论文检索

Xingzhi Zhou, Zhiliang Tian, Boyang Zhang, Yibo Zhang, Ka Chun Cheung, Simon See, Hao Yang, Yun Zhou, Nevin L. Zhang

Test-time adaptation (TTA) deals with domain shifts during inference by training models based on only unlabeled test samples. Test samples may include noisy samples, which degrade domain adaptation. Existing methods rely on the model's output prediction to detect and filter noisy samples, and further search for flat regions during optimization, which makes the optimization more robust on noisy samples. However, there are two issues: (1) the output prediction tends to be inaccurate due to domain shifts, weakening noisy-sample detection; (2) current approaches for searching flat regions focus on optimization to enhance the worst case, which ignores achieving flatness by avoiding the quick changing of losses. To address these challenges, we propose a model pruning-based test-time adaptation model for noisy data streams, named MoTTA, which leverages a new proposed filtering, output difference under pruning (ODP)-based filtering, and a flatness-aware entropy minimization (FlatEM). Specifically, to reduce the impact of inaccurate output predictions, ODP-based filtering measures the output difference of a sample before and after model pruning, which works even under inaccurate output. To improve the search for flat loss surfaces, FlatEM integrates zeroth-order flatness and first-order flatness (minimize the maximal gradient normalization with a weight perturbation constrained in a small Euclidean ball) on entropy minimization. To solve these hard maximum problems, we leverage Taylor expansion to obtain approximated results for optimization. FlatEM also adopts a parameter regularization to mitigate incorrect updates from noisy samples. The experiments show our advantages in dealing with noisy data streams at TTA comparable to existing baselines.

Qingyuan Zhou, Weidong Yang, Ben Fei, Jingyi Xu, Rui Zhang, Keyi Liu, Yeqi Luo, Ying He

Noise is an inevitable aspect of point cloud acquisition, necessitating filtering as a fundamental task within the realm of 3D vision. Existing learning-based filtering methods have shown promising capabilities on commonly used datasets. Nonetheless, the effectiveness of these methods is constrained when dealing with a substantial quantity of point clouds. This limitation primarily stems from their limited denoising capabilities for dense and large-scale point clouds and their inclination to generate noisy outliers after denoising. To deal with this challenge, we introduce 3DMambaIPF, for the first time, exploiting Selective State Space Models (SSMs) architecture to handle highly-dense and large-scale point clouds, capitalizing on its strengths in selective input processing and large context modeling capabilities. Additionally, we present a robust and fast differentiable rendering loss to constrain the noisy points around the surface. In contrast to previous methodologies, this differentiable rendering loss enhances the visual realism of denoised geometric structures and aligns point cloud boundaries more closely with those observed in real-world objects. Extensive evaluations on commonly used datasets (typically with up to 50K points) demonstrate that 3DMambaIPF achieves state-of-the-art results. Moreover, we showcase the superior scalability and efficiency of 3DMambaIPF on highly dense and large-scale point clouds with up to 500K points compared to off-the-shelf methods.

Qingyang Zhou, Yunfan Ye, Zhiping Cai

We propose a dynamic Computed Tomography (CT) reconstruction framework called STNF4D (SpatioTemporal-aware Neural Fields). First, we represent the 4D scene using four orthogonal volumes and compress these volumes into more compact hash grids. Compared to the plane decomposition method, this method enhances the model's capacity while keeping the representation compact and efficient. However, in densely predicted high-resolution dynamic CT scenes, the lack of constraints and hash conflicts in the hash grid features lead to obvious dot-like artifact and blurring in the reconstructed images. To address these issues, we propose the Spatiotemporal Transformer (ST-Former) that guides the model in selecting and optimizing features by sensing the spatiotemporal information in different hash grids, significantly improving the quality of reconstructed images. We conducted experiments on medical and industrial datasets covering various motion types, sampling modes, and reconstruction resolutions. Experimental results show that our method outperforms the second-best by 5.99 dB and 4.11 dB in medical and industrial scenes, respectively.

Nuoyan Zhou, Dawei Zhou, Decheng Liu, Nannan Wang, Xinbo Gao

Adversarial fine-tuning methods enhance adversarial robustness via fine-tuning the pre-trained model in an adversarial training manner. However, we identify that some specific latent features of adversarial samples are confused by adversarial perturbation and lead to an unexpectedly increasing gap between features in the last hidden layer of natural and adversarial samples. To address this issue, we propose a disentanglement-based approach to explicitly model and further remove the specific latent features. We introduce a feature disentangler to separate out the specific latent features from the features of the adversarial samples, thereby boosting robustness by eliminating the specific latent features. Besides, we align clean features in the pre-trained model with features of adversarial samples in the fine-tuned model, to benefit from the intrinsic features of natural samples. Empirical evaluations on three benchmark datasets demonstrate that our approach surpasses existing adversarial fine-tuning methods and adversarial training baselines.

