论文检索

Giyoung Jeon, Haedong Jeong, Jaesik Choi

Input attribution is a widely used explanation method for deep neural networks, especially in visual tasks. Among various attribution methods, Integrated Gradients (IG) is frequently used because of its model-agnostic applicability and desirable axioms. However, previous work has shown that such method often produces noisy and unreliable attributions during the integration of the gradients over the path defined in the input space. In this paper, we tackle this issue by estimating the distribution of the possible attributions according to the integrating path selection. We show that such noisy attribution can be reduced by aggregating attributions from the multiple paths instead of using a single path. Inspired by Stick-Breaking Process (SBP), we suggest a random process to generate rich and various sampling of the gradient integrating path. Using multiple input attributions obtained from randomized path, we propose a novel attribution measure using the distribution of attributions at each input features. We identify proposed method qualitatively show less-noisy and object-aligned attribution and its feasibility through the quantitative evaluations.

Xi Chen, Shuang Li, Ser-Nam Lim, Antonio Torralba, Hengshuang Zhao

Open-vocabulary segmentation is attracting increasing attention due to its critical applications in the real world. Traditional closed-vocabulary segmentation methods are not able to characterize novel objects, whereas several recent open-vocabulary attempts obtain unsatisfactory results, i.e., notable performance reduction on the closed-vocabulary and massive demand for extra training data. To this end, we propose OPSNet, an omnipotent and data-efficient framework for Open-vocabulary Panoptic Segmentation. Specifically, the exquisitely designed Embedding Modulation module, together with several meticulous components, enables adequate embedding enhancement and information exchange between the segmentation backbone and the visual-linguistic well-aligned CLIP encoder, resulting in superior segmentation performance under both open- and closed vocabulary settings and much fewer need of additional data. Extensive experimental evaluations are conducted across multiple datasets(e.g., COCO, ADE20K, Cityscapes, and PascalContext) under various circumstances, where the proposed OPSNet achieves state-of-the-art results, which demonstrates the effectiveness and generality of the proposed approach. The code and trained models will be made publicly available.

Ziyi Wang, Xumin Yu, Yongming Rao, Jie Zhou, Jiwen Lu

With the overwhelming trend of mask image modeling led by MAE, generative pre-training has shown a remarkable potential to boost the performance of fundamental models in 2D vision. However, in 3D vision, the over-reliance on Transformer-based backbones and the unordered nature of point clouds have restricted the further development of gen- erative pre-training. In this paper, we propose a novel 3D-to- 2D generative pre-training method that is adaptable to any point cloud model. We propose to generate view images from different instructed poses via the cross-attention mechanism as the pre-training scheme. Generating view images has more precise supervision than its point cloud counterpart, thus assisting 3D backbones to have a finer comprehension of the geometrical structure and stereoscopic relations of the point cloud. Experimental results have proved the su- periority of our proposed 3D-to-2D generative pre-training over previous pre-training methods. Our method is also ef- fective in boosting the performance of architecture-oriented approaches, achieving state-of-the-art performance when fine-tuning on ScanObjectNN classification and ShapeNet- Part segmentation tasks. Code is available at https: //github.com/wangzy22/TakeAPhoto.

Zhiyu Zhu, Junhui Hou, Dapeng Oliver Wu

This paper addresses the problem of cross-modal object tracking from RGB videos and event data. Rather than constructing a complex cross-modal fusion network, we explore the great potential of a pre-trained vision Transformer (ViT). Particularly, we delicately investigate plug-and-play training augmentations that encourage the ViT to bridge the vast distribution gap between the two modalities, enabling comprehensive cross-modal information interaction and thus enhancing its ability. Specifically, we propose a mask modeling strategy that randomly masks a specific modality of some tokens to enforce the interaction between tokens from different modalities interacting proactively. To mitigate network oscillations resulting from the masking strategy and further amplify its positive effect, we then theoretically propose an orthogonal high-rank loss to regularize the attention matrix. Extensive experiments demonstrate that our plug-and-play training augmentation techniques can significantly boost state-of-the-art one-stream and two-stream trackers to a large extent in terms of both tracking precision and success rate. Our new perspective and findings will potentially bring insights to the field of leveraging powerful pre-trained ViTs to model cross-modal data. The code is publicly available at https://github.com/ZHU-Zhiyu/High-Rank_RGB-Event_Tracker.

