critical challenges related to geometric accuracy and efficiency. Building on the favorable generalization capabilities of 2D Gaussian Splatting (2DGS), we address its convergence and scalability issues. Specifically, we implement a decomposed-gradient-based densification and depth regression technique to eliminate blurry artifacts and accelerate convergence. To scale up, we introduce an elongation filter that mitigates Gaussian count explosion caused by 2DGS degeneration. Furthermore, we optimize the CityGaussian pipeline for parallel training, achieving up to 10$\times$ compression, at least 25% savings in training time, and a 50% decrease in memory usage. We also established standard geometry benchmarks under large-scale scenes. Experimental results demonstrate that our method strikes a promising balance between visual quality, geometric accuracy, as well as storage and training costs. The project page is available at https://dekuliutesla.github.io/CityGaussianV2/.
Authors: Qihang Yu, Ju He, Xueqing Deng, Xiaohui Shen, Liang-Chieh Chen
Abstract: This paper presents Randomized AutoRegressive modeling (RAR) for visual generation, which sets a new state-of-the-art performance on the image generation task while maintaining full compatibility with language modeling frameworks. The proposed RAR is simple: during a standard autoregressive training process with a next-token prediction objective, the input sequence-typically ordered in raster form-is randomly permuted into different factorization orders with a probability r, where r starts at 1 and linearly decays to 0 over the course of training. This annealing training strategy enables the model to learn to maximize the expected likelihood over all factorization orders and thus effectively improve the model's capability of modeling bidirectional contexts. Importantly, RAR preserves the integrity of the autoregressive modeling framework, ensuring full compatibility with language modeling while significantly improving performance in image generation. On the ImageNet-256 benchmark, RAR achieves an FID score of 1.48, not only surpassing prior state-of-the-art autoregressive image generators but also outperforming leading diffusion-based and masked transformer-based methods. Code and models will be made available at https://github.com/bytedance/1d-tokenizer
Authors: Donghoon Kim, Gusang Lee, Kyuhong Shim, Byonghyo Shim
Abstract: Recently, we have observed that Large Multi-modal Models (LMMs) are revolutionizing the way machines interact with the world, unlocking new possibilities across various multi-modal applications. To adapt LMMs for downstream tasks, parameter-efficient fine-tuning (PEFT) which only trains additional prefix tokens or modules, has gained popularity. Nevertheless, there has been little analysis of how PEFT works in LMMs. In this paper, we delve into the strengths and weaknesses of each tuning strategy, shifting the focus from the efficiency typically associated with these approaches. We first discover that model parameter tuning methods such as LoRA and Adapters distort the feature representation space learned during pre-training and limit the full utilization of pre-trained knowledge. We also demonstrate that prefix-tuning excels at preserving the representation space, despite its lower performance on downstream tasks. These findings suggest a simple two-step PEFT strategy called Prefix-Tuned PEFT (PT-PEFT), which successively performs prefix-tuning and then PEFT (i.e., Adapter, LoRA), combines the benefits of both. Experimental results show that PT-PEFT not only improves performance in image captioning and visual question answering compared to vanilla PEFT methods but also helps preserve the representation space of the four pre-trained models.
Authors: Tang Li, Mengmeng Ma, Xi Peng
Abstract: Large pretrained foundation models demonstrate exceptional performance and, in some high-stakes applications, even surpass human experts. However, most of these models are currently evaluated primarily on prediction accuracy, overlooking the validity of the rationales behind their accurate predictions. For the safe deployment of foundation models, there is a pressing need to ensure double-correct predictions, i.e., correct prediction backed by correct rationales. To achieve this, we propose a two-phase scheme: First, we curate a new dataset that offers structured rationales for visual recognition tasks. Second, we propose a rationale-informed optimization method to guide the model in disentangling and localizing visual evidence for each rationale, without requiring manual annotations. Extensive experiments and ablation studies demonstrate that our model outperforms state-of-the-art models by up to 10.1% in prediction accuracy across a wide range of tasks. Furthermore, our method significantly improves the model's rationale correctness, improving localization by 7.5% and disentanglement by 36.5%. Our dataset, source code, and pretrained weights: https://github.com/deep-real/DCP
Authors: Ehsan Ganjidoost, Jeff Orchard
Abstract: An adversarial example is a modified input image designed to cause a Machine Learning (ML) model to make a mistake; these perturbations are often invisible or subtle to human observers and highlight vulnerabilities in a model's ability to generalize from its training data. Several adversarial attacks can create such examples, each with a different perspective, effectiveness, and perceptibility of changes. Conversely, defending against such adversarial attacks improves the robustness of ML models in image processing and other domains of deep learning. Most defence mechanisms require either a level of model awareness, changes to the model, or access to a comprehensive set of adversarial examples during training, which is impractical. Another option is to use an auxiliary model in a preprocessing manner without changing the primary model. This study presents a practical and effective solution -- using predictive coding networks (PCnets) as an auxiliary step for adversarial defence. By seamlessly integrating PCnets into feed-forward networks as a preprocessing step, we substantially bolster resilience to adversarial perturbations. Our experiments on MNIST and CIFAR10 demonstrate the remarkable effectiveness of PCnets in mitigating adversarial examples with about 82% and 65% improvements in robustness, respectively. The PCnet, trained on a small subset of the dataset, leverages its generative nature to effectively counter adversarial efforts, reverting perturbed images closer to their original forms. This innovative approach holds promise for enhancing the security and reliability of neural network classifiers in the face of the escalating threat of adversarial attacks.
Authors: Declan Campbell, Sunayana Rane, Tyler Giallanza, Nicol\`o De Sabbata, Kia Ghods, Amogh Joshi, Alexander Ku, Steven M. Frankland, Thomas L. Griffiths, Jonathan D. Cohen, Taylor W. Webb
Abstract: Recent work has documented striking heterogeneity in the performance of state-of-the-art vision language models (VLMs), including both multimodal language models and text-to-image models. These models are able to describe and generate a diverse array of complex, naturalistic images, yet they exhibit surprising failures on basic multi-object reasoning tasks -- such as counting, localization, and simple forms of visual analogy -- that humans perform with near perfect accuracy. To better understand this puzzling pattern of successes and failures, we turn to theoretical accounts of the binding problem in cognitive science and neuroscience, a fundamental problem that arises when a shared set of representational resources must be used to represent distinct entities (e.g., to represent multiple objects in an image), necessitating the use of serial processing to avoid interference. We find that many of the puzzling failures of state-of-the-art VLMs can be explained as arising due to the binding problem, and that these failure modes are strikingly similar to the limitations exhibited by rapid, feedforward processing in the human brain.
Authors: Lipismita Panigrahi, Prianka Rani Saha, Jurdana Masuma Iqrah, Sushil Prasad
Abstract: Clinical diagnosis of breast malignancy (BM) is a challenging problem in the recent era. In particular, Deep learning (DL) models have continued to offer important solutions for early BM diagnosis but their performance experiences overfitting due to the limited volume of breast ultrasound (BUS) image data. Further, large BUS datasets are difficult to manage due to privacy and legal concerns. Hence, image augmentation is a necessary and challenging step to improve the performance of the DL models. However, the current DL-based augmentation models are inadequate and operate as a black box resulting lack of information and justifications about their suitability and efficacy. Additionally, pre and post-augmentation need high-performance computational resources and time to produce the augmented image and evaluate the model performance. Thus, this study aims to develop a novel efficient augmentation approach for BUS images with advanced neural style transfer (NST) and Explainable AI (XAI) harnessing GPU-based parallel infrastructure. We scale and distribute the training of the augmentation model across 8 GPUs using the Horovod framework on a DGX cluster, achieving a 5.09 speedup while maintaining the model's accuracy. The proposed model is evaluated on 800 (348 benign and 452 malignant) BUS images and its performance is analyzed with other progressive techniques, using different quantitative analyses. The result indicates that the proposed approach can successfully augment the BUS images with 92.47% accuracy.
Authors: Grace Guo, Jenna Jiayi Kang, Raj Sanjay Shah, Hanspeter Pfister, Sashank Varma
Abstract: Vision Language Models (VLMs) have been successful at many chart comprehension tasks that require attending to both the images of charts and their accompanying textual descriptions. However, it is not well established how VLM performance profiles map to human-like behaviors. If VLMs can be shown to have human-like chart comprehension abilities, they can then be applied to a broader range of tasks, such as designing and evaluating visualizations for human readers. This paper lays the foundations for such applications by evaluating the accuracy of zero-shot prompting of VLMs on graphical perception tasks with established human performance profiles. Our findings reveal that VLMs perform similarly to humans under specific task and style combinations, suggesting that they have the potential to be used for modeling human performance. Additionally, variations to the input stimuli show that VLM accuracy is sensitive to stylistic changes such as fill color and chart contiguity, even when the underlying data and data mappings are the same.
Authors: Sina Rismanchian, Yasaman Razeghi, Sameer Singh, Shayan Doroudi
Abstract: Humans have the ability to reason about geometric patterns in images and scenes from a young age. However, developing large multimodal models (LMMs) capable of similar reasoning remains a challenge, highlighting the need for robust evaluation methods to assess these capabilities. We introduce TurtleBench, a benchmark designed to evaluate LMMs' capacity to interpret geometric patterns -- given visual examples, textual instructions, or both -- and generate precise code outputs. Inspired by turtle geometry, a notion used to teach children foundational coding and geometric concepts, TurtleBench features tasks with patterned shapes that have underlying algorithmic logic. Our evaluation reveals that leading LMMs struggle significantly with these tasks, with GPT-4o achieving only 19\% accuracy on the simplest tasks and few-shot prompting only marginally improves their performance ($<2\%$). TurtleBench highlights the gap between human and AI performance in intuitive and visual geometrical understanding, setting the stage for future research in this area. TurtleBench stands as one of the few benchmarks to evaluate the integration of visual understanding and code generation capabilities in LMMs, setting the stage for future research. Code and Dataset for this paper is provided here: https://github.com/sinaris76/TurtleBench
Authors: David A. Danhofer
Abstract: The crucial role of convolutional models, both as standalone vision models and backbones in foundation models, necessitates effective acceleration techniques. This paper proposes a novel method to learn semi-structured sparsity patterns for convolution kernels in the form of maskings enabling the utilization of readily available hardware accelerations. The approach accelerates convolutional models more than two-fold during inference without decreasing model performance. At the same time, the original model weights and structure remain unchanged keeping the model thus easily updatable. Beyond the immediate practical use, the effect of maskings on prediction is easily quantifiable. Therefore, guarantees on model predictions under maskings are derived showing stability bounds for learned maskings even after updating the original underlying model.
Authors: Dogyun Park, Sojin Lee, Sihyeon Kim, Taehoon Lee, Youngjoon Hong, Hyunwoo J. Kim
Abstract: Rectified flow and reflow procedures have significantly advanced fast generation by progressively straightening ordinary differential equation (ODE) flows. They operate under the assumption that image and noise pairs, known as couplings, can be approximated by straight trajectories with constant velocity. However, we observe that modeling with constant velocity and using reflow procedures have limitations in accurately learning straight trajectories between pairs, resulting in suboptimal performance in few-step generation. To address these limitations, we introduce Constant Acceleration Flow (CAF), a novel framework based on a simple constant acceleration equation. CAF introduces acceleration as an additional learnable variable, allowing for more expressive and accurate estimation of the ODE flow. Moreover, we propose two techniques to further improve estimation accuracy: initial velocity conditioning for the acceleration model and a reflow process for the initial velocity. Our comprehensive studies on toy datasets, CIFAR-10, and ImageNet 64x64 demonstrate that CAF outperforms state-of-the-art baselines for one-step generation. We also show that CAF dramatically improves few-step coupling preservation and inversion over Rectified flow. Code is available at \href{https://github.com/mlvlab/CAF}{https://github.com/mlvlab/CAF}.
URLs: https://github.com/mlvlab/CAF, https://github.com/mlvlab/CAF
Authors: Samuel J. Simons, Bart{\l}omiej W. Papie\.z
Abstract: This paper introduces SpineFM, a novel pipeline that achieves state-of-the-art performance in the automatic segmentation and identification of vertebral bodies in cervical and lumbar spine radiographs. SpineFM leverages the regular geometry of the spine, employing a novel inductive process to sequentially infer the location of each vertebra along the spinal column. Vertebrae are segmented using Medical-SAM-Adaptor, a robust foundation model that diverges from commonly used CNN-based models. We achieved outstanding results on two publicly available spine X-Ray datasets, with successful identification of 97.8\% and 99.6\% of annotated vertebrae, respectively. Of which, our segmentation reached an average Dice of 0.942 and 0.921, surpassing previous state-of-the-art methods.
