Authors: Arwin Gansekoele, Emiel Hess, Sandjai Bhulai
Abstract: The growing privacy concerns surrounding face image data demand new techniques that can guarantee user privacy. One such face recognition technique that claims to achieve better user privacy is Federated Face Recognition (FRR), a subfield of Federated Learning (FL). However, FFR faces challenges due to the heterogeneity of the data, given the large number of classes that need to be handled. To overcome this problem, solutions are sought in the field of personalized FL. This work introduces three new data partitions based on the CelebA dataset, each with a different form of data heterogeneity. It also proposes Hessian-Free Model Agnostic Meta-Learning (HF-MAML) in an FFR setting. We show that HF-MAML scores higher in verification tests than current FFR models on three different CelebA data partitions. In particular, the verification scores improve the most in heterogeneous data partitions. To balance personalization with the development of an effective global model, an embedding regularization term is introduced for the loss function. This term can be combined with HF-MAML and is shown to increase global model verification performance. Lastly, this work performs a fairness analysis, showing that HF-MAML and its embedding regularization extension can improve fairness by reducing the standard deviation over the client evaluation scores.
Authors: Ziyi Zhang, Nicolas Roussel, Wenzel Jakob
Abstract: Discontinuous visibility changes remain a major bottleneck when optimizing surfaces within a physically-based inverse renderer. Many previous works have proposed sophisticated algorithms and data structures to sample visibility silhouettes more efficiently. Our work presents another solution: instead of differentiating a tentative surface locally, we differentiate a volumetric perturbation of a surface. We refer this as a many-worlds representation because it models a non-interacting superposition of conflicting explanations (worlds) of the input dataset. Each world is optically isolated from others, leading to a new transport law that distinguishes our method from prior work based on exponential random media. The resulting Monte Carlo algorithm is simpler and more efficient than prior methods. We demonstrate that our method promotes rapid convergence, both in terms of the total iteration count and the cost per iteration.
Authors: Hengyi Wang, Lourdes Agapito
Abstract: We present Spann3R, a novel approach for dense 3D reconstruction from ordered or unordered image collections. Built on the DUSt3R paradigm, Spann3R uses a transformer-based architecture to directly regress pointmaps from images without any prior knowledge of the scene or camera parameters. Unlike DUSt3R, which predicts per image-pair pointmaps each expressed in its local coordinate frame, Spann3R can predict per-image pointmaps expressed in a global coordinate system, thus eliminating the need for optimization-based global alignment. The key idea of Spann3R is to manage an external spatial memory that learns to keep track of all previous relevant 3D information. Spann3R then queries this spatial memory to predict the 3D structure of the next frame in a global coordinate system. Taking advantage of DUSt3R's pre-trained weights, and further fine-tuning on a subset of datasets, Spann3R shows competitive performance and generalization ability on various unseen datasets and can process ordered image collections in real time. Project page: \url{https://hengyiwang.github.io/projects/spanner}
Authors: Osama Mustafa, Muhammad Khizer Ali, Momina Moetesum, Imran Siddiqi
Abstract: The widespread use of charts and infographics as a means of data visualization in various domains has inspired recent research in automated chart understanding. However, information extraction from chart images is a complex multitasked process due to style variations and, as a consequence, it is challenging to design an end-to-end system. In this study, we propose a deep learning-based framework that provides a solution for key steps in the chart information extraction pipeline. The proposed framework utilizes hierarchal vision transformers for the tasks of chart-type and text-role classification, while YOLOv7 for text detection. The detected text is then enhanced using Super Resolution Generative Adversarial Networks to improve the recognition output of the OCR. Experimental results on a benchmark dataset show that our proposed framework achieves excellent performance at every stage with F1-scores of 0.97 for chart-type classification, 0.91 for text-role classification, and a mean Average Precision of 0.95 for text detection.
Authors: Dilermando Queiroz, Andr\'e Anjos, Lilian Berton
Abstract: Ensuring consistent performance across diverse populations and incorporating fairness into machine learning models are crucial for advancing medical image diagnostics and promoting equitable healthcare. However, many databases do not provide protected attributes or contain unbalanced representations of demographic groups, complicating the evaluation of model performance across different demographics and the application of bias mitigation techniques that rely on these attributes. This study aims to investigate the effectiveness of using the backbone of Foundation Models as an embedding extractor for creating groups that represent protected attributes, such as gender and age. We propose utilizing these groups in different stages of bias mitigation, including pre-processing, in-processing, and evaluation. Using databases in and out-of-distribution scenarios, it is possible to identify that the method can create groups that represent gender in both databases and reduce in 4.44% the difference between the gender attribute in-distribution and 6.16% in out-of-distribution. However, the model lacks robustness in handling age attributes, underscoring the need for more fundamentally fair and robust Foundation models. These findings suggest a role in promoting fairness assessment in scenarios where we lack knowledge of attributes, contributing to the development of more equitable medical diagnostics.
Authors: Dilermando Queiroz, Anderson Carlos, Ma\'ira Fatoretto, Andr\'e Anjos, Lilian Berton, Luis Filipe Nakayama
Abstract: Foundation models have emerged as robust models with label efficiency in diverse domains. In medical imaging, these models contribute to the advancement of medical diagnoses due to the difficulty in obtaining labeled data. However, it is unclear whether using a large amount of unlabeled data, biased by the presence of sensitive attributes during pre-training, influences the fairness of the model. This research examines the bias in the Foundation model (RetFound) when it is applied to fine-tune the Brazilian Multilabel Ophthalmological Dataset (BRSET), which has a different population than the pre-training dataset. The model evaluation, in comparison with supervised learning, shows that the Foundation Model has the potential to reduce the gap between the maximum AUC and minimum AUC evaluations across gender and age groups. However, in a data-efficient generalization, the model increases the bias when the data amount decreases. These findings suggest that when deploying a Foundation Model in real-life scenarios with limited data, the possibility of fairness issues should be considered.
Authors: M. Maruf, Arka Daw, Kazi Sajeed Mehrab, Harish Babu Manogaran, Abhilash Neog, Medha Sawhney, Mridul Khurana, James P. Balhoff, Yasin Bakis, Bahadir Altintas, Matthew J. Thompson, Elizabeth G. Campolongo, Josef C. Uyeda, Hilmar Lapp, Henry L. Bart, Paula M. Mabee, Yu Su, Wei-Lun Chao, Charles Stewart, Tanya Berger-Wolf, Wasila Dahdul, Anuj Karpatne
Abstract: Images are increasingly becoming the currency for documenting biodiversity on the planet, providing novel opportunities for accelerating scientific discoveries in the field of organismal biology, especially with the advent of large vision-language models (VLMs). We ask if pre-trained VLMs can aid scientists in answering a range of biologically relevant questions without any additional fine-tuning. In this paper, we evaluate the effectiveness of 12 state-of-the-art (SOTA) VLMs in the field of organismal biology using a novel dataset, VLM4Bio, consisting of 469K question-answer pairs involving 30K images from three groups of organisms: fishes, birds, and butterflies, covering five biologically relevant tasks. We also explore the effects of applying prompting techniques and tests for reasoning hallucination on the performance of VLMs, shedding new light on the capabilities of current SOTA VLMs in answering biologically relevant questions using images. The code and datasets for running all the analyses reported in this paper can be found at https://github.com/sammarfy/VLM4Bio.
Authors: Tanner D. Harms, Steven L. Brunton, Beverley J. McKeon
Abstract: Interpreting motion captured in image sequences is crucial for a wide range of computer vision applications. Typical estimation approaches include optical flow (OF), which approximates the apparent motion instantaneously in a scene, and multiple object tracking (MOT), which tracks the motion of subjects over time. Often, the motion of objects in a scene is governed by some underlying dynamical system which could be inferred by analyzing the motion of groups of objects. Standard motion analyses, however, are not designed to intuit flow dynamics from trajectory data, making such measurements difficult in practice. The goal of this work is to extend gradient-based dynamical systems analyses to real-world applications characterized by complex, feature-rich image sequences with imperfect tracers. The tracer trajectories are tracked using deep vision networks and gradients are approximated using Lagrangian gradient regression (LGR), a tool designed to estimate spatial gradients from sparse data. From gradients, dynamical features such as regions of coherent rotation and transport barriers are identified. The proposed approach is affordably implemented and enables advanced studies including the motion analysis of two distinct object classes in a single image sequence. Two examples of the method are presented on data sets for which standard gradient-based analyses do not apply.
Authors: Zeyi Bo (Harbin Institute of Technology), Wuxi Sun (Harbin Institute of Technology), Ye Jin (Harbin Institute of Technology)
Abstract: In recent years, the parameters of backbones of Video Understanding tasks continue to increase and even reach billion-level. Whether fine-tuning a specific task on the Video Foundation Model or pre-training the model designed for the specific task, incurs a lot of overhead. How to make these models play other values than their own tasks becomes a worthy question. Multi-Task Learning(MTL) makes the visual task acquire the rich shareable knowledge from other tasks while joint training. It is fully explored in Image Recognition tasks especially dense predict tasks. Nevertheless, it is rarely used in video domain due to the lack of multi-labels video data. In this paper, a heterogenous data video multi-task prompt learning (VMTL) method is proposed to address above problem. It's different from it in image domain, a Double-Layers Mapper(DLM) is proposed to extract the shareable knowledge into visual promptS and align it with representation of primary task. Extensive experiments prove that our DLM-VMTL performs better than baselines on 6 different video understanding tasks and 11 datasets.
Authors: Zichen Yu, Quanli Liu, Wei Wang, Liyong Zhang, Xiaoguang Zhao
Abstract: Recently, LSS-based multi-view 3D object detection provides an economical and deployment-friendly solution for autonomous driving. However, all the existing LSS-based methods transform multi-view image features into a Cartesian Bird's-Eye-View(BEV) representation, which does not take into account the non-uniform image information distribution and hardly exploits the view symmetry. In this paper, in order to adapt the image information distribution and preserve the view symmetry by regular convolution, we propose to employ the polar BEV representation to substitute the Cartesian BEV representation. To achieve this, we elaborately tailor three modules: a polar view transformer to generate the polar BEV representation, a polar temporal fusion module for fusing historical polar BEV features and a polar detection head to predict the polar-parameterized representation of the object. In addition, we design a 2D auxiliary detection head and a spatial attention enhancement module to improve the quality of feature extraction in perspective view and BEV, respectively. Finally, we integrate the above improvements into a novel multi-view 3D object detector, PolarBEVDet. Experiments on nuScenes show that PolarBEVDet achieves the superior performance. The code is available at https://github.com/Yzichen/PolarBEVDet.git.
Authors: Jiayu Liu, Shancong Mou, Nathan Gaw, Yinan Wang
Abstract: Anomaly detection is a long-standing challenge in manufacturing systems. Traditionally, anomaly detection has relied on human inspectors. However, 3D point clouds have gained attention due to their robustness to environmental factors and their ability to represent geometric data. Existing 3D anomaly detection methods generally fall into two categories. One compares scanned 3D point clouds with design files, assuming these files are always available. However, such assumptions are often violated in many real-world applications where model-free products exist, such as fresh produce (i.e., ``Cookie", ``Potato", etc.), dentures, bone, etc. The other category compares patches of scanned 3D point clouds with a library of normal patches named memory bank. However, those methods usually fail to detect incomplete shapes, which is a fairly common defect type (i.e., missing pieces of different products). The main challenge is that missing areas in 3D point clouds represent the absence of scanned points. This makes it infeasible to compare the missing region with existing point cloud patches in the memory bank. To address these two challenges, we proposed a unified, unsupervised 3D anomaly detection framework capable of identifying all types of defects on model-free products. Our method integrates two detection modules: a feature-based detection module and a reconstruction-based detection module. Feature-based detection covers geometric defects, such as dents, holes, and cracks, while the reconstruction-based method detects missing regions. Additionally, we employ a One-class Support Vector Machine (OCSVM) to fuse the detection results from both modules. The results demonstrate that (1) our proposed method outperforms the state-of-the-art methods in identifying incomplete shapes and (2) it still maintains comparable performance with the SOTA methods in detecting all other types of anomalies.
Authors: Jonggwon Park, Soobum Kim, Byungmu Yoon, Jihun Hyun, Kyoyun Choi
Abstract: The rapid evolution of artificial intelligence, especially in large language models (LLMs), has significantly impacted various domains, including healthcare. In chest X-ray (CXR) analysis, previous studies have employed LLMs, but with limitations: either underutilizing the multi-tasking capabilities of LLMs or lacking clinical accuracy. This paper presents M4CXR, a multi-modal LLM designed to enhance CXR interpretation. The model is trained on a visual instruction-following dataset that integrates various task-specific datasets in a conversational format. As a result, the model supports multiple tasks such as medical report generation (MRG), visual grounding, and visual question answering (VQA). M4CXR achieves state-of-the-art clinical accuracy in MRG by employing a chain-of-thought prompting strategy, in which it identifies findings in CXR images and subsequently generates corresponding reports. The model is adaptable to various MRG scenarios depending on the available inputs, such as single-image, multi-image, and multi-study contexts. In addition to MRG, M4CXR performs visual grounding at a level comparable to specialized models and also demonstrates outstanding performance in VQA. Both quantitative and qualitative assessments reveal M4CXR's versatility in MRG, visual grounding, and VQA, while consistently maintaining clinical accuracy.
Authors: Minghang Zheng, Xinhao Cai, Qingchao Chen, Yuxin Peng, Yang Liu
Abstract: Video temporal grounding aims to identify video segments within untrimmed videos that are most relevant to a given natural language query. Existing video temporal localization models rely on specific datasets for training and have high data collection costs, but they exhibit poor generalization capability under the across-dataset and out-of-distribution (OOD) settings. In this paper, we propose a Training-Free Video Temporal Grounding (TFVTG) approach that leverages the ability of pre-trained large models. A naive baseline is to enumerate proposals in the video and use the pre-trained visual language models (VLMs) to select the best proposal according to the vision-language alignment. However, most existing VLMs are trained on image-text pairs or trimmed video clip-text pairs, making it struggle to (1) grasp the relationship and distinguish the temporal boundaries of multiple events within the same video; (2) comprehend and be sensitive to the dynamic transition of events (the transition from one event to another) in the video. To address these issues, we propose leveraging large language models (LLMs) to analyze multiple sub-events contained in the query text and analyze the temporal order and relationships between these events. Secondly, we split a sub-event into dynamic transition and static status parts and propose the dynamic and static scoring functions using VLMs to better evaluate the relevance between the event and the description. Finally, for each sub-event description, we use VLMs to locate the top-k proposals and leverage the order and relationships between sub-events provided by LLMs to filter and integrate these proposals. Our method achieves the best performance on zero-shot video temporal grounding on Charades-STA and ActivityNet Captions datasets without any training and demonstrates better generalization capabilities in cross-dataset and OOD settings.
Authors: Jingyi Wang, Jianzhong Ju, Jian Luan, Zhidong Deng
Abstract: Recent advances in large vision-language models (VLMs) typically employ vision encoders based on the Vision Transformer (ViT) architecture. The division of the images into patches by ViT results in a fragmented perception, thereby hindering the visual understanding capabilities of VLMs. In this paper, we propose an innovative enhancement to address this limitation by introducing a Scene Graph Expression (SGE) module in VLMs. This module extracts and structurally expresses the complex semantic information within images, thereby improving the foundational perception and understanding abilities of VLMs. Extensive experiments demonstrate that integrating our SGE module significantly enhances the VLM's performance in vision-language tasks, indicating its effectiveness in preserving intricate semantic details and facilitating better visual understanding. Code and data would be available.
Authors: Zhijie Shen, Chunyu Lin, Lang Nie, Kang Liao
Abstract: Depth estimation from a monocular 360 image is important to the perception of the entire 3D environment. However, the inherent distortion and large field of view (FoV) in 360 images pose great challenges for this task. To this end, existing mainstream solutions typically introduce additional perspective-based 360 representations (\textit{e.g.}, Cubemap) to achieve effective feature extraction. Nevertheless, regardless of the introduced representations, they eventually need to be unified into the equirectangular projection (ERP) format for the subsequent depth estimation, which inevitably reintroduces the troublesome distortions. In this work, we propose an oriented distortion-aware Gabor Fusion framework (PGFuse) to address the above challenges. First, we introduce Gabor filters that analyze texture in the frequency domain, thereby extending the receptive fields and enhancing depth cues. To address the reintroduced distortions, we design a linear latitude-aware distortion representation method to generate customized, distortion-aware Gabor filters (PanoGabor filters). Furthermore, we design a channel-wise and spatial-wise unidirectional fusion module (CS-UFM) that integrates the proposed PanoGabor filters to unify other representations into the ERP format, delivering effective and distortion-free features. Considering the orientation sensitivity of the Gabor transform, we introduce a spherical gradient constraint to stabilize this sensitivity. Experimental results on three popular indoor 360 benchmarks demonstrate the superiority of the proposed PGFuse to existing state-of-the-art solutions. Code can be available upon acceptance.
Authors: Kshitij Pathania
Abstract: In the realm of image synthesis, achieving fidelity to a reference image while adhering to conditional prompts remains a significant challenge. This paper proposes a novel approach that integrates a diffusion model with latent space manipulation and gradient-based selective attention mechanisms to address this issue. Leveraging Grad-SAM (Gradient-based Selective Attention Manipulation), we analyze the cross attention maps of the cross attention layers and gradients for the denoised latent vector, deriving importance scores of elements of denoised latent vector related to the subject of interest. Using this information, we create masks at specific timesteps during denoising to preserve subjects while seamlessly integrating the reference image features. This approach ensures the faithful formation of subjects based on conditional prompts, while concurrently refining the background for a more coherent composition. Our experiments on places365 dataset demonstrate promising results, with our proposed model achieving the lowest mean and median Frechet Inception Distance (FID) scores compared to baseline models, indicating superior fidelity preservation. Furthermore, our model exhibits competitive performance in aligning the generated images with provided textual descriptions, as evidenced by high CLIP scores. These results highlight the effectiveness of our approach in both fidelity preservation and textual context preservation, offering a significant advancement in text-to-image synthesis tasks.
Authors: Shiguang Wang, Tao Xie, Haijun Liu, Xingcheng Zhang, Jian Cheng
Abstract: Channel Pruning is one of the most widespread techniques used to compress deep neural networks while maintaining their performances. Currently, a typical pruning algorithm leverages neural architecture search to directly find networks with a configurable width, the key step of which is to identify representative subnet for various pruning ratios by training a supernet. However, current methods mainly follow a serial training strategy to optimize supernet, which is very time-consuming. In this work, we introduce PSE-Net, a novel parallel-subnets estimator for efficient channel pruning. Specifically, we propose a parallel-subnets training algorithm that simulate the forward-backward pass of multiple subnets by droping extraneous features on batch dimension, thus various subnets could be trained in one round. Our proposed algorithm facilitates the efficiency of supernet training and equips the network with the ability to interpolate the accuracy of unsampled subnets, enabling PSE-Net to effectively evaluate and rank the subnets. Over the trained supernet, we develop a prior-distributed-based sampling algorithm to boost the performance of classical evolutionary search. Such algorithm utilizes the prior information of supernet training phase to assist in the search of optimal subnets while tackling the challenge of discovering samples that satisfy resource constraints due to the long-tail distribution of network configuration. Extensive experiments demonstrate PSE-Net outperforms previous state-of-the-art channel pruning methods on the ImageNet dataset while retaining superior supernet training efficiency. For example, under 300M FLOPs constraint, our pruned MobileNetV2 achieves 75.2% Top-1 accuracy on ImageNet dataset, exceeding the original MobileNetV2 by 2.6 units while only cost 30%/16% times than BCNet/AutoAlim.
Authors: Ye Yu, Fengxin Chen, Jun Yu, Zhen Kan
Abstract: While recent low-light image enhancement (LLIE) methods have made significant advancements, they still face challenges in terms of low visual quality and weak generalization ability when applied to complex scenarios. To address these issues, we propose a semi-supervised method based on latent mean-teacher and Gaussian process, named LMT-GP. We first design a latent mean-teacher framework that integrates both labeled and unlabeled data, as well as their latent vectors, into model training. Meanwhile, we use a mean-teacher-assisted Gaussian process learning strategy to establish a connection between the latent and pseudo-latent vectors obtained from the labeled and unlabeled data. To guide the learning process, we utilize an assisted Gaussian process regression (GPR) loss function. Furthermore, we design a pseudo-label adaptation module (PAM) to ensure the reliability of the network learning. To demonstrate our method's generalization ability and effectiveness, we apply it to multiple LLIE datasets and high-level vision tasks. Experiment results demonstrate that our method achieves high generalization performance and image quality. The code is available at https://github.com/HFUT-CV/LMT-GP.
Authors: Shaolei Yang, Shen Cheng, Mingbo Hong, Haoqiang Fan, Xing Wei, Shuaicheng Liu
Abstract: In this paper, we propose Neural Spectrum Decomposition, a generic decomposition framework for dataset distillation. Unlike previous methods, we consider the entire dataset as a high-dimensional observation that is low-rank across all dimensions. We aim to discover the low-rank representation of the entire dataset and perform distillation efficiently. Toward this end, we learn a set of spectrum tensors and transformation matrices, which, through simple matrix multiplication, reconstruct the data distribution. Specifically, a spectrum tensor can be mapped back to the image space by a transformation matrix, and efficient information sharing during the distillation learning process is achieved through pairwise combinations of different spectrum vectors and transformation matrices. Furthermore, we integrate a trajectory matching optimization method guided by a real distribution. Our experimental results demonstrate that our approach achieves state-of-the-art performance on benchmarks, including CIFAR10, CIFAR100, Tiny Imagenet, and ImageNet Subset. Our code are available at \url{https://github.com/slyang2021/NSD}.