MingSheng Zhou, MingMing Kong

Learned image lossy compression techniques have surpassed traditional methods in both subjective vision and quantitative evaluation. However, current models are only applicable to three-channel image formats, limiting their practical application due to the diversity and complexity of image formats. We propose a high-performance learned image compression model for general image formats. We first introduce a transfer method to unify any-channel image formats, enhancing the applicability of neural networks. This method's effectiveness is demonstrated through image information entropy and image homomorphism theory. Then, we introduce an adaptive attention residual block into the entropy model to give it better generalization ability. Meanwhile, we propose an evenly grouped cross-channel context module for progressive preview image decoding. Experimental results demonstrate that our method achieves state-of-the-art (SOTA) in the field of learned image compression in terms of PSNR and MS-SSIM. This work extends the applicability of learned image compression techniques to more practical production environments.

Mengqi Zhou, Yuxi Wang, Jun Hou, Shougao Zhang, Yiwei Li, Chuanchen Luo, Junran Peng, Zhaoxiang Zhang

Developing comprehensive explicit world models is crucial for understanding and simulating real-world scenarios. Recently, Procedural Controllable Generation (PCG) has gained significant attention in large-scale scene generation by enabling the creation of scalable, high-quality assets. However, PCG faces challenges such as limited modular diversity, high expertise requirements, and challenges in managing the diverse elements and structures in complex scenes. In this paper, we introduce a large-scale scene generation framework, SceneX, which can automatically produce high-quality procedural models according to designers' textual descriptions. Specifically, the proposed method comprises two components, PCGHub and PCGPlanner. The former encompasses an extensive collection of accessible procedural assets and thousands of hand-craft API documents to perform as a standard protocol for PCG controller. The latter aims to generate executable actions for Blender to produce controllable and precise 3D assets guided by the user's instructions. Extensive experiments demonstrated the capability of our method in controllable large-scale scene generation, including nature scenes and unbounded cities, as well as scene editing such as asset placement and season translation.

Jingyi Zhou, Haoyu Zhang, Jiakang Yuan, Peng Ye, Tao Chen, Hao Jiang, Meiya Chen, Yangyang Zhang

As a fundamental vision task, stereo matching has made remarkable progress. While recent iterative optimization-based methods have achieved promising performance, their feature extraction capabilities still have room for improvement. Inspired by the ability of vision foundation models (VFMs) to extract general representations, in this work, we propose AIO-Stereo which can flexibly select and transfer knowledge from multiple heterogeneous VFMs to a single stereo matching model. To better reconcile features between heterogeneous VFMs and the stereo matching model and fully exploit prior knowledge from VFMs, we proposed a dual-level feature utilization mechanism that aligns heterogeneous features and transfers multi-level knowledge. Based on the mechanism, a dual-level selective knowledge transfer module is designed to selectively transfer knowledge and integrate the advantages of multiple VFMs. Experimental results show that AIO-Stereo achieves start-of-the-art performance on multiple datasets and ranks 1st on the Middlebury dataset and outperforms all the published work on the ETH3D benchmark.

Jingqiu Zhou, Lue Fan, Xuesong Chen, Linjiang Huang, Si Liu, Hongsheng Li

In this paper, we present GaussianPainter, the first method to paint a point cloud into 3D Gaussians given a reference image. GaussianPainter introduces an innovative feed-forward approach to overcome the limitations of time-consuming test-time optimization in 3D Gaussian splatting. Our method addresses a critical challenge in the field: the non-uniqueness problem inherent in the large parameter space of 3D Gaussian splatting. This space, encompassing rotation, anisotropic scales, and spherical harmonic coefficients, introduces the challenge of rendering similar images from substantially different Gaussian fields. As a result, feed-forward networks face instability when attempting to directly predict high-quality Gaussian fields, struggling to converge on consistent parameters for a given output. To address this issue, we propose to estimate a surface normal for each point to determine its Gaussian rotation. This strategy enables the network to effectively predict the remaining Gaussian parameters in the constrained space. We further enhance our approach with an appearance injection module, incorporating reference image appearance into Gaussian fields via a multiscale triplane representation. Our method successfully balances efficiency and fidelity in 3D Gaussian generation, achieving high-quality, diverse, and robust 3D content creation from point clouds in a single forward pass. A video is provided in our supplementary material for a more detailed explanation of our method.