Authors: Daichi Tajima, Yoshihiro Kanamori, Yuki Endo
Abstract: Relighting of human images enables post-photography editing of lighting effects in portraits. The current mainstream approach uses neural networks to approximate lighting effects without explicitly accounting for the principle of physical shading. As a result, it often has difficulty representing high-frequency shadows and shading. In this paper, we propose a two-stage relighting method that can reproduce physically-based shadows and shading from low to high frequencies. The key idea is to approximate an environment light source with a set of a fixed number of area light sources. The first stage employs supervised inverse rendering from a single image using neural networks and calculates physically-based shading. The second stage then calculates shadow for each area light and sums up to render the final image. We propose to make soft shadow mapping differentiable for the area-light approximation of environment lighting. We demonstrate that our method can plausibly reproduce all-frequency shadows and shading caused by environment illumination, which have been difficult to reproduce using existing methods.
Authors: Yeonsung Jung, Jaeyun Song, June Yong Yang, Jin-Hwa Kim, Sung-Yub Kim, Eunho Yang
Abstract: Learning generalized models from biased data is an important undertaking toward fairness in deep learning. To address this issue, recent studies attempt to identify and leverage bias-conflicting samples free from spurious correlations without prior knowledge of bias or an unbiased set. However, spurious correlation remains an ongoing challenge, primarily due to the difficulty in precisely detecting these samples. In this paper, inspired by the similarities between mislabeled samples and bias-conflicting samples, we approach this challenge from a novel perspective of mislabeled sample detection. Specifically, we delve into Influence Function, one of the standard methods for mislabeled sample detection, for identifying bias-conflicting samples and propose a simple yet effective remedy for biased models by leveraging them. Through comprehensive analysis and experiments on diverse datasets, we demonstrate that our new perspective can boost the precision of detection and rectify biased models effectively. Furthermore, our approach is complementary to existing methods, showing performance improvement even when applied to models that have already undergone recent debiasing techniques.
Authors: Heiko Hoffmann
Abstract: Scalar variables, e.g., the orientation of a shape in an image, are commonly predicted using a single output neuron in a neural network. In contrast, the mammalian cortex represents variables with a population of neurons. In this population code, each neuron is most active at its preferred value and shows partial activity for other values. Here, we investigate the benefit of using a population code for the output layer of a neural network. We compare population codes against single-neuron outputs and one-hot vectors. First, we show theoretically and in experiments with synthetic data that population codes improve robustness to input noise in networks of stacked linear layers. Second, we demonstrate the benefit of population codes to encode ambiguous outputs, as found for symmetric objects. Using the T-LESS dataset of feature-less real-world objects, we show that population codes improve the accuracy of predicting object orientation from RGB-image input.
Authors: Xuchen Xie, Yiqiao Qiu, Run Lin, Weishi Zheng, Ruixuan Wang
Abstract: This study focuses on incremental learning for image classification, exploring how to reduce catastrophic forgetting of all learned knowledge when access to old data is restricted due to memory or privacy constraints. The challenge of incremental learning lies in achieving an optimal balance between plasticity, the ability to learn new knowledge, and stability, the ability to retain old knowledge. Based on whether the task identifier (task-ID) of an image can be obtained during the test stage, incremental learning for image classifcation is divided into two main paradigms, which are task incremental learning (TIL) and class incremental learning (CIL). The TIL paradigm has access to the task-ID, allowing it to use multiple task-specific classification heads selected based on the task-ID. Consequently, in CIL, where the task-ID is unavailable, TIL methods must predict the task-ID to extend their application to the CIL paradigm. Our previous method for TIL adds task-specific batch normalization and classification heads incrementally. This work extends the method by predicting task-ID through an "unknown" class added to each classification head. The head with the lowest "unknown" probability is selected, enabling task-ID prediction and making the method applicable to CIL. The task-specific batch normalization (BN) modules effectively adjust the distribution of output feature maps across different tasks, enhancing the model's plasticity.Moreover, since BN has much fewer parameters compared to convolutional kernels, by only modifying the BN layers as new tasks arrive, the model can effectively manage parameter growth while ensuring stability across tasks. The innovation of this study lies in the first-time introduction of task-specific BN into CIL and verifying the feasibility of extending TIL methods to CIL through task-ID prediction with state-of-the-art performance on multiple datasets.
Authors: Vanessa Wirth, Johanna Br\"aunig, Martin Vossiek, Tim Weyrich, Marc Stamminger
Abstract: Utilizing the complementary strengths of wavelength-specific range or depth sensors is crucial for robust computer-assisted tasks such as autonomous driving. Despite this, there is still little research done at the intersection of optical depth sensors and radars operating close range, where the target is decimeters away from the sensors. Together with a growing interest in high-resolution imaging radars operating in the near field, the question arises how these sensors behave in comparison to their traditional optical counterparts. In this work, we take on the unique challenge of jointly characterizing depth imagers from both, the optical and radio-frequency domain using a multimodal spatial calibration. We collect data from four depth imagers, with three optical sensors of varying operation principle and an imaging radar. We provide a comprehensive evaluation of their depth measurements with respect to distinct object materials, geometries, and object-to-sensor distances. Specifically, we reveal scattering effects of partially transmissive materials and investigate the response of radio-frequency signals. All object measurements will be made public in form of a multimodal dataset, called MAROON.
Authors: Mianyong Ding, Matteo Maspero, Annemieke S Littooij, Martine van Grotel, Raquel Davila Fajardo, Max M van Noesel, Marry M van den Heuvel-Eibrink, Geert O Janssens
Abstract: Purposes: This study aimed to develop a computed tomography (CT)-based multi-organ segmentation model for delineating organs-at-risk (OARs) in pediatric upper abdominal tumors and evaluate its robustness across multiple datasets. Materials and methods: In-house postoperative CTs from pediatric patients with renal tumors and neuroblastoma (n=189) and a public dataset (n=189) with CTs covering thoracoabdominal regions were used. Seventeen OARs were delineated: nine by clinicians (Type 1) and eight using TotalSegmentator (Type 2). Auto-segmentation models were trained using in-house (ModelPMC-UMCU) and a combined dataset of public data (Model-Combined). Performance was assessed with Dice Similarity Coefficient (DSC), 95% Hausdorff Distance (HD95), and mean surface distance (MSD). Two clinicians rated clinical acceptability on a 5-point Likert scale across 15 patient contours. Model robustness was evaluated against sex, age, intravenous contrast, and tumor type. Results: Model-PMC-UMCU achieved mean DSC values above 0.95 for five of nine OARs, while spleen and heart ranged between 0.90 and 0.95. The stomach-bowel and pancreas exhibited DSC values below 0.90. Model-Combined demonstrated improved robustness across both datasets. Clinical evaluation revealed good usability, with both clinicians rating six of nine Type 1 OARs above four and six of eight Type 2 OARs above three. Significant performance 2 differences were only found across age groups in both datasets, specifically in the left lung and pancreas. The 0-2 age group showed the lowest performance. Conclusion: A multi-organ segmentation model was developed, showcasing enhanced robustness when trained on combined datasets. This model is suitable for various OARs and can be applied to multiple datasets in clinical settings.
Authors: Matthew C. Bendel, Rizwan Ahmad, Philip Schniter
Abstract: In ill-posed imaging inverse problems, there can exist many hypotheses that fit both the observed measurements and prior knowledge of the true image. Rather than returning just one hypothesis of that image, posterior samplers aim to explore the full solution space by generating many probable hypotheses, which can later be used to quantify uncertainty or construct recoveries that appropriately navigate the perception/distortion trade-off. In this work, we propose a fast and accurate posterior-sampling conditional generative adversarial network (cGAN) that, through a novel form of regularization, aims for correctness in the posterior mean as well as the trace and K principal components of the posterior covariance matrix. Numerical experiments demonstrate that our method outperforms contemporary cGANs and diffusion models in imaging inverse problems like denoising, large-scale inpainting, and accelerated MRI recovery. The code for our model can be found here: https://github.com/matt-bendel/pcaGAN.
Authors: Sara Ketabi, Matthias W. Wagner, Cynthia Hawkins, Uri Tabori, Birgit Betina Ertl-Wagner, Farzad Khalvati
Abstract: Despite the promising performance of convolutional neural networks (CNNs) in brain tumor diagnosis from magnetic resonance imaging (MRI), their integration into the clinical workflow has been limited. That is mainly due to the fact that the features contributing to a model's prediction are unclear to radiologists and hence, clinically irrelevant, i.e., lack of explainability. As the invaluable sources of radiologists' knowledge and expertise, radiology reports can be integrated with MRI in a contrastive learning (CL) framework, enabling learning from image-report associations, to improve CNN explainability. In this work, we train a multimodal CL architecture on 3D brain MRI scans and radiology reports to learn informative MRI representations. Furthermore, we integrate tumor location, salient to several brain tumor analysis tasks, into this framework to improve its generalizability. We then apply the learnt image representations to improve explainability and performance of genetic marker classification of pediatric Low-grade Glioma, the most prevalent brain tumor in children, as a downstream task. Our results indicate a Dice score of 31.1% between the model's attention maps and manual tumor segmentation (as an explainability measure) with test classification performance of 87.7%, significantly outperforming the baselines. These enhancements can build trust in our model among radiologists, facilitating its integration into clinical practices for more efficient tumor diagnosis.
Authors: Xiaotong Zhang, Alexander Broersen, Gonnie CM van Erp, Silvia L. Pintea, Jouke Dijkstra
Abstract: Improving connectivity and completeness are the most challenging aspects of small liver vessel segmentation. It is difficult for existing methods to obtain segmented liver vessel trees simultaneously with continuous geometry and detail in small vessels. We proposed a diffusion model-based method with a multi-scale graph attention guidance to break through the bottleneck to segment the liver vessels. Experiments show that the proposed method outperforms the other state-of-the-art methods used in this study on two public datasets of 3D-ircadb-01 and LiVS. Dice coefficient and Sensitivity are improved by at least 11.67% and 24.21% on 3D-ircadb-01 dataset, and are improved by at least 3.21% and 9.11% on LiVS dataset. Connectivity is also quantitatively evaluated in this study and our method performs best. The proposed method is reliable for small liver vessel segmentation.
Authors: Evangelos Papoutsellis, Zeljko Kereta, Kostas Papafitsoros
Abstract: Regularisation is commonly used in iterative methods for solving imaging inverse problems. Many algorithms involve the evaluation of the proximal operator of the regularisation term in every iteration, leading to a significant computational overhead since such evaluation can be costly. In this context, the ProxSkip algorithm, recently proposed for federated learning purposes, emerges as an solution. It randomly skips regularisation steps, reducing the computational time of an iterative algorithm without affecting its convergence. Here we explore for the first time the efficacy of ProxSkip to a variety of imaging inverse problems and we also propose a novel PDHGSkip version. Extensive numerical results highlight the potential of these methods to accelerate computations while maintaining high-quality reconstructions.
Authors: Fan Xiao, Junlin Hou, Ruiwei Zhao, Rui Feng, Haidong Zou, Lina Lu, Yi Xu, Juzhao Zhang
Abstract: Diabetic retinopathy (DR) is a leading cause of blindness worldwide and a common complication of diabetes. As two different imaging tools for DR grading, color fundus photography (CFP) and infrared fundus photography (IFP) are highly-correlated and complementary in clinical applications. To the best of our knowledge, this is the first study that explores a novel multi-modal deep learning framework to fuse the information from CFP and IFP towards more accurate DR grading. Specifically, we construct a dual-stream architecture Cross-Fundus Transformer (CFT) to fuse the ViT-based features of two fundus image modalities. In particular, a meticulously engineered Cross-Fundus Attention (CFA) module is introduced to capture the correspondence between CFP and IFP images. Moreover, we adopt both the single-modality and multi-modality supervisions to maximize the overall performance for DR grading. Extensive experiments on a clinical dataset consisting of 1,713 pairs of multi-modal fundus images demonstrate the superiority of our proposed method. Our code will be released for public access.
Authors: Akhila Krishna, Nikhil Cherian Kurian, Abhijeet Patil, Amruta Parulekar, Amit Sethi
Abstract: Accurate survival prediction is essential for personalized cancer treatment. However, genomic data - often a more powerful predictor than pathology data - is costly and inaccessible. We present the cross-modal genomic feature translation and alignment network for enhanced survival prediction from histopathology images (PathoGen-X). It is a deep learning framework that leverages both genomic and imaging data during training, relying solely on imaging data at testing. PathoGen-X employs transformer-based networks to align and translate image features into the genomic feature space, enhancing weaker imaging signals with stronger genomic signals. Unlike other methods, PathoGen-X translates and aligns features without projecting them to a shared latent space and requires fewer paired samples. Evaluated on TCGA-BRCA, TCGA-LUAD, and TCGA-GBM datasets, PathoGen-X demonstrates strong survival prediction performance, emphasizing the potential of enriched imaging models for accessible cancer prognosis.
