arXiv daily

Computer Vision and Pattern Recognition (cs.CV)

Tue, 23 May 2023

Other arXiv digests in this category:Thu, 14 Sep 2023; Wed, 13 Sep 2023; Tue, 12 Sep 2023; Mon, 11 Sep 2023; Fri, 08 Sep 2023; Tue, 05 Sep 2023; Fri, 01 Sep 2023; Thu, 31 Aug 2023; Wed, 30 Aug 2023; Tue, 29 Aug 2023; Mon, 28 Aug 2023; Fri, 25 Aug 2023; Thu, 24 Aug 2023; Wed, 23 Aug 2023; Tue, 22 Aug 2023; Mon, 21 Aug 2023; Fri, 18 Aug 2023; Thu, 17 Aug 2023; Wed, 16 Aug 2023; Tue, 15 Aug 2023; Mon, 14 Aug 2023; Fri, 11 Aug 2023; Thu, 10 Aug 2023; Wed, 09 Aug 2023; Tue, 08 Aug 2023; Mon, 07 Aug 2023; Fri, 04 Aug 2023; Thu, 03 Aug 2023; Wed, 02 Aug 2023; Tue, 01 Aug 2023; Mon, 31 Jul 2023; Fri, 28 Jul 2023; Thu, 27 Jul 2023; Wed, 26 Jul 2023; Tue, 25 Jul 2023; Mon, 24 Jul 2023; Fri, 21 Jul 2023; Thu, 20 Jul 2023; Wed, 19 Jul 2023; Tue, 18 Jul 2023; Mon, 17 Jul 2023; Fri, 14 Jul 2023; Thu, 13 Jul 2023; Wed, 12 Jul 2023; Tue, 11 Jul 2023; Mon, 10 Jul 2023; Fri, 07 Jul 2023; Thu, 06 Jul 2023; Wed, 05 Jul 2023; Tue, 04 Jul 2023; Mon, 03 Jul 2023; Fri, 30 Jun 2023; Thu, 29 Jun 2023; Wed, 28 Jun 2023; Tue, 27 Jun 2023; Mon, 26 Jun 2023; Fri, 23 Jun 2023; Thu, 22 Jun 2023; Wed, 21 Jun 2023; Tue, 20 Jun 2023; Fri, 16 Jun 2023; Thu, 15 Jun 2023; Tue, 13 Jun 2023; Mon, 12 Jun 2023; Fri, 09 Jun 2023; Thu, 08 Jun 2023; Wed, 07 Jun 2023; Tue, 06 Jun 2023; Mon, 05 Jun 2023; Fri, 02 Jun 2023; Thu, 01 Jun 2023; Wed, 31 May 2023; Tue, 30 May 2023; Mon, 29 May 2023; Fri, 26 May 2023; Thu, 25 May 2023; Wed, 24 May 2023; Mon, 22 May 2023; Fri, 19 May 2023; Thu, 18 May 2023; Wed, 17 May 2023; Tue, 16 May 2023; Mon, 15 May 2023; Fri, 12 May 2023; Thu, 11 May 2023; Wed, 10 May 2023; Tue, 09 May 2023; Mon, 08 May 2023; Fri, 05 May 2023; Thu, 04 May 2023; Wed, 03 May 2023; Tue, 02 May 2023; Mon, 01 May 2023; Fri, 28 Apr 2023; Thu, 27 Apr 2023; Wed, 26 Apr 2023; Tue, 25 Apr 2023; Mon, 24 Apr 2023; Fri, 21 Apr 2023; Thu, 20 Apr 2023; Wed, 19 Apr 2023; Tue, 18 Apr 2023; Mon, 17 Apr 2023; Fri, 14 Apr 2023; Thu, 13 Apr 2023; Wed, 12 Apr 2023; Tue, 11 Apr 2023; Mon, 10 Apr 2023
1.FlowChroma -- A Deep Recurrent Neural Network for Video Colorization

Authors:Thejan Wijesinghe, Chamath Abeysinghe, Chanuka Wijayakoon, Lahiru Jayathilake, Uthayasanker Thayasivam

Abstract: We develop an automated video colorization framework that minimizes the flickering of colors across frames. If we apply image colorization techniques to successive frames of a video, they treat each frame as a separate colorization task. Thus, they do not necessarily maintain the colors of a scene consistently across subsequent frames. The proposed solution includes a novel deep recurrent encoder-decoder architecture which is capable of maintaining temporal and contextual coherence between consecutive frames of a video. We use a high-level semantic feature extractor to automatically identify the context of a scenario including objects, with a custom fusion layer that combines the spatial and temporal features of a frame sequence. We demonstrate experimental results, qualitatively showing that recurrent neural networks can be successfully used to improve color consistency in video colorization.

2.DiffHand: End-to-End Hand Mesh Reconstruction via Diffusion Models

Authors:Lijun Li, Li'an Zhuo, Bang Zhang, Liefeng Bo, Chen Chen

Abstract: Hand mesh reconstruction from the monocular image is a challenging task due to its depth ambiguity and severe occlusion, there remains a non-unique mapping between the monocular image and hand mesh. To address this, we develop DiffHand, the first diffusion-based framework that approaches hand mesh reconstruction as a denoising diffusion process. Our one-stage pipeline utilizes noise to model the uncertainty distribution of the intermediate hand mesh in a forward process. We reformulate the denoising diffusion process to gradually refine noisy hand mesh and then select mesh with the highest probability of being correct based on the image itself, rather than relying on 2D joints extracted beforehand. To better model the connectivity of hand vertices, we design a novel network module called the cross-modality decoder. Extensive experiments on the popular benchmarks demonstrate that our method outperforms the state-of-the-art hand mesh reconstruction approaches by achieving 5.8mm PA-MPJPE on the Freihand test set, 4.98mm PA-MPJPE on the DexYCB test set.

3.Pulling Target to Source: A New Perspective on Domain Adaptive Semantic Segmentation

Authors:Haochen Wang, Yujun Shen, Jingjing Fei, Wei Li, Liwei Wu, Yuxi Wang, Zhaoxiang Zhang

Abstract: Domain adaptive semantic segmentation aims to transfer knowledge from a labeled source domain to an unlabeled target domain. However, existing methods primarily focus on directly learning qualified target features, making it challenging to guarantee their discrimination in the absence of target labels. This work provides a new perspective. We observe that the features learned with source data manage to keep categorically discriminative during training, thereby enabling us to implicitly learn adequate target representations by simply \textbf{pulling target features close to source features for each category}. To this end, we propose T2S-DA, which we interpret as a form of pulling Target to Source for Domain Adaptation, encouraging the model in learning similar cross-domain features. Also, considering the pixel categories are heavily imbalanced for segmentation datasets, we come up with a dynamic re-weighting strategy to help the model concentrate on those underperforming classes. Extensive experiments confirm that T2S-DA learns a more discriminative and generalizable representation, significantly surpassing the state-of-the-art. We further show that our method is quite qualified for the domain generalization task, verifying its domain-invariant property.

4.Mixup-Privacy: A simple yet effective approach for privacy-preserving segmentation

Authors:Bach Kim, Jose Dolz, Pierre-Marc Jodoin, Christian Desrosiers

Abstract: Privacy protection in medical data is a legitimate obstacle for centralized machine learning applications. Here, we propose a client-server image segmentation system which allows for the analysis of multi-centric medical images while preserving patient privacy. In this approach, the client protects the to-be-segmented patient image by mixing it to a reference image. As shown in our work, it is challenging to separate the image mixture to exact original content, thus making the data unworkable and unrecognizable for an unauthorized person. This proxy image is sent to a server for processing. The server then returns the mixture of segmentation maps, which the client can revert to a correct target segmentation. Our system has two components: 1) a segmentation network on the server side which processes the image mixture, and 2) a segmentation unmixing network which recovers the correct segmentation map from the segmentation mixture. Furthermore, the whole system is trained end-to-end. The proposed method is validated on the task of MRI brain segmentation using images from two different datasets. Results show that the segmentation accuracy of our method is comparable to a system trained on raw images, and outperforms other privacy-preserving methods with little computational overhead.

5.Human Body Pose Estimation for Gait Identification: A Comprehensive Survey of Datasets and Models

Authors:Luke K. Topham, Wasiq Khan, Dhiya Al-Jumeily, Abir Hussain

Abstract: Person identification is a problem that has received substantial attention, particularly in security domains. Gait recognition is one of the most convenient approaches enabling person identification at a distance without the need of high-quality images. There are several review studies addressing person identification such as the utilization of facial images, silhouette images, and wearable sensor. Despite skeleton-based person identification gaining popularity while overcoming the challenges of traditional approaches, existing survey studies lack the comprehensive review of skeleton-based approaches to gait identification. We present a detailed review of the human pose estimation and gait analysis that make the skeleton-based approaches possible. The study covers various types of related datasets, tools, methodologies, and evaluation metrics with associated challenges, limitations, and application domains. Detailed comparisons are presented for each of these aspects with recommendations for potential research and alternatives. A common trend throughout this paper is the positive impact that deep learning techniques are beginning to have on topics such as human pose estimation and gait identification. The survey outcomes might be useful for the related research community and other stakeholders in terms of performance analysis of existing methodologies, potential research gaps, application domains, and possible contributions in the future.

6.MIPI 2023 Challenge on Nighttime Flare Removal: Methods and Results

Authors:Yuekun Dai, Chongyi Li, Shangchen Zhou, Ruicheng Feng, Qingpeng Zhu, Qianhui Sun, Wenxiu Sun, Chen Change Loy, Jinwei Gu

Abstract: Developing and integrating advanced image sensors with novel algorithms in camera systems are prevalent with the increasing demand for computational photography and imaging on mobile platforms. However, the lack of high-quality data for research and the rare opportunity for in-depth exchange of views from industry and academia constrain the development of mobile intelligent photography and imaging (MIPI). With the success of the 1st MIPI Workshop@ECCV 2022, we introduce the second MIPI challenge including four tracks focusing on novel image sensors and imaging algorithms. In this paper, we summarize and review the Nighttime Flare Removal track on MIPI 2023. In total, 120 participants were successfully registered, and 11 teams submitted results in the final testing phase. The developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal. A detailed description of all models developed in this challenge is provided in this paper. More details of this challenge and the link to the dataset can be found at .