Authors: Yiran Xu, Haoxiang Zhong, Kai Wu, Jialin Li, Yong Liu, Chengjie Wang, Shu-Tao Xia, Hongen Liao
Abstract: Object detectors have shown outstanding performance on various public datasets. However, annotating a new dataset for a new task is usually unavoidable in real, since 1) a single existing dataset usually does not contain all object categories needed; 2) using multiple datasets usually suffers from annotation incompletion and heterogeneous features. We propose a novel problem as "Annotation-incomplete Multi-dataset Detection", and develop an end-to-end multi-task learning architecture which can accurately detect all the object categories with multiple partially annotated datasets. Specifically, we propose an attention feature extractor which helps to mine the relations among different datasets. Besides, a knowledge amalgamation training strategy is incorporated to accommodate heterogeneous features from different sources. Extensive experiments on different object detection datasets demonstrate the effectiveness of our methods and an improvement of 2.17%, 2.10% in mAP can be achieved on COCO and VOC respectively.
Authors: Kanghao Chen, Guoqiang Liang, Hangyu Li, Yunfan Lu, Lin Wang
Abstract: Event cameras offer significant advantages for low-light video enhancement, primarily due to their high dynamic range. Current research, however, is severely limited by the absence of large-scale, real-world, and spatio-temporally aligned event-video datasets. To address this, we introduce a large-scale dataset with over 30,000 pairs of frames and events captured under varying illumination. This dataset was curated using a robotic arm that traces a consistent non-linear trajectory, achieving spatial alignment precision under 0.03mm and temporal alignment with errors under 0.01s for 90% of the dataset. Based on the dataset, we propose \textbf{EvLight++}, a novel event-guided low-light video enhancement approach designed for robust performance in real-world scenarios. Firstly, we design a multi-scale holistic fusion branch to integrate structural and textural information from both images and events. To counteract variations in regional illumination and noise, we introduce Signal-to-Noise Ratio (SNR)-guided regional feature selection, enhancing features from high SNR regions and augmenting those from low SNR regions by extracting structural information from events. To incorporate temporal information and ensure temporal coherence, we further introduce a recurrent module and temporal loss in the whole pipeline. Extensive experiments on our and the synthetic SDSD dataset demonstrate that EvLight++ significantly outperforms both single image- and video-based methods by 1.37 dB and 3.71 dB, respectively. To further explore its potential in downstream tasks like semantic segmentation and monocular depth estimation, we extend our datasets by adding pseudo segmentation and depth labels via meticulous annotation efforts with foundation models. Experiments under diverse low-light scenes show that the enhanced results achieve a 15.97% improvement in mIoU for semantic segmentation.
Authors: Yu Liang, Xiucheng Zhang, Juepeng Zheng, Jianxi Huang, Haohuan Fu
Abstract: Although the Unsupervised Domain Adaptation (UDA) method has improved the effect of remote sensing image classification tasks, most of them are still limited by access to the source domain (SD) data. Designs such as Source-free Domain Adaptation (SFDA) solve the challenge of a lack of SD data, however, they still rely on a large amount of target domain data and thus cannot achieve fast adaptations, which seriously hinders their further application in broader scenarios. The real-world applications of cross-domain remote sensing image classification require a balance of speed and accuracy at the same time. Therefore, we propose a novel and comprehensive test time adaptation (TTA) method -- Low Saturation Confidence Distribution Test Time Adaptation (LSCD-TTA), which is the first attempt to solve such scenarios through the idea of TTA. LSCD-TTA specifically considers the distribution characteristics of remote sensing images, including three main parts that concentrate on different optimization directions: First, low saturation distribution (LSD) considers the dominance of low-confidence samples during the later TTA stage. Second, weak-category cross-entropy (WCCE) increases the weight of categories that are more difficult to classify with less prior knowledge. Finally, diverse categories confidence (DIV) comprehensively considers the category diversity to alleviate the deviation of the sample distribution. By weighting the abovementioned three modules, the model can widely, quickly and accurately adapt to the target domain without much prior target distributions, repeated data access, and manual annotation. We evaluate LSCD-TTA on three remote-sensing image datasets. The experimental results show that LSCD-TTA achieves a significant gain of 4.96%-10.51% with Resnet-50 and 5.33%-12.49% with Resnet-101 in average accuracy compared to other state-of-the-art DA and TTA methods.
Authors: Yanghao Wang, Long Chen
Abstract: Data Augmentation (DA), \ie, synthesizing faithful and diverse samples to expand the original training set, is a prevalent and effective strategy to improve various visual recognition tasks. With the powerful image generation ability, diffusion-based DA has shown strong performance gains on different benchmarks. In this paper, we analyze today's diffusion-based DA methods, and argue that they cannot take account of both faithfulness and diversity, which are two critical keys for generating high-quality samples and boosting final classification performance. To this end, we propose a novel Diffusion-based Inversion Interpolation DA method: Diff-II. Specifically, Diff-II consists of three main steps: 1) Category concepts learning: Learning concept embeddings for each category. 2) Inversion interpolation: Calculating the inversion for each image, and conducting spherical interpolation for two randomly sampled inversions from the same category. 3) Two-stage denoising: Using different prompts to generate synthesized images in a coarse-to-fine manner. Extensive experiments on multiple image classification tasks (\eg, few-shot, long-tailed, and out-of-distribution classification) have demonstrated its effectiveness over state-of-the-art diffusion-based DA methods.
Authors: Shiguang Wang, Zhongyu Zhang, Jian Cheng
Abstract: Dataset distillation synthesizes a small dataset such that a model trained on this set approximates the performance of the original dataset. Recent studies on dataset distillation focused primarily on the design of the optimization process, with methods such as gradient matching, feature alignment, and training trajectory matching. However, little attention has been given to the issue of underutilized regions in synthetic images. In this paper, we propose UDD, a novel approach to identify and exploit the underutilized regions to make them informative and discriminate, and thus improve the utilization of the synthetic dataset. Technically, UDD involves two underutilized regions searching policies for different conditions, i.e., response-based policy and data jittering-based policy. Compared with previous works, such two policies are utilization-sensitive, equipping with the ability to dynamically adjust the underutilized regions during the training process. Additionally, we analyze the current model optimization problem and design a category-wise feature contrastive loss, which can enhance the distinguishability of different categories and alleviate the shortcomings of the existing multi-formation methods. Experimentally, our method improves the utilization of the synthetic dataset and outperforms the state-of-the-art methods on various datasets, such as MNIST, FashionMNIST, SVHN, CIFAR-10, and CIFAR-100. For example, the improvements on CIFAR-10 and CIFAR-100 are 4.0\% and 3.7\% over the next best method with IPC=1, by mining the underutilized regions.
Authors: Kaijing Ma, Haojian Huang, Jin Chen, Haodong Chen, Pengliang Ji, Xianghao Zang, Han Fang, Chao Ban, Hao Sun, Mulin Chen, Xuelong Li
Abstract: Existing Video Temporal Grounding (VTG) models excel in accuracy but often overlook open-world challenges posed by open-vocabulary queries and untrimmed videos. This leads to unreliable predictions for noisy, corrupted, and out-of-distribution data. Adapting VTG models to dynamically estimate uncertainties based on user input can address this issue. To this end, we introduce SRAM, a robust network module that benefits from a two-stage cross-modal alignment task. More importantly, it integrates Deep Evidential Regression (DER) to explicitly and thoroughly quantify uncertainty during training, thus allowing the model to say "I do not know" in scenarios beyond its handling capacity. However, the direct application of traditional DER theory and its regularizer reveals structural flaws, leading to unintended constraints in VTG tasks. In response, we develop a simple yet effective Geom-regularizer that enhances the uncertainty learning framework from the ground up. To the best of our knowledge, this marks the first successful attempt of DER in VTG. Our extensive quantitative and qualitative results affirm the effectiveness, robustness, and interpretability of our modules and the uncertainty learning paradigm in VTG tasks. The code will be made available.
Authors: Guangxi Li, Yinsheng Song, Mingkai Zheng
Abstract: Long-tailed distributions in image recognition pose a considerable challenge due to the severe imbalance between a few dominant classes with numerous examples and many minority classes with few samples. Recently, the use of large generative models to create synthetic data for image classification has been realized, but utilizing synthetic data to address the challenge of long-tailed recognition remains relatively unexplored. In this work, we proposed the use of synthetic data as a complement to long-tailed datasets to eliminate the impact of data imbalance. To tackle this real-synthetic mixed dataset, we designed a two-branch model that contains Synthetic-Aware and Unaware branches (SAU). The core ideas are (1) a synthetic-unaware branch for classification that mixes real and synthetic data and treats all data equally without distinguishing between them. (2) A synthetic-aware branch for improving the robustness of the feature extractor by distinguishing between real and synthetic data and learning their discrepancies. Extensive experimental results demonstrate that our method can improve the accuracy of long-tailed image recognition. Notably, our approach achieves state-of-the-art Top-1 accuracy and significantly surpasses other methods on CIFAR-10-LT and CIFAR-100-LT datasets across various imbalance factors. Our code is available at https://github.com/lgX1123/gm4lt.
Authors: Yaping He, Linhao Jiang, Di Wu
Abstract: Deep neural networks typically impose significant computational loads and memory consumption. Moreover, the large parameters pose constraints on deploying the model on edge devices such as embedded systems. Tensor decomposition offers a clear advantage in compressing large-scale weight tensors. Nevertheless, direct utilization of low-rank decomposition typically leads to significant accuracy loss. This paper proposes a model compression method that integrates Variational Bayesian Matrix Factorization (VBMF) with orthogonal regularization. Initially, the model undergoes over-parameterization and training, with orthogonal regularization applied to enhance its likelihood of achieving the accuracy of the original model. Secondly, VBMF is employed to estimate the rank of the weight tensor at each layer. Our framework is sufficiently general to apply to other convolutional neural networks and easily adaptable to incorporate other tensor decomposition methods. Experimental results show that for both high and low compression ratios, our compression model exhibits advanced performance.
Authors: Kengo Nakata, Daisuke Miyashita, Youyang Ng, Yasuto Hoshi, Jun Deguchi
Abstract: In this paper, we rethink sparse lexical representations for image retrieval. By utilizing multi-modal large language models (M-LLMs) that support visual prompting, we can extract image features and convert them into textual data, enabling us to utilize efficient sparse retrieval algorithms employed in natural language processing for image retrieval tasks. To assist the LLM in extracting image features, we apply data augmentation techniques for key expansion and analyze the impact with a metric for relevance between images and textual data. We empirically show the superior precision and recall performance of our image retrieval method compared to conventional vision-language model-based methods on the MS-COCO, PASCAL VOC, and NUS-WIDE datasets in a keyword-based image retrieval scenario, where keywords serve as search queries. We also demonstrate that the retrieval performance can be improved by iteratively incorporating keywords into search queries.
Authors: Mian Zou, Baosheng Yu, Yibing Zhan, Siwei Lyu, Kede Ma
Abstract: In recent years, the multimedia forensics and security community has seen remarkable progress in multitask learning for DeepFake (i.e., face forgery) detection. The prevailing strategy has been to frame DeepFake detection as a binary classification problem augmented by manipulation-oriented auxiliary tasks. This strategy focuses on learning features specific to face manipulations, which exhibit limited generalizability. In this paper, we delve deeper into semantics-oriented multitask learning for DeepFake detection, leveraging the relationships among face semantics via joint embedding. We first propose an automatic dataset expansion technique that broadens current face forgery datasets to support semantics-oriented DeepFake detection tasks at both the global face attribute and local face region levels. Furthermore, we resort to joint embedding of face images and their corresponding labels (depicted by textual descriptions) for prediction. This approach eliminates the need for manually setting task-agnostic and task-specific parameters typically required when predicting labels directly from images. In addition, we employ a bi-level optimization strategy to dynamically balance the fidelity loss weightings of various tasks, making the training process fully automated. Extensive experiments on six DeepFake datasets show that our method improves the generalizability of DeepFake detection and, meanwhile, renders some degree of model interpretation by providing human-understandable explanations.
Authors: Luyao Tang, Yuxuan Yuan, Chaoqi Chen, Kunze Huang, Xinghao Ding, Yue Huang
Abstract: Foundation models have made incredible strides in achieving zero-shot or few-shot generalization, leveraging prompt engineering to mimic the problem-solving approach of human intelligence. However, when it comes to some foundation models like Segment Anything, there is still a challenge in performing well on out-of-distribution data, including camouflaged and medical images. Inconsistent prompting strategies during fine-tuning and testing further compound the issue, leading to decreased performance. Drawing inspiration from how human cognition processes new environments, we introduce SlotSAM, a method that reconstructs features from the encoder in a self-supervised manner to create object-centric representations. These representations are then integrated into the foundation model, bolstering its object-level perceptual capabilities while reducing the impact of distribution-related variables. The beauty of SlotSAM lies in its simplicity and adaptability to various tasks, making it a versatile solution that significantly enhances the generalization abilities of foundation models. Through limited parameter fine-tuning in a bootstrap manner, our approach paves the way for improved generalization in novel environments. The code is available at github.com/lytang63/SlotSAM.
Authors: Yukang Huo, Mingyuan Yao, Qingbin Tian, Tonghao Wang, Ruifeng Wang, Haihua Wang
Abstract: Over the past few years, the YOLO series of models has emerged as one of the dominant methodologies in the realm of object detection. Many studies have advanced these baseline models by modifying their architectures, enhancing data quality, and developing new loss functions. However, current models still exhibit deficiencies in processing feature maps, such as overlooking the fusion of cross-scale features and a static fusion approach that lacks the capability for dynamic feature adjustment. To address these issues, this paper introduces an efficient Fine-grained Multi-scale Dynamic Selection Module (FMDS Module), which applies a more effective dynamic feature selection and fusion method on fine-grained multi-scale feature maps, significantly enhancing the detection accuracy of small, medium, and large-sized targets in complex environments. Furthermore, this paper proposes an Adaptive Gated Multi-branch Focus Fusion Module (AGMF Module), which utilizes multiple parallel branches to perform complementary fusion of various features captured by the gated unit branch, FMDS Module branch, and TripletAttention branch. This approach further enhances the comprehensiveness, diversity, and integrity of feature fusion. This paper has integrated the FMDS Module, AGMF Module, into Yolov9 to develop a novel object detection model named FA-YOLO. Extensive experimental results show that under identical experimental conditions, FA-YOLO achieves an outstanding 66.1% mean Average Precision (mAP) on the PASCAL VOC 2007 dataset, representing 1.0% improvement over YOLOv9's 65.1%. Additionally, the detection accuracies of FA-YOLO for small, medium, and large targets are 44.1%, 54.6%, and 70.8%, respectively, showing improvements of 2.0%, 3.1%, and 0.9% compared to YOLOv9's 42.1%, 51.5%, and 69.9%.
Authors: Minghang Zheng, Jiahua Zhang, Qingchao Chen, Yuxin Peng, Yang Liu
Abstract: Visual grounding aims to localize the object referred to in an image based on a natural language query. Although progress has been made recently, accurately localizing target objects within multiple-instance distractions (multiple objects of the same category as the target) remains a significant challenge. Existing methods demonstrate a significant performance drop when there are multiple distractions in an image, indicating an insufficient understanding of the fine-grained semantics and spatial relationships between objects. In this paper, we propose a novel approach, the Relation and Semantic-sensitive Visual Grounding (ResVG) model, to address this issue. Firstly, we enhance the model's understanding of fine-grained semantics by injecting semantic prior information derived from text queries into the model. This is achieved by leveraging text-to-image generation models to produce images representing the semantic attributes of target objects described in queries. Secondly, we tackle the lack of training samples with multiple distractions by introducing a relation-sensitive data augmentation method. This method generates additional training data by synthesizing images containing multiple objects of the same category and pseudo queries based on their spatial relationships. The proposed ReSVG model significantly improves the model's ability to comprehend both object semantics and spatial relations, leading to enhanced performance in visual grounding tasks, particularly in scenarios with multiple-instance distractions. We conduct extensive experiments to validate the effectiveness of our methods on five datasets. Code is available at https://github.com/minghangz/ResVG.
Authors: Manuel Alejandro Diaz-Zapata (CHROMA), Wenqian Liu (CHROMA, UGA), Robin Baruffa (CHROMA), Christian Laugier (CHROMA, E-MOTION, Inria)
Abstract: Current research in semantic bird's-eye view segmentation for autonomous driving focuses solely on optimizing neural network models using a single dataset, typically nuScenes. This practice leads to the development of highly specialized models that may fail when faced with different environments or sensor setups, a problem known as domain shift. In this paper, we conduct a comprehensive cross-dataset evaluation of state-of-the-art BEV segmentation models to assess their performance across different training and testing datasets and setups, as well as different semantic categories. We investigate the influence of different sensors, such as cameras and LiDAR, on the models' ability to generalize to diverse conditions and scenarios. Additionally, we conduct multi-dataset training experiments that improve models' BEV segmentation performance compared to single-dataset training. Our work addresses the gap in evaluating BEV segmentation models under cross-dataset validation. And our findings underscore the importance of enhancing model generalizability and adaptability to ensure more robust and reliable BEV segmentation approaches for autonomous driving applications.
Authors: Mathias Vogel, Keisuke Tateno, Marc Pollefeys, Federico Tombari, Marie-Julie Rakotosaona, Francis Engelmann
Abstract: In this work, we tackle the task of point cloud denoising through a novel framework that adapts Diffusion Schr\"odinger bridges to points clouds. Unlike previous approaches that predict point-wise displacements from point features or learned noise distributions, our method learns an optimal transport plan between paired point clouds. Experiments on object datasets like PU-Net and real-world datasets such as ScanNet++ and ARKitScenes show that P2P-Bridge achieves significant improvements over existing methods. While our approach demonstrates strong results using only point coordinates, we also show that incorporating additional features, such as color information or point-wise DINOv2 features, further enhances the performance. Code and pretrained models are available at https://p2p-bridge.github.io.
Authors: Yifei Chen, Shenghao Zhu, Zhaojie Fang, Chang Liu, Binfeng Zou, Yuhe Wang, Shuo Chang, Fan Jia, Feiwei Qin, Jin Fan, Yong Peng, Changmiao Wang
Abstract: Alzheimer's Disease (AD) is a complex neurodegenerative disorder marked by memory loss, executive dysfunction, and personality changes. Early diagnosis is challenging due to subtle symptoms and varied presentations, often leading to misdiagnosis with traditional unimodal diagnostic methods due to their limited scope. This study introduces an advanced multimodal classification model that integrates clinical, cognitive, neuroimaging, and EEG data to enhance diagnostic accuracy. The model incorporates a feature tagger with a tabular data coding architecture and utilizes the TimesBlock module to capture intricate temporal patterns in Electroencephalograms (EEG) data. By employing Cross-modal Attention Aggregation module, the model effectively fuses Magnetic Resonance Imaging (MRI) spatial information with EEG temporal data, significantly improving the distinction between AD, Mild Cognitive Impairment, and Normal Cognition. Simultaneously, we have constructed the first AD classification dataset that includes three modalities: EEG, MRI, and tabular data. Our innovative approach aims to facilitate early diagnosis and intervention, potentially slowing the progression of AD. The source code and our private ADMC dataset are available at https://github.com/JustlfC03/MSTNet.
Authors: Shijia Yang, Bohan Zhai, Quanzeng You, Jianbo Yuan, Hongxia Yang, Chenfeng Xu
Abstract: We present the "Law of Vision Representation" in multimodal large language models (MLLMs). It reveals a strong correlation between the combination of cross-modal alignment, correspondence in vision representation, and MLLM performance. We quantify the two factors using the cross-modal Alignment and Correspondence score (AC score). Through extensive experiments involving thirteen different vision representation settings and evaluations across eight benchmarks, we find that the AC score is linearly correlated to model performance. By leveraging this relationship, we are able to identify and train the optimal vision representation only, which does not require finetuning the language model every time, resulting in a 99.7% reduction in computational cost.
Authors: Lucrezia Tosato, Victor Fortier, Isabelle Bloch, Catherine Pelachaud
Abstract: Studies in human human interaction have introduced the concept of F formation to describe the spatial arrangement of participants during social interactions. This paper has two objectives. It aims at detecting F formations in video sequences and predicting the next speaker in a group conversation. The proposed approach exploits time information and human multimodal signals in video sequences. In particular, we rely on measuring the engagement level of people as a feature of group belonging. Our approach makes use of a recursive neural network, the Long Short Term Memory (LSTM), to predict who will take the speaker's turn in a conversation group. Experiments on the MatchNMingle dataset led to 85% true positives in group detection and 98% accuracy in predicting the next speaker.
Authors: Pardis Afshar, Sajjad Hashembeiki, Pouya Khani, Emad Fatemizadeh, Mohammad Hossein Rohban
Abstract: Histopathological image analysis is crucial for accurate cancer diagnosis and treatment planning. While deep learning models, especially convolutional neural networks, have advanced this field, their "black-box" nature raises concerns about interpretability and trustworthiness. Explainable Artificial Intelligence (XAI) techniques aim to address these concerns, but evaluating their effectiveness remains challenging. A significant issue with current occlusion-based XAI methods is that they often generate Out-of-Distribution (OoD) samples, leading to inaccurate evaluations. In this paper, we introduce Inpainting-Based Occlusion (IBO), a novel occlusion strategy that utilizes a Denoising Diffusion Probabilistic Model to inpaint occluded regions in histopathological images. By replacing cancerous areas with realistic, non-cancerous tissue, IBO minimizes OoD artifacts and preserves data integrity. We evaluate our method on the CAMELYON16 dataset through two phases: first, by assessing perceptual similarity using the Learned Perceptual Image Patch Similarity (LPIPS) metric, and second, by quantifying the impact on model predictions through Area Under the Curve (AUC) analysis. Our results demonstrate that IBO significantly improves perceptual fidelity, achieving nearly twice the improvement in LPIPS scores compared to the best existing occlusion strategy. Additionally, IBO increased the precision of XAI performance prediction from 42% to 71% compared to traditional methods. These results demonstrate IBO's potential to provide more reliable evaluations of XAI techniques, benefiting histopathology and other applications. The source code for this study is available at https://github.com/a-fsh-r/IBO.