Authors: Feng Lu, Ziqiang Chen
Abstract: We present algebraic projective geometry definitions of 3D rotations so as to bridge a small gap between the applications and the definitions of 3D rotations in homogeneous matrix form. A general homogeneous matrix formulation to 3D rotation geometric transformations is proposed which suits for the cases when the rotation axis is unnecessarily through the coordinate system origin given their rotation axes and rotation angles. General three-dimensional rotation formula~\eqref{eqn:3D homogeneous roation} and~\eqref{eqn:3D rotation matrix vector Euclidean} similar to the Euler-Rodrigues formula were presented. The matrix-vector form of 3D rotation in Euclidean space is especially suited for numerical applications where gimbal lock is a concern.}
Authors: Naitik Khandelwal, Xiao Liu, Mengmi Zhang
Abstract: Scene graph generation (SGG) analyzes images to extract meaningful information about objects and their relationships. In the dynamic visual world, it is crucial for AI systems to continuously detect new objects and establish their relationships with existing ones. Recently, numerous studies have focused on continual learning within the domains of object detection and image recognition. However, a limited amount of research focuses on a more challenging continual learning problem in SGG. This increased difficulty arises from the intricate interactions and dynamic relationships among objects, and their associated contexts. Thus, in continual learning, SGG models are often required to expand, modify, retain, and reason scene graphs within the process of adaptive visual scene understanding. To systematically explore Continual Scene Graph Generation (CSEGG), we present a comprehensive benchmark comprising three learning regimes: relationship incremental, scene incremental, and relationship generalization. Moreover, we introduce a ``Replays via Analysis by Synthesis" method named RAS. This approach leverages the scene graphs, decomposes and re-composes them to represent different scenes, and replays the synthesized scenes based on these compositional scene graphs. The replayed synthesized scenes act as a means to practice and refine proficiency in SGG in known and unknown environments. Our experimental results not only highlight the challenges of directly combining existing continual learning methods with SGG backbones but also demonstrate the effectiveness of our proposed approach, enhancing CSEGG efficiency while simultaneously preserving privacy and memory usage. All data and source code are publicly available online.
Authors: Chang'an Yi, Haotian Chen, Yifan Zhang, Yonghui Xu, Yan Zhou, Lizhen Cui
Abstract: Test-time adaptation (TTA) aims to adapt a model, initially trained on training data, to test data with potential distribution shifts. Most existing TTA methods focus on classification problems. The pronounced success of classification might lead numerous newcomers and engineers to assume that classic TTA techniques can be directly applied to the more challenging task of semantic segmentation. However, this belief is still an open question. In this paper, we investigate the applicability of existing classic TTA strategies in semantic segmentation. Our comprehensive results have led to three key observations. First, the classic normalization updating strategy only brings slight performance improvement, and in some cases, it might even adversely affect the results. Even with the application of advanced distribution estimation techniques like batch renormalization, the problem remains unresolved. Second, although the teacher-student scheme does enhance the training stability for segmentation TTA in the presence of noisy pseudo-labels and temporal correlation, it cannot directly result in performance improvement compared to the original model without TTA under complex data distribution. Third, segmentation TTA suffers a severe long-tailed class-imbalance problem, which is substantially more complex than that in TTA for classification. This long-tailed challenge negatively affects segmentation TTA performance, even when the accuracy of pseudo-labels is high. Besides those observations, we find that visual prompt tuning (VisPT) is promising in segmentation TTA and propose a novel method named TTAP. The outstanding performance of TTAP has also been verified. We hope the community can give more attention to this challenging, yet important, segmentation TTA task in the future. The source code is available at: \textit{https://github.com/ycarobot/TTAP
Authors: Nikolaos Ioannis Bountos, Maria Sdraka, Angelos Zavras, Ilektra Karasante, Andreas Karavias, Themistocles Herekakis, Angeliki Thanasou, Dimitrios Michail, Ioannis Papoutsis
Abstract: Global floods, exacerbated by climate change, pose severe threats to human life, infrastructure, and the environment. Recent catastrophic events in Pakistan and New Zealand underscore the urgent need for precise flood mapping to guide restoration efforts, understand vulnerabilities, and prepare for future occurrences. While Synthetic Aperture Radar (SAR) remote sensing offers day-and-night, all-weather imaging capabilities, its application in deep learning for flood segmentation is limited by the lack of large annotated datasets. To address this, we introduce Kuro Siwo, a manually annotated multi-temporal dataset, spanning 43 flood events globally. Our dataset maps more than 338 billion $m^2$ of land, with 33 billion designated as either flooded areas or permanent water bodies. Kuro Siwo includes a highly processed product optimized for flood mapping based on SAR Ground Range Detected, and a primal SAR Single Look Complex product with minimal preprocessing, designed to promote research on the exploitation of both the phase and amplitude information and to offer maximum flexibility for downstream task preprocessing. To leverage advances in large scale self-supervised pretraining methods for remote sensing data, we augment Kuro Siwo with a large unlabeled set of SAR samples. Finally, we provide an extensive benchmark, namely BlackBench, offering strong baselines for a diverse set of flood events from Europe, America, Africa, Asia and Australia.
Authors: Tianhong Li, Dina Katabi, Kaiming He
Abstract: Unconditional generation -- the problem of modeling data distribution without relying on human-annotated labels -- is a long-standing and fundamental challenge in generative models, creating a potential of learning from large-scale unlabeled data. In the literature, the generation quality of an unconditional method has been much worse than that of its conditional counterpart. This gap can be attributed to the lack of semantic information provided by labels. In this work, we show that one can close this gap by generating semantic representations in the representation space produced by a self-supervised encoder. These representations can be used to condition the image generator. This framework, called Representation-Conditioned Generation (RCG), provides an effective solution to the unconditional generation problem without using labels. Through comprehensive experiments, we observe that RCG significantly improves unconditional generation quality: e.g., it achieves a new state-of-the-art FID of 2.15 on ImageNet 256x256, largely reducing the previous best of 5.91 by a relative 64%. Our unconditional results are situated in the same tier as the leading class-conditional ones. We hope these encouraging observations will attract the community's attention to the fundamental problem of unconditional generation. Code is available at https://github.com/LTH14/rcg.
Authors: Shengsheng Qian, Dizhan Xue, Yifei Wang, Shengjie Zhang, Huaiwen Zhang, Changsheng Xu
Abstract: Self-Supervised Learning (SSL) is an effective paradigm for learning representations from unlabeled data, such as text, images, and videos. However, researchers have recently found that SSL is vulnerable to backdoor attacks. The attacker can embed hidden SSL backdoors via a few poisoned examples in the training dataset and maliciously manipulate the behavior of downstream models. To defend against SSL backdoor attacks, a feasible route is to detect and remove the poisonous samples in the training set. However, the existing SSL backdoor defense method fails to detect the poisonous samples precisely. In this paper, we propose to erase the SSL backdoor by cluster activation masking and propose a novel PoisonCAM method. After obtaining the threat model trained on the poisoned dataset, our method can precisely detect poisonous samples based on the assumption that masking the backdoor trigger can effectively change the activation of a downstream clustering model. In experiments, our PoisonCAM achieves 96\% accuracy for backdoor trigger detection compared to 3\% of the state-of-the-art method on poisoned ImageNet-100. Moreover, our proposed PoisonCAM significantly improves the performance of the trained SSL model under backdoor attacks compared to the state-of-the-art method. Our code, data, and trained models will be open once this paper is accepted.
Authors: Rohith Peddi, Shivvrat Arya, Bharath Challa, Likhitha Pallapothula, Akshay Vyas, Bhavya Gouripeddi, Jikai Wang, Qifan Zhang, Vasundhara Komaragiri, Eric Ragan, Nicholas Ruozzi, Yu Xiang, Vibhav Gogate
Abstract: Following step-by-step procedures is an essential component of various activities carried out by individuals in their daily lives. These procedures serve as a guiding framework that helps to achieve goals efficiently, whether it is assembling furniture or preparing a recipe. However, the complexity and duration of procedural activities inherently increase the likelihood of making errors. Understanding such procedural activities from a sequence of frames is a challenging task that demands an accurate interpretation of visual information and the ability to reason about the structure of the activity. To this end, we collect a new egocentric 4D dataset, CaptainCook4D, comprising 384 recordings (94.5 hours) of people performing recipes in real kitchen environments. This dataset consists of two distinct types of activity: one in which participants adhere to the provided recipe instructions and another in which they deviate and induce errors. We provide 5.3K step annotations and 10K fine-grained action annotations and benchmark the dataset for the following tasks: supervised error recognition, multistep localization, and procedure learning
Authors: Kiyohiro Nakayama, Mikaela Angelina Uy, Yang You, Ke Li, Leonidas J. Guibas
Abstract: Neural radiance fields (NeRFs) have gained popularity with multiple works showing promising results across various applications. However, to the best of our knowledge, existing works do not explicitly model the distribution of training camera poses, or consequently the triangulation quality, a key factor affecting reconstruction quality dating back to classical vision literature. We close this gap with ProvNeRF, an approach that models the \textbf{provenance} for each point -- i.e., the locations where it is likely visible -- of NeRFs as a stochastic field. We achieve this by extending implicit maximum likelihood estimation (IMLE) to functional space with an optimizable objective. We show that modeling per-point provenance during the NeRF optimization enriches the model with information on triangulation leading to improvements in novel view synthesis and uncertainty estimation under the challenging sparse, unconstrained view setting against competitive baselines.
Authors: Nemanja Ra\v{s}ajski, Chintan Trivedi, Konstantinos Makantasis, Antonios Liapis, Georgios N. Yannakakis
Abstract: Domain randomisation enhances the transferability of vision models across visually distinct domains with similar content. However, current methods heavily depend on intricate simulation engines, hampering feasibility and scalability. This paper introduces BehAVE, a video understanding framework that utilises existing commercial video games for domain randomisation without accessing their simulation engines. BehAVE taps into the visual diversity of video games for randomisation and uses textual descriptions of player actions to align videos with similar content. We evaluate BehAVE across 25 first-person shooter (FPS) games using various video and text foundation models, demonstrating its robustness in domain randomisation. BehAVE effectively aligns player behavioural patterns and achieves zero-shot transfer to multiple unseen FPS games when trained on just one game. In a more challenging scenario, BehAVE enhances the zero-shot transferability of foundation models to unseen FPS games, even when trained on a game of a different genre, with improvements of up to 22%. BehAVE is available online at https://github.com/nrasajski/BehAVE.
Authors: Zeqi Xiao, Yifan Zhou, Shuai Yang, Xingang Pan
Abstract: Video generation primarily aims to model authentic and customized motion across frames, making understanding and controlling the motion a crucial topic. Most diffusion-based studies on video motion focus on motion customization with training-based paradigms, which, however, demands substantial training resources and necessitates retraining for diverse models. Crucially, these approaches do not explore how video diffusion models encode cross-frame motion information in their features, lacking interpretability and transparency in their effectiveness. To answer this question, this paper introduces a novel perspective to understand, localize, and manipulate motion-aware features in video diffusion models. Through analysis using Principal Component Analysis (PCA), our work discloses that robust motion-aware feature already exists in video diffusion models. We present a new MOtion FeaTure (MOFT) by eliminating content correlation information and filtering motion channels. MOFT provides a distinct set of benefits, including the ability to encode comprehensive motion information with clear interpretability, extraction without the need for training, and generalizability across diverse architectures. Leveraging MOFT, we propose a novel training-free video motion control framework. Our method demonstrates competitive performance in generating natural and faithful motion, providing architecture-agnostic insights and applicability in a variety of downstream tasks.