Kang Ma, Ying Fu, Dezhi Zheng, Yunjie Peng, Chunshui Cao, Yongzhen Huang

Gait recognition has emerged as a promising technique for the long-range retrieval of pedestrians, providing numerous advantages such as accurate identification in challenging conditions and non-intrusiveness, making it highly desirable for improving public safety and security. However, the high cost of labeling datasets, which is a prerequisite for most existing fully supervised approaches, poses a significant obstacle to the development of gait recognition. Recently, some unsupervised methods for gait recognition have shown promising results. However, these methods mainly rely on a fine-tuning approach that does not sufficiently consider the relationship between source and target domains, leading to the catastrophic forgetting of source domain knowledge. This paper presents a novel perspective that adjacent-view sequences exhibit overlapping views, which can be leveraged by the network to gradually attain cross-view and cross-dressing capabilities without pre-training on the labeled source domain. Specifically, we propose a fine-grained Unsupervised Domain Adaptation (UDA) framework that iteratively alternates between two stages. The initial stage involves offline clustering, which transfers knowledge from the labeled source domain to the unlabeled target domain and adaptively generates pseudo-labels according to the expressiveness of each part. Subsequently, the second stage encompasses online training, which further achieves cross-dressing capabilities by continuously learning to distinguish numerous features of source and target domains. The effectiveness of the proposed method is demonstrated through extensive experiments conducted on widely-used public gait datasets.

Akshaya Athwale, Arman Afrasiyabi, Justin Lagüe, Ichrak Shili, Ola Ahmad, Jean-François Lalonde

Wide-angle lenses are commonly used in perception tasks requiring a large field of view. Unfortunately, these lenses produce significant distortions making conventional models that ignore the distortion effects unable to adapt to wide-angle images. In this paper, we present a novel transformer-based model that automatically adapts to the distortion produced by wide-angle lenses. We leverage the physical characteristics of such lenses, which are analytically defined by the radial distortion profile (assumed to be known), to develop a distortion aware radial swin transformer (DarSwin). In contrast to conventional transformer-based architectures, DarSwin comprises a radial patch partitioning, a distortion-based sampling technique for creating token embeddings, and an angular position encoding for radial patch merging. We validate our method on classification tasks using synthetically distorted ImageNet data and show through extensive experiments that DarSwin can perform zero-shot adaptation to unseen distortions of different wide-angle lenses. Compared to other baselines, DarSwin achieves the best results (in terms of Top-1 accuracy) with significant gains when trained on bounded levels of distortions (very-low, low, medium, and high) and tested on all including out-of-distribution distortions. The code and models are publicly available at https://lvsn.github.io/darswin/

Zhaoqi Su, Liangxiao Hu, Siyou Lin, Hongwen Zhang, Shengping Zhang, Justus Thies, Yebin Liu

We present CaPhy, a novel method for reconstructing animatable human avatars with realistic dynamic properties for clothing. Specifically, we aim for capturing the geometric and physical properties of the clothing from real observations. This allows us to apply novel poses to the human avatar with physically correct deformations and wrinkles of the clothing. To this end, we combine unsupervised training with physics-based losses and 3D-supervised training using scanned data to reconstruct a dynamic model of clothing that is physically realistic and conforms to the human scans. We also optimize the physical parameters of the underlying physical model from the scans by introducing gradient constraints of the physics-based losses. In contrast to previous work on 3D avatar reconstruction, our method is able to generalize to novel poses with realistic dynamic cloth deformations. Experiments on several subjects demonstrate that our method can estimate the physical properties of the garments, resulting in superior quantitative and qualitative results compared with previous methods.

Zoey Guo, Yiwen Tang, Ray Zhang, Dong Wang, Zhigang Wang, Bin Zhao, Xuelong Li

Understanding 3D scenes from multi-view inputs has been proven to alleviate the view discrepancy issue in 3D visual grounding. However, existing methods normally neglect the view cues embedded in the text modality and fail to weigh the relative importance of different views. In this paper, we propose ViewRefer, a multi-view framework for 3D visual grounding exploring how to grasp the view knowledge from both text and 3D modalities. For the text branch, ViewRefer leverages the diverse linguistic knowledge of large-scale language models to expand a single grounding text to multiple geometry-consistent descriptions. Meanwhile, in the 3D modality, a transformer fusion module with inter-view attention is introduced to boost the interaction of objects across views. On top of that, we further present a set of learnable multi-view prototypes, which memorize scene-agnostic knowledge for different views, and enhance the framework from two perspectives: a view-guided attention module for more robust text features, and a view-guided scoring strategy during the final prediction. With our designed paradigm, ViewRefer achieves superior performance on three benchmarks and surpasses the second-best by +2.8%, +1.5%, and +1.35% on Sr3D, Nr3D, and ScanRefer.