Authors: Massimo Bosetti, Shibingfeng Zhang, Bendetta Liberatori, Giacomo Zara, Elisa Ricci, Paolo Rota
Abstract: Vision-language models (VLMs) have demonstrated remarkable performance across various visual tasks, leveraging joint learning of visual and textual representations. While these models excel in zero-shot image tasks, their application to zero-shot video action recognition (ZSVAR) remains challenging due to the dynamic and temporal nature of actions. Existing methods for ZS-VAR typically require extensive training on specific datasets, which can be resource-intensive and may introduce domain biases. In this work, we propose Text-Enhanced Action Recognition (TEAR), a simple approach to ZS-VAR that is training-free and does not require the availability of training data or extensive computational resources. Drawing inspiration from recent findings in vision and language literature, we utilize action descriptors for decomposition and contextual information to enhance zero-shot action recognition. Through experiments on UCF101, HMDB51, and Kinetics-600 datasets, we showcase the effectiveness and applicability of our proposed approach in addressing the challenges of ZS-VAR.
Authors: Jiefeng Li, Ye Yuan, Davis Rempe, Haotian Zhang, Pavlo Molchanov, Cewu Lu, Jan Kautz, Umar Iqbal
Abstract: Estimating global human motion from moving cameras is challenging due to the entanglement of human and camera motions. To mitigate the ambiguity, existing methods leverage learned human motion priors, which however often result in oversmoothed motions with misaligned 2D projections. To tackle this problem, we propose COIN, a control-inpainting motion diffusion prior that enables fine-grained control to disentangle human and camera motions. Although pre-trained motion diffusion models encode rich motion priors, we find it non-trivial to leverage such knowledge to guide global motion estimation from RGB videos. COIN introduces a novel control-inpainting score distillation sampling method to ensure well-aligned, consistent, and high-quality motion from the diffusion prior within a joint optimization framework. Furthermore, we introduce a new human-scene relation loss to alleviate the scale ambiguity by enforcing consistency among the humans, camera, and scene. Experiments on three challenging benchmarks demonstrate the effectiveness of COIN, which outperforms the state-of-the-art methods in terms of global human motion estimation and camera motion estimation. As an illustrative example, COIN outperforms the state-of-the-art method by 33% in world joint position error (W-MPJPE) on the RICH dataset.
Authors: Deshui Miao, Yameng Gu, Xin Li, Zhenyu He, Yaowei Wang, Ming-Hsuan Yang
Abstract: Video object segmentation (VOS) is a crucial task in computer vision, but current VOS methods struggle with complex scenes and prolonged object motions. To address these challenges, the MOSE dataset aims to enhance object recognition and differentiation in complex environments, while the LVOS dataset focuses on segmenting objects exhibiting long-term, intricate movements. This report introduces a discriminative spatial-temporal VOS model that utilizes discriminative object features as query representations. The semantic understanding of spatial-semantic modules enables it to recognize object parts, while salient features highlight more distinctive object characteristics. Our model, trained on extensive VOS datasets, achieved first place (\textbf{80.90\%} $\mathcal{J \& F}$) on the test set of the 6th LSVOS challenge in the VOS Track, demonstrating its effectiveness in tackling the aforementioned challenges. The code will be available at \href{https://github.com/yahooo-m/VOS-Solution}{code}.
Authors: Mohammad Belal (Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates), Taimur Hassan (Abu Dhabi University, Abu Dhabi, United Arab Emirates), Abdelfatah Hassan (Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates), Nael Alsheikh (Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates), Noureldin Elhendawi (Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates), Irfan Hussain (Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates)
Abstract: Human activity recognition is a major field of study that employs computer vision, machine vision, and deep learning techniques to categorize human actions. The field of deep learning has made significant progress, with architectures that are extremely effective at capturing human dynamics. This study emphasizes the influence of feature fusion on the accuracy of activity recognition. This technique addresses the limitation of conventional models, which face difficulties in identifying activities because of their limited capacity to understand spatial and temporal features. The technique employs sensory data obtained from four publicly available datasets: HuGaDB, PKU-MMD, LARa, and TUG. The accuracy and F1-score of two deep learning models, specifically a Transformer model and a Parameter-Optimized Graph Convolutional Network (PO-GCN), were evaluated using these datasets. The feature fusion technique integrated the final layer features from both models and inputted them into a classifier. Empirical evidence demonstrates that PO-GCN outperforms standard models in activity recognition. HuGaDB demonstrated a 2.3% improvement in accuracy and a 2.2% increase in F1-score. TUG showed a 5% increase in accuracy and a 0.5% rise in F1-score. On the other hand, LARa and PKU-MMD achieved lower accuracies of 64% and 69% respectively. This indicates that the integration of features enhanced the performance of both the Transformer model and PO-GCN.
Authors: Sierra Bonilla, Chiara Di Vece, Rema Daher, Xinwei Ju, Danail Stoyanov, Francisco Vasconcelos, Sophia Bano
Abstract: Three-dimensional (3D) reconstruction from two-dimensional images is an active research field in computer vision, with applications ranging from navigation and object tracking to segmentation and three-dimensional modeling. Traditionally, parametric techniques have been employed for this task. However, recent advancements have seen a shift towards learning-based methods. Given the rapid pace of research and the frequent introduction of new image matching methods, it is essential to evaluate them. In this paper, we present a comprehensive evaluation of various image matching methods using a structure-from-motion pipeline. We assess the performance of these methods on both in-domain and out-of-domain datasets, identifying key limitations in both the methods and benchmarks. We also investigate the impact of edge detection as a pre-processing step. Our analysis reveals that image matching for 3D reconstruction remains an open challenge, necessitating careful selection and tuning of models for specific scenarios, while also highlighting mismatches in how metrics currently represent method performance.
Authors: Zengjie Song, Jiangshe Zhang, Yuxi Wang, Junsong Fan, Zhaoxiang Zhang
Abstract: Sound source localization aims to localize objects emitting the sound in visual scenes. Recent works obtaining impressive results typically rely on contrastive learning. However, the common practice of randomly sampling negatives in prior arts can lead to the false negative issue, where the sounds semantically similar to visual instance are sampled as negatives and incorrectly pushed away from the visual anchor/query. As a result, this misalignment of audio and visual features could yield inferior performance. To address this issue, we propose a novel audio-visual learning framework which is instantiated with two individual learning schemes: self-supervised predictive learning (SSPL) and semantic-aware contrastive learning (SACL). SSPL explores image-audio positive pairs alone to discover semantically coherent similarities between audio and visual features, while a predictive coding module for feature alignment is introduced to facilitate the positive-only learning. In this regard SSPL acts as a negative-free method to eliminate false negatives. By contrast, SACL is designed to compact visual features and remove false negatives, providing reliable visual anchor and audio negatives for contrast. Different from SSPL, SACL releases the potential of audio-visual contrastive learning, offering an effective alternative to achieve the same goal. Comprehensive experiments demonstrate the superiority of our approach over the state-of-the-arts. Furthermore, we highlight the versatility of the learned representation by extending the approach to audio-visual event classification and object detection tasks. Code and models are available at: https://github.com/zjsong/SACL.
Authors: Chaeyeon Chung, Sunghyun Park, Jeongho Kim, Jaegul Choo
Abstract: Hairstyle transfer is a challenging task in the image editing field that modifies the hairstyle of a given face image while preserving its other appearance and background features. The existing hairstyle transfer approaches heavily rely on StyleGAN, which is pre-trained on cropped and aligned face images. Hence, they struggle to generalize under challenging conditions such as extreme variations of head poses or focal lengths. To address this issue, we propose a one-stage hairstyle transfer diffusion model, HairFusion, that applies to real-world scenarios. Specifically, we carefully design a hair-agnostic representation as the input of the model, where the original hair information is thoroughly eliminated. Next, we introduce a hair align cross-attention (Align-CA) to accurately align the reference hairstyle with the face image while considering the difference in their face shape. To enhance the preservation of the face image's original features, we leverage adaptive hair blending during the inference, where the output's hair regions are estimated by the cross-attention map in Align-CA and blended with non-hair areas of the face image. Our experimental results show that our method achieves state-of-the-art performance compared to the existing methods in preserving the integrity of both the transferred hairstyle and the surrounding features. The codes are available at https://github.com/cychungg/HairFusion.
Authors: Yongcun Zhang, Jiajun Xu, Yina He, Shaozi Li, Zhiming Luo, Huangwei Lei
Abstract: Tongue diagnosis in Traditional Chinese Medicine (TCM) is a crucial diagnostic method that can reflect an individual's health status. Traditional methods for identifying tooth-marked tongues are subjective and inconsistent because they rely on practitioner experience. We propose a novel fully automated Weakly Supervised method using Vision transformer and Multiple instance learning WSVM for tongue extraction and tooth-marked tongue recognition. Our approach first accurately detects and extracts the tongue region from clinical images, removing any irrelevant background information. Then, we implement an end-to-end weakly supervised object detection method. We utilize Vision Transformer (ViT) to process tongue images in patches and employ multiple instance loss to identify tooth-marked regions with only image-level annotations. WSVM achieves high accuracy in tooth-marked tongue classification, and visualization experiments demonstrate its effectiveness in pinpointing these regions. This automated approach enhances the objectivity and accuracy of tooth-marked tongue diagnosis. It provides significant clinical value by assisting TCM practitioners in making precise diagnoses and treatment recommendations. Code is available at https://github.com/yc-zh/WSVM.
Authors: Jing Jiang, Sicheng Zhao, Jiankun Zhu, Wenbo Tang, Zhaopan Xu, Jidong Yang, Pengfei Xu, Hongxun Yao
Abstract: Panoramic semantic segmentation has received widespread attention recently due to its comprehensive 360\degree field of view. However, labeling such images demands greater resources compared to pinhole images. As a result, many unsupervised domain adaptation methods for panoramic semantic segmentation have emerged, utilizing real pinhole images or low-cost synthetic panoramic images. But, the segmentation model lacks understanding of the panoramic structure when only utilizing real pinhole images, and it lacks perception of real-world scenes when only adopting synthetic panoramic images. Therefore, in this paper, we propose a new task of multi-source domain adaptation for panoramic semantic segmentation, aiming to utilize both real pinhole and synthetic panoramic images in the source domains, enabling the segmentation model to perform well on unlabeled real panoramic images in the target domain. Further, we propose Deformation Transform Aligner for Panoramic Semantic Segmentation (DTA4PASS), which converts all pinhole images in the source domains into panoramic-like images, and then aligns the converted source domains with the target domain. Specifically, DTA4PASS consists of two main components: Unpaired Semantic Morphing (USM) and Distortion Gating Alignment (DGA). Firstly, in USM, the Semantic Dual-view Discriminator (SDD) assists in training the diffeomorphic deformation network, enabling the effective transformation of pinhole images without paired panoramic views. Secondly, DGA assigns pinhole-like and panoramic-like features to each image by gating, and aligns these two features through uncertainty estimation. DTA4PASS outperforms the previous state-of-the-art methods by 1.92% and 2.19% on the outdoor and indoor multi-source domain adaptation scenarios, respectively. The source code will be released.
Authors: Michael Adlerstein, Angelo Bratta, Jo\~ao Carlos Virgolino Soares, Giovanni Dessy, Miguel Fernandes, Matteo Gatti, Claudio Semini
Abstract: Grapevine winter pruning is a labor-intensive and repetitive process that significantly influences the quality and quantity of the grape harvest and produced wine of the following season. It requires a careful and expert detection of the point to be cut. Because of its complexity, repetitive nature and time constraint, the task requires skilled labor that needs to be trained. This extended abstract presents the computer vision pipeline employed in project Vinum, using detectron2 as a segmentation network and keypoint visual odometry to merge different observation into a single pointcloud used to make informed pruning decisions.
Authors: Linyan Yang, Lukas Hoyer, Mark Weber, Tobias Fischer, Dengxin Dai, Laura Leal-Taix\'e, Marc Pollefeys, Daniel Cremers, Luc Van Gool
Abstract: Unsupervised Domain Adaptation (UDA) is the task of bridging the domain gap between a labeled source domain, e.g., synthetic data, and an unlabeled target domain. We observe that current UDA methods show inferior results on fine structures and tend to oversegment objects with ambiguous appearance. To address these shortcomings, we propose to leverage geometric information, i.e., depth predictions, as depth discontinuities often coincide with segmentation boundaries. We show that naively incorporating depth into current UDA methods does not fully exploit the potential of this complementary information. To this end, we present MICDrop, which learns a joint feature representation by masking image encoder features while inversely masking depth encoder features. With this simple yet effective complementary masking strategy, we enforce the use of both modalities when learning the joint feature representation. To aid this process, we propose a feature fusion module to improve both global as well as local information sharing while being robust to errors in the depth predictions. We show that our method can be plugged into various recent UDA methods and consistently improve results across standard UDA benchmarks, obtaining new state-of-the-art performances.
Authors: Zhengqing Gao, Xiang Ao, Xu-Yao Zhang, Cheng-Lin Liu
Abstract: Adapting pre-trained models to open classes is a challenging problem in machine learning. Vision-language models fully explore the knowledge of text modality, demonstrating strong zero-shot recognition performance, which is naturally suited for various open-set problems. More recently, some research focuses on fine-tuning such models to downstream tasks. Prompt tuning methods achieved huge improvements by learning context vectors on few-shot data. However, through the evaluation under open-set adaptation setting with the test data including new classes, we find that there exists a dilemma that learned prompts have worse generalization abilities than hand-crafted prompts. In this paper, we consider combining the advantages of both and come up with a test-time prompt tuning approach, which leverages the maximum concept matching (MCM) scores as dynamic weights to generate an input-conditioned prompt for each image during test. Through extensive experiments on 11 different datasets, we show that our proposed method outperforms all comparison methods on average considering both base and new classes. The code is available at https://github.com/gaozhengqing/TTPT
Authors: Wenyi Hong, Weihan Wang, Ming Ding, Wenmeng Yu, Qingsong Lv, Yan Wang, Yean Cheng, Shiyu Huang, Junhui Ji, Zhao Xue, Lei Zhao, Zhuoyi Yang, Xiaotao Gu, Xiaohan Zhang, Guanyu Feng, Da Yin, Zihan Wang, Ji Qi, Xixuan Song, Peng Zhang, Debing Liu, Bin Xu, Juanzi Li, Yuxiao Dong, Jie Tang
Abstract: Beginning with VisualGLM and CogVLM, we are continuously exploring VLMs in pursuit of enhanced vision-language fusion, efficient higher-resolution architecture, and broader modalities and applications. Here we propose the CogVLM2 family, a new generation of visual language models for image and video understanding including CogVLM2, CogVLM2-Video and GLM-4V. As an image understanding model, CogVLM2 inherits the visual expert architecture with improved training recipes in both pre-training and post-training stages, supporting input resolution up to $1344 \times 1344$ pixels. As a video understanding model, CogVLM2-Video integrates multi-frame input with timestamps and proposes automated temporal grounding data construction. Notably, CogVLM2 family has achieved state-of-the-art results on benchmarks like MMBench, MM-Vet, TextVQA, MVBench and VCGBench. All models are open-sourced in https://github.com/THUDM/CogVLM2 and https://github.com/THUDM/GLM-4, contributing to the advancement of the field.
URLs: https://github.com/THUDM/CogVLM2, https://github.com/THUDM/GLM-4,
Authors: Piotr Rudol, Patrick Doherty, Mariusz Wzorek, Chattrakul Sombattheera
Abstract: The problem of reliably detecting and geolocating objects of different classes in soft real-time is essential in many application areas, such as Search and Rescue performed using Unmanned Aerial Vehicles (UAVs). This research addresses the complementary problems of system contextual vision-based detector selection, allocation, and execution, in addition to the fusion of detection results from teams of UAVs for the purpose of accurately and reliably geolocating objects of interest in a timely manner. In an offline step, an application-independent evaluation of vision-based detectors from a system perspective is first performed. Based on this evaluation, the most appropriate algorithms for online object detection for each platform are selected automatically before a mission, taking into account a number of practical system considerations, such as the available communication links, video compression used, and the available computational resources. The detection results are fused using a method for building maps of salient locations which takes advantage of a novel sensor model for vision-based detections for both positive and negative observations. A number of simulated and real flight experiments are also presented, validating the proposed method.
Authors: Chang-Hwan Son
Abstract: This study introduces a new dense pest counting problem to predict densely distributed pests captured by digital traps. Unlike traditional detection-based counting models for sparsely distributed objects, trap-based pest counting must deal with dense pest distributions that pose challenges such as severe occlusion, wide pose variation, and similar appearances in colors and textures. To address these problems, it is essential to incorporate the local attention mechanism, which identifies locally important and unimportant areas to learn locally grouped features, thereby enhancing discriminative performance. Accordingly, this study presents a novel design that integrates locally grouped and scale-guided attention into a multiscale CenterNet framework. To group local features with similar attributes, a straightforward method is introduced using the heatmap predicted by the first hourglass containing pest centroid information, which eliminates the need for complex clustering models. To enhance attentiveness, the pixel attention module transforms the heatmap into a learnable map. Subsequently, scale-guided attention is deployed to make the object and background features more discriminative, achieving multiscale feature fusion. Through experiments, the proposed model is verified to enhance object features based on local grouping and discriminative feature attention learning. Additionally, the proposed model is highly effective in overcoming occlusion and pose variation problems, making it more suitable for dense pest counting. In particular, the proposed model outperforms state-of-the-art models by a large margin, with a remarkable contribution to dense pest counting.
Authors: Nedyalko Prisadnikov, Wouter Van Gansbeke, Danda Pani Paudel, Luc Van Gool
Abstract: Generalist vision models aim for one and the same architecture for a variety of vision tasks. While such shared architecture may seem attractive, generalist models tend to be outperformed by their bespoken counterparts, especially in the case of panoptic segmentation. We address this problem by introducing two key contributions, without compromising the desirable properties of generalist models. These contributions are: (i) a positional-embedding (PE) based loss for improved centroid regressions; (ii) Edge Distance Sampling (EDS) for the better separation of instance boundaries. The PE-based loss facilitates a better per-pixel regression of the associated instance's centroid, whereas EDS contributes by carefully handling the void regions (caused by missing labels) and smaller instances. These two simple yet effective modifications significantly improve established baselines, while achieving state-of-the-art results among all generalist solutions. More specifically, our method achieves a panoptic quality(PQ) of 52.5 on the COCO dataset, which is an improvement of 10 points over the best model with similar approach (Painter), and is superior by 2 to the best performing diffusion-based method Pix2Seq-$\mathcal{D}$. Furthermore, we provide insights into and an in-depth analysis of our contributions through exhaustive experiments. Our source code and model weights will be made publicly available.
Authors: Xiaoyu Jin, Zunnan Xu, Mingwen Ou, Wenming Yang
Abstract: Character animation is a transformative field in computer graphics and vision, enabling dynamic and realistic video animations from static images. Despite advancements, maintaining appearance consistency in animations remains a challenge. Our approach addresses this by introducing a training-free framework that ensures the generated video sequence preserves the reference image's subtleties, such as physique and proportions, through a dual alignment strategy. We decouple skeletal and motion priors from pose information, enabling precise control over animation generation. Our method also improves pixel-level alignment for conditional control from the reference character, enhancing the temporal consistency and visual cohesion of animations. Our method significantly enhances the quality of video generation without the need for large datasets or expensive computational resources.
Authors: Liyao Tang, Zhe Chen, Shanshan Zhao, Chaoyue Wang, Dacheng Tao
Abstract: Label-efficient segmentation aims to perform effective segmentation on input data using only sparse and limited ground-truth labels for training. This topic is widely studied in 3D point cloud segmentation due to the difficulty of annotating point clouds densely, while it is also essential for cost-effective segmentation on 2D images. Until recently, pseudo-labels have been widely employed to facilitate training with limited ground-truth labels, and promising progress has been witnessed in both the 2D and 3D segmentation. However, existing pseudo-labeling approaches could suffer heavily from the noises and variations in unlabelled data, which would result in significant discrepancies between generated pseudo-labels and current model predictions during training. We analyze that this can further confuse and affect the model learning process, which shows to be a shared problem in label-efficient learning across both 2D and 3D modalities. To address this issue, we propose a novel learning strategy to regularize the pseudo-labels generated for training, thus effectively narrowing the gaps between pseudo-labels and model predictions. More specifically, our method introduces an Entropy Regularization loss and a Distribution Alignment loss for label-efficient learning, resulting in an ERDA learning strategy. Interestingly, by using KL distance to formulate the distribution alignment loss, ERDA reduces to a deceptively simple cross-entropy-based loss which optimizes both the pseudo-label generation module and the segmentation model simultaneously. In addition, we innovate in the pseudo-label generation to make our ERDA consistently effective across both 2D and 3D data modalities for segmentation. Enjoying simplicity and more modality-agnostic pseudo-label generation, our method has shown outstanding performance in fully utilizing all unlabeled data points for training across ...