Authors: Kushagra Pandey, Ruihan Yang, Stephan Mandt
Abstract: Constructing fast samplers for unconditional diffusion and flow-matching models has received much attention recently; however, existing methods for solving inverse problems, such as super-resolution, inpainting, or deblurring, still require hundreds to thousands of iterative steps to obtain high-quality results. We propose a plug-and-play framework for constructing efficient samplers for inverse problems, requiring only pre-trained diffusion or flow-matching models. We present Conditional Conjugate Integrators, which leverage the specific form of the inverse problem to project the respective conditional diffusion/flow dynamics into a more amenable space for sampling. Our method complements popular posterior approximation methods for solving inverse problems using diffusion/flow models. We evaluate the proposed method's performance on various linear image restoration tasks across multiple datasets, employing diffusion and flow-matching models. Notably, on challenging inverse problems like 4x super-resolution on the ImageNet dataset, our method can generate high-quality samples in as few as 5 conditional sampling steps and outperforms competing baselines requiring 20-1000 steps. Our code will be publicly available at https://github.com/mandt-lab/c-pigdm
Authors: Jaehong Yoon, Shoubin Yu, Mohit Bansal
Abstract: Recent video generative models primarily rely on carefully written text prompts for specific tasks, like inpainting or style editing. They require labor-intensive textual descriptions for input videos, hindering their flexibility to adapt personal/raw videos to user specifications. This paper proposes RACCooN, a versatile and user-friendly video-to-paragraph-to-video generative framework that supports multiple video editing capabilities such as removal, addition, and modification, through a unified pipeline. RACCooN consists of two principal stages: Video-to-Paragraph (V2P) and Paragraph-to-Video (P2V). In the V2P stage, we automatically describe video scenes in well-structured natural language, capturing both the holistic context and focused object details. Subsequently, in the P2V stage, users can optionally refine these descriptions to guide the video diffusion model, enabling various modifications to the input video, such as removing, changing subjects, and/or adding new objects. The proposed approach stands out from other methods through several significant contributions: (1) RACCooN suggests a multi-granular spatiotemporal pooling strategy to generate well-structured video descriptions, capturing both the broad context and object details without requiring complex human annotations, simplifying precise video content editing based on text for users. (2) Our video generative model incorporates auto-generated narratives or instructions to enhance the quality and accuracy of the generated content. (3) RACCooN also plans to imagine new objects in a given video, so users simply prompt the model to receive a detailed video editing plan for complex video editing. The proposed framework demonstrates impressive versatile capabilities in video-to-paragraph generation, video content editing, and can be incorporated into other SoTA video generative models for further enhancement.
Authors: Khoa Vo, Thinh Phan, Kashu Yamazaki, Minh Tran, Ngan Le
Abstract: Current video-language models (VLMs) rely extensively on instance-level alignment between video and language modalities, which presents two major limitations: (1) visual reasoning disobeys the natural perception that humans do in first-person perspective, leading to a lack of reasoning interpretation; and (2) learning is limited in capturing inherent fine-grained relationships between two modalities. In this paper, we take an inspiration from human perception and explore a compositional approach for egocentric video representation. We introduce HENASY (Hierarchical ENtities ASsemblY), which includes a spatiotemporal token grouping mechanism to explicitly assemble dynamically evolving scene entities through time and model their relationship for video representation. By leveraging compositional structure understanding, HENASY possesses strong interpretability via visual grounding with free-form text queries. We further explore a suite of multi-grained contrastive losses to facilitate entity-centric understandings. This comprises three alignment types: video-narration, noun-entity, verb-entities alignments. Our method demonstrates strong interpretability in both quantitative and qualitative experiments; while maintaining competitive performances on five downstream tasks via zero-shot transfer or as video/text representation, including video/text retrieval, action recognition, multi-choice query, natural language query, and moments query.
Authors: Zhengrui Xu, Guan'an Wang, Xiaowen Huang, Jitao Sang
Abstract: The denoising model has been proven a powerful generative model but has little exploration of discriminative tasks. Representation learning is important in discriminative tasks, which is defined as "learning representations (or features) of the data that make it easier to extract useful information when building classifiers or other predictors". In this paper, we propose a novel Denoising Model for Representation Learning (DenoiseRep) to improve feature discrimination with joint feature extraction and denoising. DenoiseRep views each embedding layer in a backbone as a denoising layer, processing the cascaded embedding layers as if we are recursively denoise features step-by-step. This unifies the frameworks of feature extraction and denoising, where the former progressively embeds features from low-level to high-level, and the latter recursively denoises features step-by-step. After that, DenoiseRep fuses the parameters of feature extraction and denoising layers, and theoretically demonstrates its equivalence before and after the fusion, thus making feature denoising computation-free. DenoiseRep is a label-free algorithm that incrementally improves features but also complementary to the label if available. Experimental results on various discriminative vision tasks, including re-identification (Market-1501, DukeMTMC-reID, MSMT17, CUHK-03, vehicleID), image classification (ImageNet, UB200, Oxford-Pet, Flowers), object detection (COCO), image segmentation (ADE20K) show stability and impressive improvements. We also validate its effectiveness on the CNN (ResNet) and Transformer (ViT, Swin, Vmamda) architectures.
Authors: Desai Xie, Sai Bi, Zhixin Shu, Kai Zhang, Zexiang Xu, Yi Zhou, S\"oren Pirk, Arie Kaufman, Xin Sun, Hao Tan
Abstract: We present LRM-Zero, a Large Reconstruction Model (LRM) trained entirely on synthesized 3D data, achieving high-quality sparse-view 3D reconstruction. The core of LRM-Zero is our procedural 3D dataset, Zeroverse, which is automatically synthesized from simple primitive shapes with random texturing and augmentations (e.g., height fields, boolean differences, and wireframes). Unlike previous 3D datasets (e.g., Objaverse) which are often captured or crafted by humans to approximate real 3D data, Zeroverse completely ignores realistic global semantics but is rich in complex geometric and texture details that are locally similar to or even more intricate than real objects. We demonstrate that our LRM-Zero, trained with our fully synthesized Zeroverse, can achieve high visual quality in the reconstruction of real-world objects, competitive with models trained on Objaverse. We also analyze several critical design choices of Zeroverse that contribute to LRM-Zero's capability and training stability. Our work demonstrates that 3D reconstruction, one of the core tasks in 3D vision, can potentially be addressed without the semantics of real-world objects. The Zeroverse's procedural synthesis code and interactive visualization are available at: https://desaixie.github.io/lrm-zero/.
Authors: Tianhong Li, Yonglong Tian, He Li, Mingyang Deng, Kaiming He
Abstract: Conventional wisdom holds that autoregressive models for image generation are typically accompanied by vector-quantized tokens. We observe that while a discrete-valued space can facilitate representing a categorical distribution, it is not a necessity for autoregressive modeling. In this work, we propose to model the per-token probability distribution using a diffusion procedure, which allows us to apply autoregressive models in a continuous-valued space. Rather than using categorical cross-entropy loss, we define a Diffusion Loss function to model the per-token probability. This approach eliminates the need for discrete-valued tokenizers. We evaluate its effectiveness across a wide range of cases, including standard autoregressive models and generalized masked autoregressive (MAR) variants. By removing vector quantization, our image generator achieves strong results while enjoying the speed advantage of sequence modeling. We hope this work will motivate the use of autoregressive generation in other continuous-valued domains and applications. Code is available at: https://github.com/LTH14/mar.
Authors: Xuweiyi Chen, Ziqiao Ma, Xuejun Zhang, Sihan Xu, Shengyi Qian, Jianing Yang, David F. Fouhey, Joyce Chai
Abstract: Large vision language models (LVLMs) often suffer from object hallucination, producing objects not present in the given images. While current benchmarks for object hallucination primarily concentrate on the presence of a single object class rather than individual entities, this work systematically investigates multi-object hallucination, examining how models misperceive (e.g., invent nonexistent objects or become distracted) when tasked with focusing on multiple objects simultaneously. We introduce Recognition-based Object Probing Evaluation (ROPE), an automated evaluation protocol that considers the distribution of object classes within a single image during testing and uses visual referring prompts to eliminate ambiguity. With comprehensive empirical studies and analysis of potential factors leading to multi-object hallucination, we found that (1). LVLMs suffer more hallucinations when focusing on multiple objects compared to a single object. (2). The tested object class distribution affects hallucination behaviors, indicating that LVLMs may follow shortcuts and spurious correlations. (3). Hallucinatory behaviors are influenced by data-specific factors, salience and frequency, and model intrinsic behaviors. We hope to enable LVLMs to recognize and reason about multiple objects that often occur in realistic visual scenes, provide insights, and quantify our progress towards mitigating the issues.
Authors: Shuang Wang, Qianwen Lu, Boxing Peng, Yihe Nie, Qingchuan Tao
Abstract: For the task of low-light image enhancement, deep learning-based algorithms have demonstrated superiority and effectiveness compared to traditional methods. However, these methods, primarily based on Retinex theory, tend to overlook the noise and color distortions in input images, leading to significant noise amplification and local color distortions in enhanced results. To address these issues, we propose the Dual-Path Error Compensation (DPEC) method, designed to improve image quality under low-light conditions by preserving local texture details while restoring global image brightness without amplifying noise. DPEC incorporates precise pixel-level error estimation to capture subtle differences and an independent denoising mechanism to prevent noise amplification. We introduce the HIS-Retinex loss to guide DPEC's training, ensuring the brightness distribution of enhanced images closely aligns with real-world conditions. To balance computational speed and resource efficiency while training DPEC for a comprehensive understanding of the global context, we integrated the VMamba architecture into its backbone. Comprehensive quantitative and qualitative experimental results demonstrate that our algorithm significantly outperforms state-of-the-art methods in low-light image enhancement. The code is publicly available online at https://github.com/wangshuang233/DPEC.
Authors: Jae Joong Lee, Bedrich Benes
Abstract: We introduce RGB2Point, an unposed single-view RGB image to a 3D point cloud generation based on Transformer. RGB2Point takes an input image of an object and generates a dense 3D point cloud. Contrary to prior works based on CNN layers and diffusion denoising approaches, we use pre-trained Transformer layers that are fast and generate high-quality point clouds with consistent quality over available categories. Our generated point clouds demonstrate high quality on a real-world dataset, as evidenced by improved Chamfer distance (51.15%) and Earth Mover's distance (45.96%) metrics compared to the current state-of-the-art. Additionally, our approach shows a better quality on a synthetic dataset, achieving better Chamfer distance (39.26%), Earth Mover's distance (26.95%), and F-score (47.16%). Moreover, our method produces 63.1% more consistent high-quality results across various object categories compared to prior works. Furthermore, RGB2Point is computationally efficient, requiring only 2.3GB of VRAM to reconstruct a 3D point cloud from a single RGB image, and our implementation generates the results 15,133x faster than a SOTA diffusion-based model.
Authors: Guiqiu Liao, Matjaz Jogan, Sai Koushik, Eric Eaton, Daniel A. Hashimoto
Abstract: Weakly supervised video object segmentation (WSVOS) enables the identification of segmentation maps without requiring an extensive training dataset of object masks, relying instead on coarse video labels indicating object presence. Current state-of-the-art methods either require multiple independent stages of processing that employ motion cues or, in the case of end-to-end trainable networks, lack in segmentation accuracy, in part due to the difficulty of learning segmentation maps from videos with transient object presence. This limits the application of WSVOS for semantic annotation of surgical videos where multiple surgical tools frequently move in and out of the field of view, a problem that is more difficult than typically encountered in WSVOS. This paper introduces Video Spatio-Temporal Disentanglement Networks (VDST-Net), a framework to disentangle spatiotemporal information using semi-decoupled knowledge distillation to predict high-quality class activation maps (CAMs). A teacher network designed to resolve temporal conflicts when specifics about object location and timing in the video are not provided works with a student network that integrates information over time by leveraging temporal dependencies. We demonstrate the efficacy of our framework on a public reference dataset and on a more challenging surgical video dataset where objects are, on average, present in less than 60\% of annotated frames. Our method outperforms state-of-the-art techniques and generates superior segmentation masks under video-level weak supervision.
Authors: Yunhao Fang, Ligeng Zhu, Yao Lu, Yan Wang, Pavlo Molchanov, Jan Kautz, Jang Hyun Cho, Marco Pavone, Song Han, Hongxu Yin
Abstract: While visual language model architectures and training infrastructures advance rapidly, data curation remains under-explored where quantity and quality become a bottleneck. Existing work either crawls extra Internet data with a loose guarantee of quality or distills from black-box proprietary models, e.g., GPT-4V / Gemini that are API frequency and performance bounded. This work enables a VLM to improve itself via data enhancement, exploiting its generative nature. We introduce a simple yet effective VLM augmentation scheme that includes a self-augment step and a specialist-augment step to iteratively improve data quality and hence, model performance. In the self-augment step, the instruction-finetuned VLM recaptions its pretraining caption datasets and then retrains from scratch leveraging refined data. Without any expensive human-in-the-loop annotation, we observe improvements in data quality and downstream accuracy boosts with three self-augmentation rounds -- a viable free lunch to the current VLM training recipe. When self-augmentation saturates, we augment the caption diversity by leveraging specialty skills picked up from instruction finetuning. We finetune VLM specialists from the self-augmented VLM with domain-specific experts, including spatial, grounding, and OCR, to fuse task-aware synthetic data into the pretraining stage. Data quality improvements and hallucination reductions are cross-checked by VLM (GPT-4V, Gemini) and human judges. Combining self-augmentation and specialist-augmented training, VILA$^2$ consistently improves the accuracy on a wide range of benchmarks over the prior art, producing a reusable pretraining dataset that is 300x more cost-efficient than human labeling.
