论文检索

Chaoran Feng, Wangbo Yu, Xinhua Cheng, Zhenyu Tang, Junwu Zhang, Li Yuan, Yonghong Tian

Compared to frame-based methods, computational neuromorphic imaging using event cameras offers significant advantages, such as minimal motion blur, enhanced temporal resolution, and high dynamic range. The multi-view consistency of Neural Radiance Fields combined with the unique benefits of event cameras, has spurred recent research into reconstructing NeRF from data captured by moving event