Authors: Yu Wang, Shaohua Wang, Yicheng Li, Mingchun Liu
Abstract: In recent years, 3D object perception has become a crucial component in the development of autonomous driving systems, providing essential environmental awareness. However, as perception tasks in autonomous driving evolve, their variants have increased, leading to diverse insights from industry and academia. Currently, there is a lack of comprehensive surveys that collect and summarize these perception tasks and their developments from a broader perspective. This review extensively summarizes traditional 3D object detection methods, focusing on camera-based, LiDAR-based, and fusion detection techniques. We provide a comprehensive analysis of the strengths and limitations of each approach, highlighting advancements in accuracy and robustness. Furthermore, we discuss future directions, including methods to improve accuracy such as temporal perception, occupancy grids, and end-to-end learning frameworks. We also explore cooperative perception methods that extend the perception range through collaborative communication. By providing a holistic view of the current state and future developments in 3D object perception, we aim to offer a more comprehensive understanding of perception tasks for autonomous driving. Additionally, we have established an active repository to provide continuous updates on the latest advancements in this field, accessible at: https://github.com/Fishsoup0/Autonomous-Driving-Perception.
URLs: https://github.com/Fishsoup0/Autonomous-Driving-Perception.
Authors: Nikita Kister, Istv\'an S\'ar\'andi, Anna Khoreva, Gerard Pons-Moll
Abstract: The estimation of 3D human poses from images has progressed tremendously over the last few years as measured on standard benchmarks. However, performance in the open world remains underexplored, as current benchmarks cannot capture its full extent. Especially in safety-critical systems, it is crucial that 3D pose estimators are audited before deployment, and their sensitivity towards single factors or attributes occurring in the operational domain is thoroughly examined. Nevertheless, we currently lack a benchmark that would enable such fine-grained analysis. We thus present STAGE, a GenAI data toolkit for auditing 3D human pose estimators. We enable a text-to-image model to control the 3D human body pose in the generated image. This allows us to create customized annotated data covering a wide range of open-world attributes. We leverage STAGE and generate a series of benchmarks to audit the sensitivity of popular pose estimators towards attributes such as gender, ethnicity, age, clothing, location, and weather. Our results show that the presence of such naturally occurring attributes can cause severe degradation in the performance of pose estimators and leads us to question if they are ready for open-world deployment.
Authors: Xiangchen Yin, Donglin Di, Lei Fan, Hao Li, Chen Wei, Xiaofei Gou, Yang Song, Xiao Sun, Xun Yang
Abstract: Recent methods using diffusion models have made significant progress in human image generation with various additional controls such as pose priors. However, existing approaches still struggle to generate high-quality images with consistent pose alignment, resulting in unsatisfactory outputs. In this paper, we propose a framework delving into the graph relations of pose priors to provide control information for human image generation. The main idea is to establish a graph topological structure between the pose priors and latent representation of diffusion models to capture the intrinsic associations between different pose parts. A Progressive Graph Integrator (PGI) is designed to learn the spatial relationships of the pose priors with the graph structure, adopting a hierarchical strategy within an Adapter to gradually propagate information across different pose parts. A pose perception loss is further introduced based on a pretrained pose estimation network to minimize the pose differences. Extensive qualitative and quantitative experiments conducted on the Human-Art and LAION-Human datasets demonstrate that our model achieves superior performance, with a 9.98% increase in pose average precision compared to the latest benchmark model. The code is released on *******.
Authors: Kevin Raj, Christopher Wewer, Raza Yunus, Eddy Ilg, Jan Eric Lenssen
Abstract: We introduce Spurfies, a novel method for sparse-view surface reconstruction that disentangles appearance and geometry information to utilize local geometry priors trained on synthetic data. Recent research heavily focuses on 3D reconstruction using dense multi-view setups, typically requiring hundreds of images. However, these methods often struggle with few-view scenarios. Existing sparse-view reconstruction techniques often rely on multi-view stereo networks that need to learn joint priors for geometry and appearance from a large amount of data. In contrast, we introduce a neural point representation that disentangles geometry and appearance to train a local geometry prior using a subset of the synthetic ShapeNet dataset only. During inference, we utilize this surface prior as additional constraint for surface and appearance reconstruction from sparse input views via differentiable volume rendering, restricting the space of possible solutions. We validate the effectiveness of our method on the DTU dataset and demonstrate that it outperforms previous state of the art by 35% in surface quality while achieving competitive novel view synthesis quality. Moreover, in contrast to previous works, our method can be applied to larger, unbounded scenes, such as Mip-NeRF 360.
Authors: Yuchen Che, Ryo Furukawa, Asako Kanezaki
Abstract: Category-level articulated object pose estimation focuses on the pose estimation of unknown articulated objects within known categories. Despite its significance, this task remains challenging due to the varying shapes and poses of objects, expensive dataset annotation costs, and complex real-world environments. In this paper, we propose a novel self-supervised approach that leverages a single-frame point cloud to solve this task. Our model consistently generates reconstruction with a canonical pose and joint state for the entire input object, and it estimates object-level poses that reduce overall pose variance and part-level poses that align each part of the input with its corresponding part of the reconstruction. Experimental results demonstrate that our approach significantly outperforms previous self-supervised methods and is comparable to the state-of-the-art supervised methods. To assess the performance of our model in real-world scenarios, we also introduce a new real-world articulated object benchmark dataset.
Authors: Eduarda Caldeira, Jaime S. Cardoso, Ana F. Sequeira, Pedro C. Neto
Abstract: As in school, one teacher to cover all subjects is insufficient to distill equally robust information to a student. Hence, each subject is taught by a highly specialised teacher. Following a similar philosophy, we propose a multiple specialized teacher framework to distill knowledge to a student network. In our approach, directed at face recognition use cases, we train four teachers on one specific ethnicity, leading to four highly specialized and biased teachers. Our strategy learns a project of these four teachers into a common space and distill that information to a student network. Our results highlighted increased performance and reduced bias for all our experiments. In addition, we further show that having biased/specialized teachers is crucial by showing that our approach achieves better results than when knowledge is distilled from four teachers trained on balanced datasets. Our approach represents a step forward to the understanding of the importance of ethnicity-specific features.
Authors: Yangxiang Zhang, Yuezun Li, Ao Luo, Jiaran Zhou, Junyu Dong
Abstract: With the rise in popularity of portable devices, the spread of falsified media on social platforms has become rampant. This necessitates the timely identification of authentic content. However, most advanced detection methods are computationally heavy, hindering their real-time application. In this paper, we describe an efficient two-stream architecture for real-time image manipulation detection. Our method consists of two-stream branches targeting the cognitive and inspective perspectives. In the cognitive branch, we propose efficient wavelet-guided Transformer blocks to capture the global manipulation traces related to frequency. This block contains an interactive wavelet-guided self-attention module that integrates wavelet transformation with efficient attention design, interacting with the knowledge from the inspective branch. The inspective branch consists of simple convolutions that capture fine-grained traces and interact bidirectionally with Transformer blocks to provide mutual support. Our method is lightweight ($\sim$ 8M) but achieves competitive performance compared to many other counterparts, demonstrating its efficacy in image manipulation detection and its potential for portable integration.
Authors: Ishwar B Balappanawar, Ashmit Chamoli, Ruwan Wickramarachchi, Aditya Mishra, Ponnurangam Kumaraguru, Amit P. Sheth
Abstract: Distracted driving is a leading cause of road accidents globally. Identification of distracted driving involves reliably detecting and classifying various forms of driver distraction (e.g., texting, eating, or using in-car devices) from in-vehicle camera feeds to enhance road safety. This task is challenging due to the need for robust models that can generalize to a diverse set of driver behaviors without requiring extensive annotated datasets. In this paper, we propose KiD3, a novel method for distracted driver detection (DDD) by infusing auxiliary knowledge about semantic relations between entities in a scene and the structural configuration of the driver's pose. Specifically, we construct a unified framework that integrates the scene graphs, and driver pose information with the visual cues in video frames to create a holistic representation of the driver's actions.Our results indicate that KiD3 achieves a 13.64% accuracy improvement over the vision-only baseline by incorporating such auxiliary knowledge with visual information.
Authors: Ripon Kumar Saha, Esen Salcin, Jihoo Kim, Joseph Smith, Suren Jayasuriya
Abstract: Images captured from a long distance suffer from dynamic image distortion due to turbulent flow of air cells with random temperatures, and thus refractive indices. This phenomenon, known as image dancing, is commonly characterized by its refractive-index structure constant $C_n^2$ as a measure of the turbulence strength. For many applications such as atmospheric forecast model, long-range/astronomy imaging, and aviation safety, optical communication technology, $C_n^2$ estimation is critical for accurately sensing the turbulent environment. Previous methods for $C_n^2$ estimation include estimation from meteorological data (temperature, relative humidity, wind shear, etc.) for single-point measurements, two-ended pathlength measurements from optical scintillometer for path-averaged $C_n^2$, and more recently estimating $C_n^2$ from passive video cameras for low cost and hardware complexity. In this paper, we present a comparative analysis of classical image gradient methods for $C_n^2$ estimation and modern deep learning-based methods leveraging convolutional neural networks. To enable this, we collect a dataset of video capture along with reference scintillometer measurements for ground truth, and we release this unique dataset to the scientific community. We observe that deep learning methods can achieve higher accuracy when trained on similar data, but suffer from generalization errors to other, unseen imagery as compared to classical methods. To overcome this trade-off, we present a novel physics-based network architecture that combines learned convolutional layers with a differentiable image gradient method that maintains high accuracy while being generalizable across image datasets.
Authors: Ryota Tanaka, Tomohiro Suzuki, Keisuke Fujii
Abstract: Understanding human actions from videos is essential in many domains, including sports. In figure skating, technical judgments are performed by watching skaters' 3D movements, and its part of the judging procedure can be regarded as a Temporal Action Segmentation (TAS) task. TAS tasks in figure skating that automatically assign temporal semantics to video are actively researched. However, there is a lack of datasets and effective methods for TAS tasks requiring 3D pose data. In this study, we first created the FS-Jump3D dataset of complex and dynamic figure skating jumps using optical markerless motion capture. We also propose a new fine-grained figure skating jump TAS dataset annotation method with which TAS models can learn jump procedures. In the experimental results, we validated the usefulness of 3D pose features as input and the fine-grained dataset for the TAS model in figure skating. FS-Jump3D Dataset is available at https://github.com/ryota-skating/FS-Jump3D.
Authors: Rohit Venkata Sai Dulam, Chandra Kambhamettu
Abstract: Salient Object Detection (SOD) has traditionally relied on feature refinement modules that utilize the features of an ImageNet pre-trained backbone. However, this approach limits the possibility of pre-training the entire network because of the distinct nature of SOD and image classification. Additionally, the architecture of these backbones originally built for Image classification is sub-optimal for a dense prediction task like SOD. To address these issues, we propose a novel encoder-decoder-style neural network called SODAWideNet++ that is designed explicitly for SOD. Inspired by the vision transformers ability to attain a global receptive field from the initial stages, we introduce the Attention Guided Long Range Feature Extraction (AGLRFE) module, which combines large dilated convolutions and self-attention. Specifically, we use attention features to guide long-range information extracted by multiple dilated convolutions, thus taking advantage of the inductive biases of a convolution operation and the input dependency brought by self-attention. In contrast to the current paradigm of ImageNet pre-training, we modify 118K annotated images from the COCO semantic segmentation dataset by binarizing the annotations to pre-train the proposed model end-to-end. Further, we supervise the background predictions along with the foreground to push our model to generate accurate saliency predictions. SODAWideNet++ performs competitively on five different datasets while only containing 35% of the trainable parameters compared to the state-of-the-art models. The code and pre-computed saliency maps are provided at https://github.com/VimsLab/SODAWideNetPlusPlus.
Authors: Yongjie Fu, Anmol Jain, Xuan Di, Xu Chen, Zhaobin Mo
Abstract: The advancement of autonomous driving technologies necessitates increasingly sophisticated methods for understanding and predicting real-world scenarios. Vision language models (VLMs) are emerging as revolutionary tools with significant potential to influence autonomous driving. In this paper, we propose the DriveGenVLM framework to generate driving videos and use VLMs to understand them. To achieve this, we employ a video generation framework grounded in denoising diffusion probabilistic models (DDPM) aimed at predicting real-world video sequences. We then explore the adequacy of our generated videos for use in VLMs by employing a pre-trained model known as Efficient In-context Learning on Egocentric Videos (EILEV). The diffusion model is trained with the Waymo open dataset and evaluated using the Fr\'echet Video Distance (FVD) score to ensure the quality and realism of the generated videos. Corresponding narrations are provided by EILEV for these generated videos, which may be beneficial in the autonomous driving domain. These narrations can enhance traffic scene understanding, aid in navigation, and improve planning capabilities. The integration of video generation with VLMs in the DriveGenVLM framework represents a significant step forward in leveraging advanced AI models to address complex challenges in autonomous driving.
Authors: Farnoosh Arefi, Amir M. Mansourian, Shohreh Kasaei
Abstract: The performance of Video Instance Segmentation (VIS) methods has improved significantly with the advent of transformer networks. However, these networks often face challenges in training due to the high annotation cost. To address this, unsupervised and weakly-supervised methods have been developed to reduce the dependency on annotations. This work introduces a novel weakly-supervised method called Eigen-cluster VIS that, without requiring any mask annotations, achieves competitive accuracy compared to other VIS approaches. This method is based on two key innovations: a Temporal Eigenvalue Loss (TEL) and a clip-level Quality Cluster Coefficient (QCC). The TEL ensures temporal coherence by leveraging the eigenvalues of the Laplacian matrix derived from graph adjacency matrices. By minimizing the mean absolute error (MAE) between the eigenvalues of adjacent frames, this loss function promotes smooth transitions and stable segmentation boundaries over time, reducing temporal discontinuities and improving overall segmentation quality. The QCC employs the K-means method to ensure the quality of spatio-temporal clusters without relying on ground truth masks. Using the Davies-Bouldin score, the QCC provides an unsupervised measure of feature discrimination, allowing the model to self-evaluate and adapt to varying object distributions, enhancing robustness during the testing phase. These enhancements are computationally efficient and straightforward, offering significant performance gains without additional annotated data. The proposed Eigen-Cluster VIS method is evaluated on the YouTube-VIS 2019/2021 and OVIS datasets, demonstrating that it effectively narrows the performance gap between the fully-supervised and weakly-supervised VIS approaches. The code is available on: https://github.com/farnooshar/EigenClusterVIS
Authors: Emilia Szymanska, Mihai Dusmanu, Jan-Willem Buurlage, Mahdi Rad, Marc Pollefeys
Abstract: Answering questions about the spatial properties of the environment poses challenges for existing language and vision foundation models due to a lack of understanding of the 3D world notably in terms of relationships between objects. To push the field forward, multiple 3D Q&A datasets were proposed which, overall, provide a variety of questions, but they individually focus on particular aspects of 3D reasoning or are limited in terms of data modalities. To address this, we present Space3D-Bench - a collection of 1000 general spatial questions and answers related to scenes of the Replica dataset which offers a variety of data modalities: point clouds, posed RGB-D images, navigation meshes and 3D object detections. To ensure that the questions cover a wide range of 3D objectives, we propose an indoor spatial questions taxonomy inspired by geographic information systems and use it to balance the dataset accordingly. Moreover, we provide an assessment system that grades natural language responses based on predefined ground-truth answers by leveraging a Vision Language Model's comprehension of both text and images to compare the responses with ground-truth textual information or relevant visual data. Finally, we introduce a baseline called RAG3D-Chat integrating the world understanding of foundation models with rich context retrieval, achieving an accuracy of 67% on the proposed dataset.
Authors: Lei Tan, Pingyang Dai, Jie Chen, Liujuan Cao, Yongjian Wu, Rongrong Ji
Abstract: Extracting robust feature representation is critical for object re-identification to accurately identify objects across non-overlapping cameras. Although having a strong representation ability, the Vision Transformer (ViT) tends to overfit on most distinct regions of training data, limiting its generalizability and attention to holistic object features. Meanwhile, due to the structural difference between CNN and ViT, fine-grained strategies that effectively address this issue in CNN do not continue to be successful in ViT. To address this issue, by observing the latent diverse representation hidden behind the multi-head attention, we present PartFormer, an innovative adaptation of ViT designed to overcome the granularity limitations in object Re-ID tasks. The PartFormer integrates a Head Disentangling Block (HDB) that awakens the diverse representation of multi-head self-attention without the typical loss of feature richness induced by concatenation and FFN layers post-attention. To avoid the homogenization of attention heads and promote robust part-based feature learning, two head diversity constraints are imposed: attention diversity constraint and correlation diversity constraint. These constraints enable the model to exploit diverse and discriminative feature representations from different attention heads. Comprehensive experiments on various object Re-ID benchmarks demonstrate the superiority of the PartFormer. Specifically, our framework significantly outperforms state-of-the-art by 2.4\% mAP scores on the most challenging MSMT17 dataset.
Authors: Zhirui Gao, Renjiao Yi, Chenyang Zhu, Ke Zhuang, Wei Chen, Kai Xu
Abstract: Radiance fields including NeRFs and 3D Gaussians demonstrate great potential in high-fidelity rendering and scene reconstruction, while they require a substantial number of posed images as inputs. COLMAP is frequently employed for preprocessing to estimate poses, while it necessitates a large number of feature matches to operate effectively, and it struggles with scenes characterized by sparse features, large baselines between images, or a limited number of input images. We aim to tackle few-view NeRF reconstruction using only 3 to 6 unposed scene images. Traditional methods often use calibration boards but they are not common in images. We propose a novel idea of utilizing everyday objects, commonly found in both images and real life, as "pose probes". The probe object is automatically segmented by SAM, whose shape is initialized from a cube. We apply a dual-branch volume rendering optimization (object NeRF and scene NeRF) to constrain the pose optimization and jointly refine the geometry. Specifically, object poses of two views are first estimated by PnP matching in an SDF representation, which serves as initial poses. PnP matching, requiring only a few features, is suitable for feature-sparse scenes. Additional views are incrementally incorporated to refine poses from preceding views. In experiments, PoseProbe achieves state-of-the-art performance in both pose estimation and novel view synthesis across multiple datasets. We demonstrate its effectiveness, particularly in few-view and large-baseline scenes where COLMAP struggles. In ablations, using different objects in a scene yields comparable performance.
Authors: Moreno D'Inc\`a, Elia Peruzzo, Massimiliano Mancini, Xingqian Xu, Humphrey Shi, Nicu Sebe
Abstract: Recent progress in Text-to-Image (T2I) generative models has enabled high-quality image generation. As performance and accessibility increase, these models are gaining significant attraction and popularity: ensuring their fairness and safety is a priority to prevent the dissemination and perpetuation of biases. However, existing studies in bias detection focus on closed sets of predefined biases (e.g., gender, ethnicity). In this paper, we propose a general framework to identify, quantify, and explain biases in an open set setting, i.e. without requiring a predefined set. This pipeline leverages a Large Language Model (LLM) to propose biases starting from a set of captions. Next, these captions are used by the target generative model for generating a set of images. Finally, Vision Question Answering (VQA) is leveraged for bias evaluation. We show two variations of this framework: OpenBias and GradBias. OpenBias detects and quantifies biases, while GradBias determines the contribution of individual prompt words on biases. OpenBias effectively detects both well-known and novel biases related to people, objects, and animals and highly aligns with existing closed-set bias detection methods and human judgment. GradBias shows that neutral words can significantly influence biases and it outperforms several baselines, including state-of-the-art foundation models. Code available here: https://github.com/Moreno98/GradBias.
Authors: Anisha Jain
Abstract: Creating editable videos that depict complex interactions between multiple objects in various artistic styles has long been a challenging task in filmmaking. Progress is often hampered by the scarcity of data sets that contain paired text descriptions and corresponding videos that showcase these interactions. This paper introduces a novel depth-conditioning approach that significantly advances this field by enabling the generation of coherent and diverse videos from just a single text-video pair using a pre-trained depth-aware Text-to-Image (T2I) model. Our method fine-tunes the pre-trained model to capture continuous motion by employing custom-designed spatial and temporal attention mechanisms. During inference, we use the DDIM inversion to provide structural guidance for video generation. This innovative technique allows for continuously controllable depth in videos, facilitating the generation of multiobject interactions while maintaining the concept generation and compositional strengths of the original T2I model across various artistic styles, such as photorealism, animation, and impressionism.
Authors: Yufeng Zhou, Wenming Cao
Abstract: The integration of Convolutional Neural Network (ConvNet) and Transformer has emerged as a strong candidate for image registration, leveraging the strengths of both models and a large parameter space. However, this hybrid model, treating brain MRI volumes as grid or sequence structures, faces challenges in accurately representing anatomical connectivity, diverse brain regions, and vital connections contributing to the brain's internal architecture. Concerns also arise regarding the computational expense and GPU memory usage associated with this model. To tackle these issues, a lightweight hybrid sparse graph attention network (H-SGANet) has been developed. This network incorporates a central mechanism, Sparse Graph Attention (SGA), based on a Vision Graph Neural Network (ViG) with predetermined anatomical connections. The SGA module expands the model's receptive field and seamlessly integrates into the network. To further amplify the advantages of the hybrid network, the Separable Self-Attention (SSA) is employed as an enhanced token mixer, integrated with depth-wise convolution to constitute SSAFormer. This strategic integration is designed to more effectively extract long-range dependencies. As a hybrid ConvNet-ViG-Transformer model, H-SGANet offers threefold benefits for volumetric medical image registration. It optimizes fixed and moving images concurrently through a hybrid feature fusion layer and an end-to-end learning framework. Compared to VoxelMorph, a model with a similar parameter count, H-SGANet demonstrates significant performance enhancements of 3.5% and 1.5% in Dice score on the OASIS dataset and LPBA40 dataset, respectively.