Authors: Fuzhao Xue, Yukang Chen, Dacheng Li, Qinghao Hu, Ligeng Zhu, Xiuyu Li, Yunhao Fang, Haotian Tang, Shang Yang, Zhijian Liu, Ethan He, Hongxu Yin, Pavlo Molchanov, Jan Kautz, Linxi Fan, Yuke Zhu, Yao Lu, Song Han
Abstract: Long-context capability is critical for multi-modal foundation models, especially for long video understanding. We introduce LongVILA, a full-stack solution for long-context visual-language models by co-designing the algorithm and system. For model training, we upgrade existing VLMs to support long video understanding by incorporating two additional stages, i.e., long context extension and long video supervised fine-tuning. However, training on long video is computationally and memory intensive. We introduce the long-context Multi-Modal Sequence Parallelism (MM-SP) system that efficiently parallelizes long video training and inference, enabling 2M context length training on 256 GPUs without any gradient checkpointing. LongVILA efficiently extends the number of video frames of VILA from 8 to 2048, improving the long video captioning score from 2.00 to 3.26 (out of 5), achieving 99.8% accuracy in 6,000-frame (more than 1 million tokens) video needle-in-a-haystack. LongVILA-7B demonstrates strong accuracy on the VideoMME benchmark, i.e., 61.8% with subtitle. Besides, MM-SP is 2.1x - 5.7x faster than ring style sequence parallelism and 1.1x - 1.4x faster than Megatron with a hybrid context and tensor parallelism. Moreover, it seamlessly integrates with Hugging Face Transformers.
Authors: Chaohua Shi, Xuan Wang, Si Shi, Xule Wang, Mingrui Zhu, Nannan Wang, Xinbo Gao
Abstract: Food image composition requires the use of existing dish images and background images to synthesize a natural new image, while diffusion models have made significant advancements in image generation, enabling the construction of end-to-end architectures that yield promising results. However, existing diffusion models face challenges in processing and fusing information from multiple images and lack access to high-quality publicly available datasets, which prevents the application of diffusion models in food image composition. In this paper, we introduce a large-scale, high-quality food image composite dataset, FC22k, which comprises 22,000 foreground, background, and ground truth ternary image pairs. Additionally, we propose a novel food image composition method, Foodfusion, which leverages the capabilities of the pre-trained diffusion models and incorporates a Fusion Module for processing and integrating foreground and background information. This fused information aligns the foreground features with the background structure by merging the global structural information at the cross-attention layer of the denoising UNet. To further enhance the content and structure of the background, we also integrate a Content-Structure Control Module. Extensive experiments demonstrate the effectiveness and scalability of our proposed method.
Authors: Md Manjurul Ahsan, Yingtao Liu, Shivakumar Raman, Zahed Siddique
Abstract: Digital Twins (DTs) are becoming popular in Additive Manufacturing (AM) due to their ability to create virtual replicas of physical components of AM machines, which helps in real-time production monitoring. Advanced techniques such as Machine Learning (ML), Augmented Reality (AR), and simulation-based models play key roles in developing intelligent and adaptable DTs in manufacturing processes. However, questions remain regarding scalability, the integration of high-quality data, and the computational power required for real-time applications in developing DTs. Understanding the current state of DTs in AM is essential to address these challenges and fully utilize their potential in advancing AM processes. Considering this opportunity, this work aims to provide a comprehensive overview of DTs in AM by addressing the following four research questions: (1) What are the key types of DTs used in AM and their specific applications? (2) What are the recent developments and implementations of DTs? (3) How are DTs employed in process improvement and hybrid manufacturing? (4) How are DTs integrated with Industry 4.0 technologies? By discussing current applications and techniques, we aim to offer a better understanding and potential future research directions for researchers and practitioners in AM and DTs.
Authors: Daniel Miao, Gilad Lerman, Joe Kileel
Abstract: The block tensor of trifocal tensors provides crucial geometric information on the three-view geometry of a scene. The underlying synchronization problem seeks to recover camera poses (locations and orientations up to a global transformation) from the block trifocal tensor. We establish an explicit Tucker factorization of this tensor, revealing a low multilinear rank of $(6,4,4)$ independent of the number of cameras under appropriate scaling conditions. We prove that this rank constraint provides sufficient information for camera recovery in the noiseless case. The constraint motivates a synchronization algorithm based on the higher-order singular value decomposition of the block trifocal tensor. Experimental comparisons with state-of-the-art global synchronization methods on real datasets demonstrate the potential of this algorithm for significantly improving location estimation accuracy. Overall this work suggests that higher-order interactions in synchronization problems can be exploited to improve performance, beyond the usual pairwise-based approaches.
Authors: Li Pang, Xiangyong Cao, Datao Tang, Shuang Xu, Xueru Bai, Feng Zhou, Deyu Meng
Abstract: Hyperspectral image (HSI) plays a vital role in various fields such as agriculture and environmental monitoring. However, due to the expensive acquisition cost, the number of hyperspectral images is limited, degenerating the performance of downstream tasks. Although some recent studies have attempted to employ diffusion models to synthesize HSIs, they still struggle with the scarcity of HSIs, affecting the reliability and diversity of the generated images. Some studies propose to incorporate multi-modal data to enhance spatial diversity, but the spectral fidelity cannot be ensured. In addition, existing HSI synthesis models are typically uncontrollable or only support single-condition control, limiting their ability to generate accurate and reliable HSIs. To alleviate these issues, we propose HSIGene, a novel HSI generation foundation model which is based on latent diffusion and supports multi-condition control, allowing for more precise and reliable HSI generation. To enhance the spatial diversity of the training data while preserving spectral fidelity, we propose a new data augmentation method based on spatial super-resolution, in which HSIs are upscaled first, and thus abundant training patches could be obtained by cropping the high-resolution HSIs. In addition, to improve the perceptual quality of the augmented data, we introduce a novel two-stage HSI super-resolution framework, which first applies RGB bands super-resolution and then utilizes our proposed Rectangular Guided Attention Network (RGAN) for guided HSI super-resolution. Experiments demonstrate that the proposed model is capable of generating a vast quantity of realistic HSIs for downstream tasks such as denoising and super-resolution. The code and models are available at https://github.com/LiPang/HSIGene.
Authors: Xun Zhu, Ying Hu, Fanbin Mo, Miao Li, Ji Wu
Abstract: Multi-modal large language models (MLLMs) have shown impressive capabilities as a general-purpose interface for various visual and linguistic tasks. However, building a unified MLLM for multi-task learning in the medical field remains a thorny challenge. To mitigate the tug-of-war problem of multi-modal multi-task optimization in MLLMs, recent advances primarily focus on improving the LLM components, while neglecting the connector that bridges the gap between modalities. In this paper, we introduce Uni-Med, a novel medical generalist foundation model which consists of a universal visual feature extraction module, a connector mixture-of-experts (CMoE) module, and an LLM. Benefiting from the proposed CMoE that leverages a well-designed router with a mixture of projection experts at the connector, Uni-Med achieves efficient solution to the tug-of-war problem and can perform six different medical tasks including question answering, visual question answering, report generation, referring expression comprehension, referring expression generation and image classification. To the best of our knowledge, Uni-Med is the first effort to tackle multi-task interference at the connector in MLLMs. Extensive ablation experiments validate the effectiveness of introducing CMoE under any configuration, with up to an average 8% performance gains. We further provide interpretation analysis of the tug-of-war problem from the perspective of gradient optimization and parameter statistics. Compared to previous state-of-the-art medical MLLMs, Uni-Med achieves competitive or superior evaluation metrics on diverse tasks. Code and resources are available at https://github.com/tsinghua-msiip/Uni-Med.
Authors: Ruiyu Mao, Sarthak Kumar Maharana, Rishabh K Iyer, Yunhui Guo
Abstract: 3D object detection is fundamentally important for various emerging applications, including autonomous driving and robotics. A key requirement for training an accurate 3D object detector is the availability of a large amount of LiDAR-based point cloud data. Unfortunately, labeling point cloud data is extremely challenging, as accurate 3D bounding boxes and semantic labels are required for each potential object. This paper proposes a unified active 3D object detection framework, for greatly reducing the labeling cost of training 3D object detectors. Our framework is based on a novel formulation of submodular optimization, specifically tailored to the problem of active 3D object detection. In particular, we address two fundamental challenges associated with active 3D object detection: data imbalance and the need to cover the distribution of the data, including LiDAR-based point cloud data of varying difficulty levels. Extensive experiments demonstrate that our method achieves state-of-the-art performance with high computational efficiency compared to existing active learning methods. The code is available at https://github.com/RuiyuM/STONE.
Authors: Leonid Pogorelyuk, Stefan T. Radev
Abstract: We propose a new contrastive objective for learning overcomplete pixel-level features that are invariant to motion blur. Other invariances (e.g., pose, illumination, or weather) can be learned by applying the corresponding transformations on unlabeled images during self-supervised training. We showcase that a simple U-Net trained with our objective can produce local features useful for aligning the frames of an unseen video captured with a moving camera under realistic and challenging conditions. Using a carefully designed toy example, we also show that the overcomplete pixels can encode the identity of objects in an image and the pixel coordinates relative to these objects.
Authors: Hamid Manoochehri, Bodong Zhang, Beatrice S. Knudsen, Tolga Tasdizen
Abstract: Self-supervised learning has become a cornerstone in various areas, particularly histopathological image analysis. Image augmentation plays a crucial role in self-supervised learning, as it generates variations in image samples. However, traditional image augmentation techniques often overlook the unique characteristics of histopathological images. In this paper, we propose a new histopathology-specific image augmentation method called stain reconstruction augmentation (SRA). We integrate our SRA with MoCo v3, a leading model in self-supervised contrastive learning, along with our additional contrastive loss terms, and call the new model SRA-MoCo v3. We demonstrate that our SRA-MoCo v3 always outperforms the standard MoCo v3 across various downstream tasks and achieves comparable or superior performance to other foundation models pre-trained on significantly larger histopathology datasets.
Authors: XuDong Wang, Shaolun Zhang, Shufan Li, Konstantinos Kallidromitis, Kehan Li, Yusuke Kato, Kazuki Kozuka, Trevor Darrell
Abstract: We present SegLLM, a novel multi-round interactive reasoning segmentation model that enhances LLM-based segmentation by exploiting conversational memory of both visual and textual outputs. By leveraging a mask-aware multimodal LLM, SegLLM re-integrates previous segmentation results into its input stream, enabling it to reason about complex user intentions and segment objects in relation to previously identified entities, including positional, interactional, and hierarchical relationships, across multiple interactions. This capability allows SegLLM to respond to visual and text queries in a chat-like manner. Evaluated on the newly curated MRSeg benchmark, SegLLM outperforms existing methods in multi-round interactive reasoning segmentation by over 20%. Additionally, we observed that training on multi-round reasoning segmentation data enhances performance on standard single-round referring segmentation and localization tasks, resulting in a 5.5% increase in cIoU for referring expression segmentation and a 4.5% improvement in Acc@0.5 for referring expression localization.
Authors: Ranjan Sapkota, Manoj Karkee
Abstract: This study conducted a comprehensive performance evaluation on YOLO11 and YOLOv8, the latest in the "You Only Look Once" (YOLO) series, focusing on their instance segmentation capabilities for immature green apples in orchard environments. YOLO11n-seg achieved the highest mask precision across all categories with a notable score of 0.831, highlighting its effectiveness in fruit detection. YOLO11m-seg and YOLO11l-seg excelled in non-occluded and occluded fruitlet segmentation with scores of 0.851 and 0.829, respectively. Additionally, YOLO11x-seg led in mask recall for all categories, achieving a score of 0.815, with YOLO11m-seg performing best for non-occluded immature green fruitlets at 0.858 and YOLOv8x-seg leading the occluded category with 0.800. In terms of mean average precision at a 50\% intersection over union (mAP@50), YOLO11m-seg consistently outperformed, registering the highest scores for both box and mask segmentation, at 0.876 and 0.860 for the "All" class and 0.908 and 0.909 for non-occluded immature fruitlets, respectively. YOLO11l-seg and YOLOv8l-seg shared the top box mAP@50 for occluded immature fruitlets at 0.847, while YOLO11m-seg achieved the highest mask mAP@50 of 0.810. Despite the advancements in YOLO11, YOLOv8n surpassed its counterparts in image processing speed, with an impressive inference speed of 3.3 milliseconds, compared to the fastest YOLO11 series model at 4.8 milliseconds, underscoring its suitability for real-time agricultural applications related to complex green fruit environments.