7.Understanding Text-driven Motion Synthesis with Keyframe Collaboration via Diffusion Models

Authors:Dong Wei, Xiaoning Sun, Huaijiang Sun, Bin Li, Shengxiang Hu, Weiqing Li, Jianfeng Lu

Abstract: The emergence of text-driven motion synthesis technique provides animators with great potential to create efficiently. However, in most cases, textual expressions only contain general and qualitative motion descriptions, while lack fine depiction and sufficient intensity, leading to the synthesized motions that either (a) semantically compliant but uncontrollable over specific pose details, or (b) even deviates from the provided descriptions, bringing animators with undesired cases. In this paper, we propose DiffKFC, a conditional diffusion model for text-driven motion synthesis with keyframes collaborated. Different from plain text-driven designs, full interaction among texts, keyframes and the rest diffused frames are conducted at training, enabling realistic generation under efficient, collaborative dual-level control: coarse guidance at semantic level, with only few keyframes for direct and fine-grained depiction down to body posture level, to satisfy animator requirements without tedious labor. Specifically, we customize efficient Dilated Mask Attention modules, where only partial valid tokens participate in local-to-global attention, indicated by the dilated keyframe mask. For user flexibility, DiffKFC supports adjustment on importance of fine-grained keyframe control. Experimental results show that our model achieves state-of-the-art performance on text-to-motion datasets HumanML3D and KIT.

8.VisorGPT: Learning Visual Prior via Generative Pre-Training

Authors:Jinheng Xie, Kai Ye, Yudong Li, Yuexiang Li, Kevin Qinghong Lin, Yefeng Zheng, Linlin Shen, Mike Zheng Shou

Abstract: Various stuff and things in visual data possess specific traits, which can be learned by deep neural networks and are implicitly represented as the visual prior, \emph{e.g.,} object location and shape, in the model. Such prior potentially impacts many vision tasks. For example, in conditional image synthesis, spatial conditions failing to adhere to the prior can result in visually inaccurate synthetic results. This work aims to explicitly learn the visual prior and enable the customization of sampling. Inspired by advances in language modeling, we propose to learn Visual prior via Generative Pre-Training, dubbed VisorGPT. By discretizing visual locations of objects, \emph{e.g.,} bounding boxes, human pose, and instance masks, into sequences, \our~can model visual prior through likelihood maximization. Besides, prompt engineering is investigated to unify various visual locations and enable customized sampling of sequential outputs from the learned prior. Experimental results demonstrate that \our~can effectively model the visual prior, which can be employed for many vision tasks, such as customizing accurate human pose for conditional image synthesis models like ControlNet. Code will be released at

9.Full Resolution Repetition Counting

Authors:Jianing Li, Bowen Chen, Zhiyong Wang, Honghai Liu

Abstract: Given an untrimmed video, repetitive actions counting aims to estimate the number of repetitions of class-agnostic actions. To handle the various length of videos and repetitive actions, also optimization challenges in end-to-end video model training, down-sampling is commonly utilized in recent state-of-the-art methods, leading to ignorance of several repetitive samples. In this paper, we attempt to understand repetitive actions from a full temporal resolution view, by combining offline feature extraction and temporal convolution networks. The former step enables us to train repetition counting network without down-sampling while preserving all repetition regardless of the video length and action frequency, and the later network models all frames in a flexible and dynamically expanding temporal receptive field to retrieve all repetitions with a global aspect. We experimentally demonstrate that our method achieves better or comparable performance in three public datasets, i.e., TransRAC, UCFRep and QUVA. We expect this work will encourage our community to think about the importance of full temporal resolution.

10.Perception Test: A Diagnostic Benchmark for Multimodal Video Models

Authors:Viorica Pătrăucean, Lucas Smaira, Ankush Gupta, Adrià Recasens Continente, Larisa Markeeva, Dylan Banarse, Skanda Koppula, Joseph Heyward, Mateusz Malinowski, Yi Yang, Carl Doersch, Tatiana Matejovicova, Yury Sulsky, Antoine Miech, Alex Frechette, Hanna Klimczak, Raphael Koster, Junlin Zhang, Stephanie Winkler, Yusuf Aytar, Simon Osindero, Dima Damen, Andrew Zisserman, João Carreira

Abstract: We propose a novel multimodal video benchmark - the Perception Test - to evaluate the perception and reasoning skills of pre-trained multimodal models (e.g. Flamingo, BEiT-3, or GPT-4). Compared to existing benchmarks that focus on computational tasks (e.g. classification, detection or tracking), the Perception Test focuses on skills (Memory, Abstraction, Physics, Semantics) and types of reasoning (descriptive, explanatory, predictive, counterfactual) across video, audio, and text modalities, to provide a comprehensive and efficient evaluation tool. The benchmark probes pre-trained models for their transfer capabilities, in a zero-shot / few-shot or limited finetuning regime. For these purposes, the Perception Test introduces 11.6k real-world videos, 23s average length, designed to show perceptually interesting situations, filmed by around 100 participants worldwide. The videos are densely annotated with six types of labels (multiple-choice and grounded video question-answers, object and point tracks, temporal action and sound segments), enabling both language and non-language evaluations. The fine-tuning and validation splits of the benchmark are publicly available (CC-BY license), in addition to a challenge server with a held-out test split. Human baseline results compared to state-of-the-art video QA models show a significant gap in performance (91.4% vs 43.6%), suggesting that there is significant room for improvement in multimodal video understanding. Dataset, baselines code, and challenge server are available at

11.Leveraging Uncertainty Quantification for Picking Robust First Break Times

Authors:Hongtao Wang, Jiangshe Zhang, Xiaoli Wei, Li Long, Chunxia Zhang

Abstract: In seismic exploration, the selection of first break times is a crucial aspect in the determination of subsurface velocity models, which in turn significantly influences the placement of wells. Many deep neural network (DNN)-based automatic first break picking methods have been proposed to speed up this picking processing. However, there has been no work on the uncertainty of the first picking results of the output of DNN. In this paper, we propose a new framework for first break picking based on a Bayesian neural network to further explain the uncertainty of the output. In a large number of experiments, we evaluate that the proposed method has better accuracy and robustness than the deterministic DNN-based model. In addition, we also verify that the uncertainty of measurement is meaningful, which can provide a reference for human decision-making.

12.Generalizable Synthetic Image Detection via Language-guided Contrastive Learning

Authors:Haiwei Wu, Jiantao Zhou, Shile Zhang

Abstract: The heightened realism of AI-generated images can be attributed to the rapid development of synthetic models, including generative adversarial networks (GANs) and diffusion models (DMs). The malevolent use of synthetic images, such as the dissemination of fake news or the creation of fake profiles, however, raises significant concerns regarding the authenticity of images. Though many forensic algorithms have been developed for detecting synthetic images, their performance, especially the generalization capability, is still far from being adequate to cope with the increasing number of synthetic models. In this work, we propose a simple yet very effective synthetic image detection method via a language-guided contrastive learning and a new formulation of the detection problem. We first augment the training images with carefully-designed textual labels, enabling us to use a joint image-text contrastive learning for the forensic feature extraction. In addition, we formulate the synthetic image detection as an identification problem, which is vastly different from the traditional classification-based approaches. It is shown that our proposed LanguAge-guided SynThEsis Detection (LASTED) model achieves much improved generalizability to unseen image generation models and delivers promising performance that far exceeds state-of-the-art competitors by +22.66% accuracy and +15.24% AUC. The code is available at

13.Online Open-set Semi-supervised Object Detection via Semi-supervised Outlier Filtering

Authors:Zerun Wang, Ling Xiao, Liuyu Xiang, Zhaotian Weng, Toshihiko Yamasaki

Abstract: Open-set semi-supervised object detection (OSSOD) methods aim to utilize practical unlabeled datasets with out-of-distribution (OOD) instances for object detection. The main challenge in OSSOD is distinguishing and filtering the OOD instances from the in-distribution (ID) instances during pseudo-labeling. The previous method uses an offline OOD detection network trained only with labeled data for solving this problem. However, the scarcity of available data limits the potential for improvement. Meanwhile, training separately leads to low efficiency. To alleviate the above issues, this paper proposes a novel end-to-end online framework that improves performance and efficiency by mining more valuable instances from unlabeled data. Specifically, we first propose a semi-supervised OOD detection strategy to mine valuable ID and OOD instances in unlabeled datasets for training. Then, we constitute an online end-to-end trainable OSSOD framework by integrating the OOD detection head into the object detector, making it jointly trainable with the original detection task. Our experimental results show that our method works well on several benchmarks, including the partially labeled COCO dataset with open-set classes and the fully labeled COCO dataset with the additional large-scale open-set unlabeled dataset, OpenImages. Compared with previous OSSOD methods, our approach achieves the best performance on COCO with OpenImages by +0.94 mAP, reaching 44.07 mAP.