Xinquan Chen, Xitong Gao, Juanjuan Zhao, Kejiang Ye, Cheng-Zhong Xu

Previous work on adversarial examples typically involves a fixed norm perturbation budget, which fails to capture the way humans perceive perturbations. Recent work has shifted towards investigating natural unrestricted adversarial examples (UAEs) that breaks l_p perturbation bounds but nonetheless remain semantically plausible. Current methods use GAN or VAE to generate UAEs by perturbing latent codes. However, this leads to loss of high-level information, resulting in low-quality and unnatural UAEs. In light of this, we propose AddDiffuser, a new method for synthesizing natural UAEs using diffusion models. It can generate UAEs from scratch or conditionally based on reference images. To generate natural UAEs, we perturb predicted images to steer their latent code towards the adversarial sample space of a particular classifier. In addition, we propose adversarial inpainting based on class activation mapping to retain the salient regions of the image while perturbing less important areas. Our method achieves impressive results on CIFAR-10, CelebA and ImageNet, and we demonstrate that it can defeat the most robust models on the RobustBench leaderboard with near 100% success rates. Furthermore, The synthesized UAEs are not only more natural but also stronger compared to the current state-of-the-art attacks. Specifically, compared with GA-attack, the UAEs generated with AdvDiffuser exhibit 6xsmaller LPIPS perturbations, 2 ~ 3 xsmaller FID scores and 0.28 higher in SSIM metrics, making them perceptually stealthier. Lastly, it is capable of generating an unlimited number of natural adversarial examples. For more please visit our project page: Link to follow.

Zhizhong Wang, Lei Zhao, Wei Xing

Content and style (C-S) disentanglement is a fundamental problem and critical challenge of style transfer. Existing approaches based on explicit definitions (e.g., Gram matrix) or implicit learning (e.g., GANs) are neither interpretable nor easy to control, resulting in entangled representations and less satisfying results. In this paper, we propose a new C-S disentangled framework for style transfer without using previous assumptions. The key insight is to explicitly extract the content information and implicitly learn the complementary style information, yielding interpretable and controllable C-S disentanglement and style transfer. A simple yet effective CLIP-based style disentanglement loss coordinated with a style reconstruction prior is introduced to disentangle C-S in the CLIP image space. By further leveraging the powerful style removal and generative ability of diffusion models, our framework achieves superior results than state of the art and flexible C-S disentanglement and trade-off control. Our work provides new insights into the C-S disentanglement in style transfer and demonstrates the potential of diffusion models for learning well-disentangled C-S characteristics.

Yue Fan, Anna Kukleva, Dengxin Dai, Bernt Schiele

Semi-supervised learning (SSL) methods effectively leverage unlabeled data to improve model generalization. However, SSL models often underperform in open-set scenarios, where unlabeled data contain outliers from novel categories that do not appear in the labeled set. In this paper, we study the challenging and realistic open-set SSL setting, where the goal is to both correctly classify inliers and to detect outliers. Intuitively, the inlier classifier should be trained on inlier data only. However, we find that inlier classification performance can be largely improved by incorporating high-confidence pseudo-labeled data, regardless of whether they are inliers or outliers. Also, we propose to utilize non-linear transformations to separate the features used for inlier classification and outlier detection in the multi-task learning framework, preventing adverse effects between them. Additionally, we introduce pseudo-negative mining, which further boosts outlier detection performance. The three ingredients lead to what we call Simple but Strong Baseline (SSB) for open-set SSL. In experiments, SSB greatly improves both inlier classification and outlier detection performance, outperforming existing methods by a large margin. Our code will be released at https://github.com/YUE-FAN/SSB.

Virat Shejwalkar, Lingjuan Lyu, Amir Houmansadr

Semi-supervised learning (SSL) is gaining popularity as it reduces cost of machine learning (ML) by training high performance models using unlabeled data. In this paper, we reveal that the key feature of SSL, i.e., learning from (non-inspected) unlabeled data, exposes SSL to strong poisoning attacks that can significantly damage its security. Poisoning is a long-standing problem in conventional supervised ML, but we argue that, as SSL relies on non-inspected unlabeled data, poisoning poses a more significant threat to SSL. We demonstrate this by designing a backdoor poisoning attack on SSL that can be conducted by a weak adversary with no knowledge of the target SSL pipeline. This is unlike prior poisoning attacks on supervised ML that assume strong adversaries with impractical capabilities. We show that by poisoning only 0.2% of the unlabeled training data, our (weak) adversary can successfully cause misclassification on more than 80% of test inputs (when they contain the backdoor trigger). Our attack remains effective across different benchmark datasets and SSL algorithms, and even circumvents state-of-the-art defenses against backdoor attacks. Our work raises significant concerns about the security of SSL in real-world security critical applications.