Authors: Mingjiang Liang, Yongkang Cheng, Hualin Liang, Shaoli Huang, Wei Liu
Abstract: We present RopeTP, a novel framework that combines Robust pose estimation with a diffusion Trajectory Prior to reconstruct global human motion from videos. At the heart of RopeTP is a hierarchical attention mechanism that significantly improves context awareness, which is essential for accurately inferring the posture of occluded body parts. This is achieved by exploiting the relationships with visible anatomical structures, enhancing the accuracy of local pose estimations. The improved robustness of these local estimations allows for the reconstruction of precise and stable global trajectories. Additionally, RopeTP incorporates a diffusion trajectory model that predicts realistic human motion from local pose sequences. This model ensures that the generated trajectories are not only consistent with observed local actions but also unfold naturally over time, thereby improving the realism and stability of 3D human motion reconstruction. Extensive experimental validation shows that RopeTP surpasses current methods on two benchmark datasets, particularly excelling in scenarios with occlusions. It also outperforms methods that rely on SLAM for initial camera estimates and extensive optimization, delivering more accurate and realistic trajectories.
Authors: Phillip Y. Lee, Taehoon Yoon, Minhyuk Sung
Abstract: We introduce GrounDiT, a novel training-free spatial grounding technique for text-to-image generation using Diffusion Transformers (DiT). Spatial grounding with bounding boxes has gained attention for its simplicity and versatility, allowing for enhanced user control in image generation. However, prior training-free approaches often rely on updating the noisy image during the reverse diffusion process via backpropagation from custom loss functions, which frequently struggle to provide precise control over individual bounding boxes. In this work, we leverage the flexibility of the Transformer architecture, demonstrating that DiT can generate noisy patches corresponding to each bounding box, fully encoding the target object and allowing for fine-grained control over each region. Our approach builds on an intriguing property of DiT, which we refer to as semantic sharing. Due to semantic sharing, when a smaller patch is jointly denoised alongside a generatable-size image, the two become semantic clones. Each patch is denoised in its own branch of the generation process and then transplanted into the corresponding region of the original noisy image at each timestep, resulting in robust spatial grounding for each bounding box. In our experiments on the HRS and DrawBench benchmarks, we achieve state-of-the-art performance compared to previous training-free approaches.
Authors: Camilo Espinosa-Curilem, Millaray Curilem, Daniel Basualto
Abstract: In volcano monitoring, effective recognition of seismic events is essential for understanding volcanic activity and raising timely warning alerts. Traditional methods rely on manual analysis, which can be subjective and labor-intensive. Furthermore, current automatic approaches often tackle detection and classification separately, mostly rely on single station information and generally require tailored preprocessing and representations to perform predictions. These limitations often hinder their application to real-time monitoring and utilization across different volcano conditions. This study introduces a novel approach that utilizes Semantic Segmentation models to automate seismic event recognition by applying a straight forward transformation of multi-channel 1D signals into 2D representations, enabling their use as images. Our framework employs a data-driven, end-to-end design that integrates multi-station seismic data with minimal preprocessing, performing both detection and classification simultaneously for five seismic event classes. We evaluated four state-of-the-art segmentation models (UNet, UNet++, DeepLabV3+ and SwinUNet) on approximately 25.000 seismic events recorded at four different Chilean volcanoes: Nevados del Chill\'an Volcanic Complex, Laguna del Maule, Villarrica and Puyehue-Cord\'on Caulle. Among these models, the UNet architecture was identified as the most effective model, achieving mean F1 and Intersection over Union (IoU) scores of up to 0.91 and 0.88, respectively, and demonstrating superior noise robustness and model flexibility to unseen volcano datasets.
Authors: Tien-Huy Nguyen, Quang-Khai Tran, Anh-Tuan Quang-Hoang
Abstract: The cognitive faculty of visual reasoning necessitates the integration of multimodal perceptual processing and commonsense and external knowledge of the world. In recent years, a plethora of large vision-language models (LVLMs) have been proposed, demonstrating outstanding power and exceptional proficiency in commonsense reasoning across diverse domains and tasks. Nevertheless, training such LVLMs requires a lot of costly resources. Recent approaches, instead of training LVLMs from scratch on various large datasets, focus on exploring ways to take advantage of the capabilities of many different LVLMs, such as ensemble methods. In this work, we propose self-ensemble, a novel method that improves the generalization and visual reasoning of the model without updating any parameters, a training-free method. Our key insight is that we realized that LVLM itself can ensemble without the need for any other LVLMs, which helps to unlock their internal capabilities. Extensive experiments on various benchmarks demonstrate the effectiveness of our method in achieving state-of-the-art (SOTA) performance on SketchyVQA, Outside Knowledge VQA, and out-of-distribution VQA tasks.
Authors: Marcel Roth, Micha V. Nowak, Adrian Krenzer, Frank Puppe
Abstract: Video capsule endoscopy has transformed gastrointestinal endoscopy (GIE) diagnostics by offering a non-invasive method for capturing detailed images of the gastrointestinal tract, enabling early disease detection. However, its potential is limited by the sheer volume of images generated during the imaging procedure, which can take anywhere from 6-8 hours and often produce up to 1 million images, necessitating automated analysis. Additionally, the variability of these images, combined with the need for expert annotations and the scarcity of large, high-quality labeled datasets, constrains the effectiveness of current medical image analysis models. To address this, we introduce a novel large gastrointestinal endoscopy dataset, called EndoExtend24, created by merging and re-stratifying the train/test splits of ten existing public and private datasets, ensuring no overlap of patient data across splits. EndoExtend24 includes over 226,000 labeled images, as well as dynamic class mappings, which allow unified training across datasets with differing labeling granularity, supporting up to 123 distinct pathological findings. Further, we propose to leverage domain adaptive pre-training of foundation models in computer vision trained with self-supervision on generic image data, to adapt them to the task of GIE medical diagnosis. Specifically, the EVA-02 model, which is based on the vision transformer architecture and was trained on ImageNet-22k with masked image modeling (using EVA-CLIP as a MIM teacher), is pre-trained on the novel EndoExtend24 dataset to achieve domain adaptation, and finally trained on the Capsule Endoscopy 2024 Challenge dataset. Experimental results demonstrate strong performance with an F1 score of 0.88, an improvement of about 39% over the baseline model's F1 score of 0.49. Additionally, the model achieved a macro AUC score of 0.993 and a balanced accuracy of 89.3%.
Authors: Hongbo Li, Haikuan Zhu, Sikai Zhong, Ningna Wang, Cheng Lin, Xiaohu Guo, Shiqing Xin, Wenping Wang, Jing Hua, Zichun Zhong
Abstract: This paper introduces a new learning-based method, NASM, for anisotropic surface meshing. Our key idea is to propose a graph neural network to embed an input mesh into a high-dimensional (high-d) Euclidean embedding space to preserve curvature-based anisotropic metric by using a dot product loss between high-d edge vectors. This can dramatically reduce the computational time and increase the scalability. Then, we propose a novel feature-sensitive remeshing on the generated high-d embedding to automatically capture sharp geometric features. We define a high-d normal metric, and then derive an automatic differentiation on a high-d centroidal Voronoi tessellation (CVT) optimization with the normal metric to simultaneously preserve geometric features and curvature anisotropy that exhibit in the original 3D shapes. To our knowledge, this is the first time that a deep learning framework and a large dataset are proposed to construct a high-d Euclidean embedding space for 3D anisotropic surface meshing. Experimental results are evaluated and compared with the state-of-the-art in anisotropic surface meshing on a large number of surface models from Thingi10K dataset as well as tested on extensive unseen 3D shapes from Multi-Garment Network dataset and FAUST human dataset.
Authors: Meng Ye, Bingyu Xin, Leon Axel, Dimitris Metaxas
Abstract: Current cardiac cine magnetic resonance image (cMR) studies focus on the end diastole (ED) and end systole (ES) phases, while ignoring the abundant temporal information in the whole image sequence. This is because whole sequence segmentation is currently a tedious process and inaccurate. Conventional whole sequence segmentation approaches first estimate the motion field between frames, which is then used to propagate the mask along the temporal axis. However, the mask propagation results could be prone to error, especially for the basal and apex slices, where through-plane motion leads to significant morphology and structural change during the cardiac cycle. Inspired by recent advances in video object segmentation (VOS), based on spatio-temporal memory (STM) networks, we propose a continuous STM (CSTM) network for semi-supervised whole heart and whole sequence cMR segmentation. Our CSTM network takes full advantage of the spatial, scale, temporal and through-plane continuity prior of the underlying heart anatomy structures, to achieve accurate and fast 4D segmentation. Results of extensive experiments across multiple cMR datasets show that our method can improve the 4D cMR segmentation performance, especially for the hard-to-segment regions.
Authors: Yining Hong, Beide Liu, Maxine Wu, Yuanhao Zhai, Kai-Wei Chang, Linjie Li, Kevin Lin, Chung-Ching Lin, Jianfeng Wang, Zhengyuan Yang, Yingnian Wu, Lijuan Wang
Abstract: Human beings are endowed with a complementary learning system, which bridges the slow learning of general world dynamics with fast storage of episodic memory from a new experience. Previous video generation models, however, primarily focus on slow learning by pre-training on vast amounts of data, overlooking the fast learning phase crucial for episodic memory storage. This oversight leads to inconsistencies across temporally distant frames when generating longer videos, as these frames fall beyond the model's context window. To this end, we introduce SlowFast-VGen, a novel dual-speed learning system for action-driven long video generation. Our approach incorporates a masked conditional video diffusion model for the slow learning of world dynamics, alongside an inference-time fast learning strategy based on a temporal LoRA module. Specifically, the fast learning process updates its temporal LoRA parameters based on local inputs and outputs, thereby efficiently storing episodic memory in its parameters. We further propose a slow-fast learning loop algorithm that seamlessly integrates the inner fast learning loop into the outer slow learning loop, enabling the recall of prior multi-episode experiences for context-aware skill learning. To facilitate the slow learning of an approximate world model, we collect a large-scale dataset of 200k videos with language action annotations, covering a wide range of scenarios. Extensive experiments show that SlowFast-VGen outperforms baselines across various metrics for action-driven video generation, achieving an FVD score of 514 compared to 782, and maintaining consistency in longer videos, with an average of 0.37 scene cuts versus 0.89. The slow-fast learning loop algorithm significantly enhances performances on long-horizon planning tasks as well. Project Website: https://slowfast-vgen.github.io
Authors: Kyungjin Seo, Junghoon Seo, Hanseok Jeong, Sangpil Kim, Sang Ho Yoon
Abstract: We present PiMForce, a novel framework that enhances hand pressure estimation by leveraging 3D hand posture information to augment forearm surface electromyography (sEMG) signals. Our approach utilizes detailed spatial information from 3D hand poses in conjunction with dynamic muscle activity from sEMG to enable accurate and robust whole-hand pressure measurements under diverse hand-object interactions. We also developed a multimodal data collection system that combines a pressure glove, an sEMG armband, and a markerless finger-tracking module. We created a comprehensive dataset from 21 participants, capturing synchronized data of hand posture, sEMG signals, and exerted hand pressure across various hand postures and hand-object interaction scenarios using our collection system. Our framework enables precise hand pressure estimation in complex and natural interaction scenarios. Our approach substantially mitigates the limitations of traditional sEMG-based or vision-based methods by integrating 3D hand posture information with sEMG signals. Video demos, data, and code are available online.
Authors: Lianghua Huang, Wei Wang, Zhi-Fan Wu, Yupeng Shi, Huanzhang Dou, Chen Liang, Yutong Feng, Yu Liu, Jingren Zhou
Abstract: Recent research arXiv:2410.15027 has explored the use of diffusion transformers (DiTs) for task-agnostic image generation by simply concatenating attention tokens across images. However, despite substantial computational resources, the fidelity of the generated images remains suboptimal. In this study, we reevaluate and streamline this framework by hypothesizing that text-to-image DiTs inherently possess in-context generation capabilities, requiring only minimal tuning to activate them. Through diverse task experiments, we qualitatively demonstrate that existing text-to-image DiTs can effectively perform in-context generation without any tuning. Building on this insight, we propose a remarkably simple pipeline to leverage the in-context abilities of DiTs: (1) concatenate images instead of tokens, (2) perform joint captioning of multiple images, and (3) apply task-specific LoRA tuning using small datasets (e.g., $20\sim 100$ samples) instead of full-parameter tuning with large datasets. We name our models In-Context LoRA (IC-LoRA). This approach requires no modifications to the original DiT models, only changes to the training data. Remarkably, our pipeline generates high-fidelity image sets that better adhere to prompts. While task-specific in terms of tuning data, our framework remains task-agnostic in architecture and pipeline, offering a powerful tool for the community and providing valuable insights for further research on product-level task-agnostic generation systems. We release our code, data, and models at https://github.com/ali-vilab/In-Context-LoRA
Authors: Faisal Mehmood, Enqing Chen, Touqeer Abbas, Samah M. Alzanin
Abstract: Human Action Recognition (HAR) is an interesting research area in human-computer interaction used to monitor the activities of elderly and disabled individuals affected by physical and mental health. In the recent era, skeleton-based HAR has received much attention because skeleton data has shown that it can handle changes in striking, body size, camera views, and complex backgrounds. One key characteristic of ST-GCN is automatically learning spatial and temporal patterns from skeleton sequences. It has some limitations, as this method only works for short-range correlation due to its limited receptive field. Consequently, understanding human action requires long-range interconnection. To address this issue, we developed a spatial-temporal relative transformer ST-RTR model. The ST-RTR includes joint and relay nodes, which allow efficient communication and data transmission within the network. These nodes help to break the inherent spatial and temporal skeleton topologies, which enables the model to understand long-range human action better. Furthermore, we combine ST-RTR with a fusion model for further performance improvements. To assess the performance of the ST-RTR method, we conducted experiments on three skeleton-based HAR benchmarks: NTU RGB+D 60, NTU RGB+D 120, and UAV-Human. It boosted CS and CV by 2.11 % and 1.45% on NTU RGB+D 60, 1.25% and 1.05% on NTU RGB+D 120. On UAV-Human datasets, accuracy improved by 2.54%. The experimental outcomes explain that the proposed ST-RTR model significantly improves action recognition associated with the standard ST-GCN method.