14.NORM: Knowledge Distillation via N-to-One Representation Matching

Authors:Xiaolong Liu, Lujun Li, Chao Li, Anbang Yao

Abstract: Existing feature distillation methods commonly adopt the One-to-one Representation Matching between any pre-selected teacher-student layer pair. In this paper, we present N-to-One Representation (NORM), a new two-stage knowledge distillation method, which relies on a simple Feature Transform (FT) module consisting of two linear layers. In view of preserving the intact information learnt by the teacher network, during training, our FT module is merely inserted after the last convolutional layer of the student network. The first linear layer projects the student representation to a feature space having N times feature channels than the teacher representation from the last convolutional layer, and the second linear layer contracts the expanded output back to the original feature space. By sequentially splitting the expanded student representation into N non-overlapping feature segments having the same number of feature channels as the teacher's, they can be readily forced to approximate the intact teacher representation simultaneously, formulating a novel many-to-one representation matching mechanism conditioned on a single teacher-student layer pair. After training, such an FT module will be naturally merged into the subsequent fully connected layer thanks to its linear property, introducing no extra parameters or architectural modifications to the student network at inference. Extensive experiments on different visual recognition benchmarks demonstrate the leading performance of our method. For instance, the ResNet18|MobileNet|ResNet50-1/4 model trained by NORM reaches 72.14%|74.26%|68.03% top-1 accuracy on the ImageNet dataset when using a pre-trained ResNet34|ResNet50|ResNet50 model as the teacher, achieving an absolute improvement of 2.01%|4.63%|3.03% against the individually trained counterpart. Code is available at

15.Leveraging BEV Representation for 360-degree Visual Place Recognition

Authors:Xuecheng Xu, Yanmei Jiao, Sha Lu, Xiaqing Ding, Rong Xiong, Yue Wang

Abstract: This paper investigates the advantages of using Bird's Eye View (BEV) representation in 360-degree visual place recognition (VPR). We propose a novel network architecture that utilizes the BEV representation in feature extraction, feature aggregation, and vision-LiDAR fusion, which bridges visual cues and spatial awareness. Our method extracts image features using standard convolutional networks and combines the features according to pre-defined 3D grid spatial points. To alleviate the mechanical and time misalignments between cameras, we further introduce deformable attention to learn the compensation. Upon the BEV feature representation, we then employ the polar transform and the Discrete Fourier transform for aggregation, which is shown to be rotation-invariant. In addition, the image and point cloud cues can be easily stated in the same coordinates, which benefits sensor fusion for place recognition. The proposed BEV-based method is evaluated in ablation and comparative studies on two datasets, including on-the-road and off-the-road scenarios. The experimental results verify the hypothesis that BEV can benefit VPR by its superior performance compared to baseline methods. To the best of our knowledge, this is the first trial of employing BEV representation in this task.

16.WaveDM: Wavelet-Based Diffusion Models for Image Restoration

Authors:Yi Huang, Jiancheng Huang, Jianzhuang Liu, Yu Dong, Jiaxi Lv, Shifeng Chen

Abstract: Latest diffusion-based methods for many image restoration tasks outperform traditional models, but they encounter the long-time inference problem. To tackle it, this paper proposes a Wavelet-Based Diffusion Model (WaveDM) with an Efficient Conditional Sampling (ECS) strategy. WaveDM learns the distribution of clean images in the wavelet domain conditioned on the wavelet spectrum of degraded images after wavelet transform, which is more time-saving in each step of sampling than modeling in the spatial domain. In addition, ECS follows the same procedure as the deterministic implicit sampling in the initial sampling period and then stops to predict clean images directly, which reduces the number of total sampling steps to around 5. Evaluations on four benchmark datasets including image raindrop removal, defocus deblurring, demoir\'eing, and denoising demonstrate that WaveDM achieves state-of-the-art performance with the efficiency that is comparable to traditional one-pass methods and over 100 times faster than existing image restoration methods using vanilla diffusion models.

17.SAR-to-Optical Image Translation via Thermodynamics-inspired Network

Authors:Mingjin Zhang, Jiamin Xu, Chengyu He, Wenteng Shang, Yunsong Li, Xinbo Gao

Abstract: Synthetic aperture radar (SAR) is prevalent in the remote sensing field but is difficult to interpret in human visual perception. Recently, SAR-to-optical (S2O) image conversion methods have provided a prospective solution for interpretation. However, since there is a huge domain difference between optical and SAR images, they suffer from low image quality and geometric distortion in the produced optical images. Motivated by the analogy between pixels during the S2O image translation and molecules in a heat field, Thermodynamics-inspired Network for SAR-to-Optical Image Translation (S2O-TDN) is proposed in this paper. Specifically, we design a Third-order Finite Difference (TFD) residual structure in light of the TFD equation of thermodynamics, which allows us to efficiently extract inter-domain invariant features and facilitate the learning of the nonlinear translation mapping. In addition, we exploit the first law of thermodynamics (FLT) to devise an FLT-guided branch that promotes the state transition of the feature values from the unstable diffusion state to the stable one, aiming to regularize the feature diffusion and preserve image structures during S2O image translation. S2O-TDN follows an explicit design principle derived from thermodynamic theory and enjoys the advantage of explainability. Experiments on the public SEN1-2 dataset show the advantages of the proposed S2O-TDN over the current methods with more delicate textures and higher quantitative results.

18.Control-A-Video: Controllable Text-to-Video Generation with Diffusion Models

Authors:Weifeng Chen, Jie Wu, Pan Xie, Hefeng Wu, Jiashi Li, Xin Xia, Xuefeng Xiao, Liang Lin

Abstract: This paper presents a controllable text-to-video (T2V) diffusion model, named Video-ControlNet, that generates videos conditioned on a sequence of control signals, such as edge or depth maps. Video-ControlNet is built on a pre-trained conditional text-to-image (T2I) diffusion model by incorporating a spatial-temporal self-attention mechanism and trainable temporal layers for efficient cross-frame modeling. A first-frame conditioning strategy is proposed to facilitate the model to generate videos transferred from the image domain as well as arbitrary-length videos in an auto-regressive manner. Moreover, Video-ControlNet employs a novel residual-based noise initialization strategy to introduce motion prior from an input video, producing more coherent videos. With the proposed architecture and strategies, Video-ControlNet can achieve resource-efficient convergence and generate superior quality and consistent videos with fine-grained control. Extensive experiments demonstrate its success in various video generative tasks such as video editing and video style transfer, outperforming previous methods in terms of consistency and quality. Project Page:

19.Self-Supervised Gaussian Regularization of Deep Classifiers for Mahalanobis-Distance-Based Uncertainty Estimation

Authors:Aishwarya Venkataramanan, Assia Benbihi, Martin Laviale, Cedric Pradalier

Abstract: Recent works show that the data distribution in a network's latent space is useful for estimating classification uncertainty and detecting Out-of-distribution (OOD) samples. To obtain a well-regularized latent space that is conducive for uncertainty estimation, existing methods bring in significant changes to model architectures and training procedures. In this paper, we present a lightweight, fast, and high-performance regularization method for Mahalanobis distance-based uncertainty prediction, and that requires minimal changes to the network's architecture. To derive Gaussian latent representation favourable for Mahalanobis Distance calculation, we introduce a self-supervised representation learning method that separates in-class representations into multiple Gaussians. Classes with non-Gaussian representations are automatically identified and dynamically clustered into multiple new classes that are approximately Gaussian. Evaluation on standard OOD benchmarks shows that our method achieves state-of-the-art results on OOD detection with minimal inference time, and is very competitive on predictive probability calibration. Finally, we show the applicability of our method to a real-life computer vision use case on microorganism classification.

20.Producing a Standard Dataset of Speed Climbing Training Videos Using Deep Learning Techniques

Authors:Yufei Xie, Shaoman Li, Penghui Lin

Abstract: This dissertation presents a methodology for recording speed climbing training sessions with multiple cameras and annotating the videos with relevant data, including body position, hand and foot placement, and timing. The annotated data is then analyzed using deep learning techniques to create a standard dataset of speed climbing training videos. The results demonstrate the potential of the new dataset for improving speed climbing training and research, including identifying areas for improvement, creating personalized training plans, and analyzing the effects of different training methods.The findings will also be applied to the training process of the Jiangxi climbing team through further empirical research to test the findings and further explore the feasibility of this study.

21.MIANet: Aggregating Unbiased Instance and General Information for Few-Shot Semantic Segmentation

Authors:Yong Yang, Qiong Chen, Yuan Feng, Tianlin Huang

Abstract: Existing few-shot segmentation methods are based on the meta-learning strategy and extract instance knowledge from a support set and then apply the knowledge to segment target objects in a query set. However, the extracted knowledge is insufficient to cope with the variable intra-class differences since the knowledge is obtained from a few samples in the support set. To address the problem, we propose a multi-information aggregation network (MIANet) that effectively leverages the general knowledge, i.e., semantic word embeddings, and instance information for accurate segmentation. Specifically, in MIANet, a general information module (GIM) is proposed to extract a general class prototype from word embeddings as a supplement to instance information. To this end, we design a triplet loss that treats the general class prototype as an anchor and samples positive-negative pairs from local features in the support set. The calculated triplet loss can transfer semantic similarities among language identities from a word embedding space to a visual representation space. To alleviate the model biasing towards the seen training classes and to obtain multi-scale information, we then introduce a non-parametric hierarchical prior module (HPM) to generate unbiased instance-level information via calculating the pixel-level similarity between the support and query image features. Finally, an information fusion module (IFM) combines the general and instance information to make predictions for the query image. Extensive experiments on PASCAL-5i and COCO-20i show that MIANet yields superior performance and set a new state-of-the-art. Code is available at

22.Variational Bayesian Framework for Advanced Image Generation with Domain-Related Variables

Authors:Yuxiao Li, Santiago Mazuelas, Yuan Shen

Abstract: Deep generative models (DGMs) and their conditional counterparts provide a powerful ability for general-purpose generative modeling of data distributions. However, it remains challenging for existing methods to address advanced conditional generative problems without annotations, which can enable multiple applications like image-to-image translation and image editing. We present a unified Bayesian framework for such problems, which introduces an inference stage on latent variables within the learning process. In particular, we propose a variational Bayesian image translation network (VBITN) that enables multiple image translation and editing tasks. Comprehensive experiments show the effectiveness of our method on unsupervised image-to-image translation, and demonstrate the novel advanced capabilities for semantic editing and mixed domain translation.