Jiaying Zhou, Qingchao Chen

Novel class discovery(NCD) aims to cluster the unlabeled data with the help of a labeled set containing different but related classes. The key to solving NCD is the knowledge transfer between labeled and unlabeled sets.Since NCD requires that known classes and unknown classes are related, it is significant to explore class-level relationships between known and unknown for more effective knowledge transfer. However, most existing methods either facilitate knowledge transfer by learning a shared representation space or by modeling coarse-grained or asymmetric relationships between known and unknown, neglecting class-level relationships. To tackle these challenges, we propose a symmetric class-to-class relationship modeling and knowledge transfer method, achieving bidirectional knowledge transfer at class-level. Considering that class-level modeling often overlooks the subtle distinctions between samples, we propose pairwise similarity-based relationship modeling and consistency constraint for instance-level knowledge transfer. Extensive experiments on CIFAR100 and three fine-grained datasets demonstrate that our method achieves significant performance improvements compared to state-of-the-art methods.

Hao Zhou, Tingjin Luo, Zhangqi Jiang

Compositional visual question answering (Compositional VQA) needs to provide an answer to a compositional question, which requires the model to have advanced capabilities of multi-modal semantic understanding and logical reasoning. However, current VQA models mainly concentrate on enriching the visual representations of images and neglect the redundancy in the enriched information to bring some negative impacts. To enhance the value and availability of semantic features, we propose a novel core-to-global reasoning (CTGR) model for compositional VQA. The model first extracts both global features and core features from image and question through a feature embedding module. Then, to enhance the value of semantic features, we propose an information filtering module to align visual features and text features at the core semantic level and to filter out the redundancy carried by image and question features at the global semantic level, which can further strengthen cross-modal correlations. Besides, we design a novel core-to-global reasoning mechanism for multimodal fusion, which integrates content features from core learning and context features from global features for accurate answer predictions. Finally, extensive experimental results on GQA, GQA-sub, VQA2.0 and Visual7W demonstrate the effectiveness and superiority of CTGR.

Hang Zhou, Jiale Cai, Yuteng Ye, Yonghui Feng, Chenxing Gao, Junqing Yu, Zikai Song, Wei Yang

A recent endeavor in one class of video anomaly detection is to leverage diffusion models and posit the task as a generation problem, where the diffusion model is trained to recover normal patterns exclusively, thus reporting abnormal patterns as outliers. Yet, existing attempts neglect the various formations of anomaly and predict normal samples at the feature level regardless that abnormal objects in surveillance videos are often relatively small. To address this, a novel patch-based diffusion model is proposed, specifically engineered to capture fine-grained local information. We further observe that anomalies in videos manifest themselves as deviations in both appearance and motion. Therefore, we argue that a comprehensive solution must consider both of these aspects simultaneously to achieve accurate frame prediction. To address this, we introduce innovative motion and appearance conditions that are seamlessly integrated into our patch diffusion model. These conditions are designed to guide the model in generating coherent and contextually appropriate predictions for both semantic content and motion relations. Experimental results on four challenging video anomaly detection datasets empirically substantiate the efficacy of our proposed approach, demonstrating that it consistently outperforms most existing methods in detecting abnormal behaviors.

Han Zhou, Wei Dong, Xiaohong Liu, Yulun Zhang, Guangtao Zhai, Jun Chen

Although significant progress has been made in enhancing visibility, retrieving texture details, and mitigating noise in Low-Light (LL) images, the challenge persists in applying current Low-Light Image Enhancement (LLIE) methods to real-world scenarios, primarily due to the diverse illumination conditions encountered. Furthermore, the quest for generating enhancements that are visually realistic and attractive remains an underexplored realm. In response to these challenges, we present a novel LLIE framework with the guidance of Generative Perceptual Priors (GPP-LLIE) derived from vision-language models (VLMs). Specifically, we first propose a pipeline that guides VLMs to assess multiple visual attributes of the LL image and quantify the assessment to output the global and local perceptual priors. Subsequently, to incorporate these generative perceptual priors to benefit LLIE, we introduce a transformer-based backbone in the diffusion process, and develop a new layer normalization (GPP-LN) and an attention mechanism (LPP-Attn) guided by global and local perceptual priors. Extensive experiments demonstrate that our model outperforms current SOTA methods on paired LL datasets and exhibits superior generalization on real-world data.