Authors: Tahar Chettaoui, Naser Damer, Fadi Boutros
Abstract: Foundation models are predominantly trained in an unsupervised or self-supervised manner on highly diverse and large-scale datasets, making them broadly applicable to various downstream tasks. In this work, we investigate for the first time whether such models are suitable for the specific domain of face recognition. We further propose and demonstrate the adaptation of these models for face recognition across different levels of data availability. Extensive experiments are conducted on multiple foundation models and datasets of varying scales for training and fine-tuning, with evaluation on a wide range of benchmarks. Our results indicate that, despite their versatility, pre-trained foundation models underperform in face recognition compared to similar architectures trained specifically for this task. However, fine-tuning foundation models yields promising results, often surpassing models trained from scratch when training data is limited. Even with access to large-scale face recognition training datasets, fine-tuned foundation models perform comparably to models trained from scratch, but with lower training computational costs and without relying on the assumption of extensive data availability. Our analysis also explores bias in face recognition, with slightly higher bias observed in some settings when using foundation models.
Authors: Kai Ye, Chong Gao, Guanbin Li, Wenzheng Chen, Baoquan Chen
Abstract: We consider the problem of physically-based inverse rendering using 3D Gaussian Splatting (3DGS) representations. While recent 3DGS methods have achieved remarkable results in novel view synthesis (NVS), accurately capturing high-fidelity geometry, physically interpretable materials and lighting remains challenging, as it requires precise geometry modeling to provide accurate surface normals, along with physically-based rendering (PBR) techniques to ensure correct material and lighting disentanglement. Previous 3DGS methods resort to approximating surface normals, but often struggle with noisy local geometry, leading to inaccurate normal estimation and suboptimal material-lighting decomposition. In this paper, we introduce GeoSplatting, a novel hybrid representation that augments 3DGS with explicit geometric guidance and differentiable PBR equations. Specifically, we bridge isosurface and 3DGS together, where we first extract isosurface mesh from a scalar field, then convert it into 3DGS points and formulate PBR equations for them in a fully differentiable manner. In GeoSplatting, 3DGS is grounded on the mesh geometry, enabling precise surface normal modeling, which facilitates the use of PBR frameworks for material decomposition. This approach further maintains the efficiency and quality of NVS from 3DGS while ensuring accurate geometry from the isosurface. Comprehensive evaluations across diverse datasets demonstrate the superiority of GeoSplatting, consistently outperforming existing methods both quantitatively and qualitatively.
Authors: Tuan Duc Ngo, Peiye Zhuang, Chuang Gan, Evangelos Kalogerakis, Sergey Tulyakov, Hsin-Ying Lee, Chaoyang Wang
Abstract: Tracking dense 3D motion from monocular videos remains challenging, particularly when aiming for pixel-level precision over long sequences. We introduce DELTA, a novel method that efficiently tracks every pixel in 3D space, enabling accurate motion estimation across entire videos. Our approach leverages a joint global-local attention mechanism for reduced-resolution tracking, followed by a transformer-based upsampler to achieve high-resolution predictions. Unlike existing methods, which are limited by computational inefficiency or sparse tracking, DELTA delivers dense 3D tracking at scale, running over 8x faster than previous methods while achieving state-of-the-art accuracy. Furthermore, we explore the impact of depth representation on tracking performance and identify log-depth as the optimal choice. Extensive experiments demonstrate the superiority of DELTA on multiple benchmarks, achieving new state-of-the-art results in both 2D and 3D dense tracking tasks. Our method provides a robust solution for applications requiring fine-grained, long-term motion tracking in 3D space.
Authors: Junyu Chen, Yihao Liu, Shuwen Wei, Zhangxing Bian, Shalini Subramanian, Aaron Carass, Jerry L. Prince, Yong Du
Abstract: Deep learning technologies have dramatically reshaped the field of medical image registration over the past decade. The initial developments, such as regression-based and U-Net-based networks, established the foundation for deep learning in image registration. Subsequent progress has been made in various aspects of deep learning-based registration, including similarity measures, deformation regularizations, network architectures, and uncertainty estimation. These advancements have not only enriched the field of image registration but have also facilitated its application in a wide range of tasks, including atlas construction, multi-atlas segmentation, motion estimation, and 2D-3D registration. In this paper, we present a comprehensive overview of the most recent advancements in deep learning-based image registration. We begin with a concise introduction to the core concepts of deep learning-based image registration. Then, we delve into innovative network architectures, loss functions specific to registration, and methods for estimating registration uncertainty. Additionally, this paper explores appropriate evaluation metrics for assessing the performance of deep learning models in registration tasks. Finally, we highlight the practical applications of these novel techniques in medical imaging and discuss the future prospects of deep learning-based image registration.
Authors: Joey Wilson, Yuewei Fu, Joshua Friesen, Parker Ewen, Andrew Capodieci, Paramsothy Jayakumar, Kira Barton, Maani Ghaffari
Abstract: In this paper, we develop a modular neural network for real-time {\color{black}(> 10 Hz)} semantic mapping in uncertain environments, which explicitly updates per-voxel probabilistic distributions within a neural network layer. Our approach combines the reliability of classical probabilistic algorithms with the performance and efficiency of modern neural networks. Although robotic perception is often divided between modern differentiable methods and classical explicit methods, a union of both is necessary for real-time and trustworthy performance. We introduce a novel Convolutional Bayesian Kernel Inference (ConvBKI) layer which incorporates semantic segmentation predictions online into a 3D map through a depthwise convolution layer by leveraging conjugate priors. We compare ConvBKI against state-of-the-art deep learning approaches and probabilistic algorithms for mapping to evaluate reliability and performance. We also create a Robot Operating System (ROS) package of ConvBKI and test it on real-world perceptually challenging off-road driving data.
Authors: Zhenbo Song, Zhenyuan Zhang, Kaihao Zhang, Zhaoxin Fan, Jianfeng Lu
Abstract: This study delves into the enhancement of Under-Display Camera (UDC) image restoration models, focusing on their robustness against adversarial attacks. Despite its innovative approach to seamless display integration, UDC technology faces unique image degradation challenges exacerbated by the susceptibility to adversarial perturbations. Our research initially conducts an in-depth robustness evaluation of deep-learning-based UDC image restoration models by employing several white-box and black-box attacking methods. This evaluation is pivotal in understanding the vulnerabilities of current UDC image restoration techniques. Following the assessment, we introduce a defense framework integrating adversarial purification with subsequent fine-tuning processes. First, our approach employs diffusion-based adversarial purification, effectively neutralizing adversarial perturbations. Then, we apply the fine-tuning methodologies to refine the image restoration models further, ensuring that the quality and fidelity of the restored images are maintained. The effectiveness of our proposed approach is validated through extensive experiments, showing marked improvements in resilience against typical adversarial attacks.
Authors: Ang Bian, Wei Li, Hangjie Yuan, Chengrong Yu, Mang Wang, Zixiang Zhao, Aojun Lu, Pengliang Ji, Tao Feng
Abstract: Model generalization ability upon incrementally acquiring dynamically updating knowledge from sequentially arriving tasks is crucial to tackle the sensitivity-stability dilemma in Continual Learning (CL). Weight loss landscape sharpness minimization seeking for flat minima lying in neighborhoods with uniform low loss or smooth gradient is proven to be a strong training regime improving model generalization compared with loss minimization based optimizer like SGD. Yet only a few works have discussed this training regime for CL, proving that dedicated designed zeroth-order sharpness optimizer can improve CL performance. In this work, we propose a Continual Flatness (C-Flat) method featuring a flatter loss landscape tailored for CL. C-Flat could be easily called with only one line of code and is plug-and-play to any CL methods. A general framework of C-Flat applied to all CL categories and a thorough comparison with loss minima optimizer and flat minima based CL approaches is presented in this paper, showing that our method can boost CL performance in almost all cases. Code is available at https://github.com/WanNaa/C-Flat.
Authors: Wei Wu, Xiaoxin Feng, Ziyan Gao, Yuheng Kan
Abstract: Data-driven autonomous driving motion generation tasks are frequently impacted by the limitations of dataset size and the domain gap between datasets, which precludes their extensive application in real-world scenarios. To address this issue, we introduce SMART, a novel autonomous driving motion generation paradigm that models vectorized map and agent trajectory data into discrete sequence tokens. These tokens are then processed through a decoder-only transformer architecture to train for the next token prediction task across spatial-temporal series. This GPT-style method allows the model to learn the motion distribution in real driving scenarios. SMART achieves state-of-the-art performance across most of the metrics on the generative Sim Agents challenge, ranking 1st on the leaderboards of Waymo Open Motion Dataset (WOMD), demonstrating remarkable inference speed. Moreover, SMART represents the generative model in the autonomous driving motion domain, exhibiting zero-shot generalization capabilities: Using only the NuPlan dataset for training and WOMD for validation, SMART achieved a competitive score of 0.72 on the Sim Agents challenge. Lastly, we have collected over 1 billion motion tokens from multiple datasets, validating the model's scalability. These results suggest that SMART has initially emulated two important properties: scalability and zero-shot generalization, and preliminarily meets the needs of large-scale real-time simulation applications. We have released all the code to promote the exploration of models for motion generation in the autonomous driving field. The source code is available at https://github.com/rainmaker22/SMART.
Authors: ZeMing Gong, Austin T. Wang, Xiaoliang Huo, Joakim Bruslund Haurum, Scott C. Lowe, Graham W. Taylor, Angel X. Chang
Abstract: Measuring biodiversity is crucial for understanding ecosystem health. While prior works have developed machine learning models for taxonomic classification of photographic images and DNA separately, in this work, we introduce a multimodal approach combining both, using CLIP-style contrastive learning to align images, barcode DNA, and text-based representations of taxonomic labels in a unified embedding space. This allows for accurate classification of both known and unknown insect species without task-specific fine-tuning, leveraging contrastive learning for the first time to fuse DNA and image data. Our method surpasses previous single-modality approaches in accuracy by over 8% on zero-shot learning tasks, showcasing its effectiveness in biodiversity studies.
Authors: Hengyi Wang, Shiwei Tan, Hao Wang
Abstract: Vision transformers (ViTs) have emerged as a significant area of focus, particularly for their capacity to be jointly trained with large language models and to serve as robust vision foundation models. Yet, the development of trustworthy explanation methods for ViTs has lagged, particularly in the context of post-hoc interpretations of ViT predictions. Existing sub-image selection approaches, such as feature-attribution and conceptual models, fall short in this regard. This paper proposes five desiderata for explaining ViTs -- faithfulness, stability, sparsity, multi-level structure, and parsimony -- and demonstrates the inadequacy of current methods in meeting these criteria comprehensively. We introduce a variational Bayesian explanation framework, dubbed ProbAbilistic Concept Explainers (PACE), which models the distributions of patch embeddings to provide trustworthy post-hoc conceptual explanations. Our qualitative analysis reveals the distributions of patch-level concepts, elucidating the effectiveness of ViTs by modeling the joint distribution of patch embeddings and ViT's predictions. Moreover, these patch-level explanations bridge the gap between image-level and dataset-level explanations, thus completing the multi-level structure of PACE. Through extensive experiments on both synthetic and real-world datasets, we demonstrate that PACE surpasses state-of-the-art methods in terms of the defined desiderata.