23.Unsafe Diffusion: On the Generation of Unsafe Images and Hateful Memes From Text-To-Image Models

Authors:Yiting Qu, Xinyue Shen, Xinlei He, Michael Backes, Savvas Zannettou, Yang Zhang

Abstract: State-of-the-art Text-to-Image models like Stable Diffusion and DALLE$\cdot$2 are revolutionizing how people generate visual content. At the same time, society has serious concerns about how adversaries can exploit such models to generate unsafe images. In this work, we focus on demystifying the generation of unsafe images and hateful memes from Text-to-Image models. We first construct a typology of unsafe images consisting of five categories (sexually explicit, violent, disturbing, hateful, and political). Then, we assess the proportion of unsafe images generated by four advanced Text-to-Image models using four prompt datasets. We find that these models can generate a substantial percentage of unsafe images; across four models and four prompt datasets, 14.56% of all generated images are unsafe. When comparing the four models, we find different risk levels, with Stable Diffusion being the most prone to generating unsafe content (18.92% of all generated images are unsafe). Given Stable Diffusion's tendency to generate more unsafe content, we evaluate its potential to generate hateful meme variants if exploited by an adversary to attack a specific individual or community. We employ three image editing methods, DreamBooth, Textual Inversion, and SDEdit, which are supported by Stable Diffusion. Our evaluation result shows that 24% of the generated images using DreamBooth are hateful meme variants that present the features of the original hateful meme and the target individual/community; these generated images are comparable to hateful meme variants collected from the real world. Overall, our results demonstrate that the danger of large-scale generation of unsafe images is imminent. We discuss several mitigating measures, such as curating training data, regulating prompts, and implementing safety filters, and encourage better safeguard tools to be developed to prevent unsafe generation.

24.Cross3DVG: Baseline and Dataset for Cross-Dataset 3D Visual Grounding on Different RGB-D Scans

Authors:Taiki Miyanishi, Daichi Azuma, Shuhei Kurita, Motoki Kawanabe

Abstract: We present Cross3DVG, a novel task for cross-dataset visual grounding in 3D scenes, revealing the limitations of existing 3D visual grounding models using restricted 3D resources and thus easily overfit to a specific 3D dataset. To facilitate Cross3DVG, we have created a large-scale 3D visual grounding dataset containing more than 63k diverse descriptions of 3D objects within 1,380 indoor RGB-D scans from 3RScan with human annotations, paired with the existing 52k descriptions on ScanRefer. We perform Cross3DVG by training a model on the source 3D visual grounding dataset and then evaluating it on the target dataset constructed in different ways (e.g., different sensors, 3D reconstruction methods, and language annotators) without using target labels. We conduct comprehensive experiments using established visual grounding models, as well as a CLIP-based 2D-3D integration method, designed to bridge the gaps between 3D datasets. By performing Cross3DVG tasks, we found that (i) cross-dataset 3D visual grounding has significantly lower performance than learning and evaluation with a single dataset, suggesting much room for improvement in cross-dataset generalization of 3D visual grounding, (ii) better detectors and transformer-based localization modules for 3D grounding are beneficial for enhancing 3D grounding performance and (iii) fusing 2D-3D data using CLIP demonstrates further performance improvements. Our Cross3DVG task will provide a benchmark for developing robust 3D visual grounding models capable of handling diverse 3D scenes while leveraging deep language understanding.

25.Generalized Expectation Maximization Framework for Blind Image Super Resolution

Authors:Yuxiao Li, Zhiming Wang, Yuan Shen

Abstract: Learning-based methods for blind single image super resolution (SISR) conduct the restoration by a learned mapping between high-resolution (HR) images and their low-resolution (LR) counterparts degraded with arbitrary blur kernels. However, these methods mostly require an independent step to estimate the blur kernel, leading to error accumulation between steps. We propose an end-to-end learning framework for the blind SISR problem, which enables image restoration within a unified Bayesian framework with either full- or semi-supervision. The proposed method, namely SREMN, integrates learning techniques into the generalized expectation-maximization (GEM) algorithm and infers HR images from the maximum likelihood estimation (MLE). Extensive experiments show the superiority of the proposed method with comparison to existing work and novelty in semi-supervised learning.

26.Deep Transductive Transfer Learning for Automatic Target Recognition

Authors:Shoaib M. Sami, Nasser M. Nasrabadi, Raghuveer Rao

Abstract: One of the major obstacles in designing an automatic target recognition (ATR) algorithm, is that there are often labeled images in one domain (i.e., infrared source domain) but no annotated images in the other target domains (i.e., visible, SAR, LIDAR). Therefore, automatically annotating these images is essential to build a robust classifier in the target domain based on the labeled images of the source domain. Transductive transfer learning is an effective way to adapt a network to a new target domain by utilizing a pretrained ATR network in the source domain. We propose an unpaired transductive transfer learning framework where a CycleGAN model and a well-trained ATR classifier in the source domain are used to construct an ATR classifier in the target domain without having any labeled data in the target domain. We employ a CycleGAN model to transfer the mid-wave infrared (MWIR) images to visible (VIS) domain images (or visible to MWIR domain). To train the transductive CycleGAN, we optimize a cost function consisting of the adversarial, identity, cycle-consistency, and categorical cross-entropy loss for both the source and target classifiers. In this paper, we perform a detailed experimental analysis on the challenging DSIAC ATR dataset. The dataset consists of ten classes of vehicles at different poses and distances ranging from 1-5 kilometers on both the MWIR and VIS domains. In our experiment, we assume that the images in the VIS domain are the unlabeled target dataset. We first detect and crop the vehicles from the raw images and then project them into a common distance of 2 kilometers. Our proposed transductive CycleGAN achieves 71.56% accuracy in classifying the visible domain vehicles in the DSIAC ATR dataset.

27.WinDB: HMD-free and Distortion-free Panoptic Video Fixation Learning

Authors:Guotao Wang, Chenglizhao Chen, Aimin Hao, Hong Qin, Deng-ping Fan

Abstract: To date, the widely-adopted way to perform fixation collection in panoptic video is based on a head-mounted display (HMD), where participants' fixations are collected while wearing an HMD to explore the given panoptic scene freely. However, this widely-used data collection method is insufficient for training deep models to accurately predict which regions in a given panoptic are most important when it contains intermittent salient events. The main reason is that there always exist "blind zooms" when using HMD to collect fixations since the participants cannot keep spinning their heads to explore the entire panoptic scene all the time. Consequently, the collected fixations tend to be trapped in some local views, leaving the remaining areas to be the "blind zooms". Therefore, fixation data collected using HMD-based methods that accumulate local views cannot accurately represent the overall global importance of complex panoramic scenes. This paper introduces the auxiliary Window with a Dynamic Blurring (WinDB) fixation collection approach for panoptic video, which doesn't need HMD and is blind-zoom-free. Thus, the collected fixations can well reflect the regional-wise importance degree. Using our WinDB approach, we have released a new PanopticVideo-300 dataset, containing 300 panoptic clips covering over 225 categories. Besides, we have presented a simple baseline design to take full advantage of PanopticVideo-300 to handle the blind-zoom-free attribute-induced fixation shifting problem. Our WinDB approach, PanopticVideo-300, and tailored fixation prediction model are all publicly available at

28.Temporal Contrastive Learning for Spiking Neural Networks

Authors:Haonan Qiu, Zeyin Song, Yanqi Chen, Munan Ning, Wei Fang, Tao Sun, Zhengyu Ma, Li Yuan, Yonghong Tian

Abstract: Biologically inspired spiking neural networks (SNNs) have garnered considerable attention due to their low-energy consumption and spatio-temporal information processing capabilities. Most existing SNNs training methods first integrate output information across time steps, then adopt the cross-entropy (CE) loss to supervise the prediction of the average representations. However, in this work, we find the method above is not ideal for the SNNs training as it omits the temporal dynamics of SNNs and degrades the performance quickly with the decrease of inference time steps. One tempting method to model temporal correlations is to apply the same label supervision at each time step and treat them identically. Although it can acquire relatively consistent performance across various time steps, it still faces challenges in obtaining SNNs with high performance. Inspired by these observations, we propose Temporal-domain supervised Contrastive Learning (TCL) framework, a novel method to obtain SNNs with low latency and high performance by incorporating contrastive supervision with temporal domain information. Contrastive learning (CL) prompts the network to discern both consistency and variability in the representation space, enabling it to better learn discriminative and generalizable features. We extend this concept to the temporal domain of SNNs, allowing us to flexibly and fully leverage the correlation between representations at different time steps. Furthermore, we propose a Siamese Temporal-domain supervised Contrastive Learning (STCL) framework to enhance the SNNs via augmentation, temporal and class constraints simultaneously. Extensive experimental results demonstrate that SNNs trained by our TCL and STCL can achieve both high performance and low latency, achieving state-of-the-art performance on a variety of datasets (e.g., CIFAR-10, CIFAR-100, and DVS-CIFAR10).

29.Compositional Text-to-Image Synthesis with Attention Map Control of Diffusion Models

Authors:Ruichen Wang, Zekang Chen, Chen Chen, Jian Ma, Haonan Lu, Xiaodong Lin

Abstract: Recent text-to-image (T2I) diffusion models show outstanding performance in generating high-quality images conditioned on textual prompts. However, these models fail to semantically align the generated images with the text descriptions due to their limited compositional capabilities, leading to attribute leakage, entity leakage, and missing entities. In this paper, we propose a novel attention mask control strategy based on predicted object boxes to address these three issues. In particular, we first train a BoxNet to predict a box for each entity that possesses the attribute specified in the prompt. Then, depending on the predicted boxes, unique mask control is applied to the cross- and self-attention maps. Our approach produces a more semantically accurate synthesis by constraining the attention regions of each token in the prompt to the image. In addition, the proposed method is straightforward and effective, and can be readily integrated into existing cross-attention-diffusion-based T2I generators. We compare our approach to competing methods and demonstrate that it not only faithfully conveys the semantics of the original text to the generated content, but also achieves high availability as a ready-to-use plugin.

