Authors: Moshira Abdalla, Sajid Javed, Muaz Al Radi, Anwaar Ulhaq, Naoufel Werghi
Abstract: Video anomaly detection (VAD) holds immense importance across diverse domains such as surveillance, healthcare, and environmental monitoring. While numerous surveys focus on conventional VAD methods, they often lack depth in exploring specific approaches and emerging trends. This survey explores deep learning-based VAD, expanding beyond traditional supervised training paradigms to encompass emerging weakly supervised, self-supervised, and unsupervised approaches. A prominent feature of this review is the investigation of core challenges within the VAD paradigms including large-scale datasets, features extraction, learning methods, loss functions, regularization, and anomaly score prediction. Moreover, this review also investigates the vision language models (VLMs) as potent feature extractors for VAD. VLMs integrate visual data with textual descriptions or spoken language from videos, enabling a nuanced understanding of scenes crucial for anomaly detection. By addressing these challenges and proposing future research directions, this review aims to foster the development of robust and efficient VAD systems leveraging the capabilities of VLMs for enhanced anomaly detection in complex real-world scenarios. This comprehensive analysis seeks to bridge existing knowledge gaps, provide researchers with valuable insights, and contribute to shaping the future of VAD research.
Authors: Heesup Yun, Sassoum Lo, Christine H. Diepenbrock, Brian N. Bailey, J. Mason Earles
Abstract: Thermal cameras are an important tool for agricultural research because they allow for non-invasive measurement of plant temperature, which relates to important photochemical, hydraulic, and agronomic traits. Utilizing low-cost thermal cameras can lower the barrier to introducing thermal imaging in agricultural research and production. This paper presents an approach to improve the temperature accuracy and image quality of low-cost thermal imaging cameras for agricultural applications. Leveraging advancements in computer vision techniques, particularly deep learning networks, we propose a method, called $\textbf{VisTA-SR}$ ($\textbf{Vis}$ual \& $\textbf{T}$hermal $\textbf{A}$lignment and $\textbf{S}$uper-$\textbf{R}$esolution Enhancement) that combines RGB and thermal images to enhance the capabilities of low-resolution thermal cameras. The research includes calibration and validation of temperature measurements, acquisition of paired image datasets, and the development of a deep learning network tailored for agricultural thermal imaging. Our study addresses the challenges of image enhancement in the agricultural domain and explores the potential of low-cost thermal cameras to replace high-resolution industrial cameras. Experimental results demonstrate the effectiveness of our approach in enhancing temperature accuracy and image sharpness, paving the way for more accessible and efficient thermal imaging solutions in agriculture.
Authors: Artemis Panagopoulou, Coby Melkin, Chris Callison-Burch
Abstract: Bistable images, also known as ambiguous or reversible images, present visual stimuli that can be seen in two distinct interpretations, though not simultaneously by the observer. In this study, we conduct the most extensive examination of vision-language models using bistable images to date. We manually gathered a dataset of 29 bistable images, along with their associated labels, and subjected them to 116 different manipulations in brightness, tint, and rotation. We evaluated twelve different models in both classification and generative tasks across six model architectures. Our findings reveal that, with the exception of models from the Idefics family and LLaVA1.5-13b, there is a pronounced preference for one interpretation over another among the models, and minimal variance under image manipulations, with few exceptions on image rotations. Additionally, we compared the model preferences with humans, noting that the models do not exhibit the same continuity biases as humans and often diverge from human initial interpretations. We also investigated the influence of variations in prompts and the use of synonymous labels, discovering that these factors significantly affect model interpretations more than image manipulations showing a higher influence of the language priors on bistable image interpretations compared to image-text training data. All code and data is open sourced.
Authors: Yipu Chen, Haotian Xue, Yongxin Chen
Abstract: Diffusion models (DMs) have emerged as a promising approach for behavior cloning (BC). Diffusion policies (DP) based on DMs have elevated BC performance to new heights, demonstrating robust efficacy across diverse tasks, coupled with their inherent flexibility and ease of implementation. Despite the increasing adoption of DP as a foundation for policy generation, the critical issue of safety remains largely unexplored. While previous attempts have targeted deep policy networks, DP used diffusion models as the policy network, making it ineffective to be attacked using previous methods because of its chained structure and randomness injected. In this paper, we undertake a comprehensive examination of DP safety concerns by introducing adversarial scenarios, encompassing offline and online attacks, and global and patch-based attacks. We propose DP-Attacker, a suite of algorithms that can craft effective adversarial attacks across all aforementioned scenarios. We conduct attacks on pre-trained diffusion policies across various manipulation tasks. Through extensive experiments, we demonstrate that DP-Attacker has the capability to significantly decrease the success rate of DP for all scenarios. Particularly in offline scenarios, DP-Attacker can generate highly transferable perturbations applicable to all frames. Furthermore, we illustrate the creation of adversarial physical patches that, when applied to the environment, effectively deceive the model. Video results are put in: https://sites.google.com/view/diffusion-policy-attacker.
URLs: https://sites.google.com/view/diffusion-policy-attacker.
Authors: Marcos L\'opez-De-Castro (DATAI - Institute of Data Science and Artificial Intelligence, Universidad de Navarra, Pamplona, Spain, TECNUN School of Engineering, Universidad de Navarra, Donostia-San Sebastian, Spain), Alberto Garc\'ia-Galindo (DATAI - Institute of Data Science and Artificial Intelligence, Universidad de Navarra, Pamplona, Spain, TECNUN School of Engineering, Universidad de Navarra, Donostia-San Sebastian, Spain), Rub\'en Arma\~nanzas (DATAI - Institute of Data Science and Artificial Intelligence, Universidad de Navarra, Pamplona, Spain, TECNUN School of Engineering, Universidad de Navarra, Donostia-San Sebastian, Spain)
Abstract: Unlike traditional statistical methods, Conformal Prediction (CP) allows for the determination of valid and accurate confidence levels associated with individual predictions based only on exchangeability of the data. We here introduce a new feature selection method that takes advantage of the CP framework. Our proposal, named Conformal Recursive Feature Elimination (CRFE), identifies and recursively removes features that increase the non-conformity of a dataset. We also present an automatic stopping criterion for CRFE, as well as a new index to measure consistency between subsets of features. CRFE selections are compared to the classical Recursive Feature Elimination (RFE) method on several multiclass datasets by using multiple partitions of the data. The results show that CRFE clearly outperforms RFE in half of the datasets, while achieving similar performance in the rest. The automatic stopping criterion provides subsets of effective and non-redundant features without computing any classification performance.
Authors: Ningli Xu, Rongjun Qin
Abstract: Generating wide-area digital surface models (DSMs) requires registering a large number of individual, and partially overlapped DSMs. This presents a challenging problem for a typical registration algorithm, since when a large number of observations from these multiple DSMs are considered, it may easily cause memory overflow. Sequential registration algorithms, although can significantly reduce the computation, are especially vulnerable for small overlapped pairs, leading to a large error accumulation. In this work, we propose a novel solution that builds the DSM registration task as a motion averaging problem: pair-wise DSMs are registered to build a scene graph, with edges representing relative poses between DSMs. Specifically, based on the grid structure of the large DSM, the pair-wise registration is performed using a novel nearest neighbor search method. We show that the scene graph can be optimized via an extremely fast motion average algorithm with O(N) complexity (N refers to the number of images). Evaluation of high-resolution satellite-derived DSM demonstrates significant improvement in computation and accuracy.
Authors: Dong Li, Yidi Liu, Xueyang Fu, Senyan Xu, Zheng-Jun Zha
Abstract: Image deraining aims to remove rain streaks from rainy images and restore clear backgrounds. Currently, some research that employs the Fourier transform has proved to be effective for image deraining, due to it acting as an effective frequency prior for capturing rain streaks. However, despite there exists dependency of low frequency and high frequency in images, these Fourier-based methods rarely exploit the correlation of different frequencies for conjuncting their learning procedures, limiting the full utilization of frequency information for image deraining. Alternatively, the recently emerged Mamba technique depicts its effectiveness and efficiency for modeling correlation in various domains (e.g., spatial, temporal), and we argue that introducing Mamba into its unexplored Fourier spaces to correlate different frequencies would help improve image deraining. This motivates us to propose a new framework termed FourierMamba, which performs image deraining with Mamba in the Fourier space. Owning to the unique arrangement of frequency orders in Fourier space, the core of FourierMamba lies in the scanning encoding of different frequencies, where the low-high frequency order formats exhibit differently in the spatial dimension (unarranged in axis) and channel dimension (arranged in axis). Therefore, we design FourierMamba that correlates Fourier space information in the spatial and channel dimensions with distinct designs. Specifically, in the spatial dimension Fourier space, we introduce the zigzag coding to scan the frequencies to rearrange the orders from low to high frequencies, thereby orderly correlating the connections between frequencies; in the channel dimension Fourier space with arranged orders of frequencies in axis, we can directly use Mamba to perform frequency correlation and improve the channel information representation.
Authors: Raman Dutt, Pedro Sanchez, Ondrej Bohdal, Sotirios A. Tsaftaris, Timothy Hospedales
Abstract: Diffusion models show a remarkable ability in generating images that closely mirror the training distribution. However, these models are prone to training data memorization, leading to significant privacy, ethical, and legal concerns, particularly in sensitive fields such as medical imaging. We hypothesize that memorization is driven by the overparameterization of deep models, suggesting that regularizing model capacity during fine-tuning could be an effective mitigation strategy. Parameter-efficient fine-tuning (PEFT) methods offer a promising approach to capacity control by selectively updating specific parameters. However, finding the optimal subset of learnable parameters that balances generation quality and memorization remains elusive. To address this challenge, we propose a bi-level optimization framework that guides automated parameter selection by utilizing memorization and generation quality metrics as rewards. Our framework successfully identifies the optimal parameter set to be updated to satisfy the generation-memorization tradeoff. We perform our experiments for the specific task of medical image generation and outperform existing state-of-the-art training-time mitigation strategies by fine-tuning as few as 0.019% of model parameters. Furthermore, we show that the strategies learned through our framework are transferable across different datasets and domains. Our proposed framework is scalable to large datasets and agnostic to the choice of reward functions. Finally, we show that our framework can be combined with existing approaches for further memorization mitigation.
Authors: Meng Cao, Haoran Tang, Jinfa Huang, Peng Jin, Can Zhang, Ruyang Liu, Long Chen, Xiaodan Liang, Li Yuan, Ge Li
Abstract: Text-Video Retrieval (TVR) aims to align relevant video content with natural language queries. To date, most state-of-the-art TVR methods learn image-to-video transfer learning based on large-scale pre-trained visionlanguage models (e.g., CLIP). However, fully fine-tuning these pre-trained models for TVR incurs prohibitively expensive computation costs. To this end, we propose to conduct efficient text-video Retrieval with a sparse-andcorrelated AdaPter (RAP), i.e., fine-tuning the pre-trained model with a few parameterized layers. To accommodate the text-video scenario, we equip our RAP with two indispensable characteristics: temporal sparsity and correlation. Specifically, we propose a low-rank modulation module to refine the per-image features from the frozen CLIP backbone, which accentuates salient frames within the video features while alleviating temporal redundancy. Besides, we introduce an asynchronous self-attention mechanism that first selects the top responsive visual patches and augments the correlation modeling between them with learnable temporal and patch offsets. Extensive experiments on four TVR datasets demonstrate that RAP achieves superior or comparable performance compared to the fully fine-tuned counterpart and other parameter-efficient fine-tuning methods.
Authors: Polezhaev Ignat, Goncharenko Igor, Iurina Natalya
Abstract: In this paper, we present a novel methodology we call MDS-ViTNet (Multi Decoder Saliency by Vision Transformer Network) for enhancing visual saliency prediction or eye-tracking. This approach holds significant potential for diverse fields, including marketing, medicine, robotics, and retail. We propose a network architecture that leverages the Vision Transformer, moving beyond the conventional ImageNet backbone. The framework adopts an encoder-decoder structure, with the encoder utilizing a Swin transformer to efficiently embed most important features. This process involves a Transfer Learning method, wherein layers from the Vision Transformer are converted by the Encoder Transformer and seamlessly integrated into a CNN Decoder. This methodology ensures minimal information loss from the original input image. The decoder employs a multi-decoding technique, utilizing dual decoders to generate two distinct attention maps. These maps are subsequently combined into a singular output via an additional CNN model. Our trained model MDS-ViTNet achieves state-of-the-art results across several benchmarks. Committed to fostering further collaboration, we intend to make our code, models, and datasets accessible to the public.
Authors: Islam Osman, Mohamed S. Shehata
Abstract: Current state-of-the-art video object segmentation models have achieved great success using supervised learning with massive labeled training datasets. However, these models are trained using a single source domain and evaluated using videos sampled from the same source domain. When these models are evaluated using videos sampled from a different target domain, their performance degrades significantly due to poor domain generalization, i.e., their inability to learn from multi-domain sources simultaneously using traditional supervised learning. In this paper, We propose a dynamically growing tree of sub-networks (DGT) to learn effectively from multi-domain sources. DGT uses a novel lifelong learning technique that allows the model to continuously and effectively learn from new domains without forgetting the previously learned domains. Hence, the model can generalize to out-of-domain videos. The proposed work is evaluated using single-source in-domain (traditional video object segmentation), multi-source in-domain, and multi-source out-of-domain video object segmentation. The results of DGT show a single source in-domain performance gain of 0.2% and 3.5% on the DAVIS16 and DAVIS17 datasets, respectively. However, when DGT is evaluated using in-domain multi-sources, the results show superior performance compared to state-of-the-art video object segmentation and other lifelong learning techniques with an average performance increase in the F-score of 6.9% with minimal catastrophic forgetting. Finally, in the out-of-domain experiment, the performance of DGT is 2.7% and 4% better than state-of-the-art in 1 and 5-shots, respectively.
Authors: Lianlei Shan, Wenzhang Zhou, Wei Li, Xingyu Ding
Abstract: The goal of incremental Few-shot Semantic Segmentation (iFSS) is to extend pre-trained segmentation models to new classes via few annotated images without access to old training data. During incrementally learning novel classes, the data distribution of old classes will be destroyed, leading to catastrophic forgetting. Meanwhile, the novel classes have only few samples, making models impossible to learn the satisfying representations of novel classes. For the iFSS problem, we propose a network called OINet, i.e., the background embedding space \textbf{O}rganization and prototype \textbf{I}nherit Network. Specifically, when training base classes, OINet uses multiple classification heads for the background and sets multiple sub-class prototypes to reserve embedding space for the latent novel classes. During incrementally learning novel classes, we propose a strategy to select the sub-class prototypes that best match the current learning novel classes and make the novel classes inherit the selected prototypes' embedding space. This operation allows the novel classes to be registered in the embedding space using few samples without affecting the distribution of the base classes. Results on Pascal-VOC and COCO show that OINet achieves a new state of the art.
Authors: Vaibhav Vavilala, Florian Kluger, Seemandhar Jain, Bodo Rosenhahn, David Forsyth
Abstract: Describing a scene in terms of primitives -- geometrically simple shapes that offer a parsimonious but accurate abstraction of structure -- is an established vision problem. This is a good model of a difficult fitting problem: different scenes require different numbers of primitives and primitives interact strongly, but any proposed solution can be evaluated at inference time. The state of the art method involves a learned regression procedure to predict a start point consisting of a fixed number of primitives, followed by a descent method to refine the geometry and remove redundant primitives. Methods are evaluated by accuracy in depth and normal prediction and in scene segmentation. This paper shows that very significant improvements in accuracy can be obtained by (a) incorporating a small number of negative primitives and (b) ensembling over a number of different regression procedures. Ensembling is by refining each predicted start point, then choosing the best by fitting loss. Extensive experiments on a standard dataset confirm that negative primitives are useful in a large fraction of images, and that our refine-then-choose strategy outperforms choose-then-refine, confirming that the fitting problem is very difficult.
Authors: Hamadi Chihaoui, Abdelhak Lemkhenter, Paolo Favaro
Abstract: Recently, various methods have been proposed to solve Image Restoration (IR) tasks using a pre-trained diffusion model leading to state-of-the-art performance. However, most of these methods assume that the degradation operator in the IR task is completely known. Furthermore, a common characteristic among these approaches is that they alter the diffusion sampling process in order to satisfy the consistency with the degraded input image. This choice has recently been shown to be sub-optimal and to cause the restored image to deviate from the data manifold. To address these issues, we propose Blind Image Restoration via fast Diffusion inversion (BIRD) a blind IR method that jointly optimizes for the degradation model parameters and the restored image. To ensure that the restored images lie onto the data manifold, we propose a novel sampling technique on a pre-trained diffusion model. A key idea in our method is not to modify the reverse sampling, i.e., not to alter all the intermediate latents, once an initial noise is sampled. This is ultimately equivalent to casting the IR task as an optimization problem in the space of the input noise. Moreover, to mitigate the computational cost associated with inverting a fully unrolled diffusion model, we leverage the inherent capability of these models to skip ahead in the forward diffusion process using large time steps. We experimentally validate BIRD on several image restoration tasks and show that it achieves state of the art performance on all of them. Our code is available at https://github.com/hamadichihaoui/BIRD.
Authors: Junjie Zhang, Chenjia Bai, Haoran He, Wenke Xia, Zhigang Wang, Bin Zhao, Xiu Li, Xuelong Li
Abstract: Acquiring a multi-task imitation policy in 3D manipulation poses challenges in terms of scene understanding and action prediction. Current methods employ both 3D representation and multi-view 2D representation to predict the poses of the robot's end-effector. However, they still require a considerable amount of high-quality robot trajectories, and suffer from limited generalization in unseen tasks and inefficient execution in long-horizon reasoning. In this paper, we propose SAM-E, a novel architecture for robot manipulation by leveraging a vision-foundation model for generalizable scene understanding and sequence imitation for long-term action reasoning. Specifically, we adopt Segment Anything (SAM) pre-trained on a huge number of images and promptable masks as the foundation model for extracting task-relevant features, and employ parameter-efficient fine-tuning on robot data for a better understanding of embodied scenarios. To address long-horizon reasoning, we develop a novel multi-channel heatmap that enables the prediction of the action sequence in a single pass, notably enhancing execution efficiency. Experimental results from various instruction-following tasks demonstrate that SAM-E achieves superior performance with higher execution efficiency compared to the baselines, and also significantly improves generalization in few-shot adaptation to new tasks.
Authors: Jeffrey D. Rudie, Hui-Ming Lin, Robyn L. Ball, Sabeena Jalal, Luciano M. Prevedello, Savvas Nicolaou, Brett S. Marinelli, Adam E. Flanders, Kirti Magudia, George Shih, Melissa A. Davis, John Mongan, Peter D. Chang, Ferco H. Berger, Sebastiaan Hermans, Meng Law, Tyler Richards, Jan-Peter Grunz, Andreas Steven Kunz, Shobhit Mathur, Sandro Galea-Soler, Andrew D. Chung, Saif Afat, Chin-Chi Kuo, Layal Aweidah, Ana Villanueva Campos, Arjuna Somasundaram, Felipe Antonio Sanchez Tijmes, Attaporn Jantarangkoon, Leonardo Kayat Bittencourt, Michael Brassil, Ayoub El Hajjami, Hakan Dogan, Muris Becircic, Agrahara G. Bharatkumar, Eduardo Moreno J\'udice de Mattos Farina, Dataset Curator Group, Dataset Contributor Group, Dataset Annotator Group, Errol Colak
Abstract: The RSNA Abdominal Traumatic Injury CT (RATIC) dataset is the largest publicly available collection of adult abdominal CT studies annotated for traumatic injuries. This dataset includes 4,274 studies from 23 institutions across 14 countries. The dataset is freely available for non-commercial use via Kaggle at https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection. Created for the RSNA 2023 Abdominal Trauma Detection competition, the dataset encourages the development of advanced machine learning models for detecting abdominal injuries on CT scans. The dataset encompasses detection and classification of traumatic injuries across multiple organs, including the liver, spleen, kidneys, bowel, and mesentery. Annotations were created by expert radiologists from the American Society of Emergency Radiology (ASER) and Society of Abdominal Radiology (SAR). The dataset is annotated at multiple levels, including the presence of injuries in three solid organs with injury grading, image-level annotations for active extravasations and bowel injury, and voxelwise segmentations of each of the potentially injured organs. With the release of this dataset, we hope to facilitate research and development in machine learning and abdominal trauma that can lead to improved patient care and outcomes.
URLs: https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection.
Authors: Yujiao Jiang, Qingmin Liao, Zhaolong Wang, Xiangru Lin, Zongqing Lu, Yuxi Zhao, Hanqing Wei, Jingrui Ye, Yu Zhang, Zhijing Shao
Abstract: Recovering photorealistic and drivable full-body avatars is crucial for numerous applications, including virtual reality, 3D games, and tele-presence. Most methods, whether reconstruction or generation, require large numbers of human motion sequences and corresponding textured meshes. To easily learn a drivable avatar, a reasonable parametric body model with unified topology is paramount. However, existing human body datasets either have images or textured models and lack parametric models which fit clothes well. We propose a new parametric model SMPLX-Lite-D, which can fit detailed geometry of the scanned mesh while maintaining stable geometry in the face, hand and foot regions. We present SMPLX-Lite dataset, the most comprehensive clothing avatar dataset with multi-view RGB sequences, keypoints annotations, textured scanned meshes, and textured SMPLX-Lite-D models. With the SMPLX-Lite dataset, we train a conditional variational autoencoder model that takes human pose and facial keypoints as input, and generates a photorealistic drivable human avatar.
Authors: Wenchao Sun, Xuewu Lin, Yining Shi, Chuang Zhang, Haoran Wu, Sifa Zheng
Abstract: The well-established modular autonomous driving system is decoupled into different standalone tasks, e.g. perception, prediction and planning, suffering from information loss and error accumulation across modules. In contrast, end-to-end paradigms unify multi-tasks into a fully differentiable framework, allowing for optimization in a planning-oriented spirit. Despite the great potential of end-to-end paradigms, both the performance and efficiency of existing methods are not satisfactory, particularly in terms of planning safety. We attribute this to the computationally expensive BEV (bird's eye view) features and the straightforward design for prediction and planning. To this end, we explore the sparse representation and review the task design for end-to-end autonomous driving, proposing a new paradigm named SparseDrive. Concretely, SparseDrive consists of a symmetric sparse perception module and a parallel motion planner. The sparse perception module unifies detection, tracking and online mapping with a symmetric model architecture, learning a fully sparse representation of the driving scene. For motion prediction and planning, we review the great similarity between these two tasks, leading to a parallel design for motion planner. Based on this parallel design, which models planning as a multi-modal problem, we propose a hierarchical planning selection strategy , which incorporates a collision-aware rescore module, to select a rational and safe trajectory as the final planning output. With such effective designs, SparseDrive surpasses previous state-of-the-arts by a large margin in performance of all tasks, while achieving much higher training and inference efficiency. Code will be avaliable at https://github.com/swc-17/SparseDrive for facilitating future research.
Authors: Jos\'e Ignacio D\'iaz Villa, Patricio Loncomilla, Javier Ruiz-del-Solar
Abstract: This paper introduces YotoR (You Only Transform One Representation), a novel deep learning model for object detection that combines Swin Transformers and YoloR architectures. Transformers, a revolutionary technology in natural language processing, have also significantly impacted computer vision, offering the potential to enhance accuracy and computational efficiency. YotoR combines the robust Swin Transformer backbone with the YoloR neck and head. In our experiments, YotoR models TP5 and BP4 consistently outperform YoloR P6 and Swin Transformers in various evaluations, delivering improved object detection performance and faster inference speeds than Swin Transformer models. These results highlight the potential for further model combinations and improvements in real-time object detection with Transformers. The paper concludes by emphasizing the broader implications of YotoR, including its potential to enhance transformer-based models for image-related tasks.
Authors: Xinyang Huang, Chuang Zhu, Kebin Liu, Ruiying Ren, Shengjie Liu
Abstract: Existing few-shot segmentation (FSS) only considers learning support-query correlation and segmenting unseen categories under the precise pixel masks. However, the cost of a large number of pixel masks during training is expensive. This paper considers a more challenging scenario, weakly-supervised few-shot segmentation (WS-FSS), which only provides category ($i.e.$ image-level) labels. It requires the model to learn robust support-query information when the generated mask is inaccurate. In this work, we design a Correlation Enhancement Network (CORENet) with foundation model, which utilizes multi-information guidance to learn robust correlation. Specifically, correlation-guided transformer (CGT) utilizes self-supervised ViT tokens to learn robust correlation from both local and global perspectives. From the perspective of semantic categories, the class-guided module (CGM) guides the model to locate valuable correlations through the pre-trained CLIP. Finally, the embedding-guided module (EGM) implicitly guides the model to supplement the inevitable information loss during the correlation learning by the original appearance embedding and finally generates the query mask. Extensive experiments on PASCAL-5$^i$ and COCO-20$^i$ have shown that CORENet exhibits excellent performance compared to existing methods.
Authors: Ryo Fujii, Masashi Hatano, Hideo Saito, Hiroki Kajita
Abstract: Surgical phase recognition has gained significant attention due to its potential to offer solutions to numerous demands of the modern operating room. However, most existing methods concentrate on minimally invasive surgery (MIS), leaving surgical phase recognition for open surgery understudied. This discrepancy is primarily attributed to the scarcity of publicly available open surgery video datasets for surgical phase recognition. To address this issue, we introduce a new egocentric open surgery video dataset for phase recognition, named EgoSurgery-Phase. This dataset comprises 15 hours of real open surgery videos spanning 9 distinct surgical phases all captured using an egocentric camera attached to the surgeon's head. In addition to video, the EgoSurgery-Phase offers eye gaze. As far as we know, it is the first real open surgery video dataset for surgical phase recognition publicly available. Furthermore, inspired by the notable success of masked autoencoders (MAEs) in video understanding tasks (e.g., action recognition), we propose a gaze-guided masked autoencoder (GGMAE). Considering the regions where surgeons' gaze focuses are often critical for surgical phase recognition (e.g., surgical field), in our GGMAE, the gaze information acts as an empirical semantic richness prior to guiding the masking process, promoting better attention to semantically rich spatial regions. GGMAE significantly improves the previous state-of-the-art recognition method (6.4% in Jaccard) and the masked autoencoder-based method (3.1% in Jaccard) on EgoSurgery-Phase. The dataset will be released at https://github.com/Fujiry0/EgoSurgery.
Authors: David Ferman, Pablo Garrido, Gaurav Bharaj
Abstract: 3D facial landmark localization has proven to be of particular use for applications, such as face tracking, 3D face modeling, and image-based 3D face reconstruction. In the supervised learning case, such methods usually rely on 3D landmark datasets derived from 3DMM-based registration that often lack spatial definition alignment, as compared with that chosen by hand-labeled human consensus, e.g., how are eyebrow landmarks defined? This creates a gap between landmark datasets generated via high-quality 2D human labels and 3DMMs, and it ultimately limits their effectiveness. To address this issue, we introduce a novel semi-supervised learning approach that learns 3D landmarks by directly lifting (visible) hand-labeled 2D landmarks and ensures better definition alignment, without the need for 3D landmark datasets. To lift 2D landmarks to 3D, we leverage 3D-aware GANs for better multi-view consistency learning and in-the-wild multi-frame videos for robust cross-generalization. Empirical experiments demonstrate that our method not only achieves better definition alignment between 2D-3D landmarks but also outperforms other supervised learning 3D landmark localization methods on both 3DMM labeled and photogrammetric ground truth evaluation datasets. Project Page: https://davidcferman.github.io/FaceLift
Authors: Xiaolong Yu, Cong Tian
Abstract: Although deep convolutional neural networks have achieved rapid development, it is challenging to widely promote and apply these models on low-power devices, due to computational and storage limitations. To address this issue, researchers have proposed techniques such as model compression, activation sparsity induction, and hardware accelerators. This paper presents a method to induce the sparsity of activation maps based on Transformed $\ell1$ regularization, so as to improve the research in the field of activation sparsity induction. Further, the method is innovatively combined with traditional pruning, constituting a dual sparse training framework. Compared to previous methods, Transformed $\ell1$ can achieve higher sparsity and better adapt to different network structures. Experimental results show that the method achieves improvements by more than 20\% in activation map sparsity on most models and corresponding datasets without compromising the accuracy. Specifically, it achieves a 27.52\% improvement for ResNet18 on the ImageNet dataset, and a 44.04\% improvement for LeNet5 on the MNIST dataset. In addition, the dual sparse training framework can greatly reduce the computational load and provide potential for reducing the required storage during runtime. Specifically, the ResNet18 and ResNet50 models obtained by the dual sparse training framework respectively reduce 81.7\% and 84.13\% of multiplicative floating-point operations, while maintaining accuracy and a low pruning rate.
Authors: Wei Sun, Qi Zhang, Yanzhao Zhou, Qixiang Ye, Jianbin Jiao, Yuan Li
Abstract: 3D Gaussian splatting has demonstrated impressive performance in real-time novel view synthesis. However, achieving successful reconstruction from RGB images generally requires multiple input views captured under static conditions. To address the challenge of sparse input views, previous approaches have incorporated depth supervision into the training of 3D Gaussians to mitigate overfitting, using dense predictions from pretrained depth networks as pseudo-ground truth. Nevertheless, depth predictions from monocular depth estimation models inherently exhibit significant uncertainty in specific areas. Relying solely on pixel-wise L2 loss may inadvertently incorporate detrimental noise from these uncertain areas. In this work, we introduce a novel method to supervise the depth distribution of 3D Gaussians, utilizing depth priors with integrated uncertainty estimates. To address these localized errors in depth predictions, we integrate a patch-wise optimal transport strategy to complement traditional L2 loss in depth supervision. Extensive experiments conducted on the LLFF, DTU, and Blender datasets demonstrate that our approach, UGOT, achieves superior novel view synthesis and consistently outperforms state-of-the-art methods.