Bingkun Huang, Zhiyu Zhao, Guozhen Zhang, Yu Qiao, Limin Wang

Masked autoencoding has shown excellent performance on self-supervised video representation learning. Temporal redundancy has led to a high masking ratio and customized masking strategy in VideoMAE. In this paper, we aim to further improve the performance of video masked autoencoding by introducing a motion guided masking strategy. Our key insight is that motion is a general and unique prior in video, which should be taken into account during masked pre-training. Our motion guided masking explicitly incorporates motion information to build temporal consistent masking volume. Based on this masking volume, we can track the unmasked tokens in time and sample a set of temporal consistent cubes from videos. These temporal aligned unmasked tokens will further relieve the information leakage issue in time and encourage the MGMAE to learn more useful structure information. We implement our MGMAE with an online efficient optical flow estimator and backward masking map warping strategy. We perform experiments on the datasets of Something-Something V2 and Kinetics-400, demonstrating the superior performance of our MGMAE to the original VideoMAE. In addition, we provide the visualization analysis to illustrate that our MGMAE can sample temporal consistent cubes in a motion-adaptive manner for more effective video pre-training.

Shenghao Fu, Junkai Yan, Yipeng Gao, Xiaohua Xie, Wei-Shi Zheng

Recent sparse detectors with multiple, e.g. six, decoder layers achieve promising performance but much inference time due to complex heads. Previous works have explored using dense priors as initialization and built one-decoder-layer detectors. Although they gain remarkable acceleration, their performance still lags behind their six-decoder-layer counterparts by a large margin. In this work, we aim to bridge this performance gap while retaining fast speed. We find that the architecture discrepancy between dense and sparse detectors leads to feature conflict, hampering the performance of one-decoder-layer detectors. Thus we propose Adaptive Sparse Anchor Generator (ASAG) which predicts dynamic anchors on patches rather than grids in a sparse way so that it alleviates the feature conflict problem. For each image, ASAG dynamically selects which feature maps and which locations to predict, forming a fully adaptive way to generate image-specific anchors. Further, a simple and effective Query Weighting method eases the training instability from adaptiveness. Extensive experiments show that our method outperforms dense-initialized ones and achieves a better speed-accuracy trade-off. The code is available at https://github.com/iSEE-Laboratory/ASAG.

Guhnoo Yun, Juhan Yoo, Kijung Kim, Jeongho Lee, Dong Hwan Kim

Recent studies show that self-attentions behave like low-pass filters (as opposed to convolutions) and enhancing their high-pass filtering capability improves model performance. Contrary to this idea, we investigate existing convolution-based models with spectral analysis and observe that improving the low-pass filtering in convolution operations also leads to performance improvement. To account for this observation, we hypothesize that utilizing optimal token mixers that capture balanced representations of both high- and low-frequency components can enhance the performance of models. We verify this by decomposing visual features into the frequency domain and combining them in a balanced manner. To handle this, we replace the balancing problem with a mask filtering problem in the frequency domain. Then, we introduce a novel token-mixer named SPAM and leverage it to derive a MetaFormer model termed as SPANet. Experimental results show that the proposed method provides a way to achieve this balance, and the balanced representations of both high- and low-frequency components can improve the performance of models on multiple computer vision tasks. Our code is available at https://doranlyong.github.io/projects/spanet/.

Jae-Hyeok Lee, Dae-Shik Kim

Neural Radiance Fields (NeRFs) have gained considerable attention for their high-quality results in 3D scene reconstruction and rendering. Recently, there have been active studies on various tasks such as novel view synthesis and scene editing. However, editing NeRFs is challenging as accurately decomposing the desired area of 3D space and ensuring the consistency of edited results from different angles is difficult. In this paper, we propose ICE-NeRF, an Interactive Color Editing framework that performs color editing by taking a pre-trained NeRF and a rough user mask as input. Our proposed method performs the entire color editing process in only under a minute using a partial fine-tuning approach. To perform effective color editing, we address two issues: (1) the entanglement of the implicit representation that causes unwanted color changes in undesired areas when learning weights, and (2) the loss of multi-view consistency when fine-tuning for a single or a few views. To address these issues, we introduce two techniques: Activation Field-based Regularization (AFR) and Single-mask Multi-view Rendering (SMR). The AFR performs weight regularization during fine-tuning based on the assumption that not all weights have an equal impact on the desired area. The SMR maps the 2D mask to 3D space through inverse projection and renders it from other views to generate multi-view masks. ICE-NeRF not only enables well-decomposed, multi-view consistent color editing but also significantly reduces processing time compared to existing methods.