30.Decoupled Kullback-Leibler Divergence Loss

Authors:Jiequan Cui, Zhuotao Tian, Zhisheng Zhong, Xiaojuan Qi, Bei Yu, Hanwang Zhang

Abstract: In this paper, we delve deeper into the Kullback-Leibler (KL) Divergence loss and observe that it is equivalent to the Doupled Kullback-Leibler (DKL) Divergence loss that consists of 1) a weighted Mean Square Error (wMSE) loss and 2) a Cross-Entropy loss incorporating soft labels. From our analysis of the DKL loss, we have identified two areas for improvement. Firstly, we address the limitation of DKL in scenarios like knowledge distillation by breaking its asymmetry property in training optimization. This modification ensures that the wMSE component is always effective during training, providing extra constructive cues. Secondly, we introduce global information into DKL for intra-class consistency regularization. With these two enhancements, we derive the Improved Kullback-Leibler (IKL) Divergence loss and evaluate its effectiveness by conducting experiments on CIFAR-10/100 and ImageNet datasets, focusing on adversarial training and knowledge distillation tasks. The proposed approach achieves new state-of-the-art performance on both tasks, demonstrating the substantial practical merits. Code and models will be available soon at

31.From Model-Based to Data-Driven Simulation: Challenges and Trends in Autonomous Driving

Authors:Ferdinand Mütsch, Helen Gremmelmaier, Nicolas Becker, Daniel Bogdoll, Marc René Zofka, J. Marius Zöllner

Abstract: Simulation is an integral part in the process of developing autonomous vehicles and advantageous for training, validation, and verification of driving functions. Even though simulations come with a series of benefits compared to real-world experiments, various challenges still prevent virtual testing from entirely replacing physical test-drives. Our work provides an overview of these challenges with regard to different aspects and types of simulation and subsumes current trends to overcome them. We cover aspects around perception-, behavior- and content-realism as well as general hurdles in the domain of simulation. Among others, we observe a trend of data-driven, generative approaches and high-fidelity data synthesis to increasingly replace model-based simulation.

32.Metrics Matter in Surgical Phase Recognition

Authors:Isabel Funke, Dominik Rivoir, Stefanie Speidel

Abstract: Surgical phase recognition is a basic component for different context-aware applications in computer- and robot-assisted surgery. In recent years, several methods for automatic surgical phase recognition have been proposed, showing promising results. However, a meaningful comparison of these methods is difficult due to differences in the evaluation process and incomplete reporting of evaluation details. In particular, the details of metric computation can vary widely between different studies. To raise awareness of potential inconsistencies, this paper summarizes common deviations in the evaluation of phase recognition algorithms on the Cholec80 benchmark. In addition, a structured overview of previously reported evaluation results on Cholec80 is provided, taking known differences in evaluation protocols into account. Greater attention to evaluation details could help achieve more consistent and comparable results on the surgical phase recognition task, leading to more reliable conclusions about advancements in the field and, finally, translation into clinical practice.

33.Multi-Echo Denoising in Adverse Weather

Authors:Alvari Seppänen, Risto Ojala, Kari Tammi

Abstract: Adverse weather can cause noise to light detection and ranging (LiDAR) data. This is a problem since it is used in many outdoor applications, e.g. object detection and mapping. We propose the task of multi-echo denoising, where the goal is to pick the echo that represents the objects of interest and discard other echoes. Thus, the idea is to pick points from alternative echoes that are not available in standard strongest echo point clouds due to the noise. In an intuitive sense, we are trying to see through the adverse weather. To achieve this goal, we propose a novel self-supervised deep learning method and the characteristics similarity regularization method to boost its performance. Based on extensive experiments on a semi-synthetic dataset, our method achieves superior performance compared to the state-of-the-art in self-supervised adverse weather denoising (23% improvement). Moreover, the experiments with a real multi-echo adverse weather dataset prove the efficacy of multi-echo denoising. Our work enables more reliable point cloud acquisition in adverse weather and thus promises safer autonomous driving and driving assistance systems in such conditions. The code is available at

34.CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model

Authors:Shuai Zhao, Xiaohan Wang, Linchao Zhu, Yi Yang

Abstract: Pre-trained vision-language models are the de-facto foundation models for various downstream tasks. However, this trend has not extended to the field of scene text recognition (STR), despite the potential of CLIP to serve as a powerful scene text reader. CLIP can robustly identify regular (horizontal) and irregular (rotated, curved, blurred, or occluded) text in natural images. With such merits, we introduce CLIP4STR, a simple yet effective STR method built upon image and text encoders of CLIP. It has two encoder-decoder branches: a visual branch and a cross-modal branch. The visual branch provides an initial prediction based on the visual feature, and the cross-modal branch refines this prediction by addressing the discrepancy between the visual feature and text semantics. To fully leverage the capabilities of both branches, we design a dual predict-and-refine decoding scheme for inference. CLIP4STR achieves new state-of-the-art performance on 11 STR benchmarks. Additionally, a comprehensive empirical study is provided to enhance the understanding of the adaptation of CLIP to STR. We believe our method establishes a simple but strong baseline for future STR research with VL models.

35.Faster Video Moment Retrieval with Point-Level Supervision

Authors:Xun Jiang, Zailei Zhou, Xing Xu, Yang Yang, Guoqing Wang, Heng Tao Shen

Abstract: Video Moment Retrieval (VMR) aims at retrieving the most relevant events from an untrimmed video with natural language queries. Existing VMR methods suffer from two defects: (1) massive expensive temporal annotations are required to obtain satisfying performance; (2) complicated cross-modal interaction modules are deployed, which lead to high computational cost and low efficiency for the retrieval process. To address these issues, we propose a novel method termed Cheaper and Faster Moment Retrieval (CFMR), which well balances the retrieval accuracy, efficiency, and annotation cost for VMR. Specifically, our proposed CFMR method learns from point-level supervision where each annotation is a single frame randomly located within the target moment. It is 6 times cheaper than the conventional annotations of event boundaries. Furthermore, we also design a concept-based multimodal alignment mechanism to bypass the usage of cross-modal interaction modules during the inference process, remarkably improving retrieval efficiency. The experimental results on three widely used VMR benchmarks demonstrate the proposed CFMR method establishes new state-of-the-art with point-level supervision. Moreover, it significantly accelerates the retrieval speed with more than 100 times FLOPs compared to existing approaches with point-level supervision.

36.Sparse4D v2: Recurrent Temporal Fusion with Sparse Model

Authors:Xuewu Lin, Tianwei Lin, Zixiang Pei, Lichao Huang, Zhizhong Su

Abstract: Sparse algorithms offer great flexibility for multi-view temporal perception tasks. In this paper, we present an enhanced version of Sparse4D, in which we improve the temporal fusion module by implementing a recursive form of multi-frame feature sampling. By effectively decoupling image features and structured anchor features, Sparse4D enables a highly efficient transformation of temporal features, thereby facilitating temporal fusion solely through the frame-by-frame transmission of sparse features. The recurrent temporal fusion approach provides two main benefits. Firstly, it reduces the computational complexity of temporal fusion from $O(T)$ to $O(1)$, resulting in significant improvements in inference speed and memory usage. Secondly, it enables the fusion of long-term information, leading to more pronounced performance improvements due to temporal fusion. Our proposed approach, Sparse4Dv2, further enhances the performance of the sparse perception algorithm and achieves state-of-the-art results on the nuScenes 3D detection benchmark. Code will be available at \url{}.

37.Realistic Noise Synthesis with Diffusion Models

Authors:Qi Wu, Mingyan Han, Ting Jiang, Haoqiang Fan, Bing Zeng, Shuaicheng Liu

Abstract: Deep learning-based approaches have achieved remarkable performance in single-image denoising. However, training denoising models typically requires a large amount of data, which can be difficult to obtain in real-world scenarios. Furthermore, synthetic noise used in the past has often produced significant differences compared to real-world noise due to the complexity of the latter and the poor modeling ability of noise distributions of Generative Adversarial Network (GAN) models, resulting in residual noise and artifacts within denoising models. To address these challenges, we propose a novel method for synthesizing realistic noise using diffusion models. This approach enables us to generate large amounts of high-quality data for training denoising models by controlling camera settings to simulate different environmental conditions and employing guided multi-scale content information to ensure that our method is more capable of generating real noise with multi-frequency spatial correlations. In particular, we design an inversion mechanism for the setting, which extends our method to more public datasets without setting information. Based on the noise dataset we synthesized, we have conducted sufficient experiments on multiple benchmarks, and experimental results demonstrate that our method outperforms state-of-the-art methods on multiple benchmarks and metrics, demonstrating its effectiveness in synthesizing realistic noise for training denoising models.

38.Why semantics matters: A deep study on semantic particle-filtering localization in a LiDAR semantic pole-map

Authors:Yuming Huang, Yi Gu, Chengzhong Xu, Hui Kong

Abstract: In most urban and suburban areas, pole-like structures such as tree trunks or utility poles are ubiquitous. These structural landmarks are very useful for the localization of autonomous vehicles given their geometrical locations in maps and measurements from sensors. In this work, we aim at creating an accurate map for autonomous vehicles or robots with pole-like structures as the dominant localization landmarks, hence called pole-map. In contrast to the previous pole-based mapping or localization methods, we exploit the semantics of pole-like structures. Specifically, semantic segmentation is achieved by a new mask-range transformer network in a mask-classfication paradigm. With the semantics extracted for the pole-like structures in each frame, a multi-layer semantic pole-map is created by aggregating the detected pole-like structures from all frames. Given the semantic pole-map, we propose a semantic particle-filtering localization scheme for vehicle localization. Theoretically, we have analyzed why the semantic information can benefit the particle-filter localization, and empirically it is validated on the public SemanticKITTI dataset that the particle-filtering localization with semantics achieves much better performance than the counterpart without semantics when each particle's odometry prediction and/or the online observation is subject to uncertainties at significant levels.

39.Learning a Single Convolutional Layer Model for Low Light Image Enhancement

Authors:Yuantong Zhang, Baoxin Teng, Daiqin Yang, Zhenzhong Chen, Haichuan Ma, Gang Li, Wenpeng Ding

Abstract: Low-light image enhancement (LLIE) aims to improve the illuminance of images due to insufficient light exposure. Recently, various lightweight learning-based LLIE methods have been proposed to handle the challenges of unfavorable prevailing low contrast, low brightness, etc. In this paper, we have streamlined the architecture of the network to the utmost degree. By utilizing the effective structural re-parameterization technique, a single convolutional layer model (SCLM) is proposed that provides global low-light enhancement as the coarsely enhanced results. In addition, we introduce a local adaptation module that learns a set of shared parameters to accomplish local illumination correction to address the issue of varied exposure levels in different image regions. Experimental results demonstrate that the proposed method performs favorably against the state-of-the-art LLIE methods in both objective metrics and subjective visual effects. Additionally, our method has fewer parameters and lower inference complexity compared to other learning-based schemes.