Authors: Yilin Liu, Xuezhou Guo, Xinqi Wang, Fangzhou Du
Abstract: Our project proposes an end-to-end 3D face alignment and reconstruction network. The backbone of our model is built by Bottle-Neck structure via Depth-wise Separable Convolution. We integrate Coordinate Attention mechanism and Spatial Group-wise Enhancement to extract more representative features. For more stable training process and better convergence, we jointly use Wing loss and the Weighted Parameter Distance Cost to learn parameters for 3D Morphable model and 3D vertices. Our proposed model outperforms all baseline models both quantitatively and qualitatively.
Authors: Jiawei Chen, Xiao Yang, Zhengwei Fang, Yu Tian, Yinpeng Dong, Zhaoxia Yin, Hang Su
Abstract: Despite the widespread application of large language models (LLMs) across various tasks, recent studies indicate that they are susceptible to jailbreak attacks, which can render their defense mechanisms ineffective. However, previous jailbreak research has frequently been constrained by limited universality, suboptimal efficiency, and a reliance on manual crafting. In response, we rethink the approach to jailbreaking LLMs and formally define three essential properties from the attacker' s perspective, which contributes to guiding the design of jailbreak methods. We further introduce AutoBreach, a novel method for jailbreaking LLMs that requires only black-box access. Inspired by the versatility of wordplay, AutoBreach employs a wordplay-guided mapping rule sampling strategy to generate a variety of universal mapping rules for creating adversarial prompts. This generation process leverages LLMs' automatic summarization and reasoning capabilities, thus alleviating the manual burden. To boost jailbreak success rates, we further suggest sentence compression and chain-of-thought-based mapping rules to correct errors and wordplay misinterpretations in target LLMs. Additionally, we propose a two-stage mapping rule optimization strategy that initially optimizes mapping rules before querying target LLMs to enhance the efficiency of AutoBreach. AutoBreach can efficiently identify security vulnerabilities across various LLMs, including three proprietary models: Claude-3, GPT-3.5, GPT-4 Turbo, and two LLMs' web platforms: Bingchat, GPT-4 Web, achieving an average success rate of over 80% with fewer than 10 queries
Authors: Lei Xiong, Xin Luo, Zihao Wang, Chaofan He, Shuyuan Zhu, Bing Zeng
Abstract: With the rapid development of machine vision technology in recent years, many researchers have begun to focus on feature compression that is better suited for machine vision tasks. The target of feature compression is deep features, which arise from convolution in the middle layer of a pre-trained convolutional neural network. However, due to the large volume of data and high level of abstraction of deep features, their application is primarily limited to machine-centric scenarios, which poses significant constraints in situations requiring human-computer interaction. This paper investigates features and textures and proposes a texture-guided feature compression strategy based on their characteristics. Specifically, the strategy comprises feature layers and texture layers. The feature layers serve the machine, including a feature selection module and a feature reconstruction network. With the assistance of texture images, they selectively compress and transmit channels relevant to visual tasks, reducing feature data while providing high-quality features for the machine. The texture layers primarily serve humans and consist of an image reconstruction network. This image reconstruction network leverages features and texture images to reconstruct preview images for humans. Our method fully exploits the characteristics of texture and features. It eliminates feature redundancy, reconstructs high-quality preview images for humans, and supports decision-making. The experimental results demonstrate excellent performance when employing our proposed method to compress the deep features.
Authors: Haodong Xiang, Xinghui Li, Xiansong Lai, Wanting Zhang, Zhichao Liao, Kai Cheng, Xueping Liu
Abstract: Recently, 3D Gaussian Splatting(3DGS) has revolutionized neural rendering with its high-quality rendering and real-time speed. However, when it comes to indoor scenes with a significant number of textureless areas, 3DGS yields incomplete and noisy reconstruction results due to the poor initialization of the point cloud and under-constrained optimization. Inspired by the continuity of signed distance field (SDF), which naturally has advantages in modeling surfaces, we present a unified optimizing framework integrating neural SDF with 3DGS. This framework incorporates a learnable neural SDF field to guide the densification and pruning of Gaussians, enabling Gaussians to accurately model scenes even with poor initialized point clouds. At the same time, the geometry represented by Gaussians improves the efficiency of the SDF field by piloting its point sampling. Additionally, we regularize the optimization with normal and edge priors to eliminate geometry ambiguity in textureless areas and improve the details. Extensive experiments in ScanNet and ScanNet++ show that our method achieves state-of-the-art performance in both surface reconstruction and novel view synthesis.
Authors: Ankush Gajanan Arudkar, Bernard J. E. Evans
Abstract: Accurate detection of colorectal cancer and early prevention heavily rely on precise polyp identification during gastrointestinal colonoscopy. Due to limited data, many current state-of-the-art deep learning methods for polyp segmentation often rely on post-processing of masks to reduce noise and enhance results. In this study, we propose an approach that integrates mask refinement and binary semantic segmentation, leveraging a novel collaborative training strategy that surpasses current widely-used refinement strategies. We demonstrate the superiority of our approach through comprehensive evaluation on established benchmark datasets and its successful application across various medical image segmentation architectures.
Authors: Aisha Urooj Khan, John Garrett, Tyler Bradshaw, Lonie Salkowski, Jiwoong Jason Jeong, Amara Tariq, Imon Banerjee
Abstract: A visual-language model (VLM) pre-trained on natural images and text pairs poses a significant barrier when applied to medical contexts due to domain shift. Yet, adapting or fine-tuning these VLMs for medical use presents considerable hurdles, including domain misalignment, limited access to extensive datasets, and high-class imbalances. Hence, there is a pressing need for strategies to effectively adapt these VLMs to the medical domain, as such adaptations would prove immensely valuable in healthcare applications. In this study, we propose a framework designed to adeptly tailor VLMs to the medical domain, employing selective sampling and hard-negative mining techniques for enhanced performance in retrieval tasks. We validate the efficacy of our proposed approach by implementing it across two distinct VLMs: the in-domain VLM (MedCLIP) and out-of-domain VLMs (ALBEF). We assess the performance of these models both in their original off-the-shelf state and after undergoing our proposed training strategies, using two extensive datasets containing mammograms and their corresponding reports. Our evaluation spans zero-shot, few-shot, and supervised scenarios. Through our approach, we observe a notable enhancement in Recall@K performance for the image-text retrieval task.
Authors: Haodi He, Colton Stearns, Adam W. Harley, Leonidas J. Guibas
Abstract: Large-scale vision foundation models such as Segment Anything (SAM) demonstrate impressive performance in zero-shot image segmentation at multiple levels of granularity. However, these zero-shot predictions are rarely 3D-consistent. As the camera viewpoint changes in a scene, so do the segmentation predictions, as well as the characterizations of ``coarse" or ``fine" granularity. In this work, we address the challenging task of lifting multi-granular and view-inconsistent image segmentations into a hierarchical and 3D-consistent representation. We learn a novel feature field within a Neural Radiance Field (NeRF) representing a 3D scene, whose segmentation structure can be revealed at different scales by simply using different thresholds on feature distance. Our key idea is to learn an ultrametric feature space, which unlike a Euclidean space, exhibits transitivity in distance-based grouping, naturally leading to a hierarchical clustering. Put together, our method takes view-inconsistent multi-granularity 2D segmentations as input and produces a hierarchy of 3D-consistent segmentations as output. We evaluate our method and several baselines on synthetic datasets with multi-view images and multi-granular segmentation, showcasing improved accuracy and viewpoint-consistency. We additionally provide qualitative examples of our model's 3D hierarchical segmentations in real world scenes.\footnote{The code and dataset are available at:
Authors: Hongbin Lin, Yifan Zhang, Shuaicheng Niu, Shuguang Cui, Zhen Li
Abstract: Monocular 3D object detection (Mono 3Det) aims to identify 3D objects from a single RGB image. However, existing methods often assume training and test data follow the same distribution, which may not hold in real-world test scenarios. To address the out-of-distribution (OOD) problems, we explore a new adaptation paradigm for Mono 3Det, termed Fully Test-time Adaptation. It aims to adapt a well-trained model to unlabeled test data by handling potential data distribution shifts at test time without access to training data and test labels. However, applying this paradigm in Mono 3Det poses significant challenges due to OOD test data causing a remarkable decline in object detection scores. This decline conflicts with the pre-defined score thresholds of existing detection methods, leading to severe object omissions (i.e., rare positive detections and many false negatives). Consequently, the limited positive detection and plenty of noisy predictions cause test-time adaptation to fail in Mono 3Det. To handle this problem, we propose a novel Monocular Test-Time Adaptation (MonoTTA) method, based on two new strategies. 1) Reliability-driven adaptation: we empirically find that high-score objects are still reliable and the optimization of high-score objects can enhance confidence across all detections. Thus, we devise a self-adaptive strategy to identify reliable objects for model adaptation, which discovers potential objects and alleviates omissions. 2) Noise-guard adaptation: since high-score objects may be scarce, we develop a negative regularization term to exploit the numerous low-score objects via negative learning, preventing overfitting to noise and trivial solutions. Experimental results show that MonoTTA brings significant performance gains for Mono 3Det models in OOD test scenarios, approximately 190% gains by average on KITTI and 198% gains on nuScenes.
Authors: Xiaofeng Cong, Yu Zhao, Jie Gui, Junming Hou, Dacheng Tao
Abstract: Underwater image enhancement (UIE) is a challenging research task in the field of computer vision. Although hundreds of UIE algorithms have been proposed, a comprehensive and systematic review is still lacking. To promote future research, we summarize the UIE task from multiple perspectives. First, the physical models, data construction processes, evaluation metrics, and loss functions are introduced. Second, according to the contributions brought by different literatures, recent proposed algorithms are discussed and classified from six perspectives, namely network architecture, learning strategy, learning stage, assistance task, domain perspective and disentanglement fusion, respectively. Third, considering the inconsistencies in experimental settings in different literatures, a comprehensive and fair comparison does not yet exist. To this end, we quantitatively and qualitatively evaluate state-of-the-art algorithms on multiple benchmark datasets. Finally, issues worthy of further research in the UIE task are raised. A collection of useful materials is available at https://github.com/YuZhao1999/UIE.
Authors: Haitao Cao, Baoping Cheng, Qiran Pu, Haocheng Zhang, Bin Luo, Yixiang Zhuang, Juncong Lin, Liyan Chen, Xuan Cheng
Abstract: Parametric 3D models have enabled a wide variety of computer vision and graphics tasks, such as modeling human faces, bodies and hands. In 3D face modeling, 3DMM is the most widely used parametric model, but can't generate fine geometric details solely from identity and expression inputs. To tackle this limitation, we propose a neural parametric model named DNPM for the facial geometric details, which utilizes deep neural network to extract latent codes from facial displacement maps encoding details and wrinkles. Built upon DNPM, a novel 3DMM named Detailed3DMM is proposed, which augments traditional 3DMMs by including the synthesis of facial details only from the identity and expression inputs. Moreover, we show that DNPM and Detailed3DMM can facilitate two downstream applications: speech-driven detailed 3D facial animation and 3D face reconstruction from a degraded image. Extensive experiments have shown the usefulness of DNPM and Detailed3DMM, and the progressiveness of two proposed applications.
Authors: Xuan Wu, Hongxiang Li, Yuanjiang Luo, Xuxin Cheng, Xianwei Zhuang, Meng Cao, Keren Fu
Abstract: Sign language video retrieval plays a key role in facilitating information access for the deaf community. Despite significant advances in video-text retrieval, the complexity and inherent uncertainty of sign language preclude the direct application of these techniques. Previous methods achieve the mapping between sign language video and text through fine-grained modal alignment. However, due to the scarcity of fine-grained annotation, the uncertainty inherent in sign language video is underestimated, limiting the further development of sign language retrieval tasks. To address this challenge, we propose a novel Uncertainty-aware Probability Distribution Retrieval (UPRet), that conceptualizes the mapping process of sign language video and text in terms of probability distributions, explores their potential interrelationships, and enables flexible mappings. Experiments on three benchmarks demonstrate the effectiveness of our method, which achieves state-of-the-art results on How2Sign (59.1%), PHOENIX-2014T (72.0%), and CSL-Daily (78.4%).
Authors: Qizao Wang, Xuelin Qian, Bin Li, Xiangyang Xue
Abstract: In real-world scenarios, person Re-IDentification (Re-ID) systems need to be adaptable to changes in space and time. Therefore, the adaptation of Re-ID models to new domains while preserving previously acquired knowledge is crucial, known as Lifelong person Re-IDentification (LReID). Advanced LReID methods rely on replaying exemplars from old domains and applying knowledge distillation in logits with old models. However, due to privacy concerns, retaining previous data is inappropriate. Additionally, the fine-grained and open-set characteristics of Re-ID limit the effectiveness of the distillation paradigm for accumulating knowledge. We argue that a Re-ID model trained on diverse and challenging pedestrian images at a large scale can acquire robust and general human semantic knowledge. These semantics can be readily utilized as shared knowledge for lifelong applications. In this paper, we identify the challenges and discrepancies associated with adapting a pre-trained model to each application domain, and introduce the Distribution Aligned Semantics Adaption (DASA) framework. It efficiently adjusts Batch Normalization (BN) to mitigate interference from data distribution discrepancy and freezes the pre-trained convolutional layers to preserve shared knowledge. Additionally, we propose the lightweight Semantics Adaption (SA) module, which effectively adapts learned semantics to enhance pedestrian representations. Extensive experiments demonstrate the remarkable superiority of our proposed framework over advanced LReID methods, and it exhibits significantly reduced storage consumption. DASA presents a novel and cost-effective perspective on effectively adapting pre-trained models for LReID.
Authors: Haoxing Chen, Yan Hong, Zizheng Huang, Zhuoer Xu, Zhangxuan Gu, Yaohui Li, Jun Lan, Huijia Zhu, Jianfu Zhang, Weiqiang Wang, Huaxiong Li
Abstract: Recently, video generation techniques have advanced rapidly. Given the popularity of video content on social media platforms, these models intensify concerns about the spread of fake information. Therefore, there is a growing demand for detectors capable of distinguishing between fake AI-generated videos and mitigating the potential harm caused by fake information. However, the lack of large-scale datasets from the most advanced video generators poses a barrier to the development of such detectors. To address this gap, we introduce the first AI-generated video detection dataset, GenVideo. It features the following characteristics: (1) a large volume of videos, including over one million AI-generated and real videos collected; (2) a rich diversity of generated content and methodologies, covering a broad spectrum of video categories and generation techniques. We conducted extensive studies of the dataset and proposed two evaluation methods tailored for real-world-like scenarios to assess the detectors' performance: the cross-generator video classification task assesses the generalizability of trained detectors on generators; the degraded video classification task evaluates the robustness of detectors to handle videos that have degraded in quality during dissemination. Moreover, we introduced a plug-and-play module, named Detail Mamba (DeMamba), designed to enhance the detectors by identifying AI-generated videos through the analysis of inconsistencies in temporal and spatial dimensions. Our extensive experiments demonstrate DeMamba's superior generalizability and robustness on GenVideo compared to existing detectors. We believe that the GenVideo dataset and the DeMamba module will significantly advance the field of AI-generated video detection. Our code and dataset will be aviliable at \url{https://github.com/chenhaoxing/DeMamba}.
Authors: Jia Li, Lijie Hu, Zhixian He, Jingfeng Zhang, Tianhang Zheng, Di Wang
Abstract: With the advancement of image-to-image diffusion models guided by text, significant progress has been made in image editing. However, a persistent challenge remains in seamlessly incorporating objects into images based on textual instructions, without relying on extra user-provided guidance. Text and images are inherently distinct modalities, bringing out difficulties in fully capturing the semantic intent conveyed through language and accurately translating that into the desired visual modifications. Therefore, text-guided image editing models often produce generations with residual object attributes that do not fully align with human expectations. To address this challenge, the models should comprehend the image content effectively away from a disconnect between the provided textual editing prompts and the actual modifications made to the image. In our paper, we propose a novel method called Locate and Forget (LaF), which effectively locates potential target concepts in the image for modification by comparing the syntactic trees of the target prompt and scene descriptions in the input image, intending to forget their existence clues in the generated image. Compared to the baselines, our method demonstrates its superiority in text-guided image editing tasks both qualitatively and quantitatively.
Authors: Alessandro Sanvito, Andrea Ramazzina, Stefanie Walz, Mario Bijelic, Felix Heide
Abstract: No augmented application is possible without animated humanoid avatars. At the same time, generating human replicas from real-world monocular hand-held or robotic sensor setups is challenging due to the limited availability of views. Previous work showed the feasibility of virtual avatars but required the presence of 360 degree views of the targeted subject. To address this issue, we propose HINT, a NeRF-based algorithm able to learn a detailed and complete human model from limited viewing angles. We achieve this by introducing a symmetry prior, regularization constraints, and training cues from large human datasets. In particular, we introduce a sagittal plane symmetry prior to the appearance of the human, directly supervise the density function of the human model using explicit 3D body modeling, and leverage a co-learned human digitization network as additional supervision for the unseen angles. As a result, our method can reconstruct complete humans even from a few viewing angles, increasing performance by more than 15% PSNR compared to previous state-of-the-art algorithms.
Authors: Yihe Deng, Pan Lu, Fan Yin, Ziniu Hu, Sheng Shen, James Zou, Kai-Wei Chang, Wei Wang
Abstract: Large vision language models (LVLMs) integrate large language models (LLMs) with pre-trained vision encoders, thereby activating the perception capability of the model to understand image inputs for different queries and conduct subsequent reasoning. Improving this capability requires high-quality vision-language data, which is costly and labor-intensive to acquire. Self-training approaches have been effective in single-modal settings to alleviate the need for labeled data by leveraging model's own generation. However, effective self-training remains a challenge regarding the unique visual perception and reasoning capability of LVLMs. To address this, we introduce Self-Training on Image Comprehension (STIC), which emphasizes a self-training approach specifically for image comprehension. First, the model self-constructs a preference dataset for image descriptions using unlabeled images. Preferred responses are generated through a step-by-step prompt, while dis-preferred responses are generated from either corrupted images or misleading prompts. To further self-improve reasoning on the extracted visual information, we let the model reuse a small portion of existing instruction-tuning data and append its self-generated image descriptions to the prompts. We validate the effectiveness of STIC across seven different benchmarks, demonstrating substantial performance gains of 4.0% on average while using 70% less supervised fine-tuning data than the current method. Further studies investigate various components of STIC and highlight its potential to leverage vast quantities of unlabeled images for self-training. Code and data are made publicly available.
Authors: Yuxing Duan, Shihan Peng, Lin Zhu, Wei Zhang, Yi Chang, Sheng Zhong, Luxin Yan
Abstract: Event camera has significant advantages in capturing dynamic scene information while being prone to noise interference, particularly in challenging conditions like low threshold and low illumination. However, most existing research focuses on gentle situations, hindering event camera applications in realistic complex scenarios. To tackle this limitation and advance the field, we construct a new paired real-world event denoising dataset (LED), including 3K sequences with 18K seconds of high-resolution (1200*680) event streams and showing three notable distinctions compared to others: diverse noise levels and scenes, larger-scale with high-resolution, and high-quality GT. Specifically, it contains stepped parameters and varying illumination with diverse scenarios. Moreover, based on the property of noise events inconsistency and signal events consistency, we propose a novel effective denoising framework(DED) using homogeneous dual events to generate the GT with better separating noise from the raw. Furthermore, we design a bio-inspired baseline leveraging Leaky-Integrate-and-Fire (LIF) neurons with dynamic thresholds to realize accurate denoising. The experimental results demonstrate that the remarkable performance of the proposed approach on different datasets.The dataset and code are at https://github.com/Yee-Sing/led.
Authors: Xuan-Bac Nguyen, Hoang-Quan Nguyen, Samuel Yen-Chi Chen, Samee U. Khan, Hugh Churchill, Khoa Luu
Abstract: Unsupervised vision clustering, a cornerstone in computer vision, has been studied for decades, yielding significant outcomes across numerous vision tasks. However, these algorithms involve substantial computational demands when confronted with vast amounts of unlabeled data. Conversely, Quantum computing holds promise in expediting unsupervised algorithms when handling large-scale databases. In this study, we introduce QClusformer, a pioneering Transformer-based framework leveraging Quantum machines to tackle unsupervised vision clustering challenges. Specifically, we design the Transformer architecture, including the self-attention module and transformer blocks, from a Quantum perspective to enable execution on Quantum hardware. In addition, we present QClusformer, a variant based on the Transformer architecture, tailored for unsupervised vision clustering tasks. By integrating these elements into an end-to-end framework, QClusformer consistently outperforms previous methods running on classical computers. Empirical evaluations across diverse benchmarks, including MS-Celeb-1M and DeepFashion, underscore the superior performance of QClusformer compared to state-of-the-art methods.
Authors: Thong Thanh Nguyen, Zhiyuan Hu, Xiaobao Wu, Cong-Duy T Nguyen, See-Kiong Ng, Anh Tuan Luu
Abstract: Seeking answers effectively for long videos is essential to build video question answering (videoQA) systems. Previous methods adaptively select frames and regions from long videos to save computations. However, this fails to reason over the whole sequence of video, leading to sub-optimal performance. To address this problem, we introduce a state space layer (SSL) into multi-modal Transformer to efficiently integrate global semantics of the video, which mitigates the video information loss caused by frame and region selection modules. Our SSL includes a gating unit to enable controllability over the flow of global semantics into visual representations. To further enhance the controllability, we introduce a cross-modal compositional congruence (C^3) objective to encourage global semantics aligned with the question. To rigorously evaluate long-form videoQA capacity, we construct two new benchmarks Ego-QA and MAD-QA featuring videos of considerably long length, i.e. 17.5 minutes and 1.9 hours, respectively. Extensive experiments demonstrate the superiority of our framework on these new as well as existing datasets.
Authors: Feng Chen, Zhen Yang, Bohan Zhuang, Qi Wu
Abstract: We present a novel task called online video editing, which is designed to edit \textbf{streaming} frames while maintaining temporal consistency. Unlike existing offline video editing assuming all frames are pre-established and accessible, online video editing is tailored to real-life applications such as live streaming and online chat, requiring (1) fast continual step inference, (2) long-term temporal modeling, and (3) zero-shot video editing capability. To solve these issues, we propose Streaming Video Diffusion (SVDiff), which incorporates the compact spatial-aware temporal recurrence into off-the-shelf Stable Diffusion and is trained with the segment-level scheme on large-scale long videos. This simple yet effective setup allows us to obtain a single model that is capable of executing a broad range of videos and editing each streaming frame with temporal coherence. Our experiments indicate that our model can edit long, high-quality videos with remarkable results, achieving a real-time inference speed of 15.2 FPS at a resolution of 512x512.
Authors: Koki Endo, Shuhei Tsuchida, Tsukasa Fukusato, Takeo Igarashi
Abstract: Segmenting dance video into short movements is a popular way to easily understand dance choreography. However, it is currently done manually and requires a significant amount of effort by experts. That is, even if many dance videos are available on social media (e.g., TikTok and YouTube), it remains difficult for people, especially novices, to casually watch short video segments to practice dance choreography. In this paper, we propose a method to automatically segment a dance video into each movement. Given a dance video as input, we first extract visual and audio features: the former is computed from the keypoints of the dancer in the video, and the latter is computed from the Mel spectrogram of the music in the video. Next, these features are passed to a Temporal Convolutional Network (TCN), and segmentation points are estimated by picking peaks of the network output. To build our training dataset, we annotate segmentation points to dance videos in the AIST Dance Video Database, which is a shared database containing original street dance videos with copyright-cleared dance music. The evaluation study shows that the proposed method (i.e., combining the visual and audio features) can estimate segmentation points with high accuracy. In addition, we developed an application to help dancers practice choreography using the proposed method.
Authors: Zixian Guo, Ming Liu, Zhilong Ji, Jinfeng Bai, Yiwen Guo, Wangmeng Zuo
Abstract: Learning a skill generally relies on both practical experience by doer and insightful high-level guidance by instructor. Will this strategy also work well for solving complex non-convex optimization problems? Here, a common gradient-based optimizer acts like a disciplined doer, making locally optimal update at each step. Recent methods utilize large language models (LLMs) to optimize solutions for concrete problems by inferring from natural language instructions, akin to a high-level instructor. In this paper, we show that these two optimizers are complementary to each other, suggesting a collaborative optimization approach. The gradient-based optimizer and LLM-based optimizer are combined in an interleaved manner. We instruct LLMs using task descriptions and timely optimization trajectories recorded during gradient-based optimization. Inferred results from LLMs are used as restarting points for the next stage of gradient optimization. By leveraging both the locally rigorous gradient-based optimizer and the high-level deductive LLM-based optimizer, our combined optimization method consistently yields improvements over competitive baseline prompt tuning methods. Our results demonstrate the synergistic effect of conventional gradient-based optimization and the inference ability of LLMs. The code is released at https://github.com/guozix/LLM-catalyst.
Authors: Yong-Qiang Mao, Hanbo Bi, Xuexue Li, Kaiqiang Chen, Zhirui Wang, Xian Sun, Kun Fu
Abstract: Thanks to the application of deep learning technology in point cloud processing of the remote sensing field, point cloud segmentation has become a research hotspot in recent years, which can be applied to real-world 3D, smart cities, and other fields. Although existing solutions have made unprecedented progress, they ignore the inherent characteristics of point clouds in remote sensing fields that are strictly arranged according to latitude, longitude, and altitude, which brings great convenience to the segmentation of point clouds in remote sensing fields. To consider this property cleverly, we propose novel convolution operators, termed Twin Deformable point Convolutions (TDConvs), which aim to achieve adaptive feature learning by learning deformable sampling points in the latitude-longitude plane and altitude direction, respectively. First, to model the characteristics of the latitude-longitude plane, we propose a Cylinder-wise Deformable point Convolution (CyDConv) operator, which generates a two-dimensional cylinder map by constructing a cylinder-like grid in the latitude-longitude direction. Furthermore, to better integrate the features of the latitude-longitude plane and the spatial geometric features, we perform a multi-scale fusion of the extracted latitude-longitude features and spatial geometric features, and realize it through the aggregation of adjacent point features of different scales. In addition, a Sphere-wise Deformable point Convolution (SpDConv) operator is introduced to adaptively offset the sampling points in three-dimensional space by constructing a sphere grid structure, aiming at modeling the characteristics in the altitude direction. Experiments on existing popular benchmarks conclude that our TDConvs achieve the best segmentation performance, surpassing the existing state-of-the-art methods.
Authors: Hyemin Ahn
Abstract: We hypothesize dance as a motion that forms a visual rhythm from music, where the visual rhythm can be perceived from an optical flow. If an agent can recognize the relationship between visual rhythm and music, it will be able to dance by generating a motion to create a visual rhythm that matches the music. Based on this, we propose a framework for any kind of non-humanoid agents to learn how to dance from human videos. Our framework works in two processes: (1) training a reward model which perceives the relationship between optical flow (visual rhythm) and music from human dance videos, (2) training the non-humanoid dancer based on that reward model, and reinforcement learning. Our reward model consists of two feature encoders for optical flow and music. They are trained based on contrastive learning which makes the higher similarity between concurrent optical flow and music features. With this reward model, the agent learns dancing by getting a higher reward when its action creates an optical flow whose feature has a higher similarity with the given music feature. Experiment results show that generated dance motion can align with the music beat properly, and user study result indicates that our framework is more preferred by humans compared to the baselines. To the best of our knowledge, our work of non-humanoid agents which learn dance from human videos is unprecedented. An example video can be found at https://youtu.be/dOUPvo-O3QY.
Authors: Boming Zhao, Yuan Li, Ziyu Sun, Lin Zeng, Yujun Shen, Rui Ma, Yinda Zhang, Hujun Bao, Zhaopeng Cui
Abstract: Forecasting future scenarios in dynamic environments is essential for intelligent decision-making and navigation, a challenge yet to be fully realized in computer vision and robotics. Traditional approaches like video prediction and novel-view synthesis either lack the ability to forecast from arbitrary viewpoints or to predict temporal dynamics. In this paper, we introduce GaussianPrediction, a novel framework that empowers 3D Gaussian representations with dynamic scene modeling and future scenario synthesis in dynamic environments. GaussianPrediction can forecast future states from any viewpoint, using video observations of dynamic scenes. To this end, we first propose a 3D Gaussian canonical space with deformation modeling to capture the appearance and geometry of dynamic scenes, and integrate the lifecycle property into Gaussians for irreversible deformations. To make the prediction feasible and efficient, a concentric motion distillation approach is developed by distilling the scene motion with key points. Finally, a Graph Convolutional Network is employed to predict the motions of key points, enabling the rendering of photorealistic images of future scenarios. Our framework shows outstanding performance on both synthetic and real-world datasets, demonstrating its efficacy in predicting and rendering future environments.