Authors: Stefan K. Nielsen, Laziz U. Abdullaev, Rachel S. Y. Teo, Tan M. Nguyen
Abstract: Pairwise dot-product self-attention is key to the success of transformers that achieve state-of-the-art performance across a variety of applications in language and vision. This dot-product self-attention computes attention weights among the input tokens using Euclidean distance, which makes the model prone to representation collapse and vulnerable to contaminated samples. In this paper, we propose using a Mahalanobis distance metric for computing the attention weights to stretch the underlying feature space in directions of high contextual relevance. In particular, we define a hyper-ellipsoidal neighborhood around each query to increase the attention weights of the tokens lying in the contextually important directions. We term this novel class of attention Elliptical Attention. Our Elliptical Attention provides two benefits: 1) reducing representation collapse and 2) enhancing the model's robustness as Elliptical Attention pays more attention to contextually relevant information rather than focusing on some small subset of informative features. We empirically demonstrate the advantages of Elliptical Attention over the baseline dot-product attention and state-of-the-art attention methods on various practical tasks, including object classification, image segmentation, and language modeling across different data modalities.
Authors: Jiahe Huang, Guandao Yang, Zichen Wang, Jeong Joon Park
Abstract: We introduce a general framework for solving partial differential equations (PDEs) using generative diffusion models. In particular, we focus on the scenarios where we do not have the full knowledge of the scene necessary to apply classical solvers. Most existing forward or inverse PDE approaches perform poorly when the observations on the data or the underlying coefficients are incomplete, which is a common assumption for real-world measurements. In this work, we propose DiffusionPDE that can simultaneously fill in the missing information and solve a PDE by modeling the joint distribution of the solution and coefficient spaces. We show that the learned generative priors lead to a versatile framework for accurately solving a wide range of PDEs under partial observation, significantly outperforming the state-of-the-art methods for both forward and inverse directions.
Authors: Jonas Ngnaw\'e, Sabyasachi Sahoo, Yann Pequignot, Fr\'ed\'eric Precioso, Christian Gagn\'e
Abstract: Despite extensive research on adversarial training strategies to improve robustness, the decisions of even the most robust deep learning models can still be quite sensitive to imperceptible perturbations, creating serious risks when deploying them for high-stakes real-world applications. While detecting such cases may be critical, evaluating a model's vulnerability at a per-instance level using adversarial attacks is computationally too intensive and unsuitable for real-time deployment scenarios. The input space margin is the exact score to detect non-robust samples and is intractable for deep neural networks. This paper introduces the concept of margin consistency -- a property that links the input space margins and the logit margins in robust models -- for efficient detection of vulnerable samples. First, we establish that margin consistency is a necessary and sufficient condition to use a model's logit margin as a score for identifying non-robust samples. Next, through comprehensive empirical analysis of various robustly trained models on CIFAR10 and CIFAR100 datasets, we show that they indicate high margin consistency with a strong correlation between their input space margins and the logit margins. Then, we show that we can effectively and confidently use the logit margin to detect brittle decisions with such models. Finally, we address cases where the model is not sufficiently margin-consistent by learning a pseudo-margin from the feature representation. Our findings highlight the potential of leveraging deep representations to assess adversarial vulnerability in deployment scenarios efficiently.
Authors: Shantanu Ghosh, Rayan Syed, Chenyu Wang, Clare B. Poynton, Shyam Visweswaran, Kayhan Batmanghelich
Abstract: Error slice discovery associates structured patterns with model errors. Existing methods discover error slices by clustering the error-prone samples with similar patterns or assigning discrete attributes to each sample for post-hoc analysis. While these methods aim for interpretability and easier mitigation through reweighting or rebalancing, they may not capture the full complexity of error patterns due to incomplete or missing attributes. Contrary to the existing approach, this paper utilizes the reasoning capabilities of the Large Language Model (LLM) to analyze complex error patterns and generate testable hypotheses. This paper proposes LADDER: Language Driven slice Discovery and Error Rectification. It first projects the model's representation into a language-aligned feature space (eg CLIP) to preserve semantics in the original model feature space. This ensures the accurate retrieval of sentences that highlight the model's errors. Next, the LLM utilizes the sentences and generates hypotheses to discover error slices. Finally, we mitigate the error by fine-tuning the classification head by creating a group-balanced dataset using the hypotheses. Our entire method does not require any attribute annotation, either explicitly or through external tagging models. We validate our method with \textbf{five} image classification datasets. The code is available (https://github.com/batmanlab/Ladder).
Authors: Wei-Cheng Tseng, Yi-Jen Shih, David Harwath, Raymond Mooney
Abstract: Audio-visual pre-trained models have gained substantial attention recently and demonstrated superior performance on various audio-visual tasks. This study investigates whether pre-trained audio-visual models demonstrate non-arbitrary associations between sounds and visual representations$\unicode{x2013}$known as sound symbolism$\unicode{x2013}$which is also observed in humans. We developed a specialized dataset with synthesized images and audio samples and assessed these models using a non-parametric approach in a zero-shot setting. Our findings reveal a significant correlation between the models' outputs and established patterns of sound symbolism, particularly in models trained on speech data. These results suggest that such models can capture sound-meaning connections akin to human language processing, providing insights into both cognitive architectures and machine learning strategies.
Authors: Hong-fu Chou, Vu Nguyen Ha, Prabhu Thiruvasagam, Thanh-Dung Le, Geoffrey Eappen, Ti Ti Nguyen, Luis M. Garces-Socarras, Jorge L. Gonzalez-Rios, Juan Carlos Merlano-Duncan, Symeon Chatzinotas
Abstract: Earth Observation (EO) systems are crucial for cartography, disaster surveillance, and resource administration. Nonetheless, they encounter considerable obstacles in the processing and transmission of extensive data, especially in specialized domains such as precision agriculture and real-time disaster response. Earth observation satellites, outfitted with remote sensing technology, gather data from onboard sensors and IoT-enabled terrestrial objects, delivering important information remotely. Domain-adapted Large Language Models (LLMs) provide a solution by enabling the integration of raw and processed EO data. Through domain adaptation, LLMs improve the assimilation and analysis of many data sources, tackling the intricacies of specialized datasets in agriculture and disaster response. This data synthesis, directed by LLMs, enhances the precision and pertinence of conveyed information. This study provides a thorough examination of using semantic inference and deep learning for sophisticated EO systems. It presents an innovative architecture for semantic communication in EO satellite networks, designed to improve data transmission efficiency using semantic processing methodologies. Recent advancements in onboard processing technologies enable dependable, adaptable, and energy-efficient data management in orbit. These improvements guarantee reliable performance in adverse space circumstances using radiation-hardened and reconfigurable technology. Collectively, these advancements enable next-generation satellite missions with improved processing capabilities, crucial for operational flexibility and real-time decision-making in 6G satellite communication.
Authors: Jose Vargas Quiros, Bart Liefers, Karin van Garderen, Jeroen Vermeulen, Eyened Reading Center, Sinergia Consortium, Caroline Klaver
Abstract: We introduce VascX models, a comprehensive set of model ensembles for analyzing retinal vasculature from color fundus images (CFIs). Annotated CFIs were aggregated from public datasets . Additional CFIs, mainly from the population-based Rotterdam Study were annotated by graders for arteries and veins at pixel level, resulting in a dataset diverse in patient demographics and imaging conditions. VascX models demonstrated superior segmentation performance across datasets, image quality levels, and anatomic regions when compared to existing, publicly available models, likely due to the increased size and variety of our training set. Important improvements were observed in artery-vein and disc segmentation performance, particularly in segmentations of these structures on CFIs of intermediate quality, common in large cohorts and clinical datasets. Importantly, these improvements translated into significantly more accurate vascular features when we compared features extracted from VascX segmentation masks with features extracted from segmentation masks generated by previous models. With VascX models we provide a robust, ready-to-use set of model ensembles and inference code aimed at simplifying the implementation and enhancing the quality of automated retinal vasculature analyses. The precise vessel parameters generated by the model can serve as starting points for the identification of disease patterns in and outside of the eye.
Authors: Maria Makarova, Daria Trinitatova, Qian Liu, Dzmitry Tsetserukou
Abstract: Many modern robotic systems operate autonomously, however they often lack the ability to accurately analyze the environment and adapt to changing external conditions, while teleoperation systems often require special operator skills. In the field of laboratory automation, the number of automated processes is growing, however such systems are usually developed to perform specific tasks. In addition, many of the objects used in this field are transparent, making it difficult to analyze them using visual channels. The contributions of this work include the development of a robotic framework with autonomous mode for manipulating liquid-filled objects with different degrees of transparency in complex pose combinations. The conducted experiments demonstrated the robustness of the designed visual perception system to accurately estimate object poses for autonomous manipulation, and confirmed the performance of the algorithms in dexterous operations such as liquid dispensing. The proposed robotic framework can be applied for laboratory automation, since it allows solving the problem of performing non-trivial manipulation tasks with the analysis of object poses of varying degrees of transparency and liquid levels, requiring high accuracy and repeatability.
Authors: Yule Wang, Chengrui Li, Weihan Li, Anqi Wu
Abstract: Understanding the neural basis of behavior is a fundamental goal in neuroscience. Current research in large-scale neuro-behavioral data analysis often relies on decoding models, which quantify behavioral information in neural data but lack details on behavior encoding. This raises an intriguing scientific question: ``how can we enable in-depth exploration of neural representations in behavioral tasks, revealing interpretable neural dynamics associated with behaviors''. However, addressing this issue is challenging due to the varied behavioral encoding across different brain regions and mixed selectivity at the population level. To tackle this limitation, our approach, named ``BeNeDiff'', first identifies a fine-grained and disentangled neural subspace using a behavior-informed latent variable model. It then employs state-of-the-art generative diffusion models to synthesize behavior videos that interpret the neural dynamics of each latent factor. We validate the method on multi-session datasets containing widefield calcium imaging recordings across the dorsal cortex. Through guiding the diffusion model to activate individual latent factors, we verify that the neural dynamics of latent factors in the disentangled neural subspace provide interpretable quantifications of the behaviors of interest. At the same time, the neural subspace in BeNeDiff demonstrates high disentanglement and neural reconstruction quality.
Authors: Yongkang Cheng, Mingjiang Liang, Shaoli Huang, Gaoge Han, Jifeng Ning, Wei Liu
Abstract: Audio-driven simultaneous gesture generation is vital for human-computer communication, AI games, and film production. While previous research has shown promise, there are still limitations. Methods based on VAEs are accompanied by issues of local jitter and global instability, whereas methods based on diffusion models are hampered by low generation efficiency. This is because the denoising process of DDPM in the latter relies on the assumption that the noise added at each step is sampled from a unimodal distribution, and the noise values are small. DDIM borrows the idea from the Euler method for solving differential equations, disrupts the Markov chain process, and increases the noise step size to reduce the number of denoising steps, thereby accelerating generation. However, simply increasing the step size during the step-by-step denoising process causes the results to gradually deviate from the original data distribution, leading to a significant drop in the quality of the generated actions and the emergence of unnatural artifacts. In this paper, we break the assumptions of DDPM and achieves breakthrough progress in denoising speed and fidelity. Specifically, we introduce a conditional GAN to capture audio control signals and implicitly match the multimodal denoising distribution between the diffusion and denoising steps within the same sampling step, aiming to sample larger noise values and apply fewer denoising steps for high-speed generation.
Authors: Hongyi Pan, Gorkem Durak, Zheyuan Zhang, Yavuz Taktak, Elif Keles, Halil Ertugrul Aktas, Alpay Medetalibeyoglu, Yury Velichko, Concetto Spampinato, Ivo Schoots, Marco J. Bruno, Rajesh N. Keswani, Pallavi Tiwari, Candice Bolan, Tamas Gonda, Michael G. Goggins, Michael B. Wallace, Ziyue Xu, Ulas Bagci
Abstract: Federated learning (FL) enables collaborative model training across institutions without sharing sensitive data, making it an attractive solution for medical imaging tasks. However, traditional FL methods, such as Federated Averaging (FedAvg), face difficulties in generalizing across domains due to variations in imaging protocols and patient demographics across institutions. This challenge is particularly evident in pancreas MRI segmentation, where anatomical variability and imaging artifacts significantly impact performance. In this paper, we conduct a comprehensive evaluation of FL algorithms for pancreas MRI segmentation and introduce a novel approach that incorporates adaptive aggregation weights. By dynamically adjusting the contribution of each client during model aggregation, our method accounts for domain-specific differences and improves generalization across heterogeneous datasets. Experimental results demonstrate that our approach enhances segmentation accuracy and reduces the impact of domain shift compared to conventional FL methods while maintaining privacy-preserving capabilities. Significant performance improvements are observed across multiple hospitals (centers).