Authors: Jihwan Kim, Miso Lee, Cheol-Ho Cho, Jihyun Lee, Jae-Pil Heo
Abstract: Temporal Action Detection (TAD) is fundamental yet challenging for real-world video applications. Leveraging the unique benefits of transformers, various DETR-based approaches have been adopted in TAD. However, it has recently been identified that the attention collapse in self-attention causes the performance degradation of DETR for TAD. Building upon previous research, this paper newly addresses the attention collapse problem in cross-attention within DETR-based TAD methods. Moreover, our findings reveal that cross-attention exhibits patterns distinct from predictions, indicating a short-cut phenomenon. To resolve this, we propose a new framework, Prediction-Feedback DETR (Pred-DETR), which utilizes predictions to restore the collapse and align the cross- and self-attention with predictions. Specifically, we devise novel prediction-feedback objectives using guidance from the relations of the predictions. As a result, Pred-DETR significantly alleviates the collapse and achieves state-of-the-art performance among DETR-based methods on various challenging benchmarks including THUMOS14, ActivityNet-v1.3, HACS, and FineAction.
Authors: Shiwei Wu, Joya Chen, Kevin Qinghong Lin, Qimeng Wang, Yan Gao, Qianli Xu, Tong Xu, Yao Hu, Enhong Chen, Mike Zheng Shou
Abstract: A well-known dilemma in large vision-language models (e.g., GPT-4, LLaVA) is that while increasing the number of vision tokens generally enhances visual understanding, it also significantly raises memory and computational costs, especially in long-term, dense video frame streaming scenarios. Although learnable approaches like Q-Former and Perceiver Resampler have been developed to reduce the vision token burden, they overlook the context causally modeled by LLMs (i.e., key-value cache), potentially leading to missed visual cues when addressing user queries. In this paper, we introduce a novel approach to reduce vision compute by leveraging redundant vision tokens "skipping layers" rather than decreasing the number of vision tokens. Our method, VideoLLM-MoD, is inspired by mixture-of-depths LLMs and addresses the challenge of numerous vision tokens in long-term or streaming video. Specifically, for each transformer layer, we learn to skip the computation for a high proportion (e.g., 80\%) of vision tokens, passing them directly to the next layer. This approach significantly enhances model efficiency, achieving approximately \textasciitilde42\% time and \textasciitilde30\% memory savings for the entire training. Moreover, our method reduces the computation in the context and avoid decreasing the vision tokens, thus preserving or even improving performance compared to the vanilla model. We conduct extensive experiments to demonstrate the effectiveness of VideoLLM-MoD, showing its state-of-the-art results on multiple benchmarks, including narration, forecasting, and summarization tasks in COIN, Ego4D, and Ego-Exo4D datasets.
Authors: Hongjun Wang, Sagar Vaze, Kai Han
Abstract: Detecting test-time distribution shift has emerged as a key capability for safely deployed machine learning models, with the question being tackled under various guises in recent years. In this paper, we aim to provide a consolidated view of the two largest sub-fields within the community: out-of-distribution (OOD) detection and open-set recognition (OSR). In particular, we aim to provide rigorous empirical analysis of different methods across settings and provide actionable takeaways for practitioners and researchers. Concretely, we make the following contributions: (i) We perform rigorous cross-evaluation between state-of-the-art methods in the OOD detection and OSR settings and identify a strong correlation between the performances of methods for them; (ii) We propose a new, large-scale benchmark setting which we suggest better disentangles the problem tackled by OOD detection and OSR, re-evaluating state-of-the-art OOD detection and OSR methods in this setting; (iii) We surprisingly find that the best performing method on standard benchmarks (Outlier Exposure) struggles when tested at scale, while scoring rules which are sensitive to the deep feature magnitude consistently show promise; and (iv) We conduct empirical analysis to explain these phenomena and highlight directions for future research. Code: \url{https://github.com/Visual-AI/Dissect-OOD-OSR}
Authors: Ziyu Chen, Jiawei Yang, Jiahui Huang, Riccardo de Lutio, Janick Martinez Esturo, Boris Ivanovic, Or Litany, Zan Gojcic, Sanja Fidler, Marco Pavone, Li Song, Yue Wang
Abstract: We introduce OmniRe, a holistic approach for efficiently reconstructing high-fidelity dynamic urban scenes from on-device logs. Recent methods for modeling driving sequences using neural radiance fields or Gaussian Splatting have demonstrated the potential of reconstructing challenging dynamic scenes, but often overlook pedestrians and other non-vehicle dynamic actors, hindering a complete pipeline for dynamic urban scene reconstruction. To that end, we propose a comprehensive 3DGS framework for driving scenes, named OmniRe, that allows for accurate, full-length reconstruction of diverse dynamic objects in a driving log. OmniRe builds dynamic neural scene graphs based on Gaussian representations and constructs multiple local canonical spaces that model various dynamic actors, including vehicles, pedestrians, and cyclists, among many others. This capability is unmatched by existing methods. OmniRe allows us to holistically reconstruct different objects present in the scene, subsequently enabling the simulation of reconstructed scenarios with all actors participating in real-time (~60Hz). Extensive evaluations on the Waymo dataset show that our approach outperforms prior state-of-the-art methods quantitatively and qualitatively by a large margin. We believe our work fills a critical gap in driving reconstruction.
Authors: Simone Foti, Stefanos Zafeiriou, Tolga Birdal
Abstract: Seams, distortions, wasted UV space, vertex-duplication, and varying resolution over the surface are the most prominent issues of the standard UV-based texturing of meshes. These issues are particularly acute when automatic UV-unwrapping techniques are used. For this reason, instead of generating textures in automatically generated UV-planes like most state-of-the-art methods, we propose to represent textures as coloured point-clouds whose colours are generated by a denoising diffusion probabilistic model constrained to operate on the surface of 3D objects. Our sampling and resolution agnostic generative model heavily relies on heat diffusion over the surface of the meshes for spatial communication between points. To enable processing of arbitrarily sampled point-cloud textures and ensure long-distance texture consistency we introduce a fast re-sampling of the mesh spectral properties used during the heat diffusion and introduce a novel heat-diffusion-based self-attention mechanism. Our code and pre-trained models are available at github.com/simofoti/UV3-TeD.
Authors: Peng Xing, Haofan Wang, Yanpeng Sun, Qixun Wang, Xu Bai, Hao Ai, Renyuan Huang, Zechao Li
Abstract: The diffusion model has shown exceptional capabilities in controlled image generation, which has further fueled interest in image style transfer. Existing works mainly focus on training free-based methods (e.g., image inversion) due to the scarcity of specific data. In this study, we present a data construction pipeline for content-style-stylized image triplets that generates and automatically cleanses stylized data triplets. Based on this pipeline, we construct a dataset IMAGStyle, the first large-scale style transfer dataset containing 210k image triplets, available for the community to explore and research. Equipped with IMAGStyle, we propose CSGO, a style transfer model based on end-to-end training, which explicitly decouples content and style features employing independent feature injection. The unified CSGO implements image-driven style transfer, text-driven stylized synthesis, and text editing-driven stylized synthesis. Extensive experiments demonstrate the effectiveness of our approach in enhancing style control capabilities in image generation. Additional visualization and access to the source code can be located on the project page: \url{https://csgo-gen.github.io/}.
Authors: Fangfu Liu, Wenqiang Sun, Hanyang Wang, Yikai Wang, Haowen Sun, Junliang Ye, Jun Zhang, Yueqi Duan
Abstract: Advancements in 3D scene reconstruction have transformed 2D images from the real world into 3D models, producing realistic 3D results from hundreds of input photos. Despite great success in dense-view reconstruction scenarios, rendering a detailed scene from insufficient captured views is still an ill-posed optimization problem, often resulting in artifacts and distortions in unseen areas. In this paper, we propose ReconX, a novel 3D scene reconstruction paradigm that reframes the ambiguous reconstruction challenge as a temporal generation task. The key insight is to unleash the strong generative prior of large pre-trained video diffusion models for sparse-view reconstruction. However, 3D view consistency struggles to be accurately preserved in directly generated video frames from pre-trained models. To address this, given limited input views, the proposed ReconX first constructs a global point cloud and encodes it into a contextual space as the 3D structure condition. Guided by the condition, the video diffusion model then synthesizes video frames that are both detail-preserved and exhibit a high degree of 3D consistency, ensuring the coherence of the scene from various perspectives. Finally, we recover the 3D scene from the generated video through a confidence-aware 3D Gaussian Splatting optimization scheme. Extensive experiments on various real-world datasets show the superiority of our ReconX over state-of-the-art methods in terms of quality and generalizability.
Authors: Ziyu Guo, Renrui Zhang, Xiangyang Zhu, Chengzhuo Tong, Peng Gao, Chunyuan Li, Pheng-Ann Heng
Abstract: We introduce SAM2Point, a preliminary exploration adapting Segment Anything Model 2 (SAM 2) for zero-shot and promptable 3D segmentation. SAM2Point interprets any 3D data as a series of multi-directional videos, and leverages SAM 2 for 3D-space segmentation, without further training or 2D-3D projection. Our framework supports various prompt types, including 3D points, boxes, and masks, and can generalize across diverse scenarios, such as 3D objects, indoor scenes, outdoor environments, and raw sparse LiDAR. Demonstrations on multiple 3D datasets, e.g., Objaverse, S3DIS, ScanNet, Semantic3D, and KITTI, highlight the robust generalization capabilities of SAM2Point. To our best knowledge, we present the most faithful implementation of SAM in 3D, which may serve as a starting point for future research in promptable 3D segmentation. Online Demo: https://huggingface.co/spaces/ZiyuG/SAM2Point . Code: https://github.com/ZiyuGuo99/SAM2Point .
URLs: https://huggingface.co/spaces/ZiyuG/SAM2Point, https://github.com/ZiyuGuo99/SAM2Point
Authors: Noor Hussein, Fahad Shamshad, Muzammal Naseer, Karthik Nandakumar
Abstract: Medical vision-language models (Med-VLMs) trained on large datasets of medical image-text pairs and later fine-tuned for specific tasks have emerged as a mainstream paradigm in medical image analysis. However, recent studies have highlighted the susceptibility of these Med-VLMs to adversarial attacks, raising concerns about their safety and robustness. Randomized smoothing is a well-known technique for turning any classifier into a model that is certifiably robust to adversarial perturbations. However, this approach requires retraining the Med-VLM-based classifier so that it classifies well under Gaussian noise, which is often infeasible in practice. In this paper, we propose a novel framework called PromptSmooth to achieve efficient certified robustness of Med-VLMs by leveraging the concept of prompt learning. Given any pre-trained Med-VLM, PromptSmooth adapts it to handle Gaussian noise by learning textual prompts in a zero-shot or few-shot manner, achieving a delicate balance between accuracy and robustness, while minimizing the computational overhead. Moreover, PromptSmooth requires only a single model to handle multiple noise levels, which substantially reduces the computational cost compared to traditional methods that rely on training a separate model for each noise level. Comprehensive experiments based on three Med-VLMs and across six downstream datasets of various imaging modalities demonstrate the efficacy of PromptSmooth. Our code and models are available at https://github.com/nhussein/promptsmooth.
Authors: Georgios Paschalidis, Romana Wilschut, Dimitrije Anti\'c, Omid Taheri, Dimitrios Tzionas
Abstract: Synthesizing 3D whole-bodies that realistically grasp objects is useful for animation, mixed reality, and robotics. This is challenging, because the hands and body need to look natural w.r.t. each other, the grasped object, as well as the local scene (i.e., a receptacle supporting the object). Only recent work tackles this, with a divide-and-conquer approach; it first generates a "guiding" right-hand grasp, and then searches for bodies that match this. However, the guiding-hand synthesis lacks controllability and receptacle awareness, so it likely has an implausible direction (i.e., a body can't match this without penetrating the receptacle) and needs corrections through major post-processing. Moreover, the body search needs exhaustive sampling and is expensive. These are strong limitations. We tackle these with a novel method called CWGrasp. Our key idea is that performing geometry-based reasoning "early on," instead of "too late," provides rich "control" signals for inference. To this end, CWGrasp first samples a plausible reaching-direction vector (used later for both the arm and hand) from a probabilistic model built via raycasting from the object and collision checking. Then, it generates a reaching body with a desired arm direction, as well as a "guiding" grasping hand with a desired palm direction that complies with the arm's one. Eventually, CWGrasp refines the body to match the "guiding" hand, while plausibly contacting the scene. Notably, generating already-compatible "parts" greatly simplifies the "whole." Moreover, CWGrasp uniquely tackles both right- and left-hand grasps. We evaluate on the GRAB and ReplicaGrasp datasets. CWGrasp outperforms baselines, at lower runtime and budget, while all components help performance. Code and models will be released.
Authors: Yu Lu, Kevin Bui, Roummel F. Marcia
Abstract: Matrix completion focuses on recovering missing or incomplete information in matrices. This problem arises in various applications, including image processing and network analysis. Previous research proposed Poisson matrix completion for count data with noise that follows a Poisson distribution, which assumes that the mean and variance are equal. Since overdispersed count data, whose variance is greater than the mean, is more likely to occur in realistic settings, we assume that the noise follows the negative binomial (NB) distribution, which can be more general than the Poisson distribution. In this paper, we introduce NB matrix completion by proposing a nuclear-norm regularized model that can be solved by proximal gradient descent. In our experiments, we demonstrate that the NB model outperforms Poisson matrix completion in various noise and missing data settings on real data.
Authors: Yu Lu, Kevin Bui, Roummel F. Marcia
Abstract: In many applications such as medical imaging, the measurement data represent counts of photons hitting a detector. Such counts in low-photon settings are often modeled using a Poisson distribution. However, this model assumes that the mean and variance of the signal's noise distribution are equal. For overdispersed data where the variance is greater than the mean, the negative binomial distribution is a more appropriate statistical model. In this paper, we propose an optimization approach for recovering images corrupted by overdispersed Poisson noise. In particular, we incorporate a weighted anisotropic-isotropic total variation regularizer, which avoids staircasing artifacts that are introduced by a regular total variation penalty. We use an alternating direction method of multipliers, where each subproblem has a closed-form solution. Numerical experiments demonstrate the effectiveness of our proposed approach, especially in very photon-limited settings.
Authors: Kaustubh Sadekar, David Maier, Atul Ingle
Abstract: Single-photon cameras present a promising avenue for high-resolution 3D imaging. They have ultra-high sensitivity -- down to individual photons -- and can record photon arrival times with extremely high (sub-nanosecond) resolution. Single-photon 3D cameras estimate the round-trip time of a laser pulse by forming equi-width (EW) histograms of detected photon timestamps. Acquiring and transferring such EW histograms requires high bandwidth and in-pixel memory, making SPCs less attractive in resource-constrained settings such as mobile devices and AR/VR headsets. In this work we propose a 3D sensing technique based on equi-depth (ED) histograms. ED histograms compress timestamp data more efficiently than EW histograms, reducing the bandwidth requirement. Moreover, to reduce the in-pixel memory requirement, we propose a lightweight algorithm to estimate ED histograms in an online fashion without explicitly storing the photon timestamps. This algorithm is amenable to future in-pixel implementations. We propose algorithms that process ED histograms to perform 3D computer-vision tasks of estimating scene distance maps and performing visual odometry under challenging conditions such as high ambient light. Our work paves the way towards lower bandwidth and reduced in-pixel memory requirements for SPCs, making them attractive for resource-constrained 3D vision applications. Project page: $\href{https://www.computational.camera/pedh}{https://www.computational.camera/pedh}$
URLs: https://www.computational.camera/pedh, https://www.computational.camera/pedh
Authors: Sizhe Hu, Wenming Wu, Yuntao Wang, Benzhu Xu, Liping Zheng
Abstract: Automating architectural floorplan design is vital for housing and interior design, offering a faster, cost-effective alternative to manual sketches by architects. However, existing methods, including rule-based and learning-based approaches, face challenges in design complexity and constrained generation with extensive post-processing, and tend to obvious geometric inconsistencies such as misalignment, overlap, and gaps. In this work, we propose a novel generative framework for vector floorplan design via structural graph generation, called GSDiff, focusing on wall junction generation and wall segment prediction to capture both geometric and semantic aspects of structural graphs. To improve the geometric rationality of generated structural graphs, we propose two innovative geometry enhancement methods. In wall junction generation, we propose a novel alignment loss function to improve geometric consistency. In wall segment prediction, we propose a random self-supervision method to enhance the model's perception of the overall geometric structure, thereby promoting the generation of reasonable geometric structures. Employing the diffusion model and the Transformer model, as well as the geometry enhancement strategies, our framework can generate wall junctions, wall segments and room polygons with structural and semantic information, resulting in structural graphs that accurately represent floorplans. Extensive experiments show that the proposed method surpasses existing techniques, enabling free generation and constrained generation, marking a shift towards structure generation in architectural design.
Authors: Xiaofeng Deng, Defu Chen, Bowen Liu, Xiwan Zhang, Haixia Qiu, Wu Yuan, Hongliang Ren
Abstract: Accurate classification of port wine stains (PWS, vascular malformations present at birth), is critical for subsequent treatment planning. However, the current method of classifying PWS based on the external skin appearance rarely reflects the underlying angiopathological heterogeneity of PWS lesions, resulting in inconsistent outcomes with the common vascular-targeted photodynamic therapy (V-PDT) treatments. Conversely, optical coherence tomography angiography (OCTA) is an ideal tool for visualizing the vascular malformations of PWS. Previous studies have shown no significant correlation between OCTA quantitative metrics and the PWS subtypes determined by the current classification approach. This study proposes a new classification approach for PWS using both OCT and OCTA. By examining the hypodermic histopathology and vascular structure of PWS, we have devised a fine-grained classification method that subdivides PWS into five distinct types. To assess the angiopathological differences of various PWS subtypes, we have analyzed six metrics related to vascular morphology and depth information of PWS lesions. The five PWS types present significant differences across all metrics compared to the conventional subtypes. Our findings suggest that an angiopathology-based classification accurately reflects the heterogeneity in PWS lesions. This research marks the first attempt to classify PWS based on angiopathology, potentially guiding more effective subtyping and treatment strategies for PWS.
Authors: Conghan Yue, Zhengwei Peng, Junlong Ma, Dongyu Zhang
Abstract: Image restoration refers to the process of restoring a damaged low-quality image back to its corresponding high-quality image. Typically, we use convolutional neural networks to directly learn the mapping from low-quality images to high-quality images achieving image restoration. Recently, a special type of diffusion bridge model has achieved more advanced results in image restoration. It can transform the direct mapping from low-quality to high-quality images into a diffusion process, restoring low-quality images through a reverse process. However, the current diffusion bridge restoration models do not emphasize the idea of conditional control, which may affect performance. This paper introduces the ECDB model enhancing the control of the diffusion bridge with low-quality images as conditions. Moreover, in response to the characteristic of diffusion models having low denoising level at larger values of \(\bm t \), we also propose a Conditional Fusion Schedule, which more effectively handles the conditional feature information of various modules. Experimental results prove that the ECDB model has achieved state-of-the-art results in many image restoration tasks, including deraining, inpainting and super-resolution. Code is avaliable at https://github.com/Hammour-steak/ECDB.
Authors: Guangyi Zhang, Hanlei Li, Yunlong Cai, Qiyu Hu, Guanding Yu, Runmin Zhang
Abstract: In this paper, we introduce an innovative hierarchical joint source-channel coding (HJSCC) framework for image transmission, utilizing a hierarchical variational autoencoder (VAE). Our approach leverages a combination of bottom-up and top-down paths at the transmitter to autoregressively generate multiple hierarchical representations of the original image. These representations are then directly mapped to channel symbols for transmission by the JSCC encoder. We extend this framework to scenarios with a feedback link, modeling transmission over a noisy channel as a probabilistic sampling process and deriving a novel generative formulation for JSCC with feedback. Compared with existing approaches, our proposed HJSCC provides enhanced adaptability by dynamically adjusting transmission bandwidth, encoding these representations into varying amounts of channel symbols. Additionally, we introduce a rate attention module to guide the JSCC encoder in optimizing its encoding strategy based on prior information. Extensive experiments on images of varying resolutions demonstrate that our proposed model outperforms existing baselines in rate-distortion performance and maintains robustness against channel noise.
Authors: Kirsten W. H. Maas, Danny Ruijters, Anna Vilanova, Nicola Pezzotti
Abstract: Dynamic three-dimensional (4D) reconstruction from two-dimensional X-ray coronary angiography (CA) remains a significant clinical problem. Challenges include sparse-view settings, intra-scan motion, and complex vessel morphology such as structure sparsity and background occlusion. Existing CA reconstruction methods often require extensive user interaction or large training datasets. On the other hand, Neural Radiance Field (NeRF), a promising deep learning technique, has successfully reconstructed high-fidelity static scenes for natural and medical scenes. Recent work, however, identified that sparse-views, background occlusion, and dynamics still pose a challenge when applying NeRF in the X-ray angiography context. Meanwhile, many successful works for natural scenes propose regularization for sparse-view reconstruction or scene decomposition to handle dynamics. However, these techniques do not directly translate to the CA context, where both challenges and background occlusion are significant. This paper introduces NeRF-CA, the first step toward a 4D CA reconstruction method that achieves reconstructions from sparse coronary angiograms with cardiac motion. We leverage the motion of the coronary artery to decouple the scene into a dynamic coronary artery component and static background. We combine this scene decomposition with tailored regularization techniques. These techniques enforce the separation of the coronary artery from the background by enforcing dynamic structure sparsity and scene smoothness. By uniquely combining these approaches, we achieve 4D reconstructions from as few as four angiogram sequences. This setting aligns with clinical workflows while outperforming state-of-the-art X-ray sparse-view NeRF reconstruction techniques. We validate our approach quantitatively and qualitatively using 4D phantom datasets and ablation studies.