Zhengyu Liang, Yingqian Wang, Longguang Wang, Jungang Yang, Shilin Zhou, Yulan Guo

Exploiting spatial-angular correlation is crucial to light field (LF) image super-resolution (SR), but is highly challenging due to its non-local property caused by the disparities among LF images. Although many deep neural networks (DNNs) have been developed for LF image SR and achieved continuously improved performance, existing methods cannot well leverage the long-range spatial-angular correlation and thus suffer a significant performance drop when handling scenes with large disparity variations. In this paper, we propose a simple yet effective method to learn the non-local spatial-angular correlation for LF image SR. In our method, we adopt the epipolar plane image (EPI) representation to project the 4D spatial-angular correlation onto multiple 2D EPI planes, and then develop a Transformer network with repetitive self-attention operations to learn the spatial-angular correlation by modeling the dependencies between each pair of EPI pixels. Our method can fully incorporate the information from all angular views while achieving a global receptive field along the epipolar line. We conduct extensive experiments with insightful visualizations to validate the effectiveness of our method. Comparative results on five public datasets show that our method not only achieves state-of-the-art SR performance, but also performs robust to disparity variations.

Yan Fang, Feng Zhu, Bowen Cheng, Luoqi Liu, Yao Zhao, Yunchao Wei

We investigate semi-supervised semantic segmentation with self-training, where a teacher model generates pseudo masks to exploit the benefits of a large amount of unlabeled images. We notice that the noisy label from the generated pseudo masks is the major obstacle to achieving good performance. Previous works all treat the noise in pixel level and ignore the contextual information of the noise. This work shows that locating the patch-wise noisy region is a better way to deal with noise. To be specific, our method, named Uncertainty-aware Patch CutMix (UPC), first estimates the uncertainty of per-pixel prediction for pseudo masks of unlabeled images. Then UPC splits the uncertainty map into patches and calculates patch-wise uncertainty. UPC selects top-k most uncertain patches to generate the uncertain regions. Finally, uncertain regions are replaced with reliable ones from labeled images. We conduct extensive experiments using standard semi-supervised settings on Pascal VOC and Cityscapes. Experiment results show that UPC can significantly boost the performance of the state-of-the-art methods. In addition, we further demonstrate that our UPC is robust to out-of-distribution unlabeled images, eg, MSCOCO.

Jianfeng Xiang, Jiaolong Yang, Binbin Huang, Xin Tong

In this paper, we introduce a novel 3D-aware image generation method that leverages 2D diffusion models. We formulate the 3D-aware image generation task as multiview 2D image set generation, and further to a sequential unconditional-conditional multiview image generation process. This allows us to utilize 2D diffusion models to boost the generative modelling power of the method. Additionally, we incorporate depth information from monocular depth estimators to construct the training data for the conditional diffusion model using only still images. We train our method on a large-scale unstructured dataset, i.e., ImageNet, which is not addressed by previous methods. It produces high-quality images that significantly outperform prior methods. Furthermore, our approach showcases its capability to generate instances with large view angles, even though the training images are diverse and unaligned, gathered from "in-the-wild" real-world environments.

Yifeng Zhang, Shi Chen, Qi Zhao

Answering visual questions requires the ability to parse visual observations and correlate them with a variety of knowledge. Existing visual question answering (VQA) models either pay little attention to the role of knowledge or do not take into account the granularity of knowledge, e.g., attaching the color of "grassland" to "ground"). They have yet to develop the capability of modeling knowledge of multiple granularity, and are also vulnerable to spurious data biases. To fill the gap, this paper makes progresses from two distinct perspectives: (1) It presents a Hierarchical Concept Graph (HCG) that discriminates and associates multi-granularity concepts with a multi-layered hierarchical structure, aligning visual observations with knowledge across different levels to alleviate data biases. (2) To facilitate a comprehensive understanding of how knowledge contributes throughout the decision-making process, we further propose an interpretable Hierarchical Concept Neural Module Network (HCNMN) It explicitly propagates multi-granularity knowledge across the hierarchical structure and incorporates them with a sequence of reasoning steps, providing a transparent interface to elaborate on the integration of observations and knowledge. Through extensive experiments on multiple challenging datasets (i.e., GQA,VQA,FVQA,OK-VQA) , we demonstrate the effectiveness of our method in answering questions in different scenarios. Our code is available at https://github.com/SuperJohnZhang/HCNMN.