40.Parts of Speech-Grounded Subspaces in Vision-Language Models

Authors:James Oldfield, Christos Tzelepis, Yannis Panagakis, Mihalis A. Nicolaou, Ioannis Patras

Abstract: Latent image representations arising from vision-language models have proved immensely useful for a variety of downstream tasks. However, their utility is limited by their entanglement with respect to different visual attributes. For instance, recent work has shown that CLIP image representations are often biased toward specific visual properties (such as objects or actions) in an unpredictable manner. In this paper, we propose to separate representations of the different visual modalities in CLIP's joint vision-language space by leveraging the association between parts of speech and specific visual modes of variation (e.g. nouns relate to objects, adjectives describe appearance). This is achieved by formulating an appropriate component analysis model that learns subspaces capturing variability corresponding to a specific part of speech, while jointly minimising variability to the rest. Such a subspace yields disentangled representations of the different visual properties of an image or text in closed form while respecting the underlying geometry of the manifold on which the representations lie. What's more, we show the proposed model additionally facilitates learning subspaces corresponding to specific visual appearances (e.g. artists' painting styles), which enables the selective removal of entire visual themes from CLIP-based text-to-image synthesis. We validate the model both qualitatively, by visualising the subspace projections with a text-to-image model and by preventing the imitation of artists' styles, and quantitatively, through class invariance metrics and improvements to baseline zero-shot classification. Our code is available at:

41.Accelerated Coordinate Encoding: Learning to Relocalize in Minutes using RGB and Poses

Authors:Eric Brachmann, Tommaso Cavallari, Victor Adrian Prisacariu

Abstract: Learning-based visual relocalizers exhibit leading pose accuracy, but require hours or days of training. Since training needs to happen on each new scene again, long training times make learning-based relocalization impractical for most applications, despite its promise of high accuracy. In this paper we show how such a system can actually achieve the same accuracy in less than 5 minutes. We start from the obvious: a relocalization network can be split in a scene-agnostic feature backbone, and a scene-specific prediction head. Less obvious: using an MLP prediction head allows us to optimize across thousands of view points simultaneously in each single training iteration. This leads to stable and extremely fast convergence. Furthermore, we substitute effective but slow end-to-end training using a robust pose solver with a curriculum over a reprojection loss. Our approach does not require privileged knowledge, such a depth maps or a 3D model, for speedy training. Overall, our approach is up to 300x faster in mapping than state-of-the-art scene coordinate regression, while keeping accuracy on par.

42.3D Open-vocabulary Segmentation with Foundation Models

Authors:Kunhao Liu, Fangneng Zhan, Jiahui Zhang, Muyu Xu, Yingchen Yu, Abdulmotaleb El Saddik, Christian Theobalt, Eric Xing, Shijian Lu

Abstract: Open-vocabulary segmentation of 3D scenes is a fundamental function of human perception and thus a crucial objective in computer vision research. However, this task is heavily impeded by the lack of large-scale and diverse 3D open-vocabulary segmentation datasets for training robust and generalizable models. Distilling knowledge from pre-trained 2D open-vocabulary segmentation models helps but it compromises the open-vocabulary feature significantly as the 2D models are mostly finetuned with close-vocabulary datasets. We tackle the challenges in 3D open-vocabulary segmentation by exploiting the open-vocabulary multimodal knowledge and object reasoning capability of pre-trained foundation models CLIP and DINO, without necessitating any fine-tuning. Specifically, we distill open-vocabulary visual and textual knowledge from CLIP into a neural radiance field (NeRF) which effectively lifts 2D features into view-consistent 3D segmentation. Furthermore, we introduce the Relevancy-Distribution Alignment loss and Feature-Distribution Alignment loss to respectively mitigate the ambiguities of CLIP features and distill precise object boundaries from DINO features, eliminating the need for segmentation annotations during training. Extensive experiments show that our method even outperforms fully supervised models trained with segmentation annotations, suggesting that 3D open-vocabulary segmentation can be effectively learned from 2D images and text-image pairs.

43.S-CLIP: Semi-supervised Vision-Language Pre-training using Few Specialist Captions

Authors:Sangwoo Mo, Minkyu Kim, Kyungmin Lee, Jinwoo Shin

Abstract: Vision-language models, such as contrastive language-image pre-training (CLIP), have demonstrated impressive results in natural image domains. However, these models often struggle when applied to specialized domains like remote sensing, and adapting to such domains is challenging due to the limited number of image-text pairs available for training. To address this, we propose S-CLIP, a semi-supervised learning method for training CLIP that utilizes additional unpaired images. S-CLIP employs two pseudo-labeling strategies specifically designed for contrastive learning and the language modality. The caption-level pseudo-label is given by a combination of captions of paired images, obtained by solving an optimal transport problem between unpaired and paired images. The keyword-level pseudo-label is given by a keyword in the caption of the nearest paired image, trained through partial label learning that assumes a candidate set of labels for supervision instead of the exact one. By combining these objectives, S-CLIP significantly enhances the training of CLIP using only a few image-text pairs, as demonstrated in various specialist domains, including remote sensing, fashion, scientific figures, and comics. For instance, S-CLIP improves CLIP by 10% for zero-shot classification and 4% for image-text retrieval on the remote sensing benchmark, matching the performance of supervised CLIP while using three times fewer image-text pairs.

44.ISP: Multi-Layered Garment Draping with Implicit Sewing Patterns

Authors:Ren Li, Benoît Guillard, Pascal Fua

Abstract: Many approaches to draping individual garments on human body models are realistic, fast, and yield outputs that are differentiable with respect to the body shape on which they are draped. However, none of them can handle multi-layered clothing, which is prevalent in everyday dress. In this paper, we introduce a parametric garment representation model that can. As in models used by clothing designers, each garment consists of individual 2D panels. Their 2D shape is defined by a Signed Distance Function and 3D shape by a 2D to 3D mapping. The 2D parameterization enables easy detection of potential collisions and the 3D parameterization handles complex shapes effectively. We show that this combination is faster and yields higher quality reconstructions than purely implicit surface representations, and makes the recovery of layered garments from images possible thanks to its differentiability. Furthermore, it supports rapid editing of garment shapes and texture by modifying individual 2D panels.

45.Federated Generalized Category Discovery

Authors:Nan Pu, Zhun Zhong, Xinyuan Ji, Nicu Sebe

Abstract: Generalized category discovery (GCD) aims at grouping unlabeled samples from known and unknown classes, given labeled data of known classes. To meet the recent decentralization trend in the community, we introduce a practical yet challenging task, namely Federated GCD (Fed-GCD), where the training data are distributively stored in local clients and cannot be shared among clients. The goal of Fed-GCD is to train a generic GCD model by client collaboration under the privacy-protected constraint. The Fed-GCD leads to two challenges: 1) representation degradation caused by training each client model with fewer data than centralized GCD learning, and 2) highly heterogeneous label spaces across different clients. To this end, we propose a novel Associated Gaussian Contrastive Learning (AGCL) framework based on learnable GMMs, which consists of a Client Semantics Association (CSA) and a global-local GMM Contrastive Learning (GCL). On the server, CSA aggregates the heterogeneous categories of local-client GMMs to generate a global GMM containing more comprehensive category knowledge. On each client, GCL builds class-level contrastive learning with both local and global GMMs. The local GCL learns robust representation with limited local data. The global GCL encourages the model to produce more discriminative representation with the comprehensive category relationships that may not exist in local data. We build a benchmark based on six visual datasets to facilitate the study of Fed-GCD. Extensive experiments show that our AGCL outperforms the FedAvg-based baseline on all datasets.

46.Learning Remote Sensing Object Detection with Single Point Supervision

Authors:Shitian He, Huanxin Zou, Yingqian Wang, Boyang Li, Xu Cao, Ning Jing

Abstract: Pointly Supervised Object Detection (PSOD) has attracted considerable interests due to its lower labeling cost as compared to box-level supervised object detection. However, the complex scenes, densely packed and dynamic-scale objects in Remote Sensing (RS) images hinder the development of PSOD methods in RS field. In this paper, we make the first attempt to achieve RS object detection with single point supervision, and propose a PSOD framework tailored with RS images. Specifically, we design a point label upgrader (PLUG) to generate pseudo box labels from single point labels, and then use the pseudo boxes to supervise the optimization of existing detectors. Moreover, to handle the challenge of the densely packed objects in RS images, we propose a sparse feature guided semantic prediction module which can generate high-quality semantic maps by fully exploiting informative cues from sparse objects. Extensive ablation studies on the DOTA dataset have validated the effectiveness of our method. Our method can achieve significantly better performance as compared to state-of-the-art image-level and point-level supervised detection methods, and reduce the performance gap between PSOD and box-level supervised object detection. Code will be available at

47.A multimodal method based on cross-attention and convolution for postoperative infection diagnosis

Authors:Xianjie Liu, Hongwei Shi

Abstract: Postoperative infection diagnosis is a common and serious complication that generally poses a high diagnostic challenge. This study focuses on PJI, a type of postoperative infection. X-ray examination is an imaging examination for suspected PJI patients that can evaluate joint prostheses and adjacent tissues, and detect the cause of pain. Laboratory examination data has high sensitivity and specificity and has significant potential in PJI diagnosis. In this study, we proposed a self-supervised masked autoencoder pre-training strategy and a multimodal fusion diagnostic network MED-NVC, which effectively implements the interaction between two modal features through the feature fusion network of CrossAttention. We tested our proposed method on our collected PJI dataset and evaluated its performance and feasibility through comparison and ablation experiments. The results showed that our method achieved an ACC of 94.71% and an AUC of 98.22%, which is better than the latest method and also reduces the number of parameters. Our proposed method has the potential to provide clinicians with a powerful tool for enhancing accuracy and efficiency.

48.Impact of Light and Shadow on Robustness of Deep Neural Networks

Authors:Chengyin Hu, Weiwen Shi, Chao Li, Jialiang Sun, Donghua Wang, Junqi Wu, Guijian Tang

Abstract: Deep neural networks (DNNs) have made remarkable strides in various computer vision tasks, including image classification, segmentation, and object detection. However, recent research has revealed a vulnerability in advanced DNNs when faced with deliberate manipulations of input data, known as adversarial attacks. Moreover, the accuracy of DNNs is heavily influenced by the distribution of the training dataset. Distortions or perturbations in the color space of input images can introduce out-of-distribution data, resulting in misclassification. In this work, we propose a brightness-variation dataset, which incorporates 24 distinct brightness levels for each image within a subset of ImageNet. This dataset enables us to simulate the effects of light and shadow on the images, so as is to investigate the impact of light and shadow on the performance of DNNs. In our study, we conduct experiments using several state-of-the-art DNN architectures on the aforementioned dataset. Through our analysis, we discover a noteworthy positive correlation between the brightness levels and the loss of accuracy in DNNs. Furthermore, we assess the effectiveness of recently proposed robust training techniques and strategies, including AugMix, Revisit, and Free Normalizer, using the ResNet50 architecture on our brightness-variation dataset. Our experimental results demonstrate that these techniques can enhance the robustness of DNNs against brightness variation, leading to improved performance when dealing with images exhibiting varying brightness levels.

49.DetGPT: Detect What You Need via Reasoning

Authors:Renjie Pi, Jiahui Gao, Shizhe Diao, Rui Pan, Hanze Dong, Jipeng Zhang, Lewei Yao, Jianhua Han, Hang Xu, Lingpeng Kong Tong Zhang

Abstract: In recent years, the field of computer vision has seen significant advancements thanks to the development of large language models (LLMs). These models have enabled more effective and sophisticated interactions between humans and machines, paving the way for novel techniques that blur the lines between human and machine intelligence. In this paper, we introduce a new paradigm for object detection that we call reasoning-based object detection. Unlike conventional object detection methods that rely on specific object names, our approach enables users to interact with the system using natural language instructions, allowing for a higher level of interactivity. Our proposed method, called DetGPT, leverages state-of-the-art multi-modal models and open-vocabulary object detectors to perform reasoning within the context of the user's instructions and the visual scene. This enables DetGPT to automatically locate the object of interest based on the user's expressed desires, even if the object is not explicitly mentioned. For instance, if a user expresses a desire for a cold beverage, DetGPT can analyze the image, identify a fridge, and use its knowledge of typical fridge contents to locate the beverage. This flexibility makes our system applicable across a wide range of fields, from robotics and automation to autonomous driving. Overall, our proposed paradigm and DetGPT demonstrate the potential for more sophisticated and intuitive interactions between humans and machines. We hope that our proposed paradigm and approach will provide inspiration to the community and open the door to more interative and versatile object detection systems. Our project page is launched at

50.TVTSv2: Learning Out-of-the-box Spatiotemporal Visual Representations at Scale

Authors:Ziyun Zeng, Yixiao Ge, Zhan Tong, Xihui Liu, Shu-Tao Xia, Ying Shan

Abstract: The ultimate goal for foundation models is realizing task-agnostic, i.e., supporting out-of-the-box usage without task-specific fine-tuning. Although breakthroughs have been made in natural language processing and image representation learning, it is still challenging for video models to reach it due to the increasing uncertainty of spatiotemporal signals. To ease training, existing works leverage image foundation models' prior knowledge and equip them with efficient temporal modules. Despite the satisfactory fine-tuning performance, we empirically find they fall short of out-of-the-box usage, given the even degraded performance in zero-shot/linear protocols compared to their baseline counterparts. In this work, we analyze the factor that leads to degradation from the perspective of language supervision distortion. We argue that tuning a text encoder end-to-end, as done in previous work, is suboptimal since it may overfit in terms of styles, thereby losing its original generalization ability to capture the semantics of various language registers. The overfitted text encoder, in turn, provides a harmful supervision signal, degrading the video representation. To tackle this issue, we propose a degradation-free pre-training strategy to retain the generalization ability of the text encoder via freezing shallow layers while enabling the task-related semantics capturing in tunable deep layers. As for the training objective, we adopted the transcript sorting task in TVTS incorporated with masking techniques to enable scalable training. As a result, we produce a series of models, dubbed TVTSv2, with up to one billion parameters. We achieve new state-of-the-arts on various video benchmarks with a frozen backbone, surpassing the recent ImageBind, InternVideo, etc. Code is available at

51.SAD: Segment Any RGBD

Authors:Jun Cen, Yizheng Wu, Kewei Wang, Xingyi Li, Jingkang Yang, Yixuan Pei, Lingdong Kong, Ziwei Liu, Qifeng Chen

Abstract: The Segment Anything Model (SAM) has demonstrated its effectiveness in segmenting any part of 2D RGB images. However, SAM exhibits a stronger emphasis on texture information while paying less attention to geometry information when segmenting RGB images. To address this limitation, we propose the Segment Any RGBD (SAD) model, which is specifically designed to extract geometry information directly from images. Inspired by the natural ability of humans to identify objects through the visualization of depth maps, SAD utilizes SAM to segment the rendered depth map, thus providing cues with enhanced geometry information and mitigating the issue of over-segmentation. We further include the open-vocabulary semantic segmentation in our framework, so that the 3D panoptic segmentation is fulfilled. The project is available on

52.DUBLIN -- Document Understanding By Language-Image Network

Authors:Kriti Aggarwal, Aditi Khandelwal, Kumar Tanmay, Owais Mohammed Khan, Qiang Liu, Monojit Choudhury, Subhojit Som, Vishrav Chaudhary, Saurabh Tiwary

Abstract: Visual document understanding is a complex task that involves analyzing both the text and the visual elements in document images. Existing models often rely on manual feature engineering or domain-specific pipelines, which limit their generalization ability across different document types and languages. In this paper, we propose DUBLIN, which is pretrained on webpages using three novel objectives that leverage the spatial and semantic information in the document images: Masked Document Content Generation Task, Bounding Box Task, and Rendered Question Answering Task. We evaluate our model on several benchmarks, such as Web-Based Structural Reading Comprehension, Document Visual Question Answering, Key Information Extraction, Diagram Understanding, and Table Question Answering. We show that our model achieves competitive or better results than the state-of-the-art models on these tasks. In particular, we show that DUBLIN is the first pixel-based model to achieve an EM of 77.75 and F1 of 84.25 on the WebSRC dataset. We also show that our model outperforms the current pixel-based SOTA models on DocVQA and AI2D datasets by significant margins, 2% and 21% increase in performance, respectively. Also, DUBLIN is the first ever pixel-based model which achieves comparable to text-based SOTA methods on XFUND dataset for Semantic Entity Recognition showcasing its multilingual capability. Moreover, we create new baselines for text-based datasets by rendering them as document images and applying this model.

53.REC-MV: REconstructing 3D Dynamic Cloth from Monocular Videos

Authors:Lingteng Qiu, Guanying Chen, Jiapeng Zhou, Mutian Xu, Junle Wang, Xiaoguang Han

Abstract: Reconstructing dynamic 3D garment surfaces with open boundaries from monocular videos is an important problem as it provides a practical and low-cost solution for clothes digitization. Recent neural rendering methods achieve high-quality dynamic clothed human reconstruction results from monocular video, but these methods cannot separate the garment surface from the body. Moreover, despite existing garment reconstruction methods based on feature curve representation demonstrating impressive results for garment reconstruction from a single image, they struggle to generate temporally consistent surfaces for the video input. To address the above limitations, in this paper, we formulate this task as an optimization problem of 3D garment feature curves and surface reconstruction from monocular video. We introduce a novel approach, called REC-MV, to jointly optimize the explicit feature curves and the implicit signed distance field (SDF) of the garments. Then the open garment meshes can be extracted via garment template registration in the canonical space. Experiments on multiple casually captured datasets show that our approach outperforms existing methods and can produce high-quality dynamic garment surfaces. The source code is available at

54.Masked Path Modeling for Vision-and-Language Navigation

Authors:Zi-Yi Dou, Feng Gao, Nanyun Peng

Abstract: Vision-and-language navigation (VLN) agents are trained to navigate in real-world environments by following natural language instructions. A major challenge in VLN is the limited availability of training data, which hinders the models' ability to generalize effectively. Previous approaches have attempted to address this issue by introducing additional supervision during training, often requiring costly human-annotated data that restricts scalability. In this paper, we introduce a masked path modeling (MPM) objective, which pretrains an agent using self-collected data for downstream navigation tasks. Our proposed method involves allowing the agent to actively explore navigation environments without a specific goal and collect the paths it traverses. Subsequently, we train the agent on this collected data to reconstruct the original path given a randomly masked subpath. This way, the agent can actively accumulate a diverse and substantial amount of data while learning conditional action generation. To evaluate the effectiveness of our technique, we conduct experiments on various VLN datasets and demonstrate the versatility of MPM across different levels of instruction complexity. Our results exhibit significant improvements in success rates, with enhancements of 1.32\%, 1.05\%, and 1.19\% on the val-unseen split of the Room-to-Room, Room-for-Room, and Room-across-Room datasets, respectively. Furthermore, we conduct an analysis that highlights the potential for additional improvements when the agent is allowed to explore unseen environments prior to testing.

55.Source-Free Domain Adaptation for RGB-D Semantic Segmentation with Vision Transformers

Authors:Giulia Rizzoli, Donald Shenaj, Pietro Zanuttigh

Abstract: With the increasing availability of depth sensors, multimodal frameworks that combine color information with depth data are attracting increasing interest. In the challenging task of semantic segmentation, depth maps allow to distinguish between similarly colored objects at different depths and provide useful geometric cues. On the other side, ground truth data for semantic segmentation is burdensome to be provided and thus domain adaptation is another significant research area. Specifically, we address the challenging source-free domain adaptation setting where the adaptation is performed without reusing source data. We propose MISFIT: MultImodal Source-Free Information fusion Transformer, a depth-aware framework which injects depth information into a segmentation module based on vision transformers at multiple stages, namely at the input, feature and output levels. Color and depth style transfer helps early-stage domain alignment while re-wiring self-attention between modalities creates mixed features allowing the extraction of better semantic content. Furthermore, a depth-based entropy minimization strategy is also proposed to adaptively weight regions at different distances. Our framework, which is also the first approach using vision transformers for source-free semantic segmentation, shows noticeable performance improvements with respect to standard strategies.

56.MOTRv3: Release-Fetch Supervision for End-to-End Multi-Object Tracking

Authors:En Yu, Tiancai Wang, Zhuoling Li, Yuang Zhang, Xiangyu Zhang, Wenbing Tao

Abstract: Although end-to-end multi-object trackers like MOTR enjoy the merits of simplicity, they suffer from the conflict between detection and association seriously, resulting in unsatisfactory convergence dynamics. While MOTRv2 partly addresses this problem, it demands an additional detection network for assistance. In this work, we serve as the first to reveal that this conflict arises from the unfair label assignment between detect queries and track queries during training, where these detect queries recognize targets and track queries associate them. Based on this observation, we propose MOTRv3, which balances the label assignment process using the developed release-fetch supervision strategy. In this strategy, labels are first released for detection and gradually fetched back for association. Besides, another two strategies named pseudo label distillation and track group denoising are designed to further improve the supervision for detection and association. Without the assistance of an extra detection network during inference, MOTRv3 achieves impressive performance across diverse benchmarks, e.g., MOT17, DanceTrack.

57.Hierarchical Adaptive Voxel-guided Sampling for Real-time Applications in Large-scale Point Clouds

Authors:Junyuan Ouyang, Xiao Liu, Haoyao Chen

Abstract: While point-based neural architectures have demonstrated their efficacy, the time-consuming sampler currently prevents them from performing real-time reasoning on scene-level point clouds. Existing methods attempt to overcome this issue by using random sampling strategy instead of the commonly-adopted farthest point sampling~(FPS), but at the expense of lower performance. So the effectiveness/efficiency trade-off remains under-explored. In this paper, we reveal the key to high-quality sampling is ensuring an even spacing between points in the subset, which can be naturally obtained through a grid. Based on this insight, we propose a hierarchical adaptive voxel-guided point sampler with linear complexity and high parallelization for real-time applications. Extensive experiments on large-scale point cloud detection and segmentation tasks demonstrate that our method achieves competitive performance with the most powerful FPS, at an amazing speed that is more than 100 times faster. This breakthrough in efficiency addresses the bottleneck of the sampling step when handling scene-level point clouds. Furthermore, our sampler can be easily integrated into existing models and achieves a 20$\sim$80\% reduction in runtime with minimal effort. The code will be available at

58.Text-guided 3D Human Generation from 2D Collections

Authors:Tsu-Jui Fu, Wenhan Xiong, Yixin Nie, Jingyu Liu, Barlas Oğuz, William Yang Wang

Abstract: 3D human modeling has been widely used for engaging interaction in gaming, film, and animation. The customization of these characters is crucial for creativity and scalability, which highlights the importance of controllability. In this work, we introduce Text-guided 3D Human Generation (\texttt{T3H}), where a model is to generate a 3D human, guided by the fashion description. There are two goals: 1) the 3D human should render articulately, and 2) its outfit is controlled by the given text. To address this \texttt{T3H} task, we propose Compositional Cross-modal Human (CCH). CCH adopts cross-modal attention to fuse compositional human rendering with the extracted fashion semantics. Each human body part perceives relevant textual guidance as its visual patterns. We incorporate the human prior and semantic discrimination to enhance 3D geometry transformation and fine-grained consistency, enabling it to learn from 2D collections for data efficiency. We conduct evaluations on DeepFashion and SHHQ with diverse fashion attributes covering the shape, fabric, and color of upper and lower clothing. Extensive experiments demonstrate that CCH achieves superior results for \texttt{T3H} with high efficiency.

59.Large Language Models are Frame-level Directors for Zero-shot Text-to-Video Generation

Authors:Susung Hong, Junyoung Seo, Sunghwan Hong, Heeseong Shin, Seungryong Kim

Abstract: In the paradigm of AI-generated content (AIGC), there has been increasing attention in extending pre-trained text-to-image (T2I) models to text-to-video (T2V) generation. Despite their effectiveness, these frameworks face challenges in maintaining consistent narratives and handling rapid shifts in scene composition or object placement from a single user prompt. This paper introduces a new framework, dubbed DirecT2V, which leverages instruction-tuned large language models (LLMs) to generate frame-by-frame descriptions from a single abstract user prompt. DirecT2V utilizes LLM directors to divide user inputs into separate prompts for each frame, enabling the inclusion of time-varying content and facilitating consistent video generation. To maintain temporal consistency and prevent object collapse, we propose a novel value mapping method and dual-softmax filtering. Extensive experimental results validate the effectiveness of the DirecT2V framework in producing visually coherent and consistent videos from abstract user prompts, addressing the challenges of zero-shot video generation.

60.Diffusion Hyperfeatures: Searching Through Time and Space for Semantic Correspondence

Authors:Grace Luo, Lisa Dunlap, Dong Huk Park, Aleksander Holynski, Trevor Darrell

Abstract: Diffusion models have been shown to be capable of generating high-quality images, suggesting that they could contain meaningful internal representations. Unfortunately, the feature maps that encode a diffusion model's internal information are spread not only over layers of the network, but also over diffusion timesteps, making it challenging to extract useful descriptors. We propose Diffusion Hyperfeatures, a framework for consolidating multi-scale and multi-timestep feature maps into per-pixel feature descriptors that can be used for downstream tasks. These descriptors can be extracted for both synthetic and real images using the generation and inversion processes. We evaluate the utility of our Diffusion Hyperfeatures on the task of semantic keypoint correspondence: our method achieves superior performance on the SPair-71k real image benchmark. We also demonstrate that our method is flexible and transferable: our feature aggregation network trained on the inversion features of real image pairs can be used on the generation features of synthetic image pairs with unseen objects and compositions. Our code is available at \url{}.

61.Prototype Adaption and Projection for Few- and Zero-shot 3D Point Cloud Semantic Segmentation

Authors:Shuting He, Xudong Jiang, Wei Jiang, Henghui Ding

Abstract: In this work, we address the challenging task of few-shot and zero-shot 3D point cloud semantic segmentation. The success of few-shot semantic segmentation in 2D computer vision is mainly driven by the pre-training on large-scale datasets like imagenet. The feature extractor pre-trained on large-scale 2D datasets greatly helps the 2D few-shot learning. However, the development of 3D deep learning is hindered by the limited volume and instance modality of datasets due to the significant cost of 3D data collection and annotation. This results in less representative features and large intra-class feature variation for few-shot 3D point cloud segmentation. As a consequence, directly extending existing popular prototypical methods of 2D few-shot classification/segmentation into 3D point cloud segmentation won't work as well as in 2D domain. To address this issue, we propose a Query-Guided Prototype Adaption (QGPA) module to adapt the prototype from support point clouds feature space to query point clouds feature space. With such prototype adaption, we greatly alleviate the issue of large feature intra-class variation in point cloud and significantly improve the performance of few-shot 3D segmentation. Besides, to enhance the representation of prototypes, we introduce a Self-Reconstruction (SR) module that enables prototype to reconstruct the support mask as well as possible. Moreover, we further consider zero-shot 3D point cloud semantic segmentation where there is no support sample. To this end, we introduce category words as semantic information and propose a semantic-visual projection model to bridge the semantic and visual spaces. Our proposed method surpasses state-of-the-art algorithms by a considerable 7.90% and 14.82% under the 2-way 1-shot setting on S3DIS and ScanNet benchmarks, respectively. Code is available at

62.Siamese Masked Autoencoders

Authors:Agrim Gupta, Jiajun Wu, Jia Deng, Li Fei-Fei

Abstract: Establishing correspondence between images or scenes is a significant challenge in computer vision, especially given occlusions, viewpoint changes, and varying object appearances. In this paper, we present Siamese Masked Autoencoders (SiamMAE), a simple extension of Masked Autoencoders (MAE) for learning visual correspondence from videos. SiamMAE operates on pairs of randomly sampled video frames and asymmetrically masks them. These frames are processed independently by an encoder network, and a decoder composed of a sequence of cross-attention layers is tasked with predicting the missing patches in the future frame. By masking a large fraction ($95\%$) of patches in the future frame while leaving the past frame unchanged, SiamMAE encourages the network to focus on object motion and learn object-centric representations. Despite its conceptual simplicity, features learned via SiamMAE outperform state-of-the-art self-supervised methods on video object segmentation, pose keypoint propagation, and semantic part propagation tasks. SiamMAE achieves competitive results without relying on data augmentation, handcrafted tracking-based pretext tasks, or other techniques to prevent representational collapse.

63.NCHO: Unsupervised Learning for Neural 3D Composition of Humans and Objects

Authors:Taeksoo Kim, Shunsuke Saito, Hanbyul Joo

Abstract: Deep generative models have been recently extended to synthesizing 3D digital humans. However, previous approaches treat clothed humans as a single chunk of geometry without considering the compositionality of clothing and accessories. As a result, individual items cannot be naturally composed into novel identities, leading to limited expressiveness and controllability of generative 3D avatars. While several methods attempt to address this by leveraging synthetic data, the interaction between humans and objects is not authentic due to the domain gap, and manual asset creation is difficult to scale for a wide variety of objects. In this work, we present a novel framework for learning a compositional generative model of humans and objects (backpacks, coats, scarves, and more) from real-world 3D scans. Our compositional model is interaction-aware, meaning the spatial relationship between humans and objects, and the mutual shape change by physical contact is fully incorporated. The key challenge is that, since humans and objects are in contact, their 3D scans are merged into a single piece. To decompose them without manual annotations, we propose to leverage two sets of 3D scans of a single person with and without objects. Our approach learns to decompose objects and naturally compose them back into a generative human model in an unsupervised manner. Despite our simple setup requiring only the capture of a single subject with objects, our experiments demonstrate the strong generalization of our model by enabling the natural composition of objects to diverse identities in various poses and the composition of multiple objects, which is unseen in training data.