Authors: Jiahang Cao, Hanzhong Guo, Ziqing Wang, Deming Zhou, Hao Cheng, Qiang Zhang, Renjing Xu
Abstract: Recent years have witnessed Spiking Neural Networks (SNNs) gaining attention for their ultra-low energy consumption and high biological plausibility compared with traditional Artificial Neural Networks (ANNs). Despite their distinguished properties, the application of SNNs in the computationally intensive field of image generation is still under exploration. In this paper, we propose the Spiking Diffusion Models (SDMs), an innovative family of SNN-based generative models that excel in producing high-quality samples with significantly reduced energy consumption. In particular, we propose a Temporal-wise Spiking Mechanism (TSM) that allows SNNs to capture more temporal features from a bio-plasticity perspective. In addition, we propose a threshold-guided strategy that can further improve the performances by up to 16.7% without any additional training. We also make the first attempt to use the ANN-SNN approach for SNN-based generation tasks. Extensive experimental results reveal that our approach not only exhibits comparable performance to its ANN counterpart with few spiking time steps, but also outperforms previous SNN-based generative models by a large margin. Moreover, we also demonstrate the high-quality generation ability of SDM on large-scale datasets, e.g., LSUN bedroom. This development marks a pivotal advancement in the capabilities of SNN-based generation, paving the way for future research avenues to realize low-energy and low-latency generative applications. Our code is available at https://github.com/AndyCao1125/SDM.
Authors: Roman Bruch, Mario Vitacolonna, Elina N\"urnberg, Simeon Sauer, R\"udiger Rudolf, Markus Reischl
Abstract: Biomedical research increasingly relies on 3D cell culture models and AI-based analysis can potentially facilitate a detailed and accurate feature extraction on a single-cell level. However, this requires for a precise segmentation of 3D cell datasets, which in turn demands high-quality ground truth for training. Manual annotation, the gold standard for ground truth data, is too time-consuming and thus not feasible for the generation of large 3D training datasets. To address this, we present a novel framework for generating 3D training data, which integrates biophysical modeling for realistic cell shape and alignment. Our approach allows the in silico generation of coherent membrane and nuclei signals, that enable the training of segmentation models utilizing both channels for improved performance. Furthermore, we present a new GAN training scheme that generates not only image data but also matching labels. Quantitative evaluation shows superior performance of biophysical motivated synthetic training data, even outperforming manual annotation and pretrained models. This underscores the potential of incorporating biophysical modeling for enhancing synthetic training data quality.
Authors: Shuaiyu Yuan, Tristan Whitmarsh, Dimitri A Kessler, Otso Arponen, Mary A McLean, Gabrielle Baxter, Frank Riemer, Aneurin J Kennerley, William J Brackenbury, Fiona J Gilbert, Joshua D Kaggie
Abstract: New multinuclear MRI techniques, such as sodium MRI, generally suffer from low image quality due to an inherently low signal. Postprocessing methods, such as image denoising, have been developed for image enhancement. However, the assessment of these enhanced images is challenging especially considering when there is a lack of high resolution and high signal images as reference, such as in sodium MRI. No-reference Image Quality Assessment (NR-IQA) metrics are approaches to solve this problem. Existing learning-based NR-IQA metrics rely on labels derived from subjective human opinions or metrics like Signal-to-Noise Ratio (SNR), which are either time-consuming or lack accurate ground truths, resulting in unreliable assessment. We note that deep learning (DL) models have a unique characteristic in that they are specialized to a characteristic training set, meaning that deviations between the input testing data from the training data will reduce prediction accuracy. Therefore, we propose a novel DL-based NR-IQA metric, the Model Specialization Metric (MSM), which does not depend on ground-truth images or labels. MSM measures the difference between the input image and the model's prediction for evaluating the quality of the input image. Experiments conducted on both simulated distorted proton T1-weighted MR images and denoised sodium MR images demonstrate that MSM exhibits a superior evaluation performance on various simulated noises and distortions. MSM also has a substantial agreement with the expert evaluations, achieving an averaged Cohen's Kappa coefficient of 0.6528, outperforming the existing NR-IQA metrics.
Authors: Yu Lu, Roummel F. Marcia
Abstract: The negative binomial model, which generalizes the Poisson distribution model, can be found in applications involving low-photon signal recovery, including medical imaging. Recent studies have explored several regularization terms for the negative binomial model, such as the $\ell_p$ quasi-norm with $0 < p < 1$, $\ell_1$ norm, and the total variation (TV) quasi-seminorm for promoting sparsity in signal recovery. These penalty terms have been shown to improve image reconstruction outcomes. In this paper, we investigate the $\ell_p$ quasi-seminorm, both isotropic and anisotropic $\ell_p$ TV quasi-seminorms, within the framework of the negative binomial statistical model. This problem can be formulated as an optimization problem, which we solve using a gradient-based approach. We present comparisons between the negative binomial and Poisson statistical models using the $\ell_p$ TV quasi-seminorm as well as common penalty terms. Our experimental results highlight the efficacy of the proposed method.
Authors: Puneet Kumar, Balasubramanian Raman
Abstract: This paper proposes a feature-based domain adaptation technique for identifying emotions in generic images, encompassing both facial and non-facial objects, as well as non-human components. This approach addresses the challenge of the limited availability of pre-trained models and well-annotated datasets for Image Emotion Recognition (IER). Initially, a deep-learning-based Facial Expression Recognition (FER) system is developed, classifying facial images into discrete emotion classes. Maintaining the same network architecture, this FER system is then adapted to recognize emotions in generic images through the application of discrepancy loss, enabling the model to effectively learn IER features while classifying emotions into categories such as 'happy,' 'sad,' 'hate,' and 'anger.' Additionally, a novel interpretability method, Divide and Conquer based Shap (DnCShap), is introduced to elucidate the visual features most relevant for emotion recognition. The proposed IER system demonstrated emotion classification accuracies of 60.98% for the IAPSa dataset, 58.86% for the ArtPhoto dataset, 69.13% for the FI dataset, and 58.06% for the EMOTIC dataset. The system effectively identifies the important visual features leading to specific emotion classifications and provides detailed embedding plots to explain the predictions, enhancing the understanding and trust in AI-driven emotion recognition systems.
Authors: Frederik Hagelskj{\ae}r, Rasmus Laurvig Haugaard
Abstract: In this paper, we present KeyMatchNet, a novel network for zero-shot pose estimation in 3D point clouds. Our method uses only depth information, making it more applicable for many industrial use cases, as color information is seldom available. The network is composed of two parallel components for computing object and scene features. The features are then combined to create matches used for pose estimation. The parallel structure allows for pre-processing of the individual parts, which decreases the run-time. Using a zero-shot network allows for a very short set-up time, as it is not necessary to train models for new objects. However, as the network is not trained for the specific object, zero-shot pose estimation methods generally have lower accuracy compared with conventional methods. To address this, we reduce the complexity of the task by including the scenario information during training. This is typically not feasible as collecting real data for new tasks drastically increases the cost. However, for zero-shot pose estimation, training for new objects is not necessary and the expensive data collection can thus be performed only once. Our method is trained on 1,500 objects and is only tested on unseen objects. We demonstrate that the trained network can not only accurately estimate poses for novel objects, but also demonstrate the ability of the network on objects outside of the trained class. Test results are also shown on real data. We believe that the presented method is valuable for many real-world scenarios. Project page available at keymatchnet.github.io
Authors: Kangjun Liu, Ke Chen, Kui Jia, Yaowei Wang
Abstract: Deep representation learning is a subfield of machine learning that focuses on learning meaningful and useful representations of data through deep neural networks. However, existing methods for semantic classification typically employ pre-defined target codes such as the one-hot and the Hadamard codes, which can either fail or be less flexible to model inter-class correlation. In light of this, this paper introduces a novel learnable target coding as an auxiliary regularization of deep representation learning, which can not only incorporate latent dependency across classes but also impose geometric properties of target codes into representation space. Specifically, a margin-based triplet loss and a correlation consistency loss on the proposed target codes are designed to encourage more discriminative representations owing to enlarging between-class margins in representation space and favoring equal semantic correlation of learnable target codes respectively. Experimental results on several popular visual classification and retrieval benchmarks can demonstrate the effectiveness of our method on improving representation learning, especially for imbalanced data. Source codes are made publicly available at \href{https://github.com/AkonLau/LTC}{https://github.com/AkonLau/LTC}.
URLs: https://github.com/AkonLau/LTC, https://github.com/AkonLau/LTC
Authors: Yiheng Li, Seth Z. Zhao, Chenfeng Xu, Chen Tang, Chenran Li, Mingyu Ding, Masayoshi Tomizuka, Wei Zhan
Abstract: Accumulating substantial volumes of real-world driving data proves pivotal in the realm of trajectory forecasting for autonomous driving. Given the heavy reliance of current trajectory forecasting models on data-driven methodologies, we aim to tackle the challenge of learning general trajectory forecasting representations under limited data availability. We propose a pipeline-level solution to mitigate the issue of data scarcity in trajectory forecasting. The solution is composed of two parts: firstly, we adopt HD map augmentation and trajectory synthesis for generating driving data, and then we learn representations by pre-training on them. Specifically, we apply vector transformations to reshape the maps, and then employ a rule-based model to generate trajectories on both original and augmented scenes; thus enlarging the driving data without collecting additional real ones. To foster the learning of general representations within this augmented dataset, we comprehensively explore the different pre-training strategies, including extending the concept of a Masked AutoEncoder (MAE) for trajectory forecasting. Without bells and whistles, our proposed pipeline-level solution is general, simple, yet effective: we conduct extensive experiments to demonstrate the effectiveness of our data expansion and pre-training strategies, which outperform the baseline prediction model by large margins, e.g. 5.04%, 3.84% and 8.30% in terms of $MR_6$, $minADE_6$ and $minFDE_6$. The pre-training dataset and the codes for pre-training and fine-tuning are released at https://github.com/yhli123/Pretraining_on_Synthetic_Driving_Data_for_Trajectory_Prediction.
URLs: https://github.com/yhli123/Pretraining_on_Synthetic_Driving_Data_for_Trajectory_Prediction.
Authors: You Zhou, Jie Wang
Abstract: We explore how to capture the significance of a sub-text block in an article and how it may be used for text mining tasks. A sub-text block is a sub-sequence of sentences in the article. We formulate the notion of content significance distribution (CSD) of sub-text blocks, referred to as CSD of the first kind and denoted by CSD-1. In particular, we leverage Hugging Face's SentenceTransformer to generate contextual sentence embeddings, and use MoverScore over text embeddings to measure how similar a sub-text block is to the entire text. To overcome the exponential blowup on the number of sub-text blocks, we present an approximation algorithm and show that the approximated CSD-1 is almost identical to the exact CSD-1. Under this approximation, we show that the average and median CSD-1's for news, scholarly research, argument, and narrative articles share the same pattern. We also show that under a certain linear transformation, the complement of the cumulative distribution function of the beta distribution with certain values of $\alpha$ and $\beta$ resembles a CSD-1 curve. We then use CSD-1's to extract linguistic features to train an SVC classifier for assessing how well an article is organized. Through experiments, we show that this method achieves high accuracy for assessing student essays. Moreover, we study CSD of sentence locations, referred to as CSD of the second kind and denoted by CSD-2, and show that average CSD-2's for different types of articles possess distinctive patterns, which either conform common perceptions of article structures or provide rectification with minor deviation.
Authors: Yuxin Du, Fan Bai, Tiejun Huang, Bo Zhao
Abstract: Precise image segmentation provides clinical study with instructive information. Despite the remarkable progress achieved in medical image segmentation, there is still an absence of a 3D foundation segmentation model that can segment a wide range of anatomical categories with easy user interaction. In this paper, we propose a 3D foundation segmentation model, named SegVol, supporting universal and interactive volumetric medical image segmentation. By scaling up training data to 90K unlabeled Computed Tomography (CT) volumes and 6K labeled CT volumes, this foundation model supports the segmentation of over 200 anatomical categories using semantic and spatial prompts. To facilitate efficient and precise inference on volumetric images, we design a zoom-out-zoom-in mechanism. Extensive experiments on 22 anatomical segmentation tasks verify that SegVol outperforms the competitors in 19 tasks, with improvements up to 37.24% compared to the runner-up methods. We demonstrate the effectiveness and importance of specific designs by ablation study. We expect this foundation model can promote the development of volumetric medical image analysis. The model and code are publicly available at: https://github.com/BAAI-DCAI/SegVol.
Authors: Ali Hatamizadeh, Jiaming Song, Guilin Liu, Jan Kautz, Arash Vahdat
Abstract: Diffusion models with their powerful expressivity and high sample quality have achieved State-Of-The-Art (SOTA) performance in the generative domain. The pioneering Vision Transformer (ViT) has also demonstrated strong modeling capabilities and scalability, especially for recognition tasks. In this paper, we study the effectiveness of ViTs in diffusion-based generative learning and propose a new model denoted as Diffusion Vision Transformers (DiffiT). Specifically, we propose a methodology for finegrained control of the denoising process and introduce the Time-dependant Multihead Self Attention (TMSA) mechanism. DiffiT is surprisingly effective in generating high-fidelity images with significantly better parameter efficiency. We also propose latent and image space DiffiT models and show SOTA performance on a variety of class-conditional and unconditional synthesis tasks at different resolutions. The Latent DiffiT model achieves a new SOTA FID score of 1.73 on ImageNet256 dataset while having 19.85%, 16.88% less parameters than other Transformer-based diffusion models such as MDT and DiT,respectively. Code: https://github.com/NVlabs/DiffiT
Authors: Zhijie Shen, Chunyu Lin, Junsong Zhang, Lang Nie, Kang Liao, Yao Zhao
Abstract: Existing panoramic layout estimation solutions tend to recover room boundaries from a vertically compressed sequence, yielding imprecise results as the compression process often muddles the semantics between various planes. Besides, these data-driven approaches impose an urgent demand for massive data annotations, which are laborious and time-consuming. For the first problem, we propose an orthogonal plane disentanglement network (termed DOPNet) to distinguish ambiguous semantics. DOPNet consists of three modules that are integrated to deliver distortion-free, semantics-clean, and detail-sharp disentangled representations, which benefit the subsequent layout recovery. For the second problem, we present an unsupervised adaptation technique tailored for horizon-depth and ratio representations. Concretely, we introduce an optimization strategy for decision-level layout analysis and a 1D cost volume construction method for feature-level multi-view aggregation, both of which are designed to fully exploit the geometric consistency across multiple perspectives. The optimizer provides a reliable set of pseudo-labels for network training, while the 1D cost volume enriches each view with comprehensive scene information derived from other perspectives. Extensive experiments demonstrate that our solution outperforms other SoTA models on both monocular layout estimation and multi-view layout estimation tasks. Cobe can be available at https://github.com/zhijieshen-bjtu/MV-DOPNet.
Authors: Ziwei Sun, Zexi Hua, Hengchao Li, Yan Li
Abstract: Aiming at the specific characteristics of flying bird objects in surveillance video, such as the typically non-obvious features in single-frame images, small size in most instances, and asymmetric shapes, this paper proposes a Flying Bird Object Detection method for Surveillance Video (FBOD-SV). Firstly, a new feature aggregation module, the Correlation Attention Feature Aggregation (Co-Attention-FA) module, is designed to aggregate the features of the flying bird object according to the bird object's correlation on multiple consecutive frames of images. Secondly, a Flying Bird Object Detection Network (FBOD-Net) with down-sampling followed by up-sampling is designed, which utilizes a large feature layer that fuses fine spatial information and large receptive field information to detect special multi-scale (mostly small-scale) bird objects. Finally, the SimOTA dynamic label allocation method is applied to One-Category object detection, and the SimOTA-OC dynamic label strategy is proposed to solve the difficult problem of label allocation caused by irregular flying bird objects. In this paper, the performance of the FBOD-SV is validated using experimental datasets of flying bird objects in traction substation surveillance videos. The experimental results show that the FBOD-SV effectively improves the detection performance of flying bird objects in surveillance video.
Authors: Apoorva Beedu, Harish Haresamudram, Karan Samel, Irfan Essa
Abstract: Anticipating future actions is a highly challenging task due to the diversity and scale of potential future actions; yet, information from different modalities help narrow down plausible action choices. Each modality can provide diverse and often complementary context for the model to learn from. While previous multi-modal methods leverage information from modalities such as video and audio, we primarily explore how text descriptions of actions and objects can also lead to more accurate action anticipation by providing additional contextual cues, e.g., about the environment and its contents. We propose a Multi-modal Contrastive Anticipative Transformer (M-CAT), a video transformer architecture that jointly learns from multi-modal features and text descriptions of actions and objects. We train our model in two stages, where the model first learns to align video clips with descriptions of future actions, and is subsequently fine-tuned to predict future actions. Compared to existing methods, M-CAT has the advantage of learning additional context from two types of text inputs: rich descriptions of future actions during pre-training, and, text descriptions for detected objects and actions during modality feature fusion. Through extensive experimental evaluation, we demonstrate that our model outperforms previous methods on the EpicKitchens datasets, and show that using simple text descriptions of actions and objects aid in more effective action anticipation. In addition, we examine the impact of object and action information obtained via text, and perform extensive ablations.
Authors: Netta Ollikka, Amro Abbas, Andrea Perin, Markku Kilpel\"ainen, St\'ephane Deny
Abstract: Deep learning is closing the gap with human vision on several object recognition benchmarks. Here we investigate this gap for challenging images where objects are seen in unusual poses. We find that humans excel at recognizing objects in such poses. In contrast, state-of-the-art deep networks for vision (EfficientNet, SWAG, ViT, SWIN, BEiT, ConvNext) and state-of-the-art large vision-language models (Claude 3.5, Gemini 1.5, GPT-4) are systematically brittle on unusual poses, with the exception of Gemini showing excellent robustness in that condition. As we limit image exposure time, human performance degrades to the level of deep networks, suggesting that additional mental processes (requiring additional time) are necessary to identify objects in unusual poses. An analysis of error patterns of humans vs. networks reveals that even time-limited humans are dissimilar to feed-forward deep networks. In conclusion, our comparison reveals that humans and deep networks rely on different mechanisms for recognizing objects in unusual poses. Understanding the nature of the mental processes taking place during extra viewing time may be key to reproduce the robustness of human vision in silico.
Authors: Jinlu Zhang, Yiyi Zhou, Qiancheng Zheng, Xiaoxiong Du, Gen Luo, Jun Peng, Xiaoshuai Sun, Rongrong Ji
Abstract: Text-to-3D-aware face (T3D Face) generation and manipulation is an emerging research hot spot in machine learning, which still suffers from low efficiency and poor quality. In this paper, we propose an End-to-End Efficient and Effective network for fast and accurate T3D face generation and manipulation, termed $E^3$-FaceNet. Different from existing complex generation paradigms, $E^3$-FaceNet resorts to a direct mapping from text instructions to 3D-aware visual space. We introduce a novel Style Code Enhancer to enhance cross-modal semantic alignment, alongside an innovative Geometric Regularization objective to maintain consistency across multi-view generations. Extensive experiments on three benchmark datasets demonstrate that $E^3$-FaceNet can not only achieve picture-like 3D face generation and manipulation, but also improve inference speed by orders of magnitudes. For instance, compared with Latent3D, $E^3$-FaceNet speeds up the five-view generations by almost 470 times, while still exceeding in generation quality. Our code is released at https://github.com/Aria-Zhangjl/E3-FaceNet.
Authors: Shuaikang Shang, Xuejing Kang, Anlong Ming
Abstract: High Dynamic Range (HDR) imaging aims to generate an artifact-free HDR image with realistic details by fusing multi-exposure Low Dynamic Range (LDR) images. Caused by large motion and severe under-/over-exposure among input LDR images, HDR imaging suffers from ghosting artifacts and fusion distortions. To address these critical issues, we propose an HDR Transformer Deformation Convolution (HDRTransDC) network to generate high-quality HDR images, which consists of the Transformer Deformable Convolution Alignment Module (TDCAM) and the Dynamic Weight Fusion Block (DWFB). To solve the ghosting artifacts, the proposed TDCAM extracts long-distance content similar to the reference feature in the entire non-reference features, which can accurately remove misalignment and fill the content occluded by moving objects. For the purpose of eliminating fusion distortions, we propose DWFB to spatially adaptively select useful information across frames to effectively fuse multi-exposed features. Extensive experiments show that our method quantitatively and qualitatively achieves state-of-the-art performance.
Authors: Ming Xu, Zilong Xie
Abstract: Most Vision-and-Language Navigation (VLN) algorithms are prone to making decision due to a lack of visual common sense and insufficient reasoning capabilities. To address this issue, we propose a Hierarchical Spatial Proximity Reasoning (HSPR) method. First, we introduce a scene understanding auxiliary task to help the agent build a knowledge base of hierarchical spatial proximity. This task utilizes panoramic views and object features to identify types of nodes and uncover the adjacency relationships between nodes, objects, and between nodes and objects. Second, we propose a multi-step reasoning navigation algorithm based on hierarchical spatial proximity knowledge base, which continuously plans feasible paths to enhance exploration efficiency. Third, we introduce a residual fusion method to improve navigation decision accuracy. Finally, we validate our approach with experiments on publicly available datasets including REVERIE, SOON, R2R, and R4R. Our code is available at https://github.com/iCityLab/HSPR.