Haitao Zhou, Chuang Wang, Rui Nie, Jinlin Liu, Dongdong Yu, Qian Yu, Changhu Wang

Recent years have seen substantial progress in diffusion-based controllable video generation. However, achieving precise control in complex scenarios, including fine-grained object parts, sophisticated motion trajectories, and coherent background movement, remains a challenge. In this paper, we introduce *TrackGo*, a novel approach that leverages free-form masks and arrows for conditional video generation. This method offers users with a flexible and precise mechanism for manipulating video content. We also propose the *TrackAdapter* for control implementation, an efficient and lightweight adapter designed to be seamlessly integrated into the temporal self-attention layers of a pretrained video generation model. This design leverages our observation that the attention map of these layers can accurately activate regions corresponding to motion in videos. Our experimental results demonstrate that our new approach, enhanced by the TrackAdapter, achieves state-of-the-art performance on key metrics such as FVD, FID, and ObjMC scores.

Feng Zhou, Qi Zhang, Ju Dai, Lei Li, Qing Fan, Junliang Xing

Point cloud completion aims to recover partial geometric and topological shapes caused by equipment defects or limited viewpoints. Current methods either solely rely on the 3D coordinates of the point cloud to complete it or incorporate additional images with well-calibrated intrinsic parameters to guide the geometric estimation of the missing parts. Although these methods have achieved excellent performance by directly predicting the location of complete points, the extracted features lack fine-grained information regarding the location of the missing area. To address this issue, we propose a rapid and efficient method to expand an unimodal framework into a multimodal framework. This approach incorporates a position-aware module designed to enhance the spatial information of the missing parts through a weighted map learning mechanism. In addition, we establish a Point-Text-Image triplet corpus PCI-TI and MVP-TI based on the existing unimodal point cloud completion dataset and use the pre-trained vision-language model CLIP to provide richer detail information for 3D shapes, thereby enhancing performance. Extensive quantitative and qualitative experiments demonstrate that our method outperforms state-of-the-art point cloud completion methods.

Chuqin Zhou, Guo Lu, Jiangchuan Li, Xiangyu Chen, Zhengxue Cheng, Li Song, Wenjun Zhang

Neural image compression often faces a challenging trade-off among rate, distortion and perception. While most existing methods typically focus on either achieving high pixel-level fidelity or optimizing for perceptual metrics, we propose a novel approach that simultaneously addresses both aspects for a fixed neural image codec. Specifically, we introduce a plug-and-play module at the decoder side that leverages a latent diffusion process to transform the decoded features, enhancing either low distortion or high perceptual quality without altering the original image compression codec. Our approach facilitates fusion of original and transformed features without additional training, enabling users to flexibly adjust the balance between distortion and perception during inference. Extensive experimental results demonstrate that our method significantly enhances the pretrained codecs with a wide, adjustable distortion-perception range while maintaining their original compression capabilities. For instance, we can achieve more than 150% improvement in LPIPS-BDRate without sacrificing more than 1 dB in PSNR.

Chu Zhou, Chao Xu, Boxin Shi

Shadow is a phenomenon that degenerates image quality and decreases the performance of downstream vision algorithms. Despite the fact that current image shadow removal methods have achieved promising progress, many of them require an externally obtained shadow mask as a necessary part of the input data, which not only introduces additional workload but also leads to degenerated performance near the shadow boundary due to the inaccuracy of the mask. Some of them do not require the shadow mask, however, they need to simultaneously consider the restoration of the brightness and color information along with the preservation of the texture and structure information inside the shadow region without external clues, which poses highly ill-posedness and makes the results prone to artifacts. In this paper, we propose Pol-ShaRe, the first Polarization-guided image Shadow Removal solution, to remove shadow in a mask-free manner with fewer artifacts. Specifically, it consists of a two-stage pipeline to relieve the ill-posedness and a neural network tailored to the pipeline to suppress the artifacts. Experimental results show that our Pol-ShaRe achieves state-of-the-art performance on both synthetic and real-world images.