Authors: Ron Keuth, Lasse Hansen, Maren Balks, Ronja J\"ager, Anne-Nele Schr\"oder, Ludger T\"ushaus, Mattias Heinrich
Abstract: Purpose: Semantic segmentation and landmark detection are fundamental tasks of medical image processing, facilitating further analysis of anatomical objects. Although deep learning-based pixel-wise classification has set a new-state-of-the-art for segmentation, it falls short in landmark detection, a strength of shape-based approaches. Methods: In this work, we propose a dense image-to-shape representation that enables the joint learning of landmarks and semantic segmentation by employing a fully convolutional architecture. Our method intuitively allows the extraction of arbitrary landmarks due to its representation of anatomical correspondences. We benchmark our method against the state-of-the-art for semantic segmentation (nnUNet), a shape-based approach employing geometric deep learning and a CNN-based method for landmark detection. Results: We evaluate our method on two medical dataset: one common benchmark featuring the lungs, heart, and clavicle from thorax X-rays, and another with 17 different bones in the paediatric wrist. While our method is on pair with the landmark detection baseline in the thorax setting (error in mm of $2.6\pm0.9$ vs $2.7\pm0.9$), it substantially surpassed it in the more complex wrist setting ($1.1\pm0.6$ vs $1.9\pm0.5$). Conclusion: We demonstrate that dense geometric shape representation is beneficial for challenging landmark detection tasks and outperforms previous state-of-the-art using heatmap regression. While it does not require explicit training on the landmarks themselves, allowing for the addition of new landmarks without necessitating retraining.}
Authors: Wenxuan Liu, Saiqian Zhang
Abstract: Diffusion Transformers (DiTs) have recently gained substantial attention in both industrial and academic fields for their superior visual generation capabilities, outperforming traditional diffusion models that use U-Net. However,the enhanced performance of DiTs also comes with high parameter counts and implementation costs, seriously restricting their use on resource-limited devices such as mobile phones. To address these challenges, we introduce the Hybrid Floating-point Quantization for DiT(HQ-DiT), an efficient post-training quantization method that utilizes 4-bit floating-point (FP) precision on both weights and activations for DiT inference. Compared to fixed-point quantization (e.g., INT8), FP quantization, complemented by our proposed clipping range selection mechanism, naturally aligns with the data distribution within DiT, resulting in a minimal quantization error. Furthermore, HQ-DiT also implements a universal identity mathematical transform to mitigate the serious quantization error caused by the outliers. The experimental results demonstrate that DiT can achieve extremely low-precision quantization (i.e., 4 bits) with negligible impact on performance. Our approach marks the first instance where both weights and activations in DiTs are quantized to just 4 bits, with only a 0.12 increase in sFID on ImageNet.
Authors: Marta Buetas Arcas, Richard Osuala, Karim Lekadir, Oliver D\'iaz
Abstract: Artificial Intelligence (AI) has emerged as a valuable tool for assisting radiologists in breast cancer detection and diagnosis. However, the success of AI applications in this domain is restricted by the quantity and quality of available data, posing challenges due to limited and costly data annotation procedures that often lead to annotation shifts. This study simulates, analyses and mitigates annotation shifts in cancer classification in the breast mammography domain. First, a high-accuracy cancer risk prediction model is developed, which effectively distinguishes benign from malignant lesions. Next, model performance is used to quantify the impact of annotation shift. We uncover a substantial impact of annotation shift on multiclass classification performance particularly for malignant lesions. We thus propose a training data augmentation approach based on single-image generative models for the affected class, requiring as few as four in-domain annotations to considerably mitigate annotation shift, while also addressing dataset imbalance. Lastly, we further increase performance by proposing and validating an ensemble architecture based on multiple models trained under different data augmentation regimes. Our study offers key insights into annotation shift in deep learning breast cancer classification and explores the potential of single-image generative models to overcome domain shift challenges.
Authors: Xingyu Wan, Chengquan Zhang, Pengyuan Lyu, Sen Fan, Zihan Ni, Kun Yao, Errui Ding, Jingdong Wang
Abstract: Existing OCR engines or document image analysis systems typically rely on training separate models for text detection in varying scenarios and granularities, leading to significant computational complexity and resource demands. In this paper, we introduce "Detect Any Text" (DAT), an advanced paradigm that seamlessly unifies scene text detection, layout analysis, and document page detection into a cohesive, end-to-end model. This design enables DAT to efficiently manage text instances at different granularities, including *word*, *line*, *paragraph* and *page*. A pivotal innovation in DAT is the across-granularity interactive attention module, which significantly enhances the representation learning of text instances at varying granularities by correlating structural information across different text queries. As a result, it enables the model to achieve mutually beneficial detection performances across multiple text granularities. Additionally, a prompt-based segmentation module refines detection outcomes for texts of arbitrary curvature and complex layouts, thereby improving DAT's accuracy and expanding its real-world applicability. Experimental results demonstrate that DAT achieves state-of-the-art performances across a variety of text-related benchmarks, including multi-oriented/arbitrarily-shaped scene text detection, document layout analysis and page detection tasks.
Authors: Zhiwen Yang, Haowei Chen, Ziniu Qian, Yang Yi, Hui Zhang, Dan Zhao, Bingzheng Wei, Yan Xu
Abstract: Although single-task medical image restoration (MedIR) has witnessed remarkable success, the limited generalizability of these methods poses a substantial obstacle to wider application. In this paper, we focus on the task of all-in-one medical image restoration, aiming to address multiple distinct MedIR tasks with a single universal model. Nonetheless, due to significant differences between different MedIR tasks, training a universal model often encounters task interference issues, where different tasks with shared parameters may conflict with each other in the gradient update direction. This task interference leads to deviation of the model update direction from the optimal path, thereby affecting the model's performance. To tackle this issue, we propose a task-adaptive routing strategy, allowing conflicting tasks to select different network paths in spatial and channel dimensions, thereby mitigating task interference. Experimental results demonstrate that our proposed \textbf{A}ll-in-one \textbf{M}edical \textbf{I}mage \textbf{R}estoration (\textbf{AMIR}) network achieves state-of-the-art performance in three MedIR tasks: MRI super-resolution, CT denoising, and PET synthesis, both in single-task and all-in-one settings. The code and data will be available at \href{https://github.com/Yaziwel/All-In-One-Medical-Image-Restoration-via-Task-Adaptive-Routing.git}{https://github.com/Yaziwel/AMIR}.
URLs: https://github.com/Yaziwel/All-In-One-Medical-Image-Restoration-via-Task-Adaptive-Routing.git, https://github.com/Yaziwel/AMIR
Authors: Tautvydas Misiunas, Hassan Mansoor, Jasper Uijlings, Oriana Riva, Victor Carbune
Abstract: Large-language models and large-vision models are increasingly capable of solving compositional reasoning tasks, as measured by breakthroughs in visual-question answering benchmarks. However, state-of-the-art solutions often involve careful construction of large pre-training and fine-tuning datasets, which can be expensive. The use of external tools, whether other ML models, search engines, or APIs, can significantly improve performance by breaking down high-level reasoning questions into sub-questions that are answerable by individual tools, but this approach has similar dataset construction costs to teach fine-tuned models how to use the available tools. We propose a technique in which existing training sets can be directly used for constructing computational environments with task metrics as rewards. This enables a model to autonomously teach itself to use itself or another model as a tool. By doing so, we augment training sets by integrating external signals. The proposed method starts with zero-shot prompts and iteratively refines them by selecting few-shot examples that maximize the task metric on the training set. Our experiments showcase how Gemini learns how to use itself, or another smaller and specialized model such as ScreenAI, to iteratively improve performance on training sets. Our approach successfully generalizes and improves upon zeroshot performance on charts, infographics, and document visual question-answering datasets
Authors: Sizhe Zheng, Pan Gao, Peng Zhou, Jie Qin
Abstract: Style transfer aims to render an image with the artistic features of a style image, while maintaining the original structure. Various methods have been put forward for this task, but some challenges still exist. For instance, it is difficult for CNN-based methods to handle global information and long-range dependencies between input images, for which transformer-based methods have been proposed. Although transformers can better model the relationship between content and style images, they require high-cost hardware and time-consuming inference. To address these issues, we design a novel transformer model that includes only the encoder, thus significantly reducing the computational cost. In addition, we also find that existing style transfer methods may lead to images under-stylied or missing content. In order to achieve better stylization, we design a content feature extractor and a style feature extractor, based on which pure content and style images can be fed to the transformer. Finally, we propose a novel network termed Puff-Net, i.e., pure content and style feature fusion network. Through qualitative and quantitative experiments, we demonstrate the advantages of our model compared to state-of-the-art ones in the literature.
Authors: Jinliang Zheng, Jianxiong Li, Sijie Cheng, Yinan Zheng, Jiaming Li, Jihao Liu, Yu Liu, Jingjing Liu, Xianyuan Zhan
Abstract: Instruction following is crucial in contemporary LLM. However, when extended to multimodal setting, it often suffers from misalignment between specific textual instruction and targeted local region of an image. To achieve more accurate and nuanced multimodal instruction following, we introduce Instruction-guided Visual Masking (IVM), a new versatile visual grounding model that is compatible with diverse multimodal models, such as LMM and robot model. By constructing visual masks for instruction-irrelevant regions, IVM-enhanced multimodal models can effectively focus on task-relevant image regions to better align with complex instructions. Specifically, we design a visual masking data generation pipeline and create an IVM-Mix-1M dataset with 1 million image-instruction pairs. We further introduce a new learning technique, Discriminator Weighted Supervised Learning (DWSL) for preferential IVM training that prioritizes high-quality data samples. Experimental results on generic multimodal tasks such as VQA and embodied robotic control demonstrate the versatility of IVM, which as a plug-and-play tool, significantly boosts the performance of diverse multimodal models, yielding new state-of-the-art results across challenging multimodal benchmarks. Code is available at https://github.com/2toinf/IVM.
Authors: Inpyo Song, Minjun Joo, Joonhyung Kwon, Jangwon Lee
Abstract: This paper addresses the daily challenges encountered by visually impaired individuals, such as limited access to information, navigation difficulties, and barriers to social interaction. To alleviate these challenges, we introduce a novel visual question answering dataset. Our dataset offers two significant advancements over previous datasets: Firstly, it features videos captured using a 360-degree egocentric wearable camera, enabling observation of the entire surroundings, departing from the static image-centric nature of prior datasets. Secondly, unlike datasets centered on singular challenges, ours addresses multiple real-life obstacles simultaneously through an innovative visual-question answering framework. We validate our dataset using various state-of-the-art VideoQA methods and diverse metrics. Results indicate that while progress has been made, satisfactory performance levels for AI-powered assistive services remain elusive for visually impaired individuals. Additionally, our evaluation highlights the distinctive features of the proposed dataset, featuring ego-motion in videos captured via 360-degree cameras across varied scenarios.
Authors: Suzana Veljanovska, Hans Dermot Doran
Abstract: Symbolic Aggregate approXimation (SAX) is a common dimensionality reduction approach for time-series data which has been employed in a variety of domains, including classification and anomaly detection in time-series data. Domains also include shape recognition where the shape outline is converted into time-series data forinstance epoch classification of archived arrowheads. In this paper we propose a dimensionality reduction and shape recognition approach based on the SAX algorithm, an application which requires responses on cost efficient, IoT-like, platforms. The challenge is largely dealing with the computational expense of the SAX algorithm in IoT-like applications, from simple time-series dimension reduction through shape recognition. The approach is based on lowering the dimensional space while capturing and preserving the most representative features of the shape. We present three scenarios of increasing computational complexity backing up our statements with measurement of performance characteristics
Authors: Chunhui Zhang, Li Liu, Guanjie Huang, Hao Wen, Xi Zhou, Yanfeng Wang
Abstract: Underwater object tracking (UOT) is a foundational task for identifying and tracing submerged entities in underwater video sequences. However, current UOT datasets suffer from limitations in scale, diversity of target categories and scenarios covered, hindering the training and evaluation of modern tracking algorithms. To bridge this gap, we take the first step and introduce WebUOT-1M, \ie, the largest public UOT benchmark to date, sourced from complex and realistic underwater environments. It comprises 1.1 million frames across 1,500 video clips filtered from 408 target categories, largely surpassing previous UOT datasets, \eg, UVOT400. Through meticulous manual annotation and verification, we provide high-quality bounding boxes for underwater targets. Additionally, WebUOT-1M includes language prompts for video sequences, expanding its application areas, \eg, underwater vision-language tracking. Most existing trackers are tailored for open-air environments, leading to performance degradation when applied to UOT due to domain gaps. Retraining and fine-tuning these trackers are challenging due to sample imbalances and limited real-world underwater datasets. To tackle these challenges, we propose a novel omni-knowledge distillation framework based on WebUOT-1M, incorporating various strategies to guide the learning of the student Transformer. To the best of our knowledge, this framework is the first to effectively transfer open-air domain knowledge to the UOT model through knowledge distillation, as demonstrated by results on both existing UOT datasets and the newly proposed WebUOT-1M. Furthermore, we comprehensively evaluate WebUOT-1M using 30 deep trackers, showcasing its value as a benchmark for UOT research by presenting new challenges and opportunities for future studies. The complete dataset, codes and tracking results, will be made publicly available.
Authors: Andrea Ramazzina, Stefanie Walz, Pragyan Dahal, Mario Bijelic, Felix Heide
Abstract: Reconstructing outdoor 3D scenes from temporal observations is a challenge that recent work on neural fields has offered a new avenue for. However, existing methods that recover scene properties, such as geometry, appearance, or radiance, solely from RGB captures often fail when handling poorly-lit or texture-deficient regions. Similarly, recovering scenes with scanning LiDAR sensors is also difficult due to their low angular sampling rate which makes recovering expansive real-world scenes difficult. Tackling these gaps, we introduce Gated Fields - a neural scene reconstruction method that utilizes active gated video sequences. To this end, we propose a neural rendering approach that seamlessly incorporates time-gated capture and illumination. Our method exploits the intrinsic depth cues in the gated videos, achieving precise and dense geometry reconstruction irrespective of ambient illumination conditions. We validate the method across day and night scenarios and find that Gated Fields compares favorably to RGB and LiDAR reconstruction methods. Our code and datasets are available at https://light.princeton.edu/gatedfields/.
Authors: Frank A. Ruis, Alma M. Liezenga, Friso G. Heslinga, Luca Ballan, Thijs A. Eker, Richard J. M. den Hollander, Martin C. van Leeuwen, Judith Dijk, Wyke Huizinga
Abstract: Collecting and annotating real-world data for the development of object detection models is a time-consuming and expensive process. In the military domain in particular, data collection can also be dangerous or infeasible. Training models on synthetic data may provide a solution for cases where access to real-world training data is restricted. However, bridging the reality gap between synthetic and real data remains a challenge. Existing methods usually build on top of baseline Convolutional Neural Network (CNN) models that have been shown to perform well when trained on real data, but have limited ability to perform well when trained on synthetic data. For example, some architectures allow for fine-tuning with the expectation of large quantities of training data and are prone to overfitting on synthetic data. Related work usually ignores various best practices from object detection on real data, e.g. by training on synthetic data from a single environment with relatively little variation. In this paper we propose a methodology for improving the performance of a pre-trained object detector when training on synthetic data. Our approach focuses on extracting the salient information from synthetic data without forgetting useful features learned from pre-training on real images. Based on the state of the art, we incorporate data augmentation methods and a Transformer backbone. Besides reaching relatively strong performance without any specialized synthetic data transfer methods, we show that our methods improve the state of the art on synthetic data trained object detection for the RarePlanes and DGTA-VisDrone datasets, and reach near-perfect performance on an in-house vehicle detection dataset.
Authors: Fengyuan Yang, Kerui Gu, Angela Yao
Abstract: 2D keypoints are commonly used as an additional cue to refine estimated 3D human meshes. Current methods optimize the pose and shape parameters with a reprojection loss on the provided 2D keypoints. Such an approach, while simple and intuitive, has limited effectiveness because the optimal solution is hard to find in ambiguous parameter space and may sacrifice depth. Additionally, divergent gradients from distal joints complicate and deviate the refinement of proximal joints in the kinematic chain. To address these, we introduce Kinematic-Tree Rotation (KITRO), a novel mesh refinement strategy that explicitly models depth and human kinematic-tree structure. KITRO treats refinement from a bone-wise perspective. Unlike previous methods which perform gradient-based optimizations, our method calculates bone directions in closed form. By accounting for the 2D pose, bone length, and parent joint's depth, the calculation results in two possible directions for each child joint. We then use a decision tree to trace binary choices for all bones along the human skeleton's kinematic-tree to select the most probable hypothesis. Our experiments across various datasets and baseline models demonstrate that KITRO significantly improves 3D joint estimation accuracy and achieves an ideal 2D fit simultaneously. Our code available at: https://github.com/MartaYang/KITRO.
Authors: Fangyi Chen, Han Zhang, Zhantao Yang, Hao Chen, Kai Hu, Marios Savvides
Abstract: Open-vocabulary object detection (OVD) requires solid modeling of the region-semantic relationship, which could be learned from massive region-text pairs. However, such data is limited in practice due to significant annotation costs. In this work, we propose RTGen to generate scalable open-vocabulary region-text pairs and demonstrate its capability to boost the performance of open-vocabulary object detection. RTGen includes both text-to-region and region-to-text generation processes on scalable image-caption data. The text-to-region generation is powered by image inpainting, directed by our proposed scene-aware inpainting guider for overall layout harmony. For region-to-text generation, we perform multiple region-level image captioning with various prompts and select the best matching text according to CLIP similarity. To facilitate detection training on region-text pairs, we also introduce a localization-aware region-text contrastive loss that learns object proposals tailored with different localization qualities. Extensive experiments demonstrate that our RTGen can serve as a scalable, semantically rich, and effective source for open-vocabulary object detection and continue to improve the model performance when more data is utilized, delivering superior performance compared to the existing state-of-the-art methods.
Authors: Riccardo Renzulli
Abstract: Capsule networks (CapsNets) were introduced to address convolutional neural networks limitations, learning object-centric representations that are more robust, pose-aware, and interpretable. They organize neurons into groups called capsules, where each capsule encodes the instantiation parameters of an object or one of its parts. Moreover, a routing algorithm connects capsules in different layers, thereby capturing hierarchical part-whole relationships in the data. This thesis investigates the intriguing aspects of CapsNets and focuses on three key questions to unlock their full potential. First, we explore the effectiveness of the routing algorithm, particularly in small-sized networks. We propose a novel method that anneals the number of routing iterations during training, enhancing performance in architectures with fewer parameters. Secondly, we investigate methods to extract more effective first-layer capsules, also known as primary capsules. By exploiting pruned backbones, we aim to improve computational efficiency by reducing the number of capsules while achieving high generalization. This approach reduces CapsNets memory requirements and computational effort. Third, we explore part-relationship learning in CapsNets. Through extensive research, we demonstrate that capsules with low entropy can extract more concise and discriminative part-whole relationships compared to traditional capsule networks, even with reasonable network sizes. Lastly, we showcase how CapsNets can be utilized in real-world applications, including autonomous localization of unmanned aerial vehicles, quaternion-based rotations prediction in synthetic datasets, and lung nodule segmentation in biomedical imaging. The findings presented in this thesis contribute to a deeper understanding of CapsNets and highlight their potential to address complex computer vision challenges.
Authors: Alessio Mazzucchelli, Adrian Garcia-Garcia, Elena Garces, Fernando Rivas-Manzaneque, Francesc Moreno-Noguer, Adrian Penate-Sanchez
Abstract: Advances in NERFs have allowed for 3D scene reconstructions and novel view synthesis. Yet, efficiently editing these representations while retaining photorealism is an emerging challenge. Recent methods face three primary limitations: they're slow for interactive use, lack precision at object boundaries, and struggle to ensure multi-view consistency. We introduce IReNe to address these limitations, enabling swift, near real-time color editing in NeRF. Leveraging a pre-trained NeRF model and a single training image with user-applied color edits, IReNe swiftly adjusts network parameters in seconds. This adjustment allows the model to generate new scene views, accurately representing the color changes from the training image while also controlling object boundaries and view-specific effects. Object boundary control is achieved by integrating a trainable segmentation module into the model. The process gains efficiency by retraining only the weights of the last network layer. We observed that neurons in this layer can be classified into those responsible for view-dependent appearance and those contributing to diffuse appearance. We introduce an automated classification approach to identify these neuron types and exclusively fine-tune the weights of the diffuse neurons. This further accelerates training and ensures consistent color edits across different views. A thorough validation on a new dataset, with edited object colors, shows significant quantitative and qualitative advancements over competitors, accelerating speeds by 5x to 500x.
Authors: Tom\'a\v{s} Voj\'i\v{r}, Jan \v{S}ochman, Ji\v{r}\'i Matas
Abstract: We propose a dense image prediction out-of-distribution detection algorithm, called PixOOD, which does not require training on samples of anomalous data and is not designed for a specific application which avoids traditional training biases. In order to model the complex intra-class variability of the in-distribution data at the pixel level, we propose an online data condensation algorithm which is more robust than standard K-means and is easily trainable through SGD. We evaluate PixOOD on a wide range of problems. It achieved state-of-the-art results on four out of seven datasets, while being competitive on the rest. The source code is available at https://github.com/vojirt/PixOOD.
Authors: Seun-An Choe, Ah-Hyung Shin, Keon-Hee Park, Jinwoo Choi, Gyeong-Moon Park
Abstract: Unsupervised domain adaptation (UDA) for semantic segmentation aims to transfer the pixel-wise knowledge from the labeled source domain to the unlabeled target domain. However, current UDA methods typically assume a shared label space between source and target, limiting their applicability in real-world scenarios where novel categories may emerge in the target domain. In this paper, we introduce Open-Set Domain Adaptation for Semantic Segmentation (OSDA-SS) for the first time, where the target domain includes unknown classes. We identify two major problems in the OSDA-SS scenario as follows: 1) the existing UDA methods struggle to predict the exact boundary of the unknown classes, and 2) they fail to accurately predict the shape of the unknown classes. To address these issues, we propose Boundary and Unknown Shape-Aware open-set domain adaptation, coined BUS. Our BUS can accurately discern the boundaries between known and unknown classes in a contrastive manner using a novel dilation-erosion-based contrastive loss. In addition, we propose OpenReMix, a new domain mixing augmentation method that guides our model to effectively learn domain and size-invariant features for improving the shape detection of the known and unknown classes. Through extensive experiments, we demonstrate that our proposed BUS effectively detects unknown classes in the challenging OSDA-SS scenario compared to the previous methods by a large margin. The code is available at https://github.com/KHU-AGI/BUS.
Authors: Huadong Li, Shichao Dong, Jin Wang, Rong Fu, Minhao Jing, Jiajun Liang, Haoqiang Fan, Renhe Ji
Abstract: This paper focuses on the area of RGB(visible)-NIR(near-infrared) cross-modality image registration, which is crucial for many downstream vision tasks to fully leverage the complementary information present in visible and infrared images. In this field, researchers face two primary challenges - the absence of a correctly-annotated benchmark with viewpoint variations for evaluating RGB-NIR cross-modality registration methods and the problem of inconsistent local features caused by the appearance discrepancy between RGB-NIR cross-modality images. To address these challenges, we first present the RGB-NIR Image Registration (RGB-NIR-IRegis) benchmark, which, for the first time, enables fair and comprehensive evaluations for the task of RGB-NIR cross-modality image registration. Evaluations of previous methods highlight the significant challenges posed by our RGB-NIR-IRegis benchmark, especially on RGB-NIR image pairs with viewpoint variations. To analyze the causes of the unsatisfying performance, we then design several metrics to reveal the toxic impact of inconsistent local features between visible and infrared images on the model performance. This further motivates us to develop a baseline method named Semantic Guidance Transformer (SGFormer), which utilizes high-level semantic guidance to mitigate the negative impact of local inconsistent features. Despite the simplicity of our motivation, extensive experimental results show the effectiveness of our method.
Authors: Masashi Hatano, Ryo Hachiuma, Ryo Fuji, Hideo Saito
Abstract: We address a novel cross-domain few-shot learning task (CD-FSL) with multimodal input and unlabeled target data for egocentric action recognition. This paper simultaneously tackles two critical challenges associated with egocentric action recognition in CD-FSL settings: (1) the extreme domain gap in egocentric videos (\eg, daily life vs. industrial domain) and (2) the computational cost for real-world applications. We propose MM-CDFSL, a domain-adaptive and computationally efficient approach designed to enhance adaptability to the target domain and improve inference speed. To address the first challenge, we propose the incorporation of multimodal distillation into the student RGB model using teacher models. Each teacher model is trained independently on source and target data for its respective modality. Leveraging only unlabeled target data during multimodal distillation enhances the student model's adaptability to the target domain. We further introduce ensemble masked inference, a technique that reduces the number of input tokens through masking. In this approach, ensemble prediction mitigates the performance degradation caused by masking, effectively addressing the second issue. Our approach outperformed the state-of-the-art CD-FSL approaches with a substantial margin on multiple egocentric datasets, improving by an average of 6.12/6.10 points for 1-shot/5-shot settings while achieving $2.2$ times faster inference speed. Project page: https://masashi-hatano.github.io/MM-CDFSL/
Authors: Angel Villar-Corrales, Moritz Austermann, Sven Behnke
Abstract: Autonomous systems, such as self-driving cars, rely on reliable semantic environment perception for decision making. Despite great advances in video semantic segmentation, existing approaches ignore important inductive biases and lack structured and interpretable internal representations. In this work, we propose MCDS-VSS, a structured filter model that learns in a self-supervised manner to estimate scene geometry and ego-motion of the camera, while also estimating the motion of external objects. Our model leverages these representations to improve the temporal consistency of semantic segmentation without sacrificing segmentation accuracy. MCDS-VSS follows a prediction-fusion approach in which scene geometry and camera motion are first used to compensate for ego-motion, then residual flow is used to compensate motion of dynamic objects, and finally the predicted scene features are fused with the current features to obtain a temporally consistent scene segmentation. Our model parses automotive scenes into multiple decoupled interpretable representations such as scene geometry, ego-motion, and object motion. Quantitative evaluation shows that MCDS-VSS achieves superior temporal consistency on video sequences while retaining competitive segmentation performance.
Authors: Xiaoyu Wu, Jiaru Zhang, Yang Hua, Bohan Lyu, Hao Wang, Tao Song, Haibing Guan
Abstract: Few-shot fine-tuning of Diffusion Models (DMs) is a key advancement, significantly reducing training costs and enabling personalized AI applications. However, we explore the training dynamics of DMs and observe an unanticipated phenomenon: during the training process, image fidelity initially improves, then unexpectedly deteriorates with the emergence of noisy patterns, only to recover later with severe overfitting. We term the stage with generated noisy patterns as corruption stage. To understand this corruption stage, we begin by theoretically modeling the one-shot fine-tuning scenario, and then extend this modeling to more general cases. Through this modeling, we identify the primary cause of this corruption stage: a narrowed learning distribution inherent in the nature of few-shot fine-tuning. To tackle this, we apply Bayesian Neural Networks (BNNs) on DMs with variational inference to implicitly broaden the learned distribution, and present that the learning target of the BNNs can be naturally regarded as an expectation of the diffusion loss and a further regularization with the pretrained DMs. This approach is highly compatible with current few-shot fine-tuning methods in DMs and does not introduce any extra inference costs. Experimental results demonstrate that our method significantly mitigates corruption, and improves the fidelity, quality and diversity of the generated images in both object-driven and subject-driven generation tasks.
Authors: Qi Zhang, Yunfei Gong, Daijie Chen, Antoni B. Chan, Hui Huang
Abstract: Recent deep learning-based multi-view people detection (MVD) methods have shown promising results on existing datasets. However, current methods are mainly trained and evaluated on small, single scenes with a limited number of multi-view frames and fixed camera views. As a result, these methods may not be practical for detecting people in larger, more complex scenes with severe occlusions and camera calibration errors. This paper focuses on improving multi-view people detection by developing a supervised view-wise contribution weighting approach that better fuses multi-camera information under large scenes. Besides, a large synthetic dataset is adopted to enhance the model's generalization ability and enable more practical evaluation and comparison. The model's performance on new testing scenes is further improved with a simple domain adaptation technique. Experimental results demonstrate the effectiveness of our approach in achieving promising cross-scene multi-view people detection performance. See code here: https://vcc.tech/research/2024/MVD.
Authors: Jianxiong Gao, Xuelin Qian, Longfei Liang, Junwei Han, Yanwei Fu
Abstract: Amodal object completion is a complex task that involves predicting the invisible parts of an object based on visible segments and background information. Learning shape priors is crucial for effective amodal completion, but traditional methods often rely on two-stage processes or additional information, leading to inefficiencies and potential error accumulation. To address these shortcomings, we introduce a novel framework named the Hyper-Transformer Amodal Network (H-TAN). This framework utilizes a hyper transformer equipped with a dynamic convolution head to directly learn shape priors and accurately predict amodal masks. Specifically, H-TAN uses a dual-branch structure to extract multi-scale features from both images and masks. The multi-scale features from the image branch guide the hyper transformer in learning shape priors and in generating the weights for dynamic convolution tailored to each instance. The dynamic convolution head then uses the features from the mask branch to predict precise amodal masks. We extensively evaluate our model on three benchmark datasets: KINS, COCOA-cls, and D2SA, where H-TAN demonstrated superior performance compared to existing methods. Additional experiments validate the effectiveness and stability of the novel hyper transformer in our framework.
Authors: Qiaowei Miao, Yawei Luo, Yi Yang
Abstract: As text-conditioned diffusion models (DMs) achieve breakthroughs in image, video, and 3D generation, the research community's focus has shifted to the more challenging task of text-to-4D synthesis, which introduces a temporal dimension to generate dynamic 3D objects. In this context, we identify Score Distillation Sampling (SDS), a widely used technique for text-to-3D synthesis, as a significant hindrance to text-to-4D performance due to its Janus-faced and texture-unrealistic problems coupled with high computational costs. In this paper, we propose \textbf{P}ixel-\textbf{L}evel \textbf{A}lignments for Text-to-\textbf{4D} Gaussian Splatting (\textbf{PLA4D}), a novel method that utilizes text-to-video frames as explicit pixel alignment targets to generate static 3D objects and inject motion into them. Specifically, we introduce Focal Alignment to calibrate camera poses for rendering and GS-Mesh Contrastive Learning to distill geometry priors from rendered image contrasts at the pixel level. Additionally, we develop Motion Alignment using a deformation network to drive changes in Gaussians and implement Reference Refinement for smooth 4D object surfaces. These techniques enable 4D Gaussian Splatting to align geometry, texture, and motion with generated videos at the pixel level. Compared to previous methods, PLA4D produces synthesized outputs with better texture details in less time and effectively mitigates the Janus-faced problem. PLA4D is fully implemented using open-source models, offering an accessible, user-friendly, and promising direction for 4D digital content creation. Our project page: \href{https://github.com/MiaoQiaowei/PLA4D.github.io}{https://github.com/MiaoQiaowei/PLA4D.github.io}.
URLs: https://github.com/MiaoQiaowei/PLA4D.github.io, https://github.com/MiaoQiaowei/PLA4D.github.io
Authors: Wenli Sun, Xinyang Jiang, Dongsheng Li, Cairong Zhao
Abstract: Person Re-Identification (ReID) systems pose a significant security risk from backdoor attacks, allowing adversaries to evade tracking or impersonate others. Beyond recognizing this issue, we investigate how backdoor attacks can be deployed in real-world scenarios, where a ReID model is typically trained on data collected in the digital domain and then deployed in a physical environment. This attack scenario requires an attack flow that embeds backdoor triggers in the digital domain realistically enough to also activate the buried backdoor in person ReID models in the physical domain. This paper realizes this attack flow by leveraging a diffusion model to generate realistic accessories on pedestrian images (e.g., bags, hats, etc.) as backdoor triggers. However, the noticeable domain gap between the triggers generated by the off-the-shelf diffusion model and their physical counterparts results in a low attack success rate. Therefore, we introduce a novel diffusion-based physical backdoor attack (DiffPhysBA) method that adopts a training-free similarity-guided sampling process to enhance the resemblance between generated and physical triggers. Consequently, DiffPhysBA can generate realistic attributes as semantic-level triggers in the digital domain and provides higher physical ASR compared to the direct paste method by 25.6% on the real-world test set. Through evaluations on newly proposed real-world and synthetic ReID test sets, DiffPhysBA demonstrates an impressive success rate exceeding 90% in both the digital and physical domains. Notably, it excels in digital stealth metrics and can effectively evade state-of-the-art defense methods.
Authors: Honghao Fu, Yufei Wang, Wenhan Yang, Bihan Wen
Abstract: Image quality assessment (IQA) plays a critical role in selecting high-quality images and guiding compression and enhancement methods in a series of applications. The blind IQA, which assesses the quality of in-the-wild images containing complex authentic distortions without reference images, poses greater challenges. Existing methods are limited to modeling a uniform distribution with local patches and are bothered by the gap between low and high-level visions (caused by widely adopted pre-trained classification networks). In this paper, we propose a novel IQA method called diffusion priors-based IQA (DP-IQA), which leverages the prior knowledge from the pre-trained diffusion model with its excellent powers to bridge semantic gaps in the perception of the visual quality of images. Specifically, we use pre-trained stable diffusion as the backbone, extract multi-level features from the denoising U-Net during the upsampling process at a specified timestep, and decode them to estimate the image quality score. The text and image adapters are adopted to mitigate the domain gap for downstream tasks and correct the information loss caused by the variational autoencoder bottleneck. Finally, we distill the knowledge in the above model into a CNN-based student model, significantly reducing the parameter to enhance applicability, with the student model performing similarly or even better than the teacher model surprisingly. Experimental results demonstrate that our DP-IQA achieves state-of-the-art results on various in-the-wild datasets with better generalization capability, which shows the superiority of our method in global modeling and utilizing the hierarchical feature clues of diffusion for evaluating image quality.
Authors: Bin Ren, Yawei Li, Jingyun Liang, Rakesh Ranjan, Mengyuan Liu, Rita Cucchiara, Luc Van Gool, Ming-Hsuan Yang, Nicu Sebe
Abstract: Image Restoration (IR), a classic low-level vision task, has witnessed significant advancements through deep models that effectively model global information. Notably, the Vision Transformers (ViTs) emergence has further propelled these advancements. When computing, the self-attention mechanism, a cornerstone of ViTs, tends to encompass all global cues, even those from semantically unrelated objects or regions. This inclusivity introduces computational inefficiencies, particularly noticeable with high input resolution, as it requires processing irrelevant information, thereby impeding efficiency. Additionally, for IR, it is commonly noted that small segments of a degraded image, particularly those closely aligned semantically, provide particularly relevant information to aid in the restoration process, as they contribute essential contextual cues crucial for accurate reconstruction. To address these challenges, we propose boosting IR's performance by sharing the key semantics via Transformer for IR (i.e., SemanIR) in this paper. Specifically, SemanIR initially constructs a sparse yet comprehensive key-semantic dictionary within each transformer stage by establishing essential semantic connections for every degraded patch. Subsequently, this dictionary is shared across all subsequent transformer blocks within the same stage. This strategy optimizes attention calculation within each block by focusing exclusively on semantically related components stored in the key-semantic dictionary. As a result, attention calculation achieves linear computational complexity within each window. Extensive experiments across 6 IR tasks confirm the proposed SemanIR's state-of-the-art performance, quantitatively and qualitatively showcasing advancements.
Authors: Michael Fuchs, Emilie Genty, Adrian Bangerter, Klaus Zuberb\"uhler, Paul Cotofrei
Abstract: This paper addresses the significant challenge of recognizing behaviors in non-human primates, specifically focusing on chimpanzees. Automated behavior recognition is crucial for both conservation efforts and the advancement of behavioral research. However, it is significantly hindered by the labor-intensive process of manual video annotation. Despite the availability of large-scale animal behavior datasets, the effective application of machine learning models across varied environmental settings poses a critical challenge, primarily due to the variability in data collection contexts and the specificity of annotations. In this paper, we introduce ChimpBehave, a novel dataset featuring over 2 hours of video (approximately 193,000 video frames) of zoo-housed chimpanzees, meticulously annotated with bounding boxes and behavior labels for action recognition. ChimpBehave uniquely aligns its behavior classes with existing datasets, allowing for the study of domain adaptation and cross-dataset generalization methods between different visual settings. Furthermore, we benchmark our dataset using a state-of-the-art CNN-based action recognition model, providing the first baseline results for both within and cross-dataset settings. The dataset, models, and code can be accessed at: https://github.com/MitchFuchs/ChimpBehave
Authors: Masashi Hatano, Ryo Hachiuma, Hideo Saito
Abstract: Predicting future human behavior from egocentric videos is a challenging but critical task for human intention understanding. Existing methods for forecasting 2D hand positions rely on visual representations and mainly focus on hand-object interactions. In this paper, we investigate the hand forecasting task and tackle two significant issues that persist in the existing methods: (1) 2D hand positions in future frames are severely affected by ego-motions in egocentric videos; (2) prediction based on visual information tends to overfit to background or scene textures, posing a challenge for generalization on novel scenes or human behaviors. To solve the aforementioned problems, we propose EMAG, an ego-motion-aware and generalizable 2D hand forecasting method. In response to the first problem, we propose a method that considers ego-motion, represented by a sequence of homography matrices of two consecutive frames. We further leverage modalities such as optical flow, trajectories of hands and interacting objects, and ego-motions, thereby alleviating the second issue. Extensive experiments on two large-scale egocentric video datasets, Ego4D and EPIC-Kitchens 55, verify the effectiveness of the proposed method. In particular, our model outperforms prior methods by $7.0$\% on cross-dataset evaluations. Project page: https://masashi-hatano.github.io/EMAG/
Authors: Pengyu Jie, Wanquan Liu, Chenqiang Gao, Yihui Wen, Rui He, Pengcheng Li, Jintao Zhang, Deyu Meng
Abstract: The lesion segmentation on endoscopic images is challenging due to its complex and ambiguous features. Fully-supervised deep learning segmentation methods can receive good performance based on entirely pixel-level labeled dataset but greatly increase experts' labeling burden. Semi-supervised and weakly supervised methods can ease labeling burden, but heavily strengthen the learning difficulty. To alleviate this difficulty, weakly semi-supervised segmentation adopts a new annotation protocol of adding a large number of point annotation samples into a few pixel-level annotation samples. However, existing methods only mine points' limited information while ignoring reliable prior surrounding the point annotations. In this paper, we propose a weakly semi-supervised method called Point-Neighborhood Learning (PNL) framework. To mine the prior of the pixels surrounding the annotated point, we transform a single-point annotation into a circular area named a point-neighborhood. We propose point-neighborhood supervision loss and pseudo-label scoring mechanism to enhance training supervision. Point-neighborhoods are also used to augment the data diversity. Our method greatly improves performance without changing the structure of segmentation network. Comprehensive experiments show the superiority of our method over the other existing methods, demonstrating its effectiveness in point-annotated medical images. The project code will be available on: https://github.com/ParryJay/PNL.
Authors: Abdelmalik Ouamane, Ammar Chouchane, Yassine Himeur, Abderrazak Debilou, Abbes Amira, Shadi Atalla, Wathiq Mansoor, Hussain Al Ahmad
Abstract: Machine learning has revolutionized the field of agricultural science, particularly in the early detection and management of plant diseases, which are crucial for maintaining crop health and productivity. Leveraging advanced algorithms and imaging technologies, researchers are now able to identify and classify plant diseases with unprecedented accuracy and speed. Effective management of tomato diseases is crucial for enhancing agricultural productivity. The development and application of tomato disease classification methods are central to this objective. This paper introduces a cutting-edge technique for the detection and classification of tomato leaf diseases, utilizing insights from the latest pre-trained Convolutional Neural Network (CNN) models. We propose a sophisticated approach within the domain of tensor subspace learning, known as Higher-Order Whitened Singular Value Decomposition (HOWSVD), designed to boost the discriminatory power of the system. Our approach to Tensor Subspace Learning is methodically executed in two phases, beginning with HOWSVD and culminating in Multilinear Discriminant Analysis (MDA). The efficacy of this innovative method was rigorously tested through comprehensive experiments on two distinct datasets, namely PlantVillage and the Taiwan dataset. The findings reveal that HOWSVD-MDA outperforms existing methods, underscoring its capability to markedly enhance the precision and dependability of diagnosing tomato leaf diseases. For instance, up to 98.36\% and 89.39\% accuracy scores have been achieved under PlantVillage and the Taiwan datasets, respectively.
Authors: Kagan Ozturk, Haiyu Wu, Kevin W. Bowyer
Abstract: Appearance of a face can be greatly altered by growing a beard and mustache. The facial hairstyles in a pair of images can cause marked changes to the impostor distribution and the genuine distribution. Also, different distributions of facial hairstyle across demographics could cause a false impression of relative accuracy across demographics. We first show that, even though larger training sets boost the recognition accuracy on all facial hairstyles, accuracy variations caused by facial hairstyles persist regardless of the size of the training set. Then, we analyze the impact of having different fractions of the training data represent facial hairstyles. We created balanced training sets using a set of identities available in Webface42M that both have clean-shaven and facial hair images. We find that, even when a face recognition model is trained with a balanced clean-shaven / facial hair training set, accuracy variation on the test data does not diminish. Next, data augmentation is employed to further investigate the effect of facial hair distribution in training data by manipulating facial hair pixels with the help of facial landmark points and a facial hair segmentation model. Our results show facial hair causes an accuracy gap between clean-shaven and facial hair images, and this impact can be significantly different between African-Americans and Caucasians.
Authors: Stavros Diolatzis, Tobias Zirr, Alexandr Kuznetsov, Georgios Kopanas, Anton Kaplanyan
Abstract: In the wake of many new ML-inspired approaches for reconstructing and representing high-quality 3D content, recent hybrid and explicitly learned representations exhibit promising performance and quality characteristics. However, their scaling to higher dimensions is challenging, e.g. when accounting for dynamic content with respect to additional parameters such as material properties, illumination, or time. In this paper, we tackle these challenges for an explicit representations based on Gaussian mixture models. With our solutions, we arrive at efficient fitting of compact N-dimensional Gaussian mixtures and enable efficient evaluation at render time: For fast fitting and evaluation, we introduce a high-dimensional culling scheme that efficiently bounds N-D Gaussians, inspired by Locality Sensitive Hashing. For adaptive refinement yet compact representation, we introduce a loss-adaptive density control scheme that incrementally guides the use of additional capacity towards missing details. With these tools we can for the first time represent complex appearance that depends on many input dimensions beyond position or viewing angle within a compact, explicit representation optimized in minutes and rendered in milliseconds.
Authors: Xiao Xu, Keyin Zhou, Yan Zhang, Yang Wang, Fei Wang, Xizhe Zhang
Abstract: Mood disorders, including depression and anxiety, often manifest through facial expressions. While previous research has explored the connection between facial features and emotions, machine learning algorithms for estimating mood disorder severity have been hindered by small datasets and limited real-world application. To address this gap, we analyzed facial videos of 11,427 participants, a dataset two orders of magnitude larger than previous studies. This comprehensive collection includes standardized facial expression videos from reading tasks, along with a detailed psychological scale that measures depression, anxiety, and stress. By examining the relationships among these emotional states and employing clustering analysis, we identified distinct subgroups embodying different emotional profiles. We then trained tree-based classifiers and deep learning models to estimate emotional states from facial features. Results indicate that models previously effective on small datasets experienced decreased performance when applied to our large dataset, highlighting the importance of data scale and mitigating overfitting in practical settings. Notably, our study identified subtle shifts in pupil dynamics and gaze orientation as potential markers of mood disorders, providing valuable information on the interaction between facial expressions and mental health. This research marks the first large-scale and comprehensive investigation of facial expressions in the context of mental health, laying the groundwork for future data-driven advancements in this field.
Authors: Kai Wu, Boyuan Jiang, Zhengkai Jiang, Qingdong He, Donghao Luo, Shengzhi Wang, Qingwen Liu, Chengjie Wang
Abstract: Multimodal large language models (MLLMs) contribute a powerful mechanism to understanding visual information building on large language models. However, MLLMs are notorious for suffering from hallucinations, especially when generating lengthy, detailed descriptions for images. Our analysis reveals that hallucinations stem from the inherent summarization mechanism of large language models, leading to excessive dependence on linguistic tokens while neglecting vision information. In this paper, we propose NoiseBoost, a broadly applicable and simple method for alleviating hallucinations for MLLMs through the integration of noise feature perturbations. Noise perturbation acts as a regularizer, facilitating a balanced distribution of attention weights among visual and linguistic tokens. Despite its simplicity, NoiseBoost consistently enhances the performance of MLLMs across common training strategies, including supervised fine-tuning and reinforcement learning. Further, NoiseBoost pioneerly enables semi-supervised learning for MLLMs, unleashing the power of unlabeled data. Comprehensive experiments demonstrate that NoiseBoost improves dense caption accuracy by 8.1% with human evaluation and achieves comparable results with 50% of the data by mining unlabeled data. Code and models are available at https://kaiwu5.github.io/noiseboost.
Authors: Muhammad Saif Ullah Khan, Dhavalkumar Limbachiya, Didier Stricker, Muhammad Zeshan Afzal
Abstract: Human pose estimation is a key task in computer vision with various applications such as activity recognition and interactive systems. However, the lack of consistency in the annotated skeletons across different datasets poses challenges in developing universally applicable models. To address this challenge, we propose a novel approach integrating multi-teacher knowledge distillation with a unified skeleton representation. Our networks are jointly trained on the COCO and MPII datasets, containing 17 and 16 keypoints, respectively. We demonstrate enhanced adaptability by predicting an extended set of 21 keypoints, 4 (COCO) and 5 (MPII) more than original annotations, improving cross-dataset generalization. Our joint models achieved an average accuracy of 70.89 and 76.40, compared to 53.79 and 55.78 when trained on a single dataset and evaluated on both. Moreover, we also evaluate all 21 predicted points by our two models by reporting an AP of 66.84 and 72.75 on the Halpe dataset. This highlights the potential of our technique to address one of the most pressing challenges in pose estimation research and application - the inconsistency in skeletal annotations.
Authors: Hao Cheng, Erjia Xiao, Jiahang Cao, Le Yang, Kaidi Xu, Jindong Gu, Renjing Xu
Abstract: Following the advent of the Artificial Intelligence (AI) era of large models, Multimodal Large Language Models (MLLMs) with the ability to understand cross-modal interactions between vision and text have attracted wide attention. Adversarial examples with human-imperceptible perturbation are shown to possess a characteristic known as transferability, which means that a perturbation generated by one model could also mislead another different model. Augmenting the diversity in input data is one of the most significant methods for enhancing adversarial transferability. This method has been certified as a way to significantly enlarge the threat impact under black-box conditions. Research works also demonstrate that MLLMs can be exploited to generate adversarial examples in the white-box scenario. However, the adversarial transferability of such perturbations is quite limited, failing to achieve effective black-box attacks across different models. In this paper, we propose the Typographic-based Semantic Transfer Attack (TSTA), which is inspired by: (1) MLLMs tend to process semantic-level information; (2) Typographic Attack could effectively distract the visual information captured by MLLMs. In the scenarios of Harmful Word Insertion and Important Information Protection, our TSTA demonstrates superior performance.
Authors: Navarro Miriam, Becerra \'Alvaro, Daza Roberto, Cobos Ruth, Morales Aythami, Fierrez Julian
Abstract: In this paper, we present an approach in the Multimodal Learning Analytics field. Within this approach, we have developed a tool to visualize and analyze eye movement data collected during learning sessions in online courses. The tool is named VAAD (an acronym for Visual Attention Analysis Dashboard). These eye movement data have been gathered using an eye-tracker and subsequently processed and visualized for interpretation. The purpose of the tool is to conduct a descriptive analysis of the data by facilitating its visualization, enabling the identification of differences and learning patterns among various learner populations. Additionally, it integrates a predictive module capable of anticipating learner activities during a learning session. Consequently, VAAD holds the potential to offer valuable insights into online learning behaviors from both descriptive and predictive perspectives.
Authors: Luca Barco, Angelica Urbanelli, Claudio Rossi
Abstract: Rapid detection and well-timed intervention are essential to mitigate the impacts of wildfires. Leveraging remote sensed data from satellite networks and advanced AI models to automatically detect hotspots (i.e., thermal anomalies caused by active fires) is an effective way to build wildfire monitoring systems. In this work, we propose a novel dataset containing time series of remotely sensed data related to European fire events and a Self-Supervised Learning (SSL)-based model able to analyse multi-temporal data and identify hotspots in potentially near real time. We train and evaluate the performance of our model using our dataset and Thraws, a dataset of thermal anomalies including several fire events, obtaining an F1 score of 63.58.
Authors: Edoardo Arnaudo, Jacopo Lungo Vaschetti, Lorenzo Innocenti, Luca Barco, Davide Lisi, Vanina Fissore, Claudio Rossi
Abstract: Very-High Resolution (VHR) remote sensing imagery is increasingly accessible, but often lacks annotations for effective machine learning applications. Recent foundation models like GroundingDINO and Segment Anything (SAM) provide opportunities to automatically generate annotations. This study introduces FMARS (Foundation Model Annotations in Remote Sensing), a methodology leveraging VHR imagery and foundation models for fast and robust annotation. We focus on disaster management and provide a large-scale dataset with labels obtained from pre-event imagery over 19 disaster events, derived from the Maxar Open Data initiative. We train segmentation models on the generated labels, using Unsupervised Domain Adaptation (UDA) techniques to increase transferability to real-world scenarios. Our results demonstrate the effectiveness of leveraging foundation models to automatically annotate remote sensing data at scale, enabling robust downstream models for critical applications. Code and dataset are available at \url{https://github.com/links-ads/igarss-fmars}.
Authors: Zhiyuan He, Pin-Yu Chen, Tsung-Yi Ho
Abstract: The rapid advances in generative AI models have empowered the creation of highly realistic images with arbitrary content, raising concerns about potential misuse and harm, such as Deepfakes. Current research focuses on training detectors using large datasets of generated images. However, these training-based solutions are often computationally expensive and show limited generalization to unseen generated images. In this paper, we propose a training-free method to distinguish between real and AI-generated images. We first observe that real images are more robust to tiny noise perturbations than AI-generated images in the representation space of vision foundation models. Based on this observation, we propose RIGID, a training-free and model-agnostic method for robust AI-generated image detection. RIGID is a simple yet effective approach that identifies whether an image is AI-generated by comparing the representation similarity between the original and the noise-perturbed counterpart. Our evaluation on a diverse set of AI-generated images and benchmarks shows that RIGID significantly outperforms existing trainingbased and training-free detectors. In particular, the average performance of RIGID exceeds the current best training-free method by more than 25%. Importantly, RIGID exhibits strong generalization across different image generation methods and robustness to image corruptions.
Authors: Prashanth Chandran, Gaspard Zoss, Paulo Gotardo, Derek Bradley
Abstract: In this paper, we examine 3 important issues in the practical use of state-of-the-art facial landmark detectors and show how a combination of specific architectural modifications can directly improve their accuracy and temporal stability. First, many facial landmark detectors require face normalization as a preprocessing step, which is accomplished by a separately-trained neural network that crops and resizes the face in the input image. There is no guarantee that this pre-trained network performs the optimal face normalization for landmark detection. We instead analyze the use of a spatial transformer network that is trained alongside the landmark detector in an unsupervised manner, and jointly learn optimal face normalization and landmark detection. Second, we show that modifying the output head of the landmark predictor to infer landmarks in a canonical 3D space can further improve accuracy. To convert the predicted 3D landmarks into screen-space, we additionally predict the camera intrinsics and head pose from the input image. As a side benefit, this allows to predict the 3D face shape from a given image only using 2D landmarks as supervision, which is useful in determining landmark visibility among other things. Finally, when training a landmark detector on multiple datasets at the same time, annotation inconsistencies across datasets forces the network to produce a suboptimal average. We propose to add a semantic correction network to address this issue. This additional lightweight neural network is trained alongside the landmark detector, without requiring any additional supervision. While the insights of this paper can be applied to most common landmark detectors, we specifically target a recently-proposed continuous 2D landmark detector to demonstrate how each of our additions leads to meaningful improvements over the state-of-the-art on standard benchmarks.
Authors: Amine Bechar, Youssef Elmir, Yassine Himeur, Rafik Medjoudj, Abbes Amira
Abstract: This review article discusses the roles of federated learning (FL) and transfer learning (TL) in cancer detection based on image analysis. These two strategies powered by machine learning have drawn a lot of attention due to their potential to increase the precision and effectiveness of cancer diagnosis in light of the growing importance of machine learning techniques in cancer detection. FL enables the training of machine learning models on data distributed across multiple sites without the need for centralized data sharing, while TL allows for the transfer of knowledge from one task to another. A comprehensive assessment of the two methods, including their strengths, and weaknesses is presented. Moving on, their applications in cancer detection are discussed, including potential directions for the future. Finally, this article offers a thorough description of the functions of TL and FL in image-based cancer detection. The authors also make insightful suggestions for additional study in this rapidly developing area.
Authors: Liyun Deng, Lei Jin, Guangcheng Wang, Quan Shi, Han Wang
Abstract: In response to the social issue of the increasing number of elderly vulnerable groups going missing due to the aggravating aging population in China, our team has developed a wearable anti-loss device and intelligent early warning system for elderly individuals with intermittent dementia using artificial intelligence and IoT technology. This system comprises an anti-loss smart helmet, a cloud computing module, and an intelligent early warning application on the caregiver's mobile device. The smart helmet integrates a miniature camera module, a GPS module, and a 5G communication module to collect first-person images and location information of the elderly. Data is transmitted remotely via 5G, FTP, and TCP protocols. In the cloud computing module, our team has proposed for the first time a multimodal dangerous state recognition network based on scene and location information to accurately assess the risk of elderly individuals going missing. Finally, the application software interface designed for the caregiver's mobile device implements multi-level early warnings. The system developed by our team requires no operation or response from the elderly, achieving fully automatic environmental perception, risk assessment, and proactive alarming. This overcomes the limitations of traditional monitoring devices, which require active operation and response, thus avoiding the issue of the digital divide for the elderly. It effectively prevents accidental loss and potential dangers for elderly individuals with dementia.
Authors: Gonca Yilmaz, Songyou Peng, Francis Engelmann, Marc Pollefeys, Hermann Blum
Abstract: The advent of Vision Language Models (VLMs) transformed image understanding from closed-set classifications to dynamic image-language interactions, enabling open-vocabulary segmentation. Despite this flexibility, VLMs often fall behind closed-set classifiers in accuracy due to their reliance on ambiguous image captions and lack of domain-specific knowledge. We, therefore, introduce a new task domain adaptation for open-vocabulary segmentation, enhancing VLMs with domain-specific priors while preserving their open-vocabulary nature. Existing adaptation methods, when applied to segmentation tasks, improve performance on training queries but can reduce VLM performance on zero-shot text inputs. To address this shortcoming, we propose an approach that combines parameter-efficient prompt tuning with a triplet-loss-based training strategy. This strategy is designed to enhance open-vocabulary generalization while adapting to the visual domain. Our results outperform other parameter-efficient adaptation strategies in open-vocabulary segment classification tasks across indoor and outdoor datasets. Notably, our approach is the only one that consistently surpasses the original VLM on zero-shot queries. Our adapted VLMs can be plug-and-play integrated into existing open-vocabulary segmentation pipelines, improving OV-Seg by +6.0% mIoU on ADE20K, and OpenMask3D by +4.1% AP on ScanNet++ Offices without any changes to the methods.
Authors: Phillip Howard, Kathleen C. Fraser, Anahita Bhiwandiwalla, Svetlana Kiritchenko
Abstract: With the advent of Large Language Models (LLMs) possessing increasingly impressive capabilities, a number of Large Vision-Language Models (LVLMs) have been proposed to augment LLMs with visual inputs. Such models condition generated text on both an input image and a text prompt, enabling a variety of use cases such as visual question answering and multimodal chat. While prior studies have examined the social biases contained in text generated by LLMs, this topic has been relatively unexplored in LVLMs. Examining social biases in LVLMs is particularly challenging due to the confounding contributions of bias induced by information contained across the text and visual modalities. To address this challenging problem, we conduct a large-scale study of text generated by different LVLMs under counterfactual changes to input images. Specifically, we present LVLMs with identical open-ended text prompts while conditioning on images from different counterfactual sets, where each set contains images which are largely identical in their depiction of a common subject (e.g., a doctor), but vary only in terms of intersectional social attributes (e.g., race and gender). We comprehensively evaluate the text produced by different models under this counterfactual generation setting at scale, producing over 57 million responses from popular LVLMs. Our multi-dimensional analysis reveals that social attributes such as race, gender, and physical characteristics depicted in input images can significantly influence the generation of toxic content, competency-associated words, harmful stereotypes, and numerical ratings of depicted individuals. We additionally explore the relationship between social bias in LVLMs and their corresponding LLMs, as well as inference-time strategies to mitigate bias.
Authors: Lukas Uzolas, Elmar Eisemann, Petr Kellnhofer
Abstract: Animation techniques bring digital 3D worlds and characters to life. However, manual animation is tedious and automated techniques are often specialized to narrow shape classes. In our work, we propose a technique for automatic re-animation of arbitrary 3D shapes based on a motion prior extracted from a video diffusion model. Unlike existing 4D generation methods, we focus solely on the motion, and we leverage an explicit mesh-based representation compatible with existing computer-graphics pipelines. Furthermore, our utilization of diffusion features enhances accuracy of our motion fitting. We analyze efficacy of these features for animation fitting and we experimentally validate our approach for two different diffusion models and four animation models. Finally, we demonstrate that our time-efficient zero-shot method achieves a superior performance re-animating a diverse set of 3D shapes when compared to existing techniques in a user study. The project website is located at https://lukas.uzolas.com/MotionDreamer.
Authors: Tommaso Monopoli, Fabio Montello, Claudio Rossi
Abstract: Landslides are one of the most critical and destructive geohazards. Widespread development of human activities and settlements combined with the effects of climate change on weather are resulting in a high increase in the frequency and destructive power of landslides, making them a major threat to human life and the economy. In this paper, we explore methodologies to map newly-occurred landslides using Sentinel-2 imagery automatically. All approaches presented are framed as a bi-temporal change detection problem, requiring only a pair of Sentinel-2 images, taken respectively before and after a landslide-triggering event. Furthermore, we introduce a novel deep learning architecture for fusing Sentinel-2 bi-temporal image pairs with Digital Elevation Model (DEM) data, showcasing its promising performances w.r.t. other change detection models in the literature. As a parallel task, we address limitations in existing datasets by creating a novel geodatabase, which includes manually validated open-access landslide inventories over heterogeneous ecoregions of the world. We release both code and dataset with an open-source license.
Authors: Yuxin Yao, Bailin Deng, Junhui Hou, Juyong Zhang
Abstract: Existing optimization-based methods for non-rigid registration typically minimize an alignment error metric based on the point-to-point or point-to-plane distance between corresponding point pairs on the source surface and target surface. However, these metrics can result in slow convergence or a loss of detail. In this paper, we propose SPARE, a novel formulation that utilizes a symmetrized point-to-plane distance for robust non-rigid registration. The symmetrized point-to-plane distance relies on both the positions and normals of the corresponding points, resulting in a more accurate approximation of the underlying geometry and can achieve higher accuracy than existing methods. To solve this optimization problem efficiently, we propose an alternating minimization solver using a majorization-minimization strategy. Moreover, for effective initialization of the solver, we incorporate a deformation graph-based coarse alignment that improves registration quality and efficiency. Extensive experiments show that the proposed method greatly improves the accuracy of non-rigid registration problems and maintains relatively high solution efficiency. The code is publicly available at https://github.com/yaoyx689/spare.
Authors: Sanghyeon Na, Yonggyu Kim, Hyunjoon Lee
Abstract: The generation of high-quality human images through text-to-image (T2I) methods is a significant yet challenging task. Distinct from general image generation, human image synthesis must satisfy stringent criteria related to human pose, anatomy, and alignment with textual prompts, making it particularly difficult to achieve realistic results. Recent advancements in T2I generation based on diffusion models have shown promise, yet challenges remain in meeting human-specific preferences. In this paper, we introduce a novel approach tailored specifically for human image generation utilizing Direct Preference Optimization (DPO). Specifically, we introduce an efficient method for constructing a specialized DPO dataset for training human image generation models without the need for costly human feedback. We also propose a modified loss function that enhances the DPO training process by minimizing artifacts and improving image fidelity. Our method demonstrates its versatility and effectiveness in generating human images, including personalized text-to-image generation. Through comprehensive evaluations, we show that our approach significantly advances the state of human image generation, achieving superior results in terms of natural anatomies, poses, and text-image alignment.
Authors: Muyao Niu, Xiaodong Cun, Xintao Wang, Yong Zhang, Ying Shan, Yinqiang Zheng
Abstract: We present MOFA-Video, an advanced controllable image animation method that generates video from the given image using various additional controllable signals (such as human landmarks reference, manual trajectories, and another even provided video) or their combinations. This is different from previous methods which only can work on a specific motion domain or show weak control abilities with diffusion prior. To achieve our goal, we design several domain-aware motion field adapters (\ie, MOFA-Adapters) to control the generated motions in the video generation pipeline. For MOFA-Adapters, we consider the temporal motion consistency of the video and generate the dense motion flow from the given sparse control conditions first, and then, the multi-scale features of the given image are wrapped as a guided feature for stable video diffusion generation. We naively train two motion adapters for the manual trajectories and the human landmarks individually since they both contain sparse information about the control. After training, the MOFA-Adapters in different domains can also work together for more controllable video generation.
Authors: Wangbo Yu, Chaoran Feng, Jiye Tang, Xu Jia, Li Yuan, Yonghong Tian
Abstract: 3D Gaussian Splatting (3D-GS) has demonstrated exceptional capabilities in 3D scene reconstruction and novel view synthesis. However, its training heavily depends on high-quality, sharp images and accurate camera poses. Fulfilling these requirements can be challenging in non-ideal real-world scenarios, where motion-blurred images are commonly encountered in high-speed moving cameras or low-light environments that require long exposure times. To address these challenges, we introduce Event Stream Assisted Gaussian Splatting (EvaGaussians), a novel approach that integrates event streams captured by an event camera to assist in reconstructing high-quality 3D-GS from blurry images. Capitalizing on the high temporal resolution and dynamic range offered by the event camera, we leverage the event streams to explicitly model the formation process of motion-blurred images and guide the deblurring reconstruction of 3D-GS. By jointly optimizing the 3D-GS parameters and recovering camera motion trajectories during the exposure time, our method can robustly facilitate the acquisition of high-fidelity novel views with intricate texture details. We comprehensively evaluated our method and compared it with previous state-of-the-art deblurring rendering methods. Both qualitative and quantitative comparisons demonstrate that our method surpasses existing techniques in restoring fine details from blurry images and producing high-fidelity novel views.
Authors: Ayyub Alzahem, Wadii Boulila, Maha Driss, Anis Koubaa
Abstract: Addressing uncertainty in Deep Learning (DL) is essential, as it enables the development of models that can make reliable predictions and informed decisions in complex, real-world environments where data may be incomplete or ambiguous. This paper introduces a novel algorithm leveraging Dempster-Shafer Theory (DST) to integrate multiple pre-trained models to form an ensemble capable of providing more reliable and enhanced classifications. The main steps of the proposed method include feature extraction, mass function calculation, fusion, and expected utility calculation. Several experiments have been conducted on CIFAR-10 and CIFAR-100 datasets, demonstrating superior classification accuracy of the proposed DST-based method, achieving improvements of 5.4% and 8.4%, respectively, compared to the best individual pre-trained models. Results highlight the potential of DST as a robust framework for managing uncertainties related to data when applying DL in real-world scenarios.
Authors: Fabio A. Faria, Mateus M. Souza, Raoni F. da S. Teixeira, Mauricio P. Segundo
Abstract: The proliferation of deep learning solutions and the scarcity of large annotated datasets pose significant challenges in real-world applications. Various strategies have been explored to overcome this challenge, with data augmentation (DA) approaches emerging as prominent solutions. DA approaches involve generating additional examples by transforming existing labeled data, thereby enriching the dataset and helping deep learning models achieve improved generalization without succumbing to overfitting. In real applications, where solutions based on deep learning are widely used, there is facial expression recognition (FER), which plays an essential role in human communication, improving a range of knowledge areas (e.g., medicine, security, and marketing). In this paper, we propose a simple and comprehensive face data augmentation approach based on mixed face component regularization that outperforms the classical DA approaches from the literature, including the MixAugment which is a specific approach for the target task in two well-known FER datasets existing in the literature.
Authors: Sijie Zhao, Yong Zhang, Xiaodong Cun, Shaoshu Yang, Muyao Niu, Xiaoyu Li, Wenbo Hu, Ying Shan
Abstract: Spatio-temporal compression of videos, utilizing networks such as Variational Autoencoders (VAE), plays a crucial role in OpenAI's SORA and numerous other video generative models. For instance, many LLM-like video models learn the distribution of discrete tokens derived from 3D VAEs within the VQVAE framework, while most diffusion-based video models capture the distribution of continuous latent extracted by 2D VAEs without quantization. The temporal compression is simply realized by uniform frame sampling which results in unsmooth motion between consecutive frames. Currently, there lacks of a commonly used continuous video (3D) VAE for latent diffusion-based video models in the research community. Moreover, since current diffusion-based approaches are often implemented using pre-trained text-to-image (T2I) models, directly training a video VAE without considering the compatibility with existing T2I models will result in a latent space gap between them, which will take huge computational resources for training to bridge the gap even with the T2I models as initialization. To address this issue, we propose a method for training a video VAE of latent video models, namely CV-VAE, whose latent space is compatible with that of a given image VAE, e.g., image VAE of Stable Diffusion (SD). The compatibility is achieved by the proposed novel latent space regularization, which involves formulating a regularization loss using the image VAE. Benefiting from the latent space compatibility, video models can be trained seamlessly from pre-trained T2I or video models in a truly spatio-temporally compressed latent space, rather than simply sampling video frames at equal intervals. With our CV-VAE, existing video models can generate four times more frames with minimal finetuning. Extensive experiments are conducted to demonstrate the effectiveness of the proposed video VAE.
Authors: Chaoyang Wang, Xiangtai Li, Lu Qi, Henghui Ding, Yunhai Tong, Ming-Hsuan Yang
Abstract: Semantic segmentation and semantic image synthesis are two representative tasks in visual perception and generation. While existing methods consider them as two distinct tasks, we propose a unified diffusion-based framework (SemFlow) and model them as a pair of reverse problems. Specifically, motivated by rectified flow theory, we train an ordinary differential equation (ODE) model to transport between the distributions of real images and semantic masks. As the training object is symmetric, samples belonging to the two distributions, images and semantic masks, can be effortlessly transferred reversibly. For semantic segmentation, our approach solves the contradiction between the randomness of diffusion outputs and the uniqueness of segmentation results. For image synthesis, we propose a finite perturbation approach to enhance the diversity of generated results without changing the semantic categories. Experiments show that our SemFlow achieves competitive results on semantic segmentation and semantic image synthesis tasks. We hope this simple framework will motivate people to rethink the unification of low-level and high-level vision. Project page: https://github.com/wang-chaoyang/SemFlow.
Authors: Minghao Guo, Bohan Wang, Kaiming He, Wojciech Matusik
Abstract: We present TetSphere splatting, an explicit, Lagrangian representation for reconstructing 3D shapes with high-quality geometry. In contrast to conventional object reconstruction methods which predominantly use Eulerian representations, including both neural implicit (e.g., NeRF, NeuS) and explicit representations (e.g., DMTet), and often struggle with high computational demands and suboptimal mesh quality, TetSphere splatting utilizes an underused but highly effective geometric primitive -- tetrahedral meshes. This approach directly yields superior mesh quality without relying on neural networks or post-processing. It deforms multiple initial tetrahedral spheres to accurately reconstruct the 3D shape through a combination of differentiable rendering and geometric energy optimization, resulting in significant computational efficiency. Serving as a robust and versatile geometry representation, Tet-Sphere splatting seamlessly integrates into diverse applications, including single-view 3D reconstruction, image-/text-to-3D content generation. Experimental results demonstrate that TetSphere splatting outperforms existing representations, delivering faster optimization speed, enhanced mesh quality, and reliable preservation of thin structures.
Authors: Jinrui Yang, Xianhang Li, Druv Pai, Yuyin Zhou, Yi Ma, Yaodong Yu, Cihang Xie
Abstract: CRATE, a white-box transformer architecture designed to learn compressed and sparse representations, offers an intriguing alternative to standard vision transformers (ViTs) due to its inherent mathematical interpretability. Despite extensive investigations into the scaling behaviors of language and vision transformers, the scalability of CRATE remains an open question which this paper aims to address. Specifically, we propose CRATE-$\alpha$, featuring strategic yet minimal modifications to the sparse coding block in the CRATE architecture design, and a light training recipe designed to improve the scalability of CRATE. Through extensive experiments, we demonstrate that CRATE-$\alpha$ can effectively scale with larger model sizes and datasets. For example, our CRATE-$\alpha$-B substantially outperforms the prior best CRATE-B model accuracy on ImageNet classification by 3.7%, achieving an accuracy of 83.2%. Meanwhile, when scaling further, our CRATE-$\alpha$-L obtains an ImageNet classification accuracy of 85.1%. More notably, these model performance improvements are achieved while preserving, and potentially even enhancing the interpretability of learned CRATE models, as we demonstrate through showing that the learned token representations of increasingly larger trained CRATE-$\alpha$ models yield increasingly higher-quality unsupervised object segmentation of images. The project page is https://rayjryang.github.io/CRATE-alpha/.
Authors: Himangi Mittal, Nakul Agarwal, Shao-Yuan Lo, Kwonjoon Lee
Abstract: We introduce PlausiVL, a large video-language model for anticipating action sequences that are plausible in the real-world. While significant efforts have been made towards anticipating future actions, prior approaches do not take into account the aspect of plausibility in an action sequence. To address this limitation, we explore the generative capability of a large video-language model in our work and further, develop the understanding of plausibility in an action sequence by introducing two objective functions, a counterfactual-based plausible action sequence learning loss and a long-horizon action repetition loss. We utilize temporal logical constraints as well as verb-noun action pair logical constraints to create implausible/counterfactual action sequences and use them to train the model with plausible action sequence learning loss. This loss helps the model to differentiate between plausible and not plausible action sequences and also helps the model to learn implicit temporal cues crucial for the task of action anticipation. The long-horizon action repetition loss puts a higher penalty on the actions that are more prone to repetition over a longer temporal window. With this penalization, the model is able to generate diverse, plausible action sequences. We evaluate our approach on two large-scale datasets, Ego4D and EPIC-Kitchens-100, and show improvements on the task of action anticipation.
Authors: Jianghao Shen, Tianfu Wu
Abstract: Learning 3D scene representation from a single-view image is a long-standing fundamental problem in computer vision, with the inherent ambiguity in predicting contents unseen from the input view. Built on the recently proposed 3D Gaussian Splatting (3DGS), the Splatter Image method has made promising progress on fast single-image novel view synthesis via learning a single 3D Gaussian for each pixel based on the U-Net feature map of an input image. However, it has limited expressive power to represent occluded components that are not observable in the input view. To address this problem, this paper presents a Hierarchical Splatter Image method in which a pixel is worth more than one 3D Gaussians. Specifically, each pixel is represented by a parent 3D Gaussian and a small number of child 3D Gaussians. Parent 3D Gaussians are learned as done in the vanilla Splatter Image. Child 3D Gaussians are learned via a lightweight Multi-Layer Perceptron (MLP) which takes as input the projected image features of a parent 3D Gaussian and the embedding of a target camera view. Both parent and child 3D Gaussians are learned end-to-end in a stage-wise way. The joint condition of input image features from eyes of the parent Gaussians and the target camera position facilitates learning to allocate child Gaussians to ``see the unseen'', recovering the occluded details that are often missed by parent Gaussians. In experiments, the proposed method is tested on the ShapeNet-SRN and CO3D datasets with state-of-the-art performance obtained, especially showing promising capabilities of reconstructing occluded contents in the input view.
Authors: Aditya Ganeshan, Ryan Y. Huang, Xianghao Xu, R. Kenny Jones, Daniel Ritchie
Abstract: The ability to edit 3D assets from natural language presents a compelling paradigm to aid in the democratization of 3D content creation. However, while natural language is often effective at communicating general intent, it is poorly suited for specifying precise manipulation. To address this gap, we introduce ParSEL, a system that enables controllable editing of high-quality 3D assets from natural language. Given a segmented 3D mesh and an editing request, ParSEL produces a parameterized editing program. Adjusting the program parameters allows users to explore shape variations with a precise control over the magnitudes of edits. To infer editing programs which align with an input edit request, we leverage the abilities of large-language models (LLMs). However, while we find that LLMs excel at identifying initial edit operations, they often fail to infer complete editing programs, and produce outputs that violate shape semantics. To overcome this issue, we introduce Analytical Edit Propagation (AEP), an algorithm which extends a seed edit with additional operations until a complete editing program has been formed. Unlike prior methods, AEP searches for analytical editing operations compatible with a range of possible user edits through the integration of computer algebra systems for geometric analysis. Experimentally we demonstrate ParSEL's effectiveness in enabling controllable editing of 3D objects through natural language requests over alternative system designs.
Authors: Sangyun Lee, Zinan Lin, Giulia Fanti
Abstract: Diffusion models have shown great promise for image and video generation, but sampling from state-of-the-art models requires expensive numerical integration of a generative ODE. One approach for tackling this problem is rectified flows, which iteratively learn smooth ODE paths that are less susceptible to truncation error. However, rectified flows still require a relatively large number of function evaluations (NFEs). In this work, we propose improved techniques for training rectified flows, allowing them to compete with knowledge distillation methods even in the low NFE setting. Our main insight is that under realistic settings, a single iteration of the Reflow algorithm for training rectified flows is sufficient to learn nearly straight trajectories; hence, the current practice of using multiple Reflow iterations is unnecessary. We thus propose techniques to improve one-round training of rectified flows, including a U-shaped timestep distribution and LPIPS-Huber premetric. With these techniques, we improve the FID of the previous 2-rectified flow by up to 72% in the 1 NFE setting on CIFAR-10. On ImageNet 64$\times$64, our improved rectified flow outperforms the state-of-the-art distillation methods such as consistency distillation and progressive distillation in both one-step and two-step settings and rivals the performance of improved consistency training (iCT) in FID. Code is available at https://github.com/sangyun884/rfpp.
Authors: Nan Huang, Xiaobao Wei, Wenzhao Zheng, Pengju An, Ming Lu, Wei Zhan, Masayoshi Tomizuka, Kurt Keutzer, Shanghang Zhang
Abstract: Photorealistic 3D reconstruction of street scenes is a critical technique for developing real-world simulators for autonomous driving. Despite the efficacy of Neural Radiance Fields (NeRF) for driving scenes, 3D Gaussian Splatting (3DGS) emerges as a promising direction due to its faster speed and more explicit representation. However, most existing street 3DGS methods require tracked 3D vehicle bounding boxes to decompose the static and dynamic elements for effective reconstruction, limiting their applications for in-the-wild scenarios. To facilitate efficient 3D scene reconstruction without costly annotations, we propose a self-supervised street Gaussian ($\textit{S}^3$Gaussian) method to decompose dynamic and static elements from 4D consistency. We represent each scene with 3D Gaussians to preserve the explicitness and further accompany them with a spatial-temporal field network to compactly model the 4D dynamics. We conduct extensive experiments on the challenging Waymo-Open dataset to evaluate the effectiveness of our method. Our $\textit{S}^3$Gaussian demonstrates the ability to decompose static and dynamic scenes and achieves the best performance without using 3D annotations. Code is available at: https://github.com/nnanhuang/S3Gaussian/.
Authors: Nicolas Dufour, Victor Besnier, Vicky Kalogeiton, David Picard
Abstract: Conditional diffusion models are powerful generative models that can leverage various types of conditional information, such as class labels, segmentation masks, or text captions. However, in many real-world scenarios, conditional information may be noisy or unreliable due to human annotation errors or weak alignment. In this paper, we propose the Coherence-Aware Diffusion (CAD), a novel method that integrates coherence in conditional information into diffusion models, allowing them to learn from noisy annotations without discarding data. We assume that each data point has an associated coherence score that reflects the quality of the conditional information. We then condition the diffusion model on both the conditional information and the coherence score. In this way, the model learns to ignore or discount the conditioning when the coherence is low. We show that CAD is theoretically sound and empirically effective on various conditional generation tasks. Moreover, we show that leveraging coherence generates realistic and diverse samples that respect conditional information better than models trained on cleaned datasets where samples with low coherence have been discarded.
Authors: Shuyuan Tu, Qi Dai, Zihao Zhang, Sicheng Xie, Zhi-Qi Cheng, Chong Luo, Xintong Han, Zuxuan Wu, Yu-Gang Jiang
Abstract: Despite impressive advancements in diffusion-based video editing models in altering video attributes, there has been limited exploration into modifying motion information while preserving the original protagonist's appearance and background. In this paper, we propose MotionFollower, a lightweight score-guided diffusion model for video motion editing. To introduce conditional controls to the denoising process, MotionFollower leverages two of our proposed lightweight signal controllers, one for poses and the other for appearances, both of which consist of convolution blocks without involving heavy attention calculations. Further, we design a score guidance principle based on a two-branch architecture, including the reconstruction and editing branches, which significantly enhance the modeling capability of texture details and complicated backgrounds. Concretely, we enforce several consistency regularizers and losses during the score estimation. The resulting gradients thus inject appropriate guidance to the intermediate latents, forcing the model to preserve the original background details and protagonists' appearances without interfering with the motion modification. Experiments demonstrate the competitive motion editing ability of MotionFollower qualitatively and quantitatively. Compared with MotionEditor, the most advanced motion editing model, MotionFollower achieves an approximately 80% reduction in GPU memory while delivering superior motion editing performance and exclusively supporting large camera movements and actions.
Authors: Chen Wang, Jiatao Gu, Xiaoxiao Long, Yuan Liu, Lingjie Liu
Abstract: 3D generation has seen remarkable progress in recent years. Existing techniques, such as score distillation methods, produce notable results but require extensive per-scene optimization, impacting time efficiency. Alternatively, reconstruction-based approaches prioritize efficiency but compromise quality due to their limited handling of uncertainty. We introduce GECO, a novel method for high-quality 3D generative modeling that operates within a second. Our approach addresses the prevalent issues of uncertainty and inefficiency in current methods through a two-stage approach. In the initial stage, we train a single-step multi-view generative model with score distillation. Then, a second-stage distillation is applied to address the challenge of view inconsistency from the multi-view prediction. This two-stage process ensures a balanced approach to 3D generation, optimizing both quality and efficiency. Our comprehensive experiments demonstrate that GECO achieves high-quality image-to-3D generation with an unprecedented level of efficiency.
Authors: Dixuan Lin, Yuxiang Zhang, Mengcheng Li, Yebin Liu, Wei Jing, Qi Yan, Qianying Wang, Hongwen Zhang
Abstract: In this paper, we introduce 4DHands, a robust approach to recovering interactive hand meshes and their relative movement from monocular inputs. Our approach addresses two major limitations of previous methods: lacking a unified solution for handling various hand image inputs and neglecting the positional relationship of two hands within images. To overcome these challenges, we develop a transformer-based architecture with novel tokenization and feature fusion strategies. Specifically, we propose a Relation-aware Two-Hand Tokenization (RAT) method to embed positional relation information into the hand tokens. In this way, our network can handle both single-hand and two-hand inputs and explicitly leverage relative hand positions, facilitating the reconstruction of intricate hand interactions in real-world scenarios. As such tokenization indicates the relative relationship of two hands, it also supports more effective feature fusion. To this end, we further develop a Spatio-temporal Interaction Reasoning (SIR) module to fuse hand tokens in 4D with attention and decode them into 3D hand meshes and relative temporal movements. The efficacy of our approach is validated on several benchmark datasets. The results on in-the-wild videos and real-world scenarios demonstrate the superior performances of our approach for interactive hand reconstruction. More video results can be found on the project page: https://4dhands.github.io.
Authors: Chinedu Innocent Nwoye, Nicolas Padoy
Abstract: Accurate tool tracking is essential for the success of computer-assisted intervention. Previous efforts often modeled tool trajectories rigidly, overlooking the dynamic nature of surgical procedures, especially tracking scenarios like out-of-body and out-of-camera views. Addressing this limitation, the new CholecTrack20 dataset provides detailed labels that account for multiple tool trajectories in three perspectives: (1) intraoperative, (2) intracorporeal, and (3) visibility, representing the different types of temporal duration of tool tracks. These fine-grained labels enhance tracking flexibility but also increase the task complexity. Re-identifying tools after occlusion or re-insertion into the body remains challenging due to high visual similarity, especially among tools of the same category. This work recognizes the critical role of the tool operators in distinguishing tool track instances, especially those belonging to the same tool category. The operators' information are however not explicitly captured in surgical videos. We therefore propose SurgiTrack, a novel deep learning method that leverages YOLOv7 for precise tool detection and employs an attention mechanism to model the originating direction of the tools, as a proxy to their operators, for tool re-identification. To handle diverse tool trajectory perspectives, SurgiTrack employs a harmonizing bipartite matching graph, minimizing conflicts and ensuring accurate tool identity association. Experimental results on CholecTrack20 demonstrate SurgiTrack's effectiveness, outperforming baselines and state-of-the-art methods with real-time inference capability. This work sets a new standard in surgical tool tracking, providing dynamic trajectories for more adaptable and precise assistance in minimally invasive surgeries.
Authors: Yao-Chih Lee, Yi-Ting Chen, Andrew Wang, Ting-Hsuan Liao, Brandon Y. Feng, Jia-Bin Huang
Abstract: We introduce VividDream, a method for generating explorable 4D scenes with ambient dynamics from a single input image or text prompt. VividDream first expands an input image into a static 3D point cloud through iterative inpainting and geometry merging. An ensemble of animated videos is then generated using video diffusion models with quality refinement techniques and conditioned on renderings of the static 3D scene from the sampled camera trajectories. We then optimize a canonical 4D scene representation using an animated video ensemble, with per-video motion embeddings and visibility masks to mitigate inconsistencies. The resulting 4D scene enables free-view exploration of a 3D scene with plausible ambient scene dynamics. Experiments demonstrate that VividDream can provide human viewers with compelling 4D experiences generated based on diverse real images and text prompts.
Authors: Jiaben Chen, Xin Yan, Yihang Chen, Siyuan Cen, Qinwei Ma, Haoyu Zhen, Kaizhi Qian, Lie Lu, Chuang Gan
Abstract: In this work, we introduce a challenging task for simultaneously generating 3D holistic body motions and singing vocals directly from textual lyrics inputs, advancing beyond existing works that typically address these two modalities in isolation. To facilitate this, we first collect the RapVerse dataset, a large dataset containing synchronous rapping vocals, lyrics, and high-quality 3D holistic body meshes. With the RapVerse dataset, we investigate the extent to which scaling autoregressive multimodal transformers across language, audio, and motion can enhance the coherent and realistic generation of vocals and whole-body human motions. For modality unification, a vector-quantized variational autoencoder is employed to encode whole-body motion sequences into discrete motion tokens, while a vocal-to-unit model is leveraged to obtain quantized audio tokens preserving content, prosodic information, and singer identity. By jointly performing transformer modeling on these three modalities in a unified way, our framework ensures a seamless and realistic blend of vocals and human motions. Extensive experiments demonstrate that our unified generation framework not only produces coherent and realistic singing vocals alongside human motions directly from textual inputs but also rivals the performance of specialized single-modality generation systems, establishing new benchmarks for joint vocal-motion generation. The project page is available for research purposes at https://vis-www.cs.umass.edu/RapVerse.
Authors: Lening Wang, Wenzhao Zheng, Yilong Ren, Han Jiang, Zhiyong Cui, Haiyang Yu, Jiwen Lu
Abstract: Understanding the evolution of 3D scenes is important for effective autonomous driving. While conventional methods mode scene development with the motion of individual instances, world models emerge as a generative framework to describe the general scene dynamics. However, most existing methods adopt an autoregressive framework to perform next-token prediction, which suffer from inefficiency in modeling long-term temporal evolutions. To address this, we propose a diffusion-based 4D occupancy generation model, OccSora, to simulate the development of the 3D world for autonomous driving. We employ a 4D scene tokenizer to obtain compact discrete spatial-temporal representations for 4D occupancy input and achieve high-quality reconstruction for long-sequence occupancy videos. We then learn a diffusion transformer on the spatial-temporal representations and generate 4D occupancy conditioned on a trajectory prompt. We conduct extensive experiments on the widely used nuScenes dataset with Occ3D occupancy annotations. OccSora can generate 16s-videos with authentic 3D layout and temporal consistency, demonstrating its ability to understand the spatial and temporal distributions of driving scenes. With trajectory-aware 4D generation, OccSora has the potential to serve as a world simulator for the decision-making of autonomous driving. Code is available at: https://github.com/wzzheng/OccSora.
Authors: Feipeng Ma, Hongwei Xue, Guangting Wang, Yizhou Zhou, Fengyun Rao, Shilin Yan, Yueyi Zhang, Siying Wu, Mike Zheng Shou, Xiaoyan Sun
Abstract: Existing Multimodal Large Language Models (MLLMs) follow the paradigm that perceives visual information by aligning visual features with the input space of Large Language Models (LLMs), and concatenating visual tokens with text tokens to form a unified sequence input for LLMs. These methods demonstrate promising results on various vision-language tasks but are limited by the high computational effort due to the extended input sequence resulting from the involvement of visual tokens. In this paper, instead of input space alignment, we propose a novel parameter space alignment paradigm that represents visual information as model weights. For each input image, we use a vision encoder to extract visual features, convert features into perceptual weights, and merge the perceptual weights with LLM's weights. In this way, the input of LLM does not require visual tokens, which reduces the length of the input sequence and greatly improves efficiency. Following this paradigm, we propose VLoRA with the perceptual weights generator. The perceptual weights generator is designed to convert visual features to perceptual weights with low-rank property, exhibiting a form similar to LoRA. The experimental results show that our VLoRA achieves comparable performance on various benchmarks for MLLMs, while significantly reducing the computational costs for both training and inference. The code and models will be made open-source.
Authors: Ling-Hao Chen, Shunlin Lu, Ailing Zeng, Hao Zhang, Benyou Wang, Ruimao Zhang, Lei Zhang
Abstract: This study delves into the realm of multi-modality (i.e., video and motion modalities) human behavior understanding by leveraging the powerful capabilities of Large Language Models (LLMs). Diverging from recent LLMs designed for video-only or motion-only understanding, we argue that understanding human behavior necessitates joint modeling from both videos and motion sequences (e.g., SMPL sequences) to capture nuanced body part dynamics and semantics effectively. In light of this, we present MotionLLM, a straightforward yet effective framework for human motion understanding, captioning, and reasoning. Specifically, MotionLLM adopts a unified video-motion training strategy that leverages the complementary advantages of existing coarse video-text data and fine-grained motion-text data to glean rich spatial-temporal insights. Furthermore, we collect a substantial dataset, MoVid, comprising diverse videos, motions, captions, and instructions. Additionally, we propose the MoVid-Bench, with carefully manual annotations, for better evaluation of human behavior understanding on video and motion. Extensive experiments show the superiority of MotionLLM in the caption, spatial-temporal comprehension, and reasoning ability.
Authors: Kailu Wu, Fangfu Liu, Zhihan Cai, Runjie Yan, Hanyang Wang, Yating Hu, Yueqi Duan, Kaisheng Ma
Abstract: In this work, we introduce Unique3D, a novel image-to-3D framework for efficiently generating high-quality 3D meshes from single-view images, featuring state-of-the-art generation fidelity and strong generalizability. Previous methods based on Score Distillation Sampling (SDS) can produce diversified 3D results by distilling 3D knowledge from large 2D diffusion models, but they usually suffer from long per-case optimization time with inconsistent issues. Recent works address the problem and generate better 3D results either by finetuning a multi-view diffusion model or training a fast feed-forward model. However, they still lack intricate textures and complex geometries due to inconsistency and limited generated resolution. To simultaneously achieve high fidelity, consistency, and efficiency in single image-to-3D, we propose a novel framework Unique3D that includes a multi-view diffusion model with a corresponding normal diffusion model to generate multi-view images with their normal maps, a multi-level upscale process to progressively improve the resolution of generated orthographic multi-views, as well as an instant and consistent mesh reconstruction algorithm called ISOMER, which fully integrates the color and geometric priors into mesh results. Extensive experiments demonstrate that our Unique3D significantly outperforms other image-to-3D baselines in terms of geometric and textural details.
Authors: Jonas Belouadi, Simone Paolo Ponzetto, Steffen Eger
Abstract: Creating high-quality scientific figures can be time-consuming and challenging, even though sketching ideas on paper is relatively easy. Furthermore, recreating existing figures that are not stored in formats preserving semantic information is equally complex. To tackle this problem, we introduce DeTikZify, a novel multimodal language model that automatically synthesizes scientific figures as semantics-preserving TikZ graphics programs based on sketches and existing figures. To achieve this, we create three new datasets: DaTikZv2, the largest TikZ dataset to date, containing over 360k human-created TikZ graphics; SketchFig, a dataset that pairs hand-drawn sketches with their corresponding scientific figures; and SciCap++, a collection of diverse scientific figures and associated metadata. We train DeTikZify on SciCap++ and DaTikZv2, along with synthetically generated sketches learned from SketchFig. We also introduce an MCTS-based inference algorithm that enables DeTikZify to iteratively refine its outputs without the need for additional training. Through both automatic and human evaluation, we demonstrate that DeTikZify outperforms commercial Claude 3 and GPT-4V in synthesizing TikZ programs, with the MCTS algorithm effectively boosting its performance. We make our code, models, and datasets publicly available.
Authors: Yuzhen Ding, Jason M. Holmes, Hongying Feng, Baoxin Li, Lisa A. McGee, Jean-Claude M. Rwigema, Sujay A. Vora, Daniel J. Ma, Robert L. Foote, Samir H. Patel, Wei Liu
Abstract: In radiotherapy, 2D orthogonally projected kV images are used for patient alignment when 3D-on-board imaging(OBI) unavailable. But tumor visibility is constrained due to the projection of patient's anatomy onto a 2D plane, potentially leading to substantial setup errors. In treatment room with 3D-OBI such as cone beam CT(CBCT), the field of view(FOV) of CBCT is limited with unnecessarily high imaging dose, thus unfavorable for pediatric patients. A solution to this dilemma is to reconstruct 3D CT from kV images obtained at the treatment position. Here, we propose a dual-models framework built with hierarchical ViT blocks. Unlike a proof-of-concept approach, our framework considers kV images as the solo input and can synthesize accurate, full-size 3D CT in real time(within milliseconds). We demonstrate the feasibility of the proposed approach on 10 patients with head and neck (H&N) cancer using image quality(MAE: <45HU), dosimetrical accuracy(Gamma passing rate (2%/2mm/10%)>97%) and patient position uncertainty(shift error: <0.4mm). The proposed framework can generate accurate 3D CT faithfully mirroring real-time patient position, thus significantly improving patient setup accuracy, keeping imaging dose minimum, and maintaining treatment veracity.
Authors: Shuai Shao, Yu Guan, Victor Sanchez
Abstract: Human Activity Recognition (HAR) has become increasingly popular with ubiquitous computing, driven by the popularity of wearable sensors in fields like healthcare and sports. While Convolutional Neural Networks (ConvNets) have significantly contributed to HAR, they often adopt a frame-by-frame analysis, concentrating on individual frames and potentially overlooking the broader temporal dynamics inherent in human activities. To address this, we propose the intra- and inter-frame attention model. This model captures both the nuances within individual frames and the broader contextual relationships across multiple frames, offering a comprehensive perspective on sequential data. We further enrich the temporal understanding by proposing a novel time-sequential batch learning strategy. This learning strategy preserves the chronological sequence of time-series data within each batch, ensuring the continuity and integrity of temporal patterns in sensor-based HAR.
Authors: Andrea Napoli, Paul White
Abstract: This paper considers the problem of model selection under domain shift. In this setting, it is proposed that a high maximum mean discrepancy (MMD) between the training and validation sets increases the generalisability of selected models. A data splitting algorithm based on kernel k-means clustering, which maximises this objective, is presented. The algorithm leverages linear programming to control the size, label, and (optionally) group distributions of the splits, and comes with convergence guarantees. The technique consistently outperforms alternative splitting strategies across a range of datasets and training algorithms, for both domain generalisation (DG) and unsupervised domain adaptation (UDA) tasks. Analysis also shows the MMD between the training and validation sets to be strongly rank-correlated ($\rho=0.63$) with test domain accuracy, further substantiating the validity of this approach.
Authors: Tugba Akinci D'Antonoli, Lucas K. Berger, Ashraya K. Indrakanti, Nathan Vishwanathan, Jakob Wei{\ss}, Matthias Jung, Zeynep Berkarda, Alexander Rau, Marco Reisert, Thomas K\"ustner, Alexandra Walter, Elmar M. Merkle, Martin Segeroth, Joshy Cyriac, Shan Yang, Jakob Wasserthal
Abstract: Purpose: To develop an open-source and easy-to-use segmentation model that can automatically and robustly segment most major anatomical structures in MR images independently of the MR sequence. Materials and Methods: In this study we extended the capabilities of TotalSegmentator to MR images. 298 MR scans and 227 CT scans were used to segment 59 anatomical structures (20 organs, 18 bones, 11 muscles, 7 vessels, 3 tissue types) relevant for use cases such as organ volumetry, disease characterization, and surgical planning. The MR and CT images were randomly sampled from routine clinical studies and thus represent a real-world dataset (different ages, pathologies, scanners, body parts, sequences, contrasts, echo times, repetition times, field strengths, slice thicknesses and sites). We trained an nnU-Net segmentation algorithm on this dataset and calculated Dice similarity coefficients (Dice) to evaluate the model's performance. Results: The model showed a Dice score of 0.824 (CI: 0.801, 0.842) on the test set, which included a wide range of clinical data with major pathologies. The model significantly outperformed two other publicly available segmentation models (Dice score, 0.824 versus 0.762; p<0.001 and 0.762 versus 0.542; p<0.001). On the CT image test set of the original TotalSegmentator paper it almost matches the performance of the original TotalSegmentator (Dice score, 0.960 versus 0.970; p<0.001). Conclusion: Our proposed model extends the capabilities of TotalSegmentator to MR images. The annotated dataset (https://zenodo.org/doi/10.5281/zenodo.11367004) and open-source toolkit (https://www.github.com/wasserth/TotalSegmentator) are publicly available.
URLs: https://zenodo.org/doi/10.5281/zenodo.11367004), https://www.github.com/wasserth/TotalSegmentator)
Authors: Haowen Lai, Gaoxiang Luo, Yifei Liu, Mingmin Zhao
Abstract: This paper introduces PanoRadar, a novel RF imaging system that brings RF resolution close to that of LiDAR, while providing resilience against conditions challenging for optical signals. Our LiDAR-comparable 3D imaging results enable, for the first time, a variety of visual recognition tasks at radio frequency, including surface normal estimation, semantic segmentation, and object detection. PanoRadar utilizes a rotating single-chip mmWave radar, along with a combination of novel signal processing and machine learning algorithms, to create high-resolution 3D images of the surroundings. Our system accurately estimates robot motion, allowing for coherent imaging through a dense grid of synthetic antennas. It also exploits the high azimuth resolution to enhance elevation resolution using learning-based methods. Furthermore, PanoRadar tackles 3D learning via 2D convolutions and addresses challenges due to the unique characteristics of RF signals. Our results demonstrate PanoRadar's robust performance across 12 buildings.
Authors: Pierre Chambon, Jean-Benoit Delbrouck, Thomas Sounack, Shih-Cheng Huang, Zhihong Chen, Maya Varma, Steven QH Truong, Chu The Chuong, Curtis P. Langlotz
Abstract: Since the release of the original CheXpert paper five years ago, CheXpert has become one of the most widely used and cited clinical AI datasets. The emergence of vision language models has sparked an increase in demands for sharing reports linked to CheXpert images, along with a growing interest among AI fairness researchers in obtaining demographic data. To address this, CheXpert Plus serves as a new collection of radiology data sources, made publicly available to enhance the scaling, performance, robustness, and fairness of models for all subsequent machine learning tasks in the field of radiology. CheXpert Plus is the largest text dataset publicly released in radiology, with a total of 36 million text tokens, including 13 million impression tokens. To the best of our knowledge, it represents the largest text de-identification effort in radiology, with almost 1 million PHI spans anonymized. It is only the second time that a large-scale English paired dataset has been released in radiology, thereby enabling, for the first time, cross-institution training at scale. All reports are paired with high-quality images in DICOM format, along with numerous image and patient metadata covering various clinical and socio-economic groups, as well as many pathology labels and RadGraph annotations. We hope this dataset will boost research for AI models that can further assist radiologists and help improve medical care. Data is available at the following URL: https://stanfordaimi.azurewebsites.net/datasets/5158c524-d3ab-4e02-96e9-6ee9efc110a1 Models are available at the following URL: https://github.com/Stanford-AIMI/chexpert-plus
URLs: https://stanfordaimi.azurewebsites.net/datasets/5158c524-d3ab-4e02-96e9-6ee9efc110a1, https://github.com/Stanford-AIMI/chexpert-plus
Authors: Yiping Wang, Yifang Chen, Wendan Yan, Alex Fang, Wenjing Zhou, Kevin Jamieson, Simon Shaolei Du
Abstract: Data selection has emerged as a core issue for large-scale visual-language model pretaining (e.g., CLIP), particularly with noisy web-curated datasets. Three main data selection approaches are: (1) leveraging external non-CLIP models to aid data selection, (2) training new CLIP-style embedding models that are more effective at selecting high-quality data than the original OpenAI CLIP model, and (3) designing better metrics or strategies universally applicable to any CLIP embedding without requiring specific model properties (e.g., CLIPScore is one popular metric). While the first two approaches have been extensively studied, the third remains under-explored. In this paper, we advance the third approach by proposing two new methods. Firstly, instead of classical CLIP scores that only consider the alignment between two modalities from a single sample, we introduce negCLIPLoss, a CLIP loss-inspired method that adds the alignment between one sample and its contrastive pairs as an extra normalization term for better quality measurement. Secondly, when downstream tasks are known, we propose a new norm-based metric, NormSim, to measure the similarity between pretraining data and target data. We test our methods on the data selection benchmark, DataComp~\cite{gadre2023datacomp}. Compared to the best baseline using only OpenAI's CLIP-L/14, our methods achieve a 5.3\% improvement on ImageNet-1k and a 2.8\% improvement on 38 downstream evaluation tasks. Moreover, both negCLIPLoss and NormSim are compatible with existing techniques. By combining our methods with the current best methods DFN~\cite{fang2023data} and HYPE~\cite{kim2024hype}, we can boost average performance on downstream tasks by 0.9\%, achieving a new state-of-the-art.
Authors: Shenghuan Sun, Gregory M. Goldgof, Alexander Schubert, Zhiqing Sun, Thomas Hartvigsen, Atul J. Butte, Ahmed Alaa
Abstract: Vision-Language Models (VLM) can support clinicians by analyzing medical images and engaging in natural language interactions to assist in diagnostic and treatment tasks. However, VLMs often exhibit "hallucinogenic" behavior, generating textual outputs not grounded in contextual multimodal information. This challenge is particularly pronounced in the medical domain, where we do not only require VLM outputs to be accurate in single interactions but also to be consistent with clinical reasoning and diagnostic pathways throughout multi-turn conversations. For this purpose, we propose a new alignment algorithm that uses symbolic representations of clinical reasoning to ground VLMs in medical knowledge. These representations are utilized to (i) generate GPT-4-guided visual instruction tuning data at scale, simulating clinician-VLM conversations with demonstrations of clinical reasoning, and (ii) create an automatic reward function that evaluates the clinical validity of VLM generations throughout clinician-VLM interactions. Our algorithm eliminates the need for human involvement in training data generation or reward model construction, reducing costs compared to standard reinforcement learning with human feedback (RLHF). We apply our alignment algorithm to develop Dr-LLaVA, a conversational VLM finetuned for analyzing bone marrow pathology slides, demonstrating strong performance in multi-turn medical conversations.
Authors: Rui-Jie Zhu, Ziqing Wang, Leilani Gilpin, Jason K. Eshraghian
Abstract: Autonomous driving demands an integrated approach that encompasses perception, prediction, and planning, all while operating under strict energy constraints to enhance scalability and environmental sustainability. We present Spiking Autonomous Driving (\name{}), the first unified Spiking Neural Network (SNN) to address the energy challenges faced by autonomous driving systems through its event-driven and energy-efficient nature. SAD is trained end-to-end and consists of three main modules: perception, which processes inputs from multi-view cameras to construct a spatiotemporal bird's eye view; prediction, which utilizes a novel dual-pathway with spiking neurons to forecast future states; and planning, which generates safe trajectories considering predicted occupancy, traffic rules, and ride comfort. Evaluated on the nuScenes dataset, SAD achieves competitive performance in perception, prediction, and planning tasks, while drawing upon the energy efficiency of SNNs. This work highlights the potential of neuromorphic computing to be applied to energy-efficient autonomous driving, a critical step toward sustainable and safety-critical automotive technology. Our code is available at \url{https://github.com/ridgerchu/SAD}.
Authors: Duhun Hwang, Suhyun Kang, Moonjung Eo, Jimyeong Kim, Wonjong Rhee
Abstract: The objective of Domain Generalization (DG) is to devise algorithms and models capable of achieving high performance on previously unseen test distributions. In the pursuit of this objective, average measure has been employed as the prevalent measure for evaluating models and comparing algorithms in the existing DG studies. Despite its significance, a comprehensive exploration of the average measure has been lacking and its suitability in approximating the true domain generalization performance has been questionable. In this study, we carefully investigate the limitations inherent in the average measure and propose worst+gap measure as a robust alternative. We establish theoretical grounds of the proposed measure by deriving two theorems starting from two different assumptions. We conduct extensive experimental investigations to compare the proposed worst+gap measure with the conventional average measure. Given the indispensable need to access the true DG performance for studying measures, we modify five existing datasets to come up with SR-CMNIST, C-Cats&Dogs, L-CIFAR10, PACS-corrupted, and VLCS-corrupted datasets. The experiment results unveil an inferior performance of the average measure in approximating the true DG performance and confirm the robustness of the theoretically supported worst+gap measure.
Authors: Xuan-Bac Nguyen, Hoang-Quan Nguyen, Hugh Churchill, Samee U. Khan, Khoa Luu
Abstract: Although quantum machine learning has been introduced for a while, its applications in computer vision are still limited. This paper, therefore, revisits the quantum visual encoding strategies, the initial step in quantum machine learning. Investigating the root cause, we uncover that the existing quantum encoding design fails to ensure information preservation of the visual features after the encoding process, thus complicating the learning process of the quantum machine learning models. In particular, the problem, termed "Quantum Information Gap" (QIG), leads to a gap of information between classical and corresponding quantum features. We provide theoretical proof and practical demonstrations of that found and underscore the significance of QIG, as it directly impacts the performance of quantum machine learning algorithms. To tackle this challenge, we introduce a simple but efficient new loss function named Quantum Information Preserving (QIP) to minimize this gap, resulting in enhanced performance of quantum machine learning algorithms. Extensive experiments validate the effectiveness of our approach, showcasing superior performance compared to current methodologies and consistently achieving state-of-the-art results in quantum modeling.
Authors: Shuhong Liu, Heng Zhou, Liuzhuozheng Li, Yun Liu, Tianchen Deng, Yiming Zhou, Mingrui Li
Abstract: Gaussian SLAM systems have made significant advancements in improving the efficiency and fidelity of real-time reconstructions. However, these systems often encounter incomplete reconstructions in complex indoor environments, characterized by substantial holes due to unobserved geometry caused by obstacles or limited view angles. To address this challenge, we present Manhattan Gaussian SLAM (MG-SLAM), an RGB-D system that leverages the Manhattan World hypothesis to enhance geometric accuracy and completeness. By seamlessly integrating fused line segments derived from structured scenes, MG-SLAM ensures robust tracking in textureless indoor areas. Moreover, The extracted lines and planar surface assumption allow strategic interpolation of new Gaussians in regions of missing geometry, enabling efficient scene completion. Extensive experiments conducted on both synthetic and real-world scenes demonstrate that these advancements enable our method to achieve state-of-the-art performance, marking a substantial improvement in the capabilities of Gaussian SLAM systems.
Authors: Francesco Petri, Luigi Asprino, Aldo Gangemi
Abstract: World modelling, i.e. building a representation of the rules that govern the world so as to predict its evolution, is an essential ability for any agent interacting with the physical world. Recent applications of the Transformer architecture to the problem of world modelling from video input show notable improvements in sample efficiency. However, existing approaches tend to work only at the image level thus disregarding that the environment is composed of objects interacting with each other. In this paper, we propose an architecture combining Transformers for world modelling with the slot-attention paradigm, an approach for learning representations of objects appearing in a scene. We describe the resulting neural architecture and report experimental results showing an improvement over the existing solutions in terms of sample efficiency and a reduction of the variation of the performance over the training examples. The code for our architecture and experiments is available at https://github.com/torchipeppo/transformers-and-slot-encoding-for-wm
URLs: https://github.com/torchipeppo/transformers-and-slot-encoding-for-wm
Authors: Andreas Koukounas, Georgios Mastrapas, Michael G\"unther, Bo Wang, Scott Martens, Isabelle Mohr, Saba Sturua, Mohammad Kalim Akram, Joan Fontanals Mart\'inez, Saahil Ognawala, Susana Guzman, Maximilian Werk, Nan Wang, Han Xiao
Abstract: Contrastive Language-Image Pretraining (CLIP) is widely used to train models to align images and texts in a common embedding space by mapping them to fixed-sized vectors. These models are key to multimodal information retrieval and related tasks. However, CLIP models generally underperform in text-only tasks compared to specialized text models. This creates inefficiencies for information retrieval systems that keep separate embeddings and models for text-only and multimodal tasks. We propose a novel, multi-task contrastive training method to address this issue, which we use to train the jina-clip-v1 model to achieve the state-of-the-art performance on both text-image and text-text retrieval tasks.
Authors: Matthew Watson, Divyashree Shivakumar Sreepathihalli, Francois Chollet, Martin Gorner, Kiranbir Sodhia, Ramesh Sampath, Tirth Patel, Haifeng Jin, Neel Kovelamudi, Gabriel Rasskin, Samaneh Saadat, Luke Wood, Chen Qian, Jonathan Bischof, Ian Stenbit
Abstract: We present the Keras domain packages KerasCV and KerasNLP, extensions of the Keras API for Computer Vision and Natural Language Processing workflows, capable of running on either JAX, TensorFlow, or PyTorch. These domain packages are designed to enable fast experimentation, with a focus on ease-of-use and performance. We adopt a modular, layered design: at the library's lowest level of abstraction, we provide building blocks for creating models and data preprocessing pipelines, and at the library's highest level of abstraction, we provide pretrained ``task" models for popular architectures such as Stable Diffusion, YOLOv8, GPT2, BERT, Mistral, CLIP, Gemma, T5, etc. Task models have built-in preprocessing, pretrained weights, and can be fine-tuned on raw inputs. To enable efficient training, we support XLA compilation for all models, and run all preprocessing via a compiled graph of TensorFlow operations using the tf.data API. The libraries are fully open-source (Apache 2.0 license) and available on GitHub.
Authors: Massimo Bini, Karsten Roth, Zeynep Akata, Anna Khoreva
Abstract: Parameter-efficient finetuning (PEFT) has become ubiquitous to adapt foundation models to downstream task requirements while retaining their generalization ability. However, the amount of additionally introduced parameters and compute for successful adaptation and hyperparameter searches can explode quickly, especially when deployed at scale to serve numerous individual requests. To ensure effective, parameter-efficient, and hyperparameter-robust adaptation, we propose the ETHER transformation family, which performs Efficient fineTuning via HypErplane Reflections. By design, ETHER transformations require a minimal number of parameters, are less likely to deteriorate model performance, and exhibit robustness to hyperparameter and learning rate choices. In particular, we introduce ETHER and its relaxation ETHER+, which match or outperform existing PEFT methods with significantly fewer parameters ($\sim$$10$-$100$ times lower than LoRA or OFT) across multiple image synthesis and natural language tasks without exhaustive hyperparameter tuning. Finally, we investigate the recent emphasis on Hyperspherical Energy retention for adaptation and raise questions on its practical utility. The code is available at https://github.com/mwbini/ether.
Authors: Weilin Lin, Li Liu, Shaokui Wei, Jianze Li, Hui Xiong
Abstract: The security threat of backdoor attacks is a central concern for deep neural networks (DNNs). Recently, without poisoned data, unlearning models with clean data and then learning a pruning mask have contributed to backdoor defense. Additionally, vanilla fine-tuning with those clean data can help recover the lost clean accuracy. However, the behavior of clean unlearning is still under-explored, and vanilla fine-tuning unintentionally induces back the backdoor effect. In this work, we first investigate model unlearning from the perspective of weight changes and gradient norms, and find two interesting observations in the backdoored model: 1) the weight changes between poison and clean unlearning are positively correlated, making it possible for us to identify the backdoored-related neurons without using poisoned data; 2) the neurons of the backdoored model are more active (i.e., larger changes in gradient norm) than those in the clean model, suggesting the need to suppress the gradient norm during fine-tuning. Then, we propose an effective two-stage defense method. In the first stage, an efficient Neuron Weight Change (NWC)-based Backdoor Reinitialization is proposed based on observation 1). In the second stage, based on observation 2), we design an Activeness-Aware Fine-Tuning to replace the vanilla fine-tuning. Extensive experiments, involving eight backdoor attacks on three benchmark datasets, demonstrate the superior performance of our proposed method compared to recent state-of-the-art backdoor defense approaches.
Authors: Yifeng Zhu, Arisrei Lim, Peter Stone, Yuke Zhu
Abstract: We present an object-centric approach to empower robots to learn vision-based manipulation skills from human videos. We investigate the problem of imitating robot manipulation from a single human video in the open-world setting, where a robot must learn to manipulate novel objects from one video demonstration. We introduce ORION, an algorithm that tackles the problem by extracting an object-centric manipulation plan from a single RGB-D video and deriving a policy that conditions on the extracted plan. Our method enables the robot to learn from videos captured by daily mobile devices such as an iPad and generalize the policies to deployment environments with varying visual backgrounds, camera angles, spatial layouts, and novel object instances. We systematically evaluate our method on both short-horizon and long-horizon tasks, demonstrating the efficacy of ORION in learning from a single human video in the open world. Videos can be found in the project website https://ut-austin-rpl.github.io/ORION-release.
Authors: David Anghelone, Cunjian Chen, Arun Ross, Antitza Dantcheva
Abstract: Cross-spectral face recognition (CFR) refers to recognizing individuals using face images stemming from different spectral bands, such as infrared vs. visible. While CFR is inherently more challenging than classical face recognition due to significant variation in facial appearance caused by the modality gap, it is useful in many scenarios including night-vision biometrics and detecting presentation attacks. Recent advances in convolutional neural networks (CNNs) have resulted in significant improvement in the performance of CFR systems. Given these developments, the contributions of this survey are three-fold. First, we provide an overview of CFR, by formalizing the CFR problem and presenting related applications. Secondly, we discuss the appropriate spectral bands for face recognition and discuss recent CFR methods, placing emphasis on deep neural networks. In particular we describe techniques that have been proposed to extract and compare heterogeneous features emerging from different spectral bands. We also discuss the datasets that have been used for evaluating CFR methods. Finally, we discuss the challenges and future lines of research on this topic.
Authors: Haoran Chen, Xintong Han, Zuxuan Wu, Yu-Gang Jiang
Abstract: Most existing methods for unsupervised domain adaptation (UDA) rely on a shared network to extract domain-invariant features. However, when facing multiple source domains, optimizing such a network involves updating the parameters of the entire network, making it both computationally expensive and challenging, particularly when coupled with min-max objectives. Inspired by recent advances in prompt learning that adapts high-capacity models for downstream tasks in a computationally economic way, we introduce Multi-Prompt Alignment (MPA), a simple yet efficient framework for multi-source UDA. Given a source and target domain pair, MPA first trains an individual prompt to minimize the domain gap through a contrastive loss. Then, MPA denoises the learned prompts through an auto-encoding process and aligns them by maximizing the agreement of all the reconstructed prompts. Moreover, we show that the resulting subspace acquired from the auto-encoding process can easily generalize to a streamlined set of target domains, making our method more efficient for practical usage. Extensive experiments show that MPA achieves state-of-the-art results on three popular datasets with an impressive average accuracy of 54.1% on DomainNet.
Authors: Junyang Wang, Yi Zhang, Jitao Sang
Abstract: The Vision-Language Pre-training (VLP) models like CLIP have gained popularity in recent years. However, many works found that the social biases hidden in CLIP easily manifest in downstream tasks, especially in image retrieval, which can have harmful effects on human society. In this work, we propose FairCLIP to eliminate the social bias in CLIP-based image retrieval without damaging the retrieval performance achieving the compatibility between the debiasing effect and the retrieval performance. FairCLIP is divided into two steps: Attribute Prototype Learning (APL) and Representation Neutralization (RN). In the first step, we extract the concepts needed for debiasing in CLIP. We use the query with learnable word vector prefixes as the extraction structure. In the second step, we first divide the attributes into target and bias attributes. By analysis, we find that both attributes have an impact on the bias. Therefore, we try to eliminate the bias by using Re-Representation Matrix (RRM) to achieve the neutralization of the representation. We compare the debiasing effect and retrieval performance with other methods, and experiments demonstrate that FairCLIP can achieve the best compatibility. Although FairCLIP is used to eliminate bias in image retrieval, it achieves the neutralization of the representation which is common to all CLIP downstream tasks. This means that FairCLIP can be applied as a general debiasing method for other fairness issues related to CLIP.
Authors: Siyu Xing, Chen Gong, Hewei Guo, Xiao-Yu Zhang, Xinwen Hou, Yu Liu
Abstract: Existing GAN inversion methods work brilliantly in reconstructing high-quality (HQ) images while struggling with more common low-quality (LQ) inputs in practical application. To address this issue, we propose Unsupervised Domain Adaptation (UDA) in the inversion process, namely UDA-inversion, for effective inversion and editing of both HQ and LQ images. Regarding unpaired HQ images as the source domain and LQ images as the unlabeled target domain, we introduce a theoretical guarantee: loss value in the target domain is upper-bounded by loss in the source domain and a novel discrepancy function measuring the difference between two domains. Following that, we can only minimize this upper bound to obtain accurate latent codes for HQ and LQ images. Thus, constructive representations of HQ images can be spontaneously learned and transformed into LQ images without supervision. UDA-Inversion achieves a better PSNR of 22.14 on FFHQ dataset and performs comparably to supervised methods.
Authors: Hamza Bouzid, Lahoucine Ballihi
Abstract: Recent technological advancements have significantly expanded the potential of human action recognition through harnessing the power of 3D data. This data provides a richer understanding of actions, including depth information that enables more accurate analysis of spatial and temporal characteristics. In this context, We study the challenge of 3D human action recognition.Unlike prior methods, that rely on sampling 2D depth images, skeleton points, or point clouds, often leading to substantial memory requirements and the ability to handle only short sequences, we introduce a novel approach for 3D human action recognition, denoted as SpATr (Spiral Auto-encoder and Transformer Network), specifically designed for fixed-topology mesh sequences. The SpATr model disentangles space and time in the mesh sequences. A lightweight auto-encoder, based on spiral convolutions, is employed to extract spatial geometrical features from each 3D mesh. These convolutions are lightweight and specifically designed for fix-topology mesh data. Subsequently, a temporal transformer, based on self-attention, captures the temporal context within the feature sequence. The self-attention mechanism enables long-range dependencies capturing and parallel processing, ensuring scalability for long sequences. The proposed method is evaluated on three prominent 3D human action datasets: Babel, MoVi, and BMLrub, from the Archive of Motion Capture As Surface Shapes (AMASS). Our results analysis demonstrates the competitive performance of our SpATr model in 3D human action recognition while maintaining efficient memory usage. The code and the training results will soon be made publicly available at https://github.com/h-bouzid/spatr.
Authors: Lucas Ventura, Antoine Yang, Cordelia Schmid, G\"ul Varol
Abstract: Composed Image Retrieval (CoIR) has recently gained popularity as a task that considers both text and image queries together, to search for relevant images in a database. Most CoIR approaches require manually annotated datasets, comprising image-text-image triplets, where the text describes a modification from the query image to the target image. However, manual curation of CoIR triplets is expensive and prevents scalability. In this work, we instead propose a scalable automatic dataset creation methodology that generates triplets given video-caption pairs, while also expanding the scope of the task to include composed video retrieval (CoVR). To this end, we mine paired videos with a similar caption from a large database, and leverage a large language model to generate the corresponding modification text. Applying this methodology to the extensive WebVid2M collection, we automatically construct our WebVid-CoVR dataset, resulting in 1.6 million triplets. Moreover, we introduce a new benchmark for CoVR with a manually annotated evaluation set, along with baseline results. Our experiments further demonstrate that training a CoVR model on our dataset effectively transfers to CoIR, leading to improved state-of-the-art performance in the zero-shot setup on both the CIRR and FashionIQ benchmarks. Our code, datasets, and models are publicly available at https://imagine.enpc.fr/~ventural/covr.
URLs: https://imagine.enpc.fr/
Authors: Rama Krishna Kandukuri, Michael Strecke, Joerg Stueckler
Abstract: Physics-based understanding of object interactions from sensory observations is an essential capability in augmented reality and robotics. It enables to capture the properties of a scene for simulation and control. In this paper, we propose a novel approach for real-to-sim which tracks rigid objects in 3D from RGB-D images and infers physical properties of the objects. We use a differentiable physics simulation as state-transition model in an Extended Kalman Filter which can model contact and friction for arbitrary mesh-based shapes and in this way estimate physically plausible trajectories. We demonstrate that our approach can filter position, orientation, velocities, and concurrently can estimate the coefficient of friction of the objects. We analyze our approach on various sliding scenarios in synthetic image sequences of single objects and colliding objects. We also demonstrate and evaluate our approach on a real-world dataset. We make our novel benchmark datasets publicly available to foster future research in this novel problem setting and comparison with our method.
Authors: Zhiquan Tan, Kaipeng Zheng, Weiran Huang
Abstract: Semi-supervised learning has made remarkable strides by effectively utilizing a limited amount of labeled data while capitalizing on the abundant information present in unlabeled data. However, current algorithms often prioritize aligning image predictions with specific classes generated through self-training techniques, thereby neglecting the inherent relationships that exist within these classes. In this paper, we present a new approach called OTMatch, which leverages semantic relationships among classes by employing an optimal transport loss function to match distributions. We conduct experiments on many standard vision and language datasets. The empirical results show improvements in our method above baseline, this demonstrates the effectiveness and superiority of our approach in harnessing semantic relationships to enhance learning performance in a semi-supervised setting.
Authors: Jie Wang, Jiu-Cheng Xie, Xianyan Li, Feng Xu, Chi-Man Pun, Hao Gao
Abstract: Constructing vivid 3D head avatars for given subjects and realizing a series of animations on them is valuable yet challenging. This paper presents GaussianHead, which models the actional human head with anisotropic 3D Gaussians. In our framework, a motion deformation field and multi-resolution tri-plane are constructed respectively to deal with the head's dynamic geometry and complex texture. Notably, we impose an exclusive derivation scheme on each Gaussian, which generates its multiple doppelgangers through a set of learnable parameters for position transformation. With this design, we can compactly and accurately encode the appearance information of Gaussians, even those fitting the head's particular components with sophisticated structures. In addition, an inherited derivation strategy for newly added Gaussians is adopted to facilitate training acceleration. Extensive experiments show that our method can produce high-fidelity renderings, outperforming state-of-the-art approaches in reconstruction, cross-identity reenactment, and novel view synthesis tasks. Our code is available at: https://github.com/chiehwangs/gaussian-head.
Authors: Simon Giebenhain, Tobias Kirschstein, Markos Georgopoulos, Martin R\"unz, Lourdes Agapito, Matthias Nie{\ss}ner
Abstract: We present Monocular Neural Parametric Head Models (MonoNPHM) for dynamic 3D head reconstructions from monocular RGB videos. To this end, we propose a latent appearance space that parameterizes a texture field on top of a neural parametric model. We constrain predicted color values to be correlated with the underlying geometry such that gradients from RGB effectively influence latent geometry codes during inverse rendering. To increase the representational capacity of our expression space, we augment our backward deformation field with hyper-dimensions, thus improving color and geometry representation in topologically challenging expressions. Using MonoNPHM as a learned prior, we approach the task of 3D head reconstruction using signed distance field based volumetric rendering. By numerically inverting our backward deformation field, we incorporated a landmark loss using facial anchor points that are closely tied to our canonical geometry representation. To evaluate the task of dynamic face reconstruction from monocular RGB videos we record 20 challenging Kinect sequences under casual conditions. MonoNPHM outperforms all baselines with a significant margin, and makes an important step towards easily accessible neural parametric face models through RGB tracking.
Authors: Feifei Wang, Zhentao Tan, Tianyi Wei, Yue Wu, Qidong Huang
Abstract: Despite the success of diffusion-based customization methods on visual content creation, increasing concerns have been raised about such techniques from both privacy and political perspectives. To tackle this issue, several anti-customization methods have been proposed in very recent months, predominantly grounded in adversarial attacks. Unfortunately, most of these methods adopt straightforward designs, such as end-to-end optimization with a focus on adversarially maximizing the original training loss, thereby neglecting nuanced internal properties intrinsic to the diffusion model, and even leading to ineffective optimization in some diffusion time steps.In this paper, we strive to bridge this gap by undertaking a comprehensive exploration of these inherent properties, to boost the performance of current anti-customization approaches. Two aspects of properties are investigated: 1) We examine the relationship between time step selection and the model's perception in the frequency domain of images and find that lower time steps can give much more contributions to adversarial noises. This inspires us to propose an adaptive greedy search for optimal time steps that seamlessly integrates with existing anti-customization methods. 2) We scrutinize the roles of features at different layers during denoising and devise a sophisticated feature-based optimization framework for anti-customization.Experiments on facial benchmarks demonstrate that our approach significantly increases identity disruption, thereby protecting user privacy and copyright. Our code is available at: https://github.com/somuchtome/SimAC.
Authors: Nanqing Liu, Xun Xu, Yongyi Su, Chengxin Liu, Peiliang Gong, Heng-Chao Li
Abstract: Domain adaptation is crucial in aerial imagery, as the visual representation of these images can significantly vary based on factors such as geographic location, time, and weather conditions. Additionally, high-resolution aerial images often require substantial storage space and may not be readily accessible to the public. To address these challenges, we propose a novel Source-Free Object Detection (SFOD) method. Specifically, our approach begins with a self-training framework, which significantly enhances the performance of baseline methods. To alleviate the noisy labels in self-training, we utilize Contrastive Language-Image Pre-training (CLIP) to guide the generation of pseudo-labels, termed CLIP-guided Aggregation (CGA). By leveraging CLIP's zero-shot classification capability, we aggregate its scores with the original predicted bounding boxes, enabling us to obtain refined scores for the pseudo-labels. To validate the effectiveness of our method, we constructed two new datasets from different domains based on the DIOR dataset, named DIOR-C and DIOR-Cloudy. Experimental results demonstrate that our method outperforms other comparative algorithms. The code is available at https://github.com/Lans1ng/SFOD-RS.
Authors: Qirui Jiao, Daoyuan Chen, Yilun Huang, Yaliang Li, Ying Shen
Abstract: Despite the impressive capabilities of Multimodal Large Language Models (MLLMs) in integrating text and image modalities, challenges remain in accurately interpreting detailed visual elements. This paper presents an empirical study on enhancing MLLMs with state-of-the-art (SOTA) object detection and Optical Character Recognition (OCR) models to improve fine-grained understanding and reduce hallucination in responses. We investigate the embedding-based infusion of textual detection information, the impact of such infusion on MLLMs' original abilities, and the interchangeability of detection models. We conduct systematic and extensive experiments with representative models such as LLaVA-1.5, DINO, PaddleOCRv2, and Grounding DINO, revealing that our simple yet general approach not only refines MLLMs' performance in fine-grained visual tasks but also maintains their original strengths. Notably, the enhanced LLaVA-1.5 outperforms its original 7B/13B models on all 10 benchmarks, achieving an improvement of up to 12.5% on the normalized average score. We release our codes to facilitate further exploration into the fine-grained multimodal capabilities of MLLMs.
Authors: Yoad Tewel, Omri Kaduri, Rinon Gal, Yoni Kasten, Lior Wolf, Gal Chechik, Yuval Atzmon
Abstract: Text-to-image models offer a new level of creative flexibility by allowing users to guide the image generation process through natural language. However, using these models to consistently portray the same subject across diverse prompts remains challenging. Existing approaches fine-tune the model to teach it new words that describe specific user-provided subjects or add image conditioning to the model. These methods require lengthy per-subject optimization or large-scale pre-training. Moreover, they struggle to align generated images with text prompts and face difficulties in portraying multiple subjects. Here, we present ConsiStory, a training-free approach that enables consistent subject generation by sharing the internal activations of the pretrained model. We introduce a subject-driven shared attention block and correspondence-based feature injection to promote subject consistency between images. Additionally, we develop strategies to encourage layout diversity while maintaining subject consistency. We compare ConsiStory to a range of baselines, and demonstrate state-of-the-art performance on subject consistency and text alignment, without requiring a single optimization step. Finally, ConsiStory can naturally extend to multi-subject scenarios, and even enable training-free personalization for common objects.
Authors: Simone Magistri, Tomaso Trinci, Albin Soutif-Cormerais, Joost van de Weijer, Andrew D. Bagdanov
Abstract: Exemplar-Free Class Incremental Learning (EFCIL) aims to learn from a sequence of tasks without having access to previous task data. In this paper, we consider the challenging Cold Start scenario in which insufficient data is available in the first task to learn a high-quality backbone. This is especially challenging for EFCIL since it requires high plasticity, which results in feature drift which is difficult to compensate for in the exemplar-free setting. To address this problem, we propose a simple and effective approach that consolidates feature representations by regularizing drift in directions highly relevant to previous tasks and employs prototypes to reduce task-recency bias. Our method, called Elastic Feature Consolidation (EFC), exploits a tractable second-order approximation of feature drift based on an Empirical Feature Matrix (EFM). The EFM induces a pseudo-metric in feature space which we use to regularize feature drift in important directions and to update Gaussian prototypes used in a novel asymmetric cross entropy loss which effectively balances prototype rehearsal with data from new tasks. Experimental results on CIFAR-100, Tiny-ImageNet, ImageNet-Subset and ImageNet-1K demonstrate that Elastic Feature Consolidation is better able to learn new tasks by maintaining model plasticity and significantly outperform the state-of-the-art.
Authors: Siddharth Karamcheti, Suraj Nair, Ashwin Balakrishna, Percy Liang, Thomas Kollar, Dorsa Sadigh
Abstract: Visually-conditioned language models (VLMs) have seen growing adoption in applications such as visual dialogue, scene understanding, and robotic task planning; adoption that has fueled a wealth of new models such as LLaVa, InstructBLIP, and PaLI-3. Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance $-$ a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps, we first compile a suite of standardized evaluations spanning visual question answering, object localization, and challenge sets that probe properties such as hallucination; evaluations that provide fine-grained insight VLM capabilities. Second, we rigorously investigate VLMs along key design axes, including pretrained visual representations and training from base vs. instruct-tuned language models, amongst others. We couple our analysis with three resource contributions: (1) a unified framework for evaluating VLMs, (2) optimized, flexible training code, and (3) checkpoints for all models, including a family of VLMs at the 7-13B scale that strictly outperform InstructBLIP and LLaVa v1.5, the state-of-the-art in open VLMs.
Authors: Tao Zhang, Xiangtai Li, Haobo Yuan, Shunping Ji, Shuicheng Yan
Abstract: Recently, state space models have exhibited strong global modeling capabilities and linear computational complexity in contrast to transformers. This research focuses on applying such architecture in point cloud analysis. In particular, for the first time, we demonstrate that Mamba-based point cloud methods can outperform previous methods based on transformer or multi-layer perceptrons (MLPs). To enable Mamba to process 3-D point cloud data more effectively, we propose a novel Consistent Traverse Serialization method to convert point clouds into 1-D point sequences while ensuring that neighboring points in the sequence are also spatially adjacent. Consistent Traverse Serialization yields six variants by permuting the order of x, y, and z coordinates, and the synergistic use of these variants aids Mamba in comprehensively observing point cloud data. Furthermore, to assist Mamba in handling point sequences with different orders more effectively, we introduce point prompts to inform Mamba of the sequence's arrangement rules. Finally, we propose positional encoding based on spatial coordinate mapping to inject positional information into point cloud sequences better. Point Cloud Mamba surpasses the state-of-the-art (SOTA) point-based method PointNeXt and achieves new SOTA performance on the ScanObjectNN, ModelNet40, ShapeNetPart, and S3DIS datasets. It is worth mentioning that when using a more powerful local feature extraction module, our PCM achieves 82.6 mIoU on S3DIS, significantly surpassing the previous SOTA models, DeLA and PTv3, by 8.5 mIoU and 7.9 mIoU, respectively. Code and model are available at https://github.com/SkyworkAI/PointCloudMamba.
Authors: Jieren Deng, Haojian Zhang, Kun Ding, Jianhua Hu, Xingxuan Zhang, Yunkuan Wang
Abstract: This paper presents Incremental Vision-Language Object Detection (IVLOD), a novel learning task designed to incrementally adapt pre-trained Vision-Language Object Detection Models (VLODMs) to various specialized domains, while simultaneously preserving their zero-shot generalization capabilities for the generalized domain. To address this new challenge, we present the Zero-interference Reparameterizable Adaptation (ZiRa), a novel method that introduces Zero-interference Loss and reparameterization techniques to tackle IVLOD without incurring additional inference costs or a significant increase in memory usage. Comprehensive experiments on COCO and ODinW-13 datasets demonstrate that ZiRa effectively safeguards the zero-shot generalization ability of VLODMs while continuously adapting to new tasks. Specifically, after training on ODinW-13 datasets, ZiRa exhibits superior performance compared to CL-DETR and iDETR, boosting zero-shot generalizability by substantial 13.91 and 8.74 AP, respectively.
Authors: Takahiro Shindo, Kein Yamada, Taiju Watanabe, Hiroshi Watanabe
Abstract: Image Coding for Machines (ICM) is an image compression technique for image recognition. This technique is essential due to the growing demand for image recognition AI. In this paper, we propose a method for ICM that focuses on encoding and decoding only the edge information of object parts in an image, which we call SA-ICM. This is an Learned Image Compression (LIC) model trained using edge information created by Segment Anything. Our method can be used for image recognition models with various tasks. SA-ICM is also robust to changes in input data, making it effective for a variety of use cases. Additionally, our method provides benefits from a privacy point of view, as it removes human facial information on the encoder's side, thus protecting one's privacy. Furthermore, this LIC model training method can be used to train Neural Representations for Videos (NeRV), which is a video compression model. By training NeRV using edge information created by Segment Anything, it is possible to create a NeRV that is effective for image recognition (SA-NeRV). Experimental results confirm the advantages of SA-ICM, presenting the best performance in image compression for image recognition. We also show that SA-NeRV is superior to ordinary NeRV in video compression for machines.
Authors: Jintao Huang, Yiu-Ming Cheung
Abstract: Partial Label Learning (PLL) grapples with learning from ambiguously labelled data, and it has been successfully applied in fields such as image recognition. Nevertheless, traditional PLL methods rely on the closed-world assumption, which can be limiting in open-world scenarios and negatively impact model performance and generalization. To tackle these challenges, our study introduces a novel method called PLL-OOD, which is the first to incorporate Out-of-Distribution (OOD) detection into the PLL framework. PLL-OOD significantly enhances model adaptability and accuracy by merging self-supervised learning with partial label loss and pioneering the Partial-Energy (PE) score for OOD detection. This approach improves data feature representation and effectively disambiguates candidate labels, using a dynamic label confidence matrix to refine predictions. The PE score, adjusted by label confidence, precisely identifies OOD instances, optimizing model training towards in-distribution data. This innovative method markedly boosts PLL model robustness and performance in open-world settings. To validate our approach, we conducted a comprehensive comparative experiment combining the existing state-of-the-art PLL model with multiple OOD scores on the CIFAR-10 and CIFAR-100 datasets with various OOD datasets. The results demonstrate that the proposed PLL-OOD framework is highly effective and effectiveness outperforms existing models, showcasing its superiority and effectiveness.
Authors: Di Wang, Jing Zhang, Minqiang Xu, Lin Liu, Dongsheng Wang, Erzhong Gao, Chengxi Han, Haonan Guo, Bo Du, Dacheng Tao, Liangpei Zhang
Abstract: Foundation models have reshaped the landscape of Remote Sensing (RS) by enhancing various image interpretation tasks. Pretraining is an active research topic, encompassing supervised and self-supervised learning methods to initialize model weights effectively. However, transferring the pretrained models to downstream tasks may encounter task discrepancy due to their formulation of pretraining as image classification or object discrimination tasks. In this study, we explore the Multi-Task Pretraining (MTP) paradigm for RS foundation models to address this issue. Using a shared encoder and task-specific decoder architecture, we conduct multi-task supervised pretraining on the SAMRS dataset, encompassing semantic segmentation, instance segmentation, and rotated object detection. MTP supports both convolutional neural networks and vision transformer foundation models with over 300 million parameters. The pretrained models are finetuned on various RS downstream tasks, such as scene classification, horizontal and rotated object detection, semantic segmentation, and change detection. Extensive experiments across 14 datasets demonstrate the superiority of our models over existing ones of similar size and their competitive performance compared to larger state-of-the-art models, thus validating the effectiveness of MTP.
Authors: Tung-Yu Wu, Sheng-Yu Huang, Yu-Chiang Frank Wang
Abstract: 3D visual grounding aims to identify the target object within a 3D point cloud scene referred to by a natural language description. Previous works usually require significant data relating to point color and their descriptions to exploit the corresponding complicated verbo-visual relations. In our work, we introduce Vigor, a novel Data-Efficient 3D Visual Grounding framework via Order-aware Referring. Vigor leverages LLM to produce a desirable referential order from the input description for 3D visual grounding. With the proposed stacked object-referring blocks, the predicted anchor objects in the above order allow one to locate the target object progressively without supervision on the identities of anchor objects or exact relations between anchor/target objects. In addition, we present an order-aware warm-up training strategy, which augments referential orders for pre-training the visual grounding framework. This allows us to better capture the complex verbo-visual relations and benefit the desirable data-efficient learning scheme. Experimental results on the NR3D and ScanRefer datasets demonstrate our superiority in low-resource scenarios. In particular, Vigor surpasses current state-of-the-art frameworks by 9.3% and 7.6% grounding accuracy under 1% data and 10% data settings on the NR3D dataset, respectively.
Authors: Lingdong Shen, Fangxin Shang, Xiaoshuang Huang, Yehui Yang, Haifeng Huang, Shiming Xiang
Abstract: In the field of medical image segmentation, tackling Out-of-Distribution (OOD) segmentation tasks in a cost-effective manner remains a significant challenge. Universal segmentation models is a solution, which aim to generalize across the diverse modality of medical images, yet their effectiveness often diminishes when applied to OOD data modalities and tasks, requiring intricate fine-tuning of model for optimal performance. Few-shot learning segmentation methods are typically designed for specific modalities of data and cannot be directly transferred for use with another modality. Therefore, we introduce SegICL, a novel approach leveraging In-Context Learning (ICL) for image segmentation. Unlike existing methods, SegICL has the capability to employ text-guided segmentation and conduct in-context learning with a small set of image-mask pairs, eliminating the need for training the model from scratch or fine-tuning for OOD tasks (including OOD modality and dataset). Extensive experimental demonstrates a positive correlation between the number of shots and segmentation performance on OOD tasks. The performance of segmentation when provided thre-shots is approximately 1.5 times better than the performance in a zero-shot setting. This indicates that SegICL effectively address new segmentation tasks based on contextual information. Additionally, SegICL also exhibits comparable performance to mainstream models on OOD and in-distribution tasks. Our code will be released after paper review.
Authors: Guoyang Zhao, Fulong Ma, Weiqing Qi, Yuxuan Liu, Ming Liu
Abstract: Curb detection is a crucial function in intelligent driving, essential for determining drivable areas on the road. However, the complexity of road environments makes curb detection challenging. This paper introduces CurbNet, a novel framework for curb detection utilizing point cloud segmentation. To address the lack of comprehensive curb datasets with 3D annotations, we have developed the 3D-Curb dataset based on SemanticKITTI, currently the largest and most diverse collection of curb point clouds. Recognizing that the primary characteristic of curbs is height variation, our approach leverages spatially rich 3D point clouds for training. To tackle the challenges posed by the uneven distribution of curb features on the xy-plane and their dependence on high-frequency features along the z-axis, we introduce the Multi-Scale and Channel Attention (MSCA) module, a customized solution designed to optimize detection performance. Additionally, we propose an adaptive weighted loss function group specifically formulated to counteract the imbalance in the distribution of curb point clouds relative to other categories. Extensive experiments conducted on 2 major datasets demonstrate that our method surpasses existing benchmarks set by leading curb detection and point cloud segmentation models. Through the post-processing refinement of the detection results, we have significantly reduced noise in curb detection, thereby improving precision by 4.5 points. Similarly, our tolerance experiments also achieved state-of-the-art results. Furthermore, real-world experiments and dataset analyses mutually validate each other, reinforcing CurbNet's superior detection capability and robust generalizability. The project website is available at: https://github.com/guoyangzhao/CurbNet/.
Authors: Yikang Zhou, Tao Zhang, Shunping Ji, Shuicheng Yan, Xiangtai Li
Abstract: Modern video segmentation methods adopt object queries to perform inter-frame association and demonstrate satisfactory performance in tracking continuously appearing objects despite large-scale motion and transient occlusion. However, they all underperform on newly emerging and disappearing objects that are common in the real world because they attempt to model object emergence and disappearance through feature transitions between background and foreground queries that have significant feature gaps. We introduce Dynamic Anchor Queries (DAQ) to shorten the transition gap between the anchor and target queries by dynamically generating anchor queries based on the features of potential candidates. Furthermore, we introduce a query-level object Emergence and Disappearance Simulation (EDS) strategy, which unleashes DAQ's potential without any additional cost. Finally, we combine our proposed DAQ and EDS with DVIS to obtain DVIS-DAQ. Extensive experiments demonstrate that DVIS-DAQ achieves a new state-of-the-art (SOTA) performance on five mainstream video segmentation benchmarks. Code and models are available at \url{https://github.com/SkyworkAI/DAQ-VS}.
Authors: Lingdong Kong, Shaoyuan Xie, Hanjiang Hu, Yaru Niu, Wei Tsang Ooi, Benoit R. Cottereau, Lai Xing Ng, Yuexin Ma, Wenwei Zhang, Liang Pan, Kai Chen, Ziwei Liu, Weichao Qiu, Wei Zhang, Xu Cao, Hao Lu, Ying-Cong Chen, Caixin Kang, Xinning Zhou, Chengyang Ying, Wentao Shang, Xingxing Wei, Yinpeng Dong, Bo Yang, Shengyin Jiang, Zeliang Ma, Dengyi Ji, Haiwen Li, Xingliang Huang, Yu Tian, Genghua Kou, Fan Jia, Yingfei Liu, Tiancai Wang, Ying Li, Xiaoshuai Hao, Yifan Yang, Hui Zhang, Mengchuan Wei, Yi Zhou, Haimei Zhao, Jing Zhang, Jinke Li, Xiao He, Xiaoqiang Cheng, Bingyang Zhang, Lirong Zhao, Dianlei Ding, Fangsheng Liu, Yixiang Yan, Hongming Wang, Nanfei Ye, Lun Luo, Yubo Tian, Yiwei Zuo, Zhe Cao, Yi Ren, Yunfan Li, Wenjie Liu, Xun Wu, Yifan Mao, Ming Li, Jian Liu, Jiayang Liu, Zihan Qin, Cunxi Chu, Jialei Xu, Wenbo Zhao, Junjun Jiang, Xianming Liu, Ziyan Wang, Chiwei Li, Shilong Li, Chendong Yuan, Songyue Yang, Wentao Liu, Peng Chen, Bin Zhou, Yubo Wang, Chi Zhang, Jianhang Sun, Hai Chen, Xiao Yang, Lizhong Wang, Dongyi Fu, Yongchun Lin, Huitong Yang, Haoang Li, Yadan Luo, Xianjing Cheng, Yong Xu
Abstract: In the realm of autonomous driving, robust perception under out-of-distribution conditions is paramount for the safe deployment of vehicles. Challenges such as adverse weather, sensor malfunctions, and environmental unpredictability can severely impact the performance of autonomous systems. The 2024 RoboDrive Challenge was crafted to propel the development of driving perception technologies that can withstand and adapt to these real-world variabilities. Focusing on four pivotal tasks -- BEV detection, map segmentation, semantic occupancy prediction, and multi-view depth estimation -- the competition laid down a gauntlet to innovate and enhance system resilience against typical and atypical disturbances. This year's challenge consisted of five distinct tracks and attracted 140 registered teams from 93 institutes across 11 countries, resulting in nearly one thousand submissions evaluated through our servers. The competition culminated in 15 top-performing solutions, which introduced a range of innovative approaches including advanced data augmentation, multi-sensor fusion, self-supervised learning for error correction, and new algorithmic strategies to enhance sensor robustness. These contributions significantly advanced the state of the art, particularly in handling sensor inconsistencies and environmental variability. Participants, through collaborative efforts, pushed the boundaries of current technologies, showcasing their potential in real-world scenarios. Extensive evaluations and analyses provided insights into the effectiveness of these solutions, highlighting key trends and successful strategies for improving the resilience of driving perception systems. This challenge has set a new benchmark in the field, providing a rich repository of techniques expected to guide future research in this field.
Authors: Wanting Xu, Yang Liu, Langping He, Xucheng Huang, Ling Jiang
Abstract: We introduce Xmodel-VLM, a cutting-edge multimodal vision language model. It is designed for efficient deployment on consumer GPU servers. Our work directly confronts a pivotal industry issue by grappling with the prohibitive service costs that hinder the broad adoption of large-scale multimodal systems. Through rigorous training, we have developed a 1B-scale language model from the ground up, employing the LLaVA paradigm for modal alignment. The result, which we call Xmodel-VLM, is a lightweight yet powerful multimodal vision language model. Extensive testing across numerous classic multimodal benchmarks has revealed that despite its smaller size and faster execution, Xmodel-VLM delivers performance comparable to that of larger models. Our model checkpoints and code are publicly available on GitHub at https://github.com/XiaoduoAILab/XmodelVLM.
Authors: Zhenwei Shao, Zhou Yu, Jun Yu, Xuecheng Ouyang, Lihao Zheng, Zhenbiao Gai, Mingyang Wang, Jiajun Ding
Abstract: By harnessing the capabilities of large language models (LLMs), recent large multimodal models (LMMs) have shown remarkable versatility in open-world multimodal understanding. Nevertheless, they are usually parameter-heavy and computation-intensive, thus hindering their applicability in resource-constrained scenarios. To this end, several lightweight LMMs have been proposed successively to maximize the capabilities under constrained scale (e.g., 3B). Despite the encouraging results achieved by these methods, most of them only focus on one or two aspects of the design space, and the key design choices that influence model capability have not yet been thoroughly investigated. In this paper, we conduct a systematic study for lightweight LMMs from the aspects of model architecture, training strategy, and training data. Based on our findings, we obtain Imp -- a family of highly capable LMMs at the 2B-4B scales. Notably, our Imp-3B model steadily outperforms all the existing lightweight LMMs of similar size, and even surpasses the state-of-the-art LMMs at the 13B scale. With low-bit quantization and resolution reduction techniques, our Imp model can be deployed on a Qualcomm Snapdragon 8Gen3 mobile chip with a high inference speed of about 13 tokens/s.
Authors: Jiahao Sun, Chunmei Qing, Xiang Xu, Lingdong Kong, Youquan Liu, Li Li, Chenming Zhu, Jingwei Zhang, Zeqi Xiao, Runnan Chen, Tai Wang, Wenwei Zhang, Kai Chen
Abstract: In the rapidly evolving field of autonomous driving, precise segmentation of LiDAR data is crucial for understanding complex 3D environments. Traditional approaches often rely on disparate, standalone codebases, hindering unified advancements and fair benchmarking across models. To address these challenges, we introduce MMDetection3D-lidarseg, a comprehensive toolbox designed for the efficient training and evaluation of state-of-the-art LiDAR segmentation models. We support a wide range of segmentation models and integrate advanced data augmentation techniques to enhance robustness and generalization. Additionally, the toolbox provides support for multiple leading sparse convolution backends, optimizing computational efficiency and performance. By fostering a unified framework, MMDetection3D-lidarseg streamlines development and benchmarking, setting new standards for research and application. Our extensive benchmark experiments on widely-used datasets demonstrate the effectiveness of the toolbox. The codebase and trained models have been publicly available, promoting further research and innovation in the field of LiDAR segmentation for autonomous driving.
Authors: Hantao Yao, Rui Zhang, Lu Yu, Changsheng Xu
Abstract: Prompt tuning based on Context Optimization (CoOp) effectively adapts visual-language models (VLMs) to downstream tasks by inferring additional learnable prompt tokens. However, these tokens are less discriminative as they are independent of the pre-trained tokens and fail to capture input-specific knowledge, such as class-aware textual or instance-aware visual knowledge. Leveraging the discriminative and generalization capabilities inherent in pre-trained tokens, we introduce a novel approach named Self-Enhanced Prompt Tuning (SEP). The core principle of SEP involves adapting the learnable prompt tokens at each encoder layer from the corresponding self-pretrained tokens, thereby explicitly incorporating discriminative prior knowledge to enhance both textual-level and visual-level embeddings. Furthermore, SEP's self-enhanced tokens not only boost discrimination but also mitigate domain shifts in unseen domains, enhancing generalization. In practice, SEP selects several representative tokens from all pre-trained tokens for each input data at every layer of the text/visual encoders. Subsequently, a Token Fusion Module (TFM) is introduced to generate a self-enhanced token by merging these representative tokens with the learnable tokens using a cross-attention mechanism. This self-enhanced token is then concatenated with all pre-trained tokens, serving as input for subsequent encoder layers to produce the relevant embeddings. Comprehensive evaluations across various benchmarks and tasks confirm SEP's efficacy in prompt tuning. Code: \href{Code}{https://github.com/htyao89/SEP}.
Authors: Xiyao Wang, Jiuhai Chen, Zhaoyang Wang, Yuhang Zhou, Yiyang Zhou, Huaxiu Yao, Tianyi Zhou, Tom Goldstein, Parminder Bhatia, Furong Huang, Cao Xiao
Abstract: Large vision-language models (LVLMs) have achieved impressive results in various visual question-answering and reasoning tasks through vision instruction tuning on specific datasets. However, there is still significant room for improvement in the alignment between visual and language modalities. Previous methods to enhance this alignment typically require external models or data, heavily depending on their capabilities and quality, which inevitably sets an upper bound on performance. In this paper, we propose SIMA, a framework that enhances visual and language modality alignment through self-improvement, eliminating the needs for external models or data. SIMA leverages prompts from existing vision instruction tuning datasets to self-generate responses and employs an in-context self-critic mechanism to select response pairs for preference tuning. The key innovation is the introduction of three vision metrics during the in-context self-critic process, which can guide the LVLM in selecting responses that enhance image comprehension. Through experiments across 14 hallucination and comprehensive benchmarks, we demonstrate that SIMA not only improves model performance across all benchmarks but also achieves superior modality alignment, outperforming previous approaches.
Authors: Mingli Zhu, Siyuan Liang, Baoyuan Wu
Abstract: Deep neural networks face persistent challenges in defending against backdoor attacks, leading to an ongoing battle between attacks and defenses. While existing backdoor defense strategies have shown promising performance on reducing attack success rates, can we confidently claim that the backdoor threat has truly been eliminated from the model? To address it, we re-investigate the characteristics of the backdoored models after defense (denoted as defense models). Surprisingly, we find that the original backdoors still exist in defense models derived from existing post-training defense strategies, and the backdoor existence is measured by a novel metric called backdoor existence coefficient. It implies that the backdoors just lie dormant rather than being eliminated. To further verify this finding, we empirically show that these dormant backdoors can be easily re-activated during inference, by manipulating the original trigger with well-designed tiny perturbation using universal adversarial attack. More practically, we extend our backdoor reactivation to black-box scenario, where the defense model can only be queried by the adversary during inference, and develop two effective methods, i.e., query-based and transfer-based backdoor re-activation attacks. The effectiveness of the proposed methods are verified on both image classification and multimodal contrastive learning (i.e., CLIP) tasks. In conclusion, this work uncovers a critical vulnerability that has never been explored in existing defense strategies, emphasizing the urgency of designing more robust and advanced backdoor defense mechanisms in the future.
Authors: Shugo Yamashita, Masaaki Ikehara
Abstract: Removing rain artifacts in images is recognized as a significant issue. In this field, deep learning-based approaches, such as convolutional neural networks (CNNs) and Transformers, have succeeded. Recently, State Space Models (SSMs) have exhibited superior performance across various tasks in both natural language processing and image processing due to their ability to model long-range dependencies. This study introduces SSM to rain removal and proposes a Deraining Frequency-Enhanced State Space Model (DFSSM). To effectively remove rain streaks, which produce high-intensity frequency components in specific directions, we employ frequency domain processing concurrently with SSM. Additionally, we develop a novel mixed-scale gated-convolutional block, which uses convolutions with multiple kernel sizes to capture various scale degradations effectively and integrates a gating mechanism to manage the flow of information. Finally, experiments on synthetic and real-world rainy image datasets show that our method surpasses state-of-the-art methods.
Authors: Yiming Li, Zehong Wang, Yue Wang, Zhiding Yu, Zan Gojcic, Marco Pavone, Chen Feng, Jose M. Alvarez
Abstract: Humans naturally retain memories of permanent elements, while ephemeral moments often slip through the cracks of memory. This selective retention is crucial for robotic perception, localization, and mapping. To endow robots with this capability, we introduce 3D Gaussian Mapping (3DGM), a self-supervised, camera-only offline mapping framework grounded in 3D Gaussian Splatting. 3DGM converts multitraverse RGB videos from the same region into a Gaussian-based environmental map while concurrently performing 2D ephemeral object segmentation. Our key observation is that the environment remains consistent across traversals, while objects frequently change. This allows us to exploit self-supervision from repeated traversals to achieve environment-object decomposition. More specifically, 3DGM formulates multitraverse environmental mapping as a robust differentiable rendering problem, treating pixels of the environment and objects as inliers and outliers, respectively. Using robust feature distillation, feature residuals mining, and robust optimization, 3DGM jointly performs 2D segmentation and 3D mapping without human intervention. We build the Mapverse benchmark, sourced from the Ithaca365 and nuPlan datasets, to evaluate our method in unsupervised 2D segmentation, 3D reconstruction, and neural rendering. Extensive results verify the effectiveness and potential of our method for self-driving and robotics.
Authors: Vida Adeli, Soroush Mehraban, Irene Ballester, Yasamin Zarghami, Andrea Sabo, Andrea Iaboni, Babak Taati
Abstract: This study investigates the application of general human motion encoders trained on large-scale human motion datasets for analyzing gait patterns in PD patients. Although these models have learned a wealth of human biomechanical knowledge, their effectiveness in analyzing pathological movements, such as parkinsonian gait, has yet to be fully validated. We propose a comparative framework and evaluate six pre-trained state-of-the-art human motion encoder models on their ability to predict the Movement Disorder Society - Unified Parkinson's Disease Rating Scale (MDS-UPDRS-III) gait scores from motion capture data. We compare these against a traditional gait feature-based predictive model in a recently released large public PD dataset, including PD patients on and off medication. The feature-based model currently shows higher weighted average accuracy, precision, recall, and F1-score. Motion encoder models with closely comparable results demonstrate promise for scalability and efficiency in clinical settings. This potential is underscored by the enhanced performance of the encoder model upon fine-tuning on PD training set. Four of the six human motion models examined provided prediction scores that were significantly different between on- and off-medication states. This finding reveals the sensitivity of motion encoder models to nuanced clinical changes. It also underscores the necessity for continued customization of these models to better capture disease-specific features, thereby reducing the reliance on labor-intensive feature engineering. Lastly, we establish a benchmark for the analysis of skeleton-based motion encoder models in clinical settings. To the best of our knowledge, this is the first study to provide a benchmark that enables state-of-the-art models to be tested and compete in a clinical context. Codes and benchmark leaderboard are available at code.
Authors: Seokil Ham, Sangmin Woo, Jin-Young Kim, Hyojun Go, Byeongjun Park, Changick Kim
Abstract: We present Diffusion Model Patching (DMP), a simple method to boost the performance of pre-trained diffusion models that have already reached convergence, with a negligible increase in parameters. DMP inserts a small, learnable set of prompts into the model's input space while keeping the original model frozen. The effectiveness of DMP is not merely due to the addition of parameters but stems from its dynamic gating mechanism, which selects and combines a subset of learnable prompts at every step of the generative process (e.g., reverse denoising steps). This strategy, which we term "mixture-of-prompts", enables the model to draw on the distinct expertise of each prompt, essentially "patching" the model's functionality at every step with minimal yet specialized parameters. Uniquely, DMP enhances the model by further training on the same dataset on which it was originally trained, even in a scenario where significant improvements are typically not expected due to model convergence. Experiments show that DMP significantly enhances the converged FID of DiT-L/2 on FFHQ 256x256 by 10.38%, achieved with only a 1.43% parameter increase and 50K additional training iterations.
Authors: Shuojue Yang, Qian Li, Daiyun Shen, Bingchen Gong, Qi Dou, Yueming Jin
Abstract: Tissue deformation poses a key challenge for accurate surgical scene reconstruction. Despite yielding high reconstruction quality, existing methods suffer from slow rendering speeds and long training times, limiting their intraoperative applicability. Motivated by recent progress in 3D Gaussian Splatting, an emerging technology in real-time 3D rendering, this work presents a novel fast reconstruction framework, termed Deform3DGS, for deformable tissues during endoscopic surgery. Specifically, we introduce 3D GS into surgical scenes by integrating a point cloud initialization to improve reconstruction. Furthermore, we propose a novel flexible deformation modeling scheme (FDM) to learn tissue deformation dynamics at the level of individual Gaussians. Our FDM can model the surface deformation with efficient representations, allowing for real-time rendering performance. More importantly, FDM significantly accelerates surgical scene reconstruction, demonstrating considerable clinical values, particularly in intraoperative settings where time efficiency is crucial. Experiments on DaVinci robotic surgery videos indicate the efficacy of our approach, showcasing superior reconstruction fidelity PSNR: (37.90) and rendering speed (338.8 FPS) while substantially reducing training time to only 1 minute/scene. Our code is available at https://github.com/jinlab-imvr/Deform3DGS.
Authors: Tianchen Zhao, Xuefei Ning, Tongcheng Fang, Enshu Liu, Guyue Huang, Zinan Lin, Shengen Yan, Guohao Dai, Yu Wang
Abstract: Diffusion models have achieved significant visual generation quality. However, their significant computational and memory costs pose challenge for their application on resource-constrained mobile devices or even desktop GPUs. Recent few-step diffusion models reduces the inference time by reducing the denoising steps. However, their memory consumptions are still excessive. The Post Training Quantization (PTQ) replaces high bit-width FP representation with low-bit integer values (INT4/8) , which is an effective and efficient technique to reduce the memory cost. However, when applying to few-step diffusion models, existing quantization methods face challenges in preserving both the image quality and text alignment. To address this issue, we propose an mixed-precision quantization framework - MixDQ. Firstly, We design specialized BOS-aware quantization method for highly sensitive text embedding quantization. Then, we conduct metric-decoupled sensitivity analysis to measure the sensitivity of each layer. Finally, we develop an integer-programming-based method to conduct bit-width allocation. While existing quantization methods fall short at W8A8, MixDQ could achieve W8A8 without performance loss, and W4A8 with negligible visual degradation. Compared with FP16, we achieve 3-4x reduction in model size and memory cost, and 1.45x latency speedup.
Authors: Jingwei Xu, Yikai Wang, Yiqun Zhao, Yanwei Fu, Shenghua Gao
Abstract: Unveiling an empty street from crowded observations captured by in-car cameras is crucial for autonomous driving. However, removing all temporarily static objects, such as stopped vehicles and standing pedestrians, presents a significant challenge. Unlike object-centric 3D inpainting, which relies on thorough observation in a small scene, street scene cases involve long trajectories that differ from previous 3D inpainting tasks. The camera-centric moving environment of captured videos further complicates the task due to the limited degree and time duration of object observation. To address these obstacles, we introduce StreetUnveiler to reconstruct an empty street. StreetUnveiler learns a 3D representation of the empty street from crowded observations. Our representation is based on the hard-label semantic 2D Gaussian Splatting (2DGS) for its scalability and ability to identify Gaussians to be removed. We inpaint rendered image after removing unwanted Gaussians to provide pseudo-labels and subsequently re-optimize the 2DGS. Given its temporal continuous movement, we divide the empty street scene into observed, partial-observed, and unobserved regions, which we propose to locate through a rendered alpha map. This decomposition helps us to minimize the regions that need to be inpainted. To enhance the temporal consistency of the inpainting, we introduce a novel time-reversal framework to inpaint frames in reverse order and use later frames as references for earlier frames to fully utilize the long-trajectory observations. Our experiments conducted on the street scene dataset successfully reconstructed a 3D representation of the empty street. The mesh representation of the empty street can be extracted for further applications. The project page and more visualizations can be found at: https://streetunveiler.github.io
Authors: Xuehao Gao, Yang Yang, Yang Wu, Shaoyi Du, Guo-Jun Qi
Abstract: Inferring 3D human motion is fundamental in many applications, including understanding human activity and analyzing one's intention. While many fruitful efforts have been made to human motion prediction, most approaches focus on pose-driven prediction and inferring human motion in isolation from the contextual environment, thus leaving the body location movement in the scene behind. However, real-world human movements are goal-directed and highly influenced by the spatial layout of their surrounding scenes. In this paper, instead of planning future human motion in a 'dark' room, we propose a Multi-Condition Latent Diffusion network (MCLD) that reformulates the human motion prediction task as a multi-condition joint inference problem based on the given historical 3D body motion and the current 3D scene contexts. Specifically, instead of directly modeling joint distribution over the raw motion sequences, MCLD performs a conditional diffusion process within the latent embedding space, characterizing the cross-modal mapping from the past body movement and current scene context condition embeddings to the future human motion embedding. Extensive experiments on large-scale human motion prediction datasets demonstrate that our MCLD achieves significant improvements over the state-of-the-art methods on both realistic and diverse predictions.
Authors: Jordi Armengol-Estap\'e, Vincent Michalski, Ramnath Kumar, Pierre-Luc St-Charles, Doina Precup, Samira Ebrahimi Kahou
Abstract: Few-shot learning aims to learn representations that can tackle novel tasks given a small number of examples. Recent studies show that cross-modal learning can improve representations for few-shot classification. More specifically, language is a rich modality that can be used to guide visual learning. In this work, we experiment with a multi-modal architecture for few-shot learning that consists of three components: a classifier, an auxiliary network, and a bridge network. While the classifier performs the main classification task, the auxiliary network learns to predict language representations from the same input, and the bridge network transforms high-level features of the auxiliary network into modulation parameters for layers of the few-shot classifier using conditional batch normalization. The bridge should encourage a form of lightweight semantic alignment between language and vision which could be useful for the classifier. However, after evaluating the proposed approach on two popular few-shot classification benchmarks we find that a) the improvements do not reproduce across benchmarks, and b) when they do, the improvements are due to the additional compute and parameters introduced by the bridge network. We contribute insights and recommendations for future work in multi-modal meta-learning, especially when using language representations.
Authors: Jiyoon Myung, Jihyeon Park
Abstract: This paper examines the limitations of advanced text-to-image models in accurately rendering unconventional concepts which are scarcely represented or absent in their training datasets. We identify how these limitations not only confine the creative potential of these models but also pose risks of reinforcing stereotypes. To address these challenges, we introduce the Inpaint Biases framework, which employs user-defined masks and inpainting techniques to enhance the accuracy of image generation, particularly for novel or inaccurately rendered objects. Through experimental validation, we demonstrate how this framework significantly improves the fidelity of generated images to the user's intent, thereby expanding the models' creative capabilities and mitigating the risk of perpetuating biases. Our study contributes to the advancement of text-to-image models as unbiased, versatile tools for creative expression.
Authors: Xuelin Qian, Ruiqi Wu, Gong Cheng, Junwei Han
Abstract: Lifelong Person Re-Identification (LReID) extends traditional ReID by requiring systems to continually learn from non-overlapping datasets across different times and locations, adapting to new identities while preserving knowledge of previous ones. Existing approaches, either rehearsal-free or rehearsal-based, still suffer from the problem of catastrophic forgetting since they try to cram diverse knowledge into one fixed model. To overcome this limitation, we introduce a novel framework AdalReID, that adopts knowledge adapters and a parameter-free auto-selection mechanism for lifelong learning. Concretely, we incrementally build distinct adapters to learn domain-specific knowledge at each step, which can effectively learn and preserve knowledge across different datasets. Meanwhile, the proposed auto-selection strategy adaptively calculates the knowledge similarity between the input set and the adapters. On the one hand, the appropriate adapters are selected for the inputs to process ReID, and on the other hand, the knowledge interaction and fusion between adapters are enhanced to improve the generalization ability of the model. Extensive experiments are conducted to demonstrate the superiority of our AdalReID, which significantly outperforms SOTAs by about 10$\sim$20\% mAP on both seen and unseen domains.
Authors: Hongyuan Dong, Jiawen Li, Bohong Wu, Jiacong Wang, Yuan Zhang, Haoyuan Guo
Abstract: Image captioning has long been regarded as a fundamental task in visual understanding. Recently, however, few large vision-language model (LVLM) research discusses model's image captioning performance because of the outdated short-caption benchmarks and unreliable evaluation metrics. In this work, we propose to benchmark detail image caption task by curating high-quality evaluation datasets annotated by human experts, GPT-4V and Gemini-1.5-Pro. We also design a more reliable caption evaluation metric called CAPTURE (CAPtion evaluation by exTracting and coUpling coRE information). CAPTURE extracts visual elements, e.g., objects, attributes and relations from captions, and then matches these elements through three stages, achieving the highest consistency with expert judgements over other rule-based or model-based caption metrics. The proposed benchmark and metric provide reliable evaluation for LVLM's detailed image captioning ability. Guided by this evaluation, we further explore to unleash LVLM's detail caption capabilities by synthesizing high-quality data through a five-stage data construction pipeline. Our pipeline only uses a given LVLM itself and other open-source tools, without any human or GPT-4V annotation in the loop. Experiments show that the proposed data construction strategy significantly improves model-generated detail caption data quality for LVLMs with leading performance, and the data quality can be further improved in a self-looping paradigm. All code and dataset will be publicly available at https://github.com/foundation-multimodal-models/CAPTURE.
URLs: https://github.com/foundation-multimodal-models/CAPTURE.
Authors: Xintong Jiang, Yaxiong Wang, Mengjian Li, Yujiao Wu, Bingwen Hu, Xueming Qian
Abstract: Composed Image Retrieval (CIR) involves searching for target images based on an image-text pair query. While current methods treat this as a query-target matching problem, we argue that CIR triplets contain additional associations beyond this primary relation. In our paper, we identify two new relations within triplets, treating each triplet as a graph node. Firstly, we introduce the concept of text-bridged image alignment, where the query text serves as a bridge between the query image and the target image. We propose a hinge-based cross-attention mechanism to incorporate this relation into network learning. Secondly, we explore complementary text reasoning, considering CIR as a form of cross-modal retrieval where two images compose to reason about complementary text. To integrate these perspectives effectively, we design a twin attention-based compositor. By combining these complementary associations with the explicit query pair-target image relation, we establish a comprehensive set of constraints for CIR. Our framework, CaLa (Complementary Association Learning for Augmenting Composed Image Retrieval), leverages these insights. We evaluate CaLa on CIRR and FashionIQ benchmarks with multiple backbones, demonstrating its superiority in composed image retrieval.
Authors: Weitian Zhang, Yichao Yan, Yunhui Liu, Xingdong Sheng, Xiaokang Yang
Abstract: This paper aims to introduce 3D Gaussian for efficient, expressive, and editable digital avatar generation. This task faces two major challenges: (1) The unstructured nature of 3D Gaussian makes it incompatible with current generation pipelines; (2) the expressive animation of 3D Gaussian in a generative setting that involves training with multiple subjects remains unexplored. In this paper, we propose a novel avatar generation method named $E^3$Gen, to effectively address these challenges. First, we propose a novel generative UV features plane representation that encodes unstructured 3D Gaussian onto a structured 2D UV space defined by the SMPL-X parametric model. This novel representation not only preserves the representation ability of the original 3D Gaussian but also introduces a shared structure among subjects to enable generative learning of the diffusion model. To tackle the second challenge, we propose a part-aware deformation module to achieve robust and accurate full-body expressive pose control. Extensive experiments demonstrate that our method achieves superior performance in avatar generation and enables expressive full-body pose control and editing. Our project page is https://olivia23333.github.io/E3Gen.
Authors: Nam Anh Dinh, Haochen Wang, Greg Shakhnarovich, Rana Hanocka
Abstract: There is no settled universal 3D representation for geometry with many alternatives such as point clouds, meshes, implicit functions, and voxels to name a few. In this work, we present a new, compelling alternative for representing shapes using a sequence of cross-sectional closed loops. The loops across all planes form an organizational hierarchy which we leverage for autoregressive shape synthesis and editing. Loops are a non-local description of the underlying shape, as simple loop manipulations (such as shifts) result in significant structural changes to the geometry. This is in contrast to manipulating local primitives such as points in a point cloud or a triangle in a triangle mesh. We further demonstrate that loops are intuitive and natural primitive for analyzing and editing shapes, both computationally and for users.
Authors: Yihao Xue, Ali Payani, Yu Yang, Baharan Mirzasoleiman
Abstract: Pretrained machine learning models need to be adapted to distribution shifts when deployed in new target environments. When obtaining labeled data from the target distribution is expensive, few-shot adaptation with only a few examples from the target distribution becomes essential. In this work, we propose MixPro, a lightweight and highly data-efficient approach for few-shot adaptation. MixPro first generates a relatively large dataset by mixing (linearly combining) pre-trained embeddings of large source data with those of the few target examples. This process preserves important features of both source and target distributions, while mitigating the specific noise in the small target data. Then, it trains a linear classifier on the mixed embeddings to effectively adapts the model to the target distribution without overfitting the small target data. Theoretically, we demonstrate the advantages of MixPro over previous methods. Our experiments, conducted across various model architectures on 8 datasets featuring different types of distribution shifts, reveal that MixPro can outperform baselines by up to 7\%, with only 2-4 target examples.
Authors: Mevan Ekanayake, Zhifeng Chen, Mehrtash Harandi, Gary Egan, Zhaolin Chen
Abstract: In Magnetic Resonance Imaging (MRI), image acquisitions are often undersampled in the measurement domain to accelerate the scanning process, at the expense of image quality. However, image quality is a crucial factor that influences the accuracy of clinical diagnosis; hence, high-quality image reconstruction from undersampled measurements has been a key area of research. Recently, deep learning (DL) methods have emerged as the state-of-the-art for MRI reconstruction, typically involving deep neural networks to transform undersampled MRI images into high-quality MRI images through data-driven processes. Nevertheless, there is clear and significant room for improvement in undersampled DL MRI reconstruction to meet the high standards required for clinical diagnosis, in terms of eliminating aliasing artifacts and reducing image noise. In this paper, we introduce a self-supervised pretraining procedure using contrastive learning to improve the accuracy of undersampled DL MRI reconstruction. We use contrastive learning to transform the MRI image representations into a latent space that maximizes mutual information among different undersampled representations and optimizes the information content at the input of the downstream DL reconstruction models. Our experiments demonstrate improved reconstruction accuracy across a range of acceleration factors and datasets, both quantitatively and qualitatively. Furthermore, our extended experiments validate the proposed framework's robustness under adversarial conditions, such as measurement noise, different k-space sampling patterns, and pathological abnormalities, and also prove the transfer learning capabilities on MRI datasets with completely different anatomy. Additionally, we conducted experiments to visualize and analyze the properties of the proposed MRI contrastive learning latent space.
Authors: Xue Yang, Zifeng Liu, Xiaohu Tang, Rongxing Lu, Bo Liu
Abstract: With the emergence of privacy leaks in federated learning, secure aggregation protocols that mainly adopt either homomorphic encryption or threshold secret sharing have been widely developed for federated learning to protect the privacy of the local training data of each client. However, these existing protocols suffer from many shortcomings, such as the dependence on a trusted third party, the vulnerability to clients being corrupted, low efficiency, the trade-off between security and fault tolerance, etc. To solve these disadvantages, we propose an efficient and multi-private key secure aggregation scheme for federated learning. Specifically, we skillfully modify the variant ElGamal encryption technique to achieve homomorphic addition operation, which has two important advantages: 1) The server and each client can freely select public and private keys without introducing a trust third party and 2) Compared to the variant ElGamal encryption, the plaintext space is relatively large, which is more suitable for the deep model. Besides, for the high dimensional deep model parameter, we introduce a super-increasing sequence to compress multi-dimensional data into 1-D, which can greatly reduce encryption and decryption times as well as communication for ciphertext transmission. Detailed security analyses show that our proposed scheme achieves the semantic security of both individual local gradients and the aggregated result while achieving optimal robustness in tolerating both client collusion and dropped clients. Extensive simulations demonstrate that the accuracy of our scheme is almost the same as the non-private approach, while the efficiency of our scheme is much better than the state-of-the-art homomorphic encryption-based secure aggregation schemes. More importantly, the efficiency advantages of our scheme will become increasingly prominent as the number of model parameters increases.
Authors: Wentao He, Yuchen Yan, Jianfeng Ren, Ruibin Bai, Xudong Jiang
Abstract: Deep neural networks have been applied to audio spectrograms for respiratory sound classification. Existing models often treat the spectrogram as a synthetic image while overlooking its physical characteristics. In this paper, a Multi-View Spectrogram Transformer (MVST) is proposed to embed different views of time-frequency characteristics into the vision transformer. Specifically, the proposed MVST splits the mel-spectrogram into different sized patches, representing the multi-view acoustic elements of a respiratory sound. These patches and positional embeddings are then fed into transformer encoders to extract the attentional information among patches through a self-attention mechanism. Finally, a gated fusion scheme is designed to automatically weigh the multi-view features to highlight the best one in a specific scenario. Experimental results on the ICBHI dataset demonstrate that the proposed MVST significantly outperforms state-of-the-art methods for classifying respiratory sounds.
Authors: Guangyi Chen, Yifan Shen, Zhenhao Chen, Xiangchen Song, Yuewen Sun, Weiran Yao, Xiao Liu, Kun Zhang
Abstract: Identifying the underlying time-delayed latent causal processes in sequential data is vital for grasping temporal dynamics and making downstream reasoning. While some recent methods can robustly identify these latent causal variables, they rely on strict assumptions about the invertible generation process from latent variables to observed data. However, these assumptions are often hard to satisfy in real-world applications containing information loss. For instance, the visual perception process translates a 3D space into 2D images, or the phenomenon of persistence of vision incorporates historical data into current perceptions. To address this challenge, we establish an identifiability theory that allows for the recovery of independent latent components even when they come from a nonlinear and non-invertible mix. Using this theory as a foundation, we propose a principled approach, CaRiNG, to learn the CAusal RepresentatIon of Non-invertible Generative temporal data with identifiability guarantees. Specifically, we utilize temporal context to recover lost latent information and apply the conditions in our theory to guide the training process. Through experiments conducted on synthetic datasets, we validate that our CaRiNG method reliably identifies the causal process, even when the generation process is non-invertible. Moreover, we demonstrate that our approach considerably improves temporal understanding and reasoning in practical applications.
Authors: Sangmin Lee, Abbas Mammadov, Jong Chul Ye
Abstract: Current theoretical and empirical research in neural networks suggests that complex datasets require large network architectures for thorough classification, yet the precise nature of this relationship remains unclear. This paper tackles this issue by defining upper and lower bounds for neural network widths, which are informed by the polytope structure of the dataset in question. We also delve into the application of these principles to simplicial complexes and specific manifold shapes, explaining how the requirement for network width varies in accordance with the geometric complexity of the dataset. Moreover, we develop an algorithm to investigate a converse situation where the polytope structure of a dataset can be inferred from its corresponding trained neural networks. Through our algorithm, it is established that popular datasets such as MNIST, Fashion-MNIST, and CIFAR10 can be efficiently encapsulated using no more than two polytopes with a small number of faces.
Authors: Mengqi Wu, Lintao Zhang, Pew-Thian Yap, Hongtu Zhu, Mingxia Liu
Abstract: Brain magnetic resonance imaging (MRI) has been extensively employed across clinical and research fields, but often exhibits sensitivity to site effects arising from non-biological variations such as differences in field strength and scanner vendors. Numerous retrospective MRI harmonization techniques have demonstrated encouraging outcomes in reducing the site effects at the image level. However, existing methods generally suffer from high computational requirements and limited generalizability, restricting their applicability to unseen MRIs. In this paper, we design a novel disentangled latent energy-based style translation (DLEST) framework for unpaired image-level MRI harmonization, consisting of (a) site-invariant image generation (SIG), (b) site-specific style translation (SST), and (c) site-specific MRI synthesis (SMS). Specifically, the SIG employs a latent autoencoder to encode MRIs into a low-dimensional latent space and reconstruct MRIs from latent codes. The SST utilizes an energy-based model to comprehend the global latent distribution of a target domain and translate source latent codes toward the target domain, while SMS enables MRI synthesis with a target-specific style. By disentangling image generation and style translation in latent space, the DLEST can achieve efficient style translation. Our model was trained on T1-weighted MRIs from a public dataset (with 3,984 subjects across 58 acquisition sites/settings) and validated on an independent dataset (with 9 traveling subjects scanned in 11 sites/settings) in four tasks: histogram and feature visualization, site classification, brain tissue segmentation, and site-specific structural MRI synthesis. Qualitative and quantitative results demonstrate the superiority of our method over several state-of-the-arts.
Authors: Mehran Hosseini, Peyman Hosseini
Abstract: Scaled Dot Product Attention (SDPA) is the backbone of many modern deep-learning models. It is so versatile that it has been used in natural language, vision, and multi-modal domains with very little change compared to its original formulation. This paper discusses why the current formulation is inefficient by delving into the mathematical details of the attention mechanism. We propose three improvements to mitigate these inefficiencies, thereby, introducing three enhanced attention mechanisms: Optimised, Efficient, and Super Attention. Optimised and Efficient Attention have one and two matrix multiplications fewer per head, respectively, and 25% and 50% fewer parameters, respectively, than standard SDPA, but perform similarly to standard SDPA in both vision and natural language tasks. They can be used in all applications where SDPA is used while offering smaller model sizes and faster training and inference without noticeable loss in performance. Super Attention introduces a new linear transformation on the values, transforming them from the left. It outperforms standard SPDA on vision and natural language tasks by up to 17% while having one fewer matrix multiplication per head and 25% fewer parameters than standard SDPA. Consequently, it is also faster than standard SDPA. Super Attention is ideal in applications where the attention layer's context length is fixed, such as Vision Transformers. In addition to providing mathematical reasoning, we evaluate the presented attention mechanisms on several datasets including MNIST, CIFAR100, ImageNet, IMDB Movie Reviews, and Amazon Reviews datasets, as well as combined Europarl and Anki English-Spanish datasets for neural machine translation.
Authors: Ekaterina Shumitskaya, Anastasia Antsiferova, Dmitriy Vatolin
Abstract: No-reference image- and video-quality metrics are widely used in video processing benchmarks. The robustness of learning-based metrics under video attacks has not been widely studied. In addition to having success, attacks that can be employed in video processing benchmarks must be fast and imperceptible. This paper introduces an Invisible One-Iteration (IOI) adversarial attack on no reference image and video quality metrics. We compared our method alongside eight prior approaches using image and video datasets via objective and subjective tests. Our method exhibited superior visual quality across various attacked metric architectures while maintaining comparable attack success and speed. We made the code available on GitHub: https://github.com/katiashh/ioi-attack.
Authors: Mohammad Shahab Sepehri, Zalan Fabian, Mahdi Soltanolkotabi
Abstract: The landscape of computational building blocks of efficient image restoration architectures is dominated by a combination of convolutional processing and various attention mechanisms. However, convolutional filters, while efficient, are inherently local and therefore struggle with modeling long-range dependencies in images. In contrast, attention excels at capturing global interactions between arbitrary image regions, but suffers from a quadratic cost in image dimension. In this work, we propose Serpent, an efficient architecture for high-resolution image restoration that combines recent advances in state space models (SSMs) with multi-scale signal processing in its core computational block. SSMs, originally introduced for sequence modeling, can maintain a global receptive field with a favorable linear scaling in input size. We propose a novel hierarchical architecture inspired by traditional signal processing principles, that converts the input image into a collection of sequences and processes them in a multi-scale fashion. Our experimental results demonstrate that Serpent can achieve reconstruction quality on par with state-of-the-art techniques, while requiring orders of magnitude less compute (up to $150$ fold reduction in FLOPS) and a factor of up to $5\times$ less GPU memory while maintaining a compact model size. The efficiency gains achieved by Serpent are especially notable at high image resolutions.
Authors: Aarush Sinha
Abstract: In this study, we propose a multi branched network approach to predict the dynamics of a physics attractor characterized by intricate and chaotic behavior. We introduce a unique neural network architecture comprised of Radial Basis Function (RBF) layers combined with an attention mechanism designed to effectively capture nonlinear inter-dependencies inherent in the attractor's temporal evolution. Our results demonstrate successful prediction of the attractor's trajectory across 100 predictions made using a real-world dataset of 36,700 time-series observations encompassing approximately 28 minutes of activity. To further illustrate the performance of our proposed technique, we provide comprehensive visualizations depicting the attractor's original and predicted behaviors alongside quantitative measures comparing observed versus estimated outcomes. Overall, this work showcases the potential of advanced machine learning algorithms in elucidating hidden structures in complex physical systems while offering practical applications in various domains requiring accurate short-term forecasting capabilities.
Authors: Zhibo Wang, Peng Kuang, Zhixuan Chu, Jingyi Wang, Kui Ren
Abstract: Spurious correlations in training data significantly hinder the generalization capability of machine learning models when faced with distribution shifts in real-world scenarios. To tackle the problem, numerous debias approaches have been proposed and benchmarked on datasets intentionally designed with severe biases. However, it remains to be asked: \textit{1. Do existing benchmarks really capture biases in the real world? 2. Can existing debias methods handle biases in the real world?} To answer the questions, we revisit biased distributions in existing benchmarks and real-world datasets, and propose a fine-grained framework for analyzing dataset bias by disentangling it into the magnitude and prevalence of bias. We observe and theoretically demonstrate that existing benchmarks poorly represent real-world biases. We further introduce two novel biased distributions to bridge this gap, forming a nuanced evaluation framework for real-world debiasing. Building upon these results, we evaluate existing debias methods with our evaluation framework. Results show that existing methods are incapable of handling real-world biases. Through in-depth analysis, we propose a simple yet effective approach that can be easily applied to existing debias methods, named Debias in Destruction (DiD). Empirical results demonstrate the superiority of DiD, improving the performance of existing methods on all types of biases within the proposed evaluation framework.