Authors: Cedric Perauer, Laurenz Adrian Heidrich, Haifan Zhang, Matthias Nie{\ss}ner, Anastasiia Kornilova, Alexey Artemov
Abstract: Recently, progress in acquisition equipment such as LiDAR sensors has enabled sensing increasingly spacious outdoor 3D environments. Making sense of such 3D acquisitions requires fine-grained scene understanding, such as constructing instance-based 3D scene segmentations. Commonly, a neural network is trained for this task; however, this requires access to a large, densely annotated dataset, which is widely known to be challenging to obtain. To address this issue, in this work we propose to predict instance segmentations for 3D scenes in an unsupervised way, without relying on ground-truth annotations. To this end, we construct a learning framework consisting of two components: (1) a pseudo-annotation scheme for generating initial unsupervised pseudo-labels; and (2) a self-training algorithm for instance segmentation to fit robust, accurate instances from initial noisy proposals. To enable generating 3D instance mask proposals, we construct a weighted proxy-graph by connecting 3D points with edges integrating multi-modal image- and point-based self-supervised features, and perform graph-cuts to isolate individual pseudo-instances. We then build on a state-of-the-art point-based architecture and train a 3D instance segmentation model, resulting in significant refinement of initial proposals. To scale to arbitrary complexity 3D scenes, we design our algorithm to operate on local 3D point chunks and construct a merging step to generate scene-level instance segmentations. Experiments on the challenging SemanticKITTI benchmark demonstrate the potential of our approach, where it attains 13.3% higher Average Precision and 9.1% higher F1 score compared to the best-performing baseline. The code will be made publicly available at https://github.com/artonson/autoinst.
Authors: Tal Hakim
Abstract: The application of machine-learning solutions to movement assessment from skeleton videos has attracted significant research attention in recent years. This advancement has made rehabilitation at home more accessible, utilizing movement assessment algorithms that can operate on affordable equipment for human pose detection and analysis from 2D or 3D videos. While the primary objective of automatic assessment tasks is to score movements, the automatic generation of feedback highlighting key movement issues has the potential to significantly enhance and accelerate the rehabilitation process. While numerous research works exist in the field of automatic movement assessment, only a handful address feedback generation. In this study, we propose terminology and criteria for the classification, evaluation, and comparison of feedback generation solutions. We discuss the challenges associated with each feedback generation approach and use our proposed criteria to classify existing solutions. To our knowledge, this is the first work that formulates feedback generation in skeletal movement assessment.
Authors: Ying Zhang, Yuezun Li, Bo Peng, Jiaran Zhou, Huiyu Zhou, Junyu Dong
Abstract: The task of video inpainting detection is to expose the pixel-level inpainted regions within a video sequence. Existing methods usually focus on leveraging spatial and temporal inconsistencies. However, these methods typically employ fixed operations to combine spatial and temporal clues, limiting their applicability in different scenarios. In this paper, we introduce a novel Multilateral Temporal-view Pyramid Transformer ({\em MumPy}) that collaborates spatial-temporal clues flexibly. Our method utilizes a newly designed multilateral temporal-view encoder to extract various collaborations of spatial-temporal clues and introduces a deformable window-based temporal-view interaction module to enhance the diversity of these collaborations. Subsequently, we develop a multi-pyramid decoder to aggregate the various types of features and generate detection maps. By adjusting the contribution strength of spatial and temporal clues, our method can effectively identify inpainted regions. We validate our method on existing datasets and also introduce a new challenging and large-scale Video Inpainting dataset based on the YouTube-VOS dataset, which employs several more recent inpainting methods. The results demonstrate the superiority of our method in both in-domain and cross-domain evaluation scenarios.
Authors: Chenyu Gao, Shunxing Bao, Michael Kim, Nancy Newlin, Praitayini Kanakaraj, Tianyuan Yao, Gaurav Rudravaram, Yuankai Huo, Daniel Moyer, Kurt Schilling, Walter Kukull, Arthur Toga, Derek Archer, Timothy Hohman, Bennett Landman, Zhiyuan Li
Abstract: Purpose: In diffusion MRI (dMRI), the volumetric and bundle analyses of whole-brain tissue microstructure and connectivity can be severely impeded by an incomplete field-of-view (FOV). This work aims to develop a method for imputing the missing slices directly from existing dMRI scans with an incomplete FOV. We hypothesize that the imputed image with complete FOV can improve the whole-brain tractography for corrupted data with incomplete FOV. Therefore, our approach provides a desirable alternative to discarding the valuable dMRI data, enabling subsequent tractography analyses that would otherwise be challenging or unattainable with corrupted data. Approach: We propose a framework based on a deep generative model that estimates the absent brain regions in dMRI scans with incomplete FOV. The model is capable of learning both the diffusion characteristics in diffusion-weighted images (DWI) and the anatomical features evident in the corresponding structural images for efficiently imputing missing slices of DWI outside of incomplete FOV. Results: For evaluating the imputed slices, on the WRAP dataset the proposed framework achieved PSNRb0=22.397, SSIMb0=0.905, PSNRb1300=22.479, SSIMb1300=0.893; on the NACC dataset it achieved PSNRb0=21.304, SSIMb0=0.892, PSNRb1300=21.599, SSIMb1300= 0.877. The proposed framework improved the tractography accuracy, as demonstrated by an increased average Dice score for 72 tracts (p < 0.001) on both the WRAP and NACC datasets. Conclusions: Results suggest that the proposed framework achieved sufficient imputation performance in dMRI data with incomplete FOV for improving whole-brain tractography, thereby repairing the corrupted data. Our approach achieved more accurate whole-brain tractography results with extended and complete FOV and reduced the uncertainty when analyzing bundles associated with Alzheimer's Disease.
Authors: Yi Xiao, Qiangqiang Yuan, Kui Jiang, Yuzeng Chen, Qiang Zhang, Chia-Wen Lin
Abstract: Recent progress in remote sensing image (RSI) super-resolution (SR) has exhibited remarkable performance using deep neural networks, e.g., Convolutional Neural Networks and Transformers. However, existing SR methods often suffer from either a limited receptive field or quadratic computational overhead, resulting in sub-optimal global representation and unacceptable computational costs in large-scale RSI. To alleviate these issues, we develop the first attempt to integrate the Vision State Space Model (Mamba) for RSI-SR, which specializes in processing large-scale RSI by capturing long-range dependency with linear complexity. To achieve better SR reconstruction, building upon Mamba, we devise a Frequency-assisted Mamba framework, dubbed FMSR, to explore the spatial and frequent correlations. In particular, our FMSR features a multi-level fusion architecture equipped with the Frequency Selection Module (FSM), Vision State Space Module (VSSM), and Hybrid Gate Module (HGM) to grasp their merits for effective spatial-frequency fusion. Considering that global and local dependencies are complementary and both beneficial for SR, we further recalibrate these multi-level features for accurate feature fusion via learnable scaling adaptors. Extensive experiments on AID, DOTA, and DIOR benchmarks demonstrate that our FMSR outperforms state-of-the-art Transformer-based methods HAT-L in terms of PSNR by 0.11 dB on average, while consuming only 28.05% and 19.08% of its memory consumption and complexity, respectively. Code will be available at https://github.com/XY-boy/FreMamba
Authors: Masane Fuchi, Tomohiro Takagi
Abstract: Generating images from text has become easier because of the scaling of diffusion models and advancements in the field of vision and language. These models are trained using vast amounts of data from the Internet. Hence, they often contain undesirable content such as copyrighted material. As it is challenging to remove such data and retrain the models, methods for erasing specific concepts from pre-trained models have been investigated. We propose a novel concept-erasure method that updates the text encoder using few-shot unlearning in which a few real images are used. The discussion regarding the generated images after erasing a concept has been lacking. While there are methods for specifying the transition destination for concepts, the validity of the specified concepts is unclear. Our method implicitly achieves this by transitioning to the latent concepts inherent in the model or the images. Our method can erase a concept within 10 s, making concept erasure more accessible than ever before. Implicitly transitioning to related concepts leads to more natural concept erasure. We applied the proposed method to various concepts and confirmed that concept erasure can be achieved tens to hundreds of times faster than with current methods. By varying the parameters to be updated, we obtained results suggesting that, like previous research, knowledge is primarily accumulated in the feed-forward networks of the text encoder. Our code is available at \url{https://github.com/fmp453/few-shot-erasing}
Authors: Zhe Huang, Yizhe Zhao, Hao Xiao, Chenyan Wu, Lingting Ge
Abstract: Recent advances in multi-view camera-only 3D object detection either rely on an accurate reconstruction of bird's-eye-view (BEV) 3D features or on traditional 2D perspective view (PV) image features. While both have their own pros and cons, few have found a way to stitch them together in order to benefit from "the best of both worlds". To this end, we explore a duo space (i.e., BEV and PV) 3D perception framework, in conjunction with some useful duo space fusion strategies that allow effective aggregation of the two feature representations. To the best of our knowledge, our proposed method, DuoSpaceNet, is the first to leverage two distinct feature spaces and achieves the state-of-the-art 3D object detection and BEV map segmentation results on nuScenes dataset.
Authors: Ting Liu, Xuyang Liu, Siteng Huang, Liangtao Shi, Zunnan Xu, Yi Xin, Quanjun Yin, Xiaohong Liu
Abstract: Parameter-efficient fine-tuning (PEFT) has emerged as a popular solution for adapting pre-trained Vision Transformer (ViT) models to downstream applications. While current PEFT methods have achieved parameter efficiency, they overlook the efficiency of computation and GPU memory during both fine-tuning and inference, falling short of practical requirements. In this paper, we propose \textbf{Sparse-Tuning}, a novel PEFT method that accounts for the information redundancy in images and videos to boost the above efficiency. By sparsely preserving the semantic-relevant tokens and merging irrelevant ones, Sparse-Tuning minimizes the quantity of tokens processed at each layer, leading to a quadratic reduction in computational and memory overhead. To align our token sparsification strategy suitably with fine-tuning purposes, we further design Dense Adapters that establish dense connections from shallow layers to deeper layers. These Dense Adapters integrate multi-level local features to enrich the current tokens, improving both token preservation and model adaptation. Empirical results on VTAB-1K, three image datasets, and two video datasets show that our Sparse-Tuning reduces GFLOPs to \textbf{62\%-70\%} of the original ViT-B while achieving state-of-the-art performance. Source code is available at \url{https://github.com/liuting20/Sparse-Tuning}.
Authors: Riccardo Benaglia, Angelo Porrello, Pietro Buzzega, Simone Calderara, Rita Cucchiara
Abstract: Trajectory forecasting is crucial for video surveillance analytics, as it enables the anticipation of future movements for a set of agents, e.g. basketball players engaged in intricate interactions with long-term intentions. Deep generative models offer a natural learning approach for trajectory forecasting, yet they encounter difficulties in achieving an optimal balance between sampling fidelity and diversity. We address this challenge by leveraging Vector Quantized Variational Autoencoders (VQ-VAEs), which utilize a discrete latent space to tackle the issue of posterior collapse. Specifically, we introduce an instance-based codebook that allows tailored latent representations for each example. In a nutshell, the rows of the codebook are dynamically adjusted to reflect contextual information (i.e., past motion patterns extracted from the observed trajectories). In this way, the discretization process gains flexibility, leading to improved reconstructions. Notably, instance-level dynamics are injected into the codebook through low-rank updates, which restrict the customization of the codebook to a lower dimension space. The resulting discrete space serves as the basis of the subsequent step, which regards the training of a diffusion-based predictive model. We show that such a two-fold framework, augmented with instance-level discretization, leads to accurate and diverse forecasts, yielding state-of-the-art performance on three established benchmarks.
Authors: Fengxiang Wang, Hongzhen Wang, Di Wang, Zonghao Guo, Zhenyu Zhong, Long Lan, Jing Zhang, Zhiyuan Liu, Maosong Sun
Abstract: Masked Image Modeling (MIM) has become an essential method for building foundational visual models in remote sensing (RS). However, the limitations in size and diversity of existing RS datasets restrict the ability of MIM methods to learn generalizable representations. Additionally, conventional MIM techniques, which require reconstructing all tokens, introduce unnecessary computational overhead. To address these issues, we present a new pre-training pipeline for RS models, featuring the creation of a large-scale RS dataset and an efficient MIM approach. We curated a high-quality dataset named OpticalRS-4M by collecting publicly available RS datasets and processing them through exclusion, slicing, and deduplication. OpticalRS-4M comprises 4 million optical images covering various RS tasks, such as object detection and pixel segmentation. To enhance efficiency, we propose SelectiveMAE, a pre-training method that dynamically encodes and reconstructs semantically rich patch tokens, thereby reducing the inefficiencies of traditional MIM models caused by redundant background pixels in RS images. Extensive experiments demonstrate that OpticalRS-4M significantly improves classification, detection, and segmentation performance, while SelectiveMAE increases training efficiency over 2 times. This highlights the effectiveness and scalability of our pipeline in developing RS foundational models.
Authors: Hui Lu, Albert Ali Salah, Ronald Poppe
Abstract: Video understanding requires the extraction of rich spatio-temporal representations, which transformer models achieve through self-attention. Unfortunately, self-attention poses a computational burden. In NLP, Mamba has surfaced as an efficient alternative for transformers. However, Mamba's successes do not trivially extend to computer vision tasks, including those in video analysis. In this paper, we theoretically analyze the differences between self-attention and Mamba. We identify two limitations in Mamba's token processing: historical decay and element contradiction. We propose VideoMambaPro (VMP) that solves the identified limitations by adding masked backward computation and elemental residual connections to a VideoMamba backbone. VideoMambaPro shows state-of-the-art video action recognition performance compared to transformer models, and surpasses VideoMamba by clear margins: 7.9% and 8.1% top-1 on Kinetics-400 and Something-Something V2, respectively. Our VideoMambaPro-M model achieves 91.9% top-1 on Kinetics-400, only 0.2% below InternVideo2-6B but with only 1.2% of its parameters. The combination of high performance and efficiency makes VideoMambaPro an interesting alternative for transformer models.
Authors: Bocheng Zou, Mu Cai, Jianrui Zhang, Yong Jae Lee
Abstract: In the realm of vision models, the primary mode of representation is using pixels to rasterize the visual world. Yet this is not always the best or unique way to represent visual content, especially for designers and artists who depict the world using geometry primitives such as polygons. Vector graphics (VG), on the other hand, offer a textual representation of visual content, which can be more concise and powerful for content like cartoons, sketches and scientific figures. Recent studies have shown promising results on processing vector graphics with capable Large Language Models (LLMs). However, such works focus solely on qualitative results, understanding, or a specific type of vector graphics. We propose VGBench, a comprehensive benchmark for LLMs on handling vector graphics through diverse aspects, including (a) both visual understanding and generation, (b) evaluation of various vector graphics formats, (c) diverse question types, (d) wide range of prompting techniques, (e) under multiple LLMs and (f) comparison with VLMs on rasterized representations. Evaluating on our collected 4279 understanding and 5845 generation samples, we find that LLMs show strong capability on both aspects while exhibiting less desirable performance on low-level formats (SVG). Both data and evaluation pipeline will be open-sourced at https://vgbench.github.io.
Authors: Yuchen Weng, Zhengwen Shen, Ruofan Chen, Qi Wang, Jun Wang
Abstract: 3D deblurring reconstruction techniques have recently seen significant advancements with the development of Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS). Although these techniques can recover relatively clear 3D reconstructions from blurry image inputs, they still face limitations in handling severe blurring and complex camera motion. To address these issues, we propose Event-assisted 3D Deblur Reconstruction with Gaussian Splatting (EaDeblur-GS), which integrates event camera data to enhance the robustness of 3DGS against motion blur. By employing an Adaptive Deviation Estimator (ADE) network to estimate Gaussian center deviations and using novel loss functions, EaDeblur-GS achieves sharp 3D reconstructions in real-time, demonstrating performance comparable to state-of-the-art methods.
Authors: Guanfeng Tang, Zhiyuan Wu, Jiahang Li, Ping Zhong, Xieyuanli Chen, Huiming Liu, Rui Fan
Abstract: Semantic segmentation and stereo matching, respectively analogous to the ventral and dorsal streams in our human brain, are two key components of autonomous driving perception systems. Addressing these two tasks with separate networks is no longer the mainstream direction in developing computer vision algorithms, particularly with the recent advances in large vision models and embodied artificial intelligence. The trend is shifting towards combining them within a joint learning framework, especially emphasizing feature sharing between the two tasks. The major contributions of this study lie in comprehensively tightening the coupling between semantic segmentation and stereo matching. Specifically, this study introduces three novelties: (1) a tightly coupled, gated feature fusion strategy, (2) a hierarchical deep supervision strategy, and (3) a coupling tightening loss function. The combined use of these technical contributions results in TiCoSS, a state-of-the-art joint learning framework that simultaneously tackles semantic segmentation and stereo matching. Through extensive experiments on the KITTI and vKITTI2 datasets, along with qualitative and quantitative analyses, we validate the effectiveness of our developed strategies and loss function, and demonstrate its superior performance compared to prior arts, with a notable increase in mIoU by over 9%. Our source code will be publicly available at mias.group/TiCoSS upon publication.
Authors: Leilei Ma, Hongxing Xie, Lei Wang, Yanping Fu, Dengdi Sun, Haifeng Zhao
Abstract: Recently, large-scale visual language pre-trained (VLP) models have demonstrated impressive performance across various downstream tasks. Motivated by these advancements, pioneering efforts have emerged in multi-label image recognition with missing labels, leveraging VLP prompt-tuning technology. However, they usually cannot match text and vision features well, due to complicated semantics gaps and missing labels in a multi-label image. To tackle this challenge, we propose $\textbf{T}$ext-$\textbf{R}$egion $\textbf{M}$atching for optimizing $\textbf{M}$ulti-$\textbf{L}$abel prompt tuning, namely TRM-ML, a novel method for enhancing meaningful cross-modal matching. Compared to existing methods, we advocate exploring the information of category-aware regions rather than the entire image or pixels, which contributes to bridging the semantic gap between textual and visual representations in a one-to-one matching manner. Concurrently, we further introduce multimodal contrastive learning to narrow the semantic gap between textual and visual modalities and establish intra-class and inter-class relationships. Additionally, to deal with missing labels, we propose a multimodal category prototype that leverages intra- and inter-category semantic relationships to estimate unknown labels, facilitating pseudo-label generation. Extensive experiments on the MS-COCO, PASCAL VOC, Visual Genome, NUS-WIDE, and CUB-200-211 benchmark datasets demonstrate that our proposed framework outperforms the state-of-the-art methods by a significant margin. Our code is available here: https://github.com/yu-gi-oh-leilei/TRM-ML.
Authors: Jinchang Zhang, Praveen Kumar Reddy, Xue-Iuan Wong, Yiannis Aloimonos, Guoyu Lu
Abstract: Depth estimation is a critical topic for robotics and vision-related tasks. In monocular depth estimation, in comparison with supervised learning that requires expensive ground truth labeling, self-supervised methods possess great potential due to no labeling cost. However, self-supervised learning still has a large gap with supervised learning in 3D reconstruction and depth estimation performance. Meanwhile, scaling is also a major issue for monocular unsupervised depth estimation, which commonly still needs ground truth scale from GPS, LiDAR, or existing maps to correct. In the era of deep learning, existing methods primarily rely on exploring image relationships to train unsupervised neural networks, while the physical properties of the camera itself such as intrinsics and extrinsics are often overlooked. These physical properties are not just mathematical parameters; they are embodiments of the camera's interaction with the physical world. By embedding these physical properties into the deep learning model, we can calculate depth priors for ground regions and regions connected to the ground based on physical principles, providing free supervision signals without the need for additional sensors. This approach is not only easy to implement but also enhances the effects of all unsupervised methods by embedding the camera's physical properties into the model, thereby achieving an embodied understanding of the real world.
Authors: Jonathan Roberts, Kai Han, Samuel Albanie
Abstract: Large multimodal models (LMMs) have exhibited proficiencies across many visual tasks. Although numerous well-known benchmarks exist to evaluate model performance, they increasingly have insufficient headroom. As such, there is a pressing need for a new generation of benchmarks challenging enough for the next generation of LMMs. One area that LMMs show potential is graph analysis, specifically, the tasks an analyst might typically perform when interpreting figures such as estimating the mean, intercepts or correlations of functions and data series. In this work, we introduce GRAB, a graph analysis benchmark, fit for current and future frontier LMMs. Our benchmark is entirely synthetic, ensuring high-quality, noise-free questions. GRAB is comprised of 2170 questions, covering four tasks and 23 graph properties. We evaluate 20 LMMs on GRAB, finding it to be a challenging benchmark, with the highest performing model attaining a score of just 21.7%. Finally, we conduct various ablations to investigate where the models succeed and struggle. We release GRAB to encourage progress in this important, growing domain.
Authors: Rundong Luo, Haoran Geng, Congyue Deng, Puhao Li, Zan Wang, Baoxiong Jia, Leonidas Guibas, Siyuan Huang
Abstract: Interactable objects are ubiquitous in our daily lives. Recent advances in 3D generative models make it possible to automate the modeling of these objects, benefiting a range of applications from 3D printing to the creation of robot simulation environments. However, while significant progress has been made in modeling 3D shapes and appearances, modeling object physics, particularly for interactable objects, remains challenging due to the physical constraints imposed by inter-part motions. In this paper, we tackle the problem of physically plausible part completion for interactable objects, aiming to generate 3D parts that not only fit precisely into the object but also allow smooth part motions. To this end, we propose a diffusion-based part generation model that utilizes geometric conditioning through classifier-free guidance and formulates physical constraints as a set of stability and mobility losses to guide the sampling process. Additionally, we demonstrate the generation of dependent parts, paving the way toward sequential part generation for objects with complex part-whole hierarchies. Experimentally, we introduce a new metric for measuring physical plausibility based on motion success rates. Our model outperforms existing baselines over shape and physical metrics, especially those that do not adequately model physical constraints. We also demonstrate our applications in 3D printing, robot manipulation, and sequential part generation, showing our strength in realistic tasks with the demand for high physical plausibility.