Baichuan Zhou, Haote Yang, Dairong Chen, Junyan Ye, Tianyi Bai, Jinhua Yu, Songyang Zhang, Dahua Lin, Conghui He, Weijia Li

Recent evaluations of Large Multimodal Models (LMMs) have explored their capabilities in various domains, with only few benchmarks specifically focusing on urban environments. Moreover, existing urban benchmarks have been limited to evaluating LMMs with basic region-level urban tasks under singular views, leading to incomplete evaluations of LMMs' abilities in urban environments. To address these issues, we present UrBench, a comprehensive benchmark designed for evaluating LMMs in complex multi-view urban scenarios. UrBench contains 11.6K meticulously curated questions at both region-level and role-level that cover 4 task dimensions: Geo-Localization, Scene Reasoning, Scene Understanding, and Object Understanding, totaling 14 task types. In constructing UrBench, we utilize data from existing datasets and additionally collect data from 11 cities, creating new annotations using a cross-view detection-matching method. With these images and annotations, we then integrate LMM-based, rule-based, and human-based methods to construct large-scale high-quality questions. Our evaluations on 21 LMMs show that current LMMs struggle in the urban environments in several aspects. Even the best performing GPT-4o lags behind humans in most tasks, ranging from simple tasks such as counting to complex tasks such as orientation, localization and object attribute recognition, with an average performance gap of 17.4%. Our benchmark also reveals that LMMs exhibit inconsistent behaviors with different urban views, especially with respect to understanding cross-view relations.

Zhizhou Zhong, Yuxi Mi, Yuge Huang, Jianqing Xu, Guodong Mu, Shouhong Ding, Jingyun Zhang, Rizen Guo, Yunsheng Wu, Shuigeng Zhou

Contemporary face recognition systems use feature templates extracted from face images to identify persons. To enhance privacy, face template protection techniques are widely employed to conceal sensitive identity and appearance information stored in the template. This paper identifies an emerging privacy attack form utilizing diffusion models that could nullify prior protection. The attack can synthesize high-quality, identity-preserving face images from templates, revealing persons' appearance. Based on studies of the diffusion model's generative capability, this paper proposes a defense by rotating templates to a noise-like distribution. This is achieved efficiently by spherically and linearly interpolating templates on their located hypersphere. This paper further proposes to group-wisely divide and drop out templates' feature dimensions, to enhance the irreversibility of rotated templates. The proposed techniques are concretized as a novel face template protection technique, SlerpFace. Extensive experiments show that SlerpFace provides satisfactory recognition accuracy and comprehensive protection against inversion and other attack forms, superior to prior arts.

Yi Zhong, Weize Quan, Dong-Ming Yan, Jie Jiang, Yingmei Wei

Point cloud completion aims to reconstruct the complete 3D shape from incomplete point clouds, and it is crucial for tasks such as 3D object detection and segmentation. Despite the continuous advances in point cloud analysis techniques, feature extraction methods are still confronted with apparent limitations. The sparse sampling of point clouds, used as inputs in most methods, often results in a certain loss of global structure information. Meanwhile, traditional local feature extraction methods usually struggle to capture the intricate geometric details. To overcome these drawbacks, we introduce PointCFormer, a transformer framework optimized for robust global retention and precise local detail capture in point cloud completion. This framework embraces several key advantages. First, we propose a relation-based local feature extraction method to perceive local delicate geometry characteristics. This approach establishes a fine-grained relationship metric between the target point and its k-nearest neighbors, quantifying each neighboring point's contribution to the target point's local features. Secondly, we introduce a progressive feature extractor that integrates our local feature perception method with self-attention. Starting with a denser sampling of points as input, it iteratively queries long-distance global dependencies and local neighborhood relationships. This extractor maintains enhanced global structure and refined local details, without generating substantial computational overhead. Additionally, we develop a correction module after generating point proxies in the latent space to reintroduce denser information from the input points, enhancing the representation capability of the point proxies. PointCFormer demonstrates state-of-the-art performance on several widely used benchmarks.

Yaoyao Zhong, Mengshi Qi, Rui Wang, Yuhan Qiu, Yang Zhang, Huadong Ma

Video Internet of Things (VIoT) has shown full potential in collecting an unprecedented volume of video data. How to schedule the domain-specific perceiving models and analyze the collected videos uniformly, efficiently, and especially intelligently to accomplish complicated tasks is challenging. To address the challenge, we build VIoTGPT, the framework based on LLMs to correctly interact with humans, query knowledge videos, and invoke vision models to analyze multimedia data collaboratively. To support VIoTGPT and related future works, we meticulously crafted the VIoT-Tool dataset, including the training dataset and the benchmark involving 11 representative vision models across three categories based on semi-automatic annotations. To guide LLM to act as the intelligent agent towards intelligent VIoT, we resort to ReAct instruction tuning method based on VIoT-Tool to learn the tool capability. Quantitative and qualitative experiments and analyses demonstrate the effectiveness of VIoTGPT. We believe VIoTGPT contributes to improving human-centered experiences in VIoT applications.