Authors: Xu Zhang, Zhipeng Xie, Haiyang Yu, Qitong Wang, Peng Wang, Wei Wang
Abstract: Handling varying computational resources is a critical issue in modern AI applications. Adaptive deep networks, featuring the dynamic employment of multiple classifier heads among different layers, have been proposed to address classification tasks under varying computing resources. Existing approaches typically utilize the last classifier supported by the available resources for inference, as they believe that the last classifier always performs better across all classes. However, our findings indicate that earlier classifier heads can outperform the last head for certain classes. Based on this observation, we introduce the Collaborative Decision Making (CDM) module, which fuses the multiple classifier heads to enhance the inference performance of adaptive deep networks. CDM incorporates an uncertainty-aware fusion method based on evidential deep learning (EDL), that utilizes the reliability (uncertainty values) from the first c-1 classifiers to improve the c-th classifier' accuracy. We also design a balance term that reduces fusion saturation and unfairness issues caused by EDL constraints to improve the fusion quality of CDM. Finally, a regularized training strategy that uses the last classifier to guide the learning process of early classifiers is proposed to further enhance the CDM module's effect, called the Guided Collaborative Decision Making (GCDM) framework. The experimental evaluation demonstrates the effectiveness of our approaches. Results on ImageNet datasets show CDM and GCDM obtain 0.4% to 2.8% accuracy improvement (under varying computing resources) on popular adaptive networks. The code is available at the link https://github.com/Meteor-Stars/GCDM_AdaptiveNet.
Authors: Joseph Cho, Samuel Schmidgall, Cyril Zakka, Mrudang Mathur, Rohan Shad, William Hiesinger
Abstract: Diffusion-based video generation models have made significant strides, producing outputs with improved visual fidelity, temporal coherence, and user control. These advancements hold great promise for improving surgical education by enabling more realistic, diverse, and interactive simulation environments. In this study, we introduce SurGen, a text-guided diffusion model tailored for surgical video synthesis, producing the highest resolution and longest duration videos among existing surgical video generation models. We validate the visual and temporal quality of the outputs using standard image and video generation metrics. Additionally, we assess their alignment to the corresponding text prompts through a deep learning classifier trained on surgical data. Our results demonstrate the potential of diffusion models to serve as valuable educational tools for surgical trainees.
Authors: Xinrui Zhang, Ailong Cai, Shaoyu Wang, Linyuan Wang, Zhizhong Zheng, Lei Li, Bin Yan
Abstract: Metal artifacts in computed tomography (CT) imaging pose significant challenges to accurate clinical diagnosis. The presence of high-density metallic implants results in artifacts that deteriorate image quality, manifesting in the forms of streaking, blurring, or beam hardening effects, etc. Nowadays, various deep learning-based approaches, particularly generative models, have been proposed for metal artifact reduction (MAR). However, these methods have limited perception ability in the diverse morphologies of different metal implants with artifacts, which may generate spurious anatomical structures and exhibit inferior generalization capability. To address the issues, we leverage visual-language model (VLM) to identify these morphological features and introduce them into a dual-domain CLIP-assisted residual optimization perception model (DuDoCROP) for MAR. Specifically, a dual-domain CLIP (DuDoCLIP) is fine-tuned on the image domain and sinogram domain using contrastive learning to extract semantic descriptions from anatomical structures and metal artifacts. Subsequently, a diffusion model is guided by the embeddings of DuDoCLIP, thereby enabling the dual-domain prior generation. Additionally, we design prompt engineering for more precise image-text descriptions that can enhance the model's perception capability. Then, a downstream task is devised for the one-step residual optimization and integration of dual-domain priors, while incorporating raw data fidelity. Ultimately, a new perceptual indicator is proposed to validate the model's perception and generation performance. With the assistance of DuDoCLIP, our DuDoCROP exhibits at least 63.7% higher generalization capability compared to the baseline model. Numerical experiments demonstrate that the proposed method can generate more realistic image structures and outperform other SOTA approaches both qualitatively and quantitatively.
Authors: Vishal Batchu, Alex Wilson, Betty Peng, Carl Elkin, Umangi Jain, Christopher Van Arsdale, Ross Goroshin, Varun Gulshan
Abstract: The transition to renewable energy, particularly solar, is key to mitigating climate change. Google's Solar API aids this transition by estimating solar potential from aerial imagery, but its impact is constrained by geographical coverage. This paper proposes expanding the API's reach using satellite imagery, enabling global solar potential assessment. We tackle challenges involved in building a Digital Surface Model (DSM) and roof instance segmentation from lower resolution and single oblique views using deep learning models. Our models, trained on aligned satellite and aerial datasets, produce 25cm DSMs and roof segments. With ~1m DSM MAE on buildings, ~5deg roof pitch error and ~56% IOU on roof segmentation, they significantly enhance the Solar API's potential to promote solar adoption.
Authors: Pavan Uttej Ravva, Behdokht Kiafar, Pinar Kullu, Jicheng Li, Anjana Bhat, Roghayeh Leila Barmaki
Abstract: Autism spectrum disorder (ASD) is characterized by significant challenges in social interaction and comprehending communication signals. Recently, therapeutic interventions for ASD have increasingly utilized Deep learning powered-computer vision techniques to monitor individual progress over time. These models are trained on private, non-public datasets from the autism community, creating challenges in comparing results across different models due to privacy-preserving data-sharing issues. This work introduces MMASD+, an enhanced version of the novel open-source dataset called Multimodal ASD (MMASD). MMASD+ consists of diverse data modalities, including 3D-Skeleton, 3D Body Mesh, and Optical Flow data. It integrates the capabilities of Yolov8 and Deep SORT algorithms to distinguish between the therapist and children, addressing a significant barrier in the original dataset. Additionally, a Multimodal Transformer framework is proposed to predict 11 action types and the presence of ASD. This framework achieves an accuracy of 95.03% for predicting action types and 96.42% for predicting ASD presence, demonstrating over a 10% improvement compared to models trained on single data modalities. These findings highlight the advantages of integrating multiple data modalities within the Multimodal Transformer framework.
Authors: Mykhailo Koshil, Tilman Wegener, Detlef Mentrup, Simone Frintrop, Christian Wilms
Abstract: Visual inspection, or industrial anomaly detection, is one of the most common quality control types in manufacturing. The task is to identify the presence of an anomaly given an image, e.g., a missing component on an image of a circuit board, for subsequent manual inspection. While industrial anomaly detection has seen a surge in recent years, most anomaly detection methods still utilize knowledge only from normal samples, failing to leverage the information from the frequently available anomalous samples. Additionally, they heavily rely on very general feature extractors pre-trained on common image classification datasets. In this paper, we address these shortcomings and propose the new anomaly detection system AnomalousPatchCore~(APC) based on a feature extractor fine-tuned with normal and anomalous in-domain samples and a subsequent memory bank for identifying unusual features. To fine-tune the feature extractor in APC, we propose three auxiliary tasks that address the different aspects of anomaly detection~(classification vs. localization) and mitigate the effect of the imbalance between normal and anomalous samples. Our extensive evaluation on the MVTec dataset shows that APC outperforms state-of-the-art systems in detecting anomalies, which is especially important in industrial anomaly detection given the subsequent manual inspection. In detailed ablation studies, we further investigate the properties of our APC.
Authors: Zhaoxuan Wang, Xu Han, Hongxin Liu, Xianzhi Li
Abstract: The rotation robustness property has drawn much attention to point cloud analysis, whereas it still poses a critical challenge in 3D object detection. When subjected to arbitrary rotation, most existing detectors fail to produce expected outputs due to the poor rotation robustness. In this paper, we present RIDE, a pioneering exploration of Rotation-Invariance for the 3D LiDAR-point-based object DEtector, with the key idea of designing rotation-invariant features from LiDAR scenes and then effectively incorporating them into existing 3D detectors. Specifically, we design a bi-feature extractor that extracts (i) object-aware features though sensitive to rotation but preserve geometry well, and (ii) rotation-invariant features, which lose geometric information to a certain extent but are robust to rotation. These two kinds of features complement each other to decode 3D proposals that are robust to arbitrary rotations. Particularly, our RIDE is compatible and easy to plug into the existing one-stage and two-stage 3D detectors, and boosts both detection performance and rotation robustness. Extensive experiments on the standard benchmarks showcase that the mean average precision (mAP) and rotation robustness can be significantly boosted by integrating with our RIDE, with +5.6% mAP and 53% rotation robustness improvement on KITTI, +5.1% and 28% improvement correspondingly on nuScenes. The code will be available soon.
Authors: Alejandro Mestre-Quereda, Juan M. Lopez-Sanchez
Abstract: Speckle suppression in synthetic aperture radar (SAR) images is a key processing step which continues to be a research topic. A wide variety of methods, using either spatially-based approaches or transform-based strategies, have been developed and have shown to provide outstanding results. However, recent advances in deep learning techniques and their application to SAR image despeckling have been demonstrated to offer state-of-the-art results. Unfortunately, they have been mostly applied to single-polarimetric images. The extension of a deep learning-based approach for speckle removal to polarimetric SAR (PolSAR) images is complicated because of the complex nature of the measured covariance matrices for every image pixel, the properties of which must be preserved during filtering. In this work, we propose a complete framework to remove speckle in polarimetric SAR images using a convolutional neural network. The methodology includes a reversible transformation of the original complex covariance matrix to obtain a set of real-valued intensity bands which are fed to the neural network. In addition, the proposed method includes a change detection strategy to avoid the neural network to learn erroneous features in areas strongly affected by temporal changes, so that the network only learns the underlying speckle component present in the data. The method is implemented and tested with dual-polarimetric images acquired by Sentinel-1. Experiments show that the proposed approach offers exceptional results in both speckle reduction and resolution preservation. More importantly, it is also shown that the neural network is not generating artifacts or introducing bias in the filtered images, making them suitable for further polarimetric processing and exploitation.
Authors: Sicheng Liu, Lintao Wang, Xiaogan Zhu, Xuequan Lu, Zhiyong Wang, Kun Hu
Abstract: Extreme Multimodal Summarization with Multimodal Output (XMSMO) becomes an attractive summarization approach by integrating various types of information to create extremely concise yet informative summaries for individual modalities. Existing methods overlook the issue that multimodal data often contains more topic irrelevant information, which can mislead the model into producing inaccurate summaries especially for extremely short ones. In this paper, we propose SITransformer, a Shared Information-guided Transformer for extreme multimodal summarization. It has a shared information guided pipeline which involves a cross-modal shared information extractor and a cross-modal interaction module. The extractor formulates semantically shared salient information from different modalities by devising a novel filtering process consisting of a differentiable top-k selector and a shared-information guided gating unit. As a result, the common, salient, and relevant contents across modalities are identified. Next, a transformer with cross-modal attentions is developed for intra- and inter-modality learning with the shared information guidance to produce the extreme summary. Comprehensive experiments demonstrate that SITransformer significantly enhances the summarization quality for both video and text summaries for XMSMO. Our code will be publicly available at https://github.com/SichengLeoLiu/MMAsia24-XMSMO.
Authors: Xueqing Zhang, Di Fu, Naihao Liu
Abstract: Video key frame extraction is important in various fields, such as video summary, retrieval, and compression. Therefore, we suggest a video key frame extraction algorithm based on shot segmentation using Von Neumann entropy. The segmentation of shots is achieved through the computation of Von Neumann entropy of the similarity matrix among frames within the video sequence. The initial frame of each shot is selected as key frames, which combines the temporal sequence information of frames. The experimental results show the extracted key frames can fully and accurately represent the original video content while minimizing the number of repeated frames.
Authors: Wei-Jhe Huang, Min-Hung Chen, Shang-Hong Lai
Abstract: Spatio-temporal action detection encompasses the tasks of localizing and classifying individual actions within a video. Recent works aim to enhance this process by incorporating interaction modeling, which captures the relationship between people and their surrounding context. However, these approaches have primarily focused on fully-supervised learning, and the current limitation lies in the lack of generalization capability to recognize unseen action categories. In this paper, we aim to adapt the pretrained image-language models to detect unseen actions. To this end, we propose a method which can effectively leverage the rich knowledge of visual-language models to perform Person-Context Interaction. Meanwhile, our Context Prompting module will utilize contextual information to prompt labels, thereby enhancing the generation of more representative text features. Moreover, to address the challenge of recognizing distinct actions by multiple people at the same timestamp, we design the Interest Token Spotting mechanism which employs pretrained visual knowledge to find each person's interest context tokens, and then these tokens will be used for prompting to generate text features tailored to each individual. To evaluate the ability to detect unseen actions, we propose a comprehensive benchmark on J-HMDB, UCF101-24, and AVA datasets. The experiments show that our method achieves superior results compared to previous approaches and can be further extended to multi-action videos, bringing it closer to real-world applications. The code and data can be found in https://webber2933.github.io/ST-CLIP-project-page.
Authors: Jiafei Duan, Wentao Yuan, Wilbert Pumacay, Yi Ru Wang, Kiana Ehsani, Dieter Fox, Ranjay Krishna
Abstract: Large-scale endeavors like and widespread community efforts such as Open-X-Embodiment have contributed to growing the scale of robot demonstration data. However, there is still an opportunity to improve the quality, quantity, and diversity of robot demonstration data. Although vision-language models have been shown to automatically generate demonstration data, their utility has been limited to environments with privileged state information, they require hand-designed skills, and are limited to interactions with few object instances. We propose Manipulate-Anything, a scalable automated generation method for real-world robotic manipulation. Unlike prior work, our method can operate in real-world environments without any privileged state information, hand-designed skills, and can manipulate any static object. We evaluate our method using two setups. First, Manipulate-Anything successfully generates trajectories for all 7 real-world and 14 simulation tasks, significantly outperforming existing methods like VoxPoser. Second, Manipulate-Anything's demonstrations can train more robust behavior cloning policies than training with human demonstrations, or from data generated by VoxPoser, Scaling-up, and Code-As-Policies. We believe Manipulate-Anything can be a scalable method for both generating data for robotics and solving novel tasks in a zero-shot setting. Project page: https://robot-ma.github.io/.
Authors: Aditya K Surikuchi, Raquel Fern\'andez, Sandro Pezzelle
Abstract: Visual storytelling consists in generating a natural language story given a temporally ordered sequence of images. This task is not only challenging for models, but also very difficult to evaluate with automatic metrics since there is no consensus about what makes a story 'good'. In this paper, we introduce a novel method that measures story quality in terms of human likeness regarding three key aspects highlighted in previous work: visual grounding, coherence, and repetitiveness. We then use this method to evaluate the stories generated by several models, showing that the foundation model LLaVA obtains the best result, but only slightly so compared to TAPM, a 50-times smaller visual storytelling model. Upgrading the visual and language components of TAPM results in a model that yields competitive performance with a relatively low number of parameters. Finally, we carry out a human evaluation study, whose results suggest that a 'good' story may require more than a human-like level of visual grounding, coherence, and repetition.
Authors: Yangmin Li, Ruiqi Zhu, Wengen Li
Abstract: Multimodal sentiment analysis is an active research area that combines multiple data modalities, e.g., text, image and audio, to analyze human emotions and benefits a variety of applications. Existing multimodal sentiment analysis methods can be classified as modality interaction-based methods, modality transformation-based methods and modality similarity-based methods. However, most of these methods highly rely on the strong correlations between modalities, and cannot fully uncover and utilize the correlations between modalities to enhance sentiment analysis. Therefore, these methods usually achieve bad performance for identifying the sentiment of multimodal data with weak correlations. To address this issue, we proposed a two-stage semi-supervised model termed Correlation-aware Multimodal Transformer (CorMulT) which consists pre-training stage and prediction stage. At the pre-training stage, a modality correlation contrastive learning module is designed to efficiently learn modality correlation coefficients between different modalities. At the prediction stage, the learned correlation coefficients are fused with modality representations to make the sentiment prediction. According to the experiments on the popular multimodal dataset CMU-MOSEI, CorMulT obviously surpasses state-of-the-art multimodal sentiment analysis methods.
Authors: Anna M. Wundram, Paul Fischer, Michael Muehlebach, Lisa M. Koch, Christian F. Baumgartner
Abstract: Recent works have introduced methods to estimate segmentation performance without ground truth, relying solely on neural network softmax outputs. These techniques hold potential for intuitive output quality control. However, such performance estimates rely on calibrated softmax outputs, which is often not the case in modern neural networks. Moreover, the estimates do not take into account inherent uncertainty in segmentation tasks. These limitations may render precise performance predictions unattainable, restricting the practical applicability of performance estimation methods. To address these challenges, we develop a novel approach for predicting performance ranges with statistical guarantees of containing the ground truth with a user specified probability. Our method leverages sampling-based segmentation uncertainty estimation to derive heuristic performance ranges, and applies split conformal prediction to transform these estimates into rigorous prediction ranges that meet the desired guarantees. We demonstrate our approach on the FIVES retinal vessel segmentation dataset and compare five commonly used sampling-based uncertainty estimation techniques. Our results show that it is possible to achieve the desired coverage with small prediction ranges, highlighting the potential of performance range prediction as a valuable tool for output quality control.
Authors: Zhaobin Li, Patrick Shafto
Abstract: Intent obfuscation is a common tactic in adversarial situations, enabling the attacker to both manipulate the target system and avoid culpability. Surprisingly, it has rarely been implemented in adversarial attacks on machine learning systems. We are the first to propose using intent obfuscation to generate adversarial examples for object detectors: by perturbing another non-overlapping object to disrupt the target object, the attacker hides their intended target. We conduct a randomized experiment on 5 prominent detectors -- YOLOv3, SSD, RetinaNet, Faster R-CNN, and Cascade R-CNN -- using both targeted and untargeted attacks and achieve success on all models and attacks. We analyze the success factors characterizing intent obfuscating attacks, including target object confidence and perturb object sizes. We then demonstrate that the attacker can exploit these success factors to increase success rates for all models and attacks. Finally, we discuss main takeaways and legal repercussions.
Authors: Fazli Wahid, Yingliang Ma, Dawar Khan, Muhammad Aamir, Syed U. K. Bukhari
Abstract: Biomedical image segmentation plays a vital role in diagnosis of diseases across various organs. Deep learning-based object detection methods are commonly used for such segmentation. There exists an extensive research in this topic. However, there is no standard review on this topic. Existing surveys often lack a standardized approach or focus on broader segmentation techniques. In this paper, we conducted a systematic literature review (SLR), collected and analysed 148 articles that explore deep learning object detection methods for biomedical image segmentation. We critically analyzed these methods, identified the key challenges, and discussed the future directions. From the selected articles we extracted the results including the deep learning models, targeted imaging modalities, targeted diseases, and the metrics for the analysis of the methods. The results have been presented in tabular and/or charted forms. The results are presented in three major categories including two stage detection models, one stage detection models and point-based detection models. Each article is individually analyzed along with its pros and cons. Finally, we discuss open challenges, potential benefits, and future research directions. This SLR aims to provide the research community with a quick yet deeper understanding of these segmentation models, ultimately facilitating the development of more powerful solutions for biomedical image analysis.
Authors: Ben Batten, Yang Zheng, Alessandro De Palma, Panagiotis Kouvaros, Alessio Lomuscio
Abstract: We address the problem of verifying neural networks against geometric transformations of the input image, including rotation, scaling, shearing, and translation. The proposed method computes provably sound piecewise linear constraints for the pixel values by using sampling and linear approximations in combination with branch-and-bound Lipschitz optimisation. The method obtains provably tighter over-approximations of the perturbation region than the present state-of-the-art. We report results from experiments on a comprehensive set of verification benchmarks on MNIST and CIFAR10. We show that our proposed implementation resolves up to 32% more verification cases than present approaches.
Authors: Cherag Aroraa, Tracy Holloway King, Jayant Kumar, Yi Lu, Sanat Sharma, Arvind Srikantan, David Uvalle, Josep Valls-Vargas, Harsha Vardhan
Abstract: As user content and queries become increasingly multi-modal, the need for effective multi-modal search systems has grown. Traditional search systems often rely on textual and metadata annotations for indexed images, while multi-modal embeddings like CLIP enable direct search using text and image embeddings. However, embedding-based approaches face challenges in integrating contextual features such as user locale and recency. Building a scalable multi-modal search system requires fine-tuning several components. This paper presents a multi-modal search architecture and a series of AB tests that optimize embeddings and multi-modal technologies in Adobe Express template search. We address considerations such as embedding model selection, the roles of embeddings in matching and ranking, and the balance between dense and sparse embeddings. Our iterative approach demonstrates how utilizing sparse, dense, and contextual features enhances short and long query search, significantly reduces null rates (over 70\%), and increases click-through rates (CTR). Our findings provide insights into developing robust multi-modal search systems, thereby enhancing relevance for complex queries.
Authors: Mikhail Kulyabin, Aline Sindel, Hilde Pedersen, Stuart Gilson, Rigmor Baraas, Andreas Maier
Abstract: Analyzing the cone photoreceptor pattern in images obtained from the living human retina using quantitative methods can be crucial for the early detection and management of various eye conditions. Confocal adaptive optics scanning light ophthalmoscope (AOSLO) imaging enables visualization of the cones from reflections of waveguiding cone photoreceptors. While there have been significant improvements in automated algorithms for segmenting cones in confocal AOSLO images, the process of labelling data remains labor-intensive and manual. This paper introduces a method based on deep learning (DL) for detecting and segmenting cones in AOSLO images. The models were trained on a semi-automatically labelled dataset of 20 AOSLO batches of images of 18 participants for 0$^{\circ}$, 1$^{\circ}$, and 2$^{\circ}$ from the foveal center. F1 scores were 0.968, 0.958, and 0.954 for 0$^{\circ}$, 1$^{\circ}$, and 2$^{\circ}$, respectively, which is better than previously reported DL approaches. Our method minimizes the need for labelled data by only necessitating a fraction of labelled cones, which is especially beneficial in the field of ophthalmology, where labelled data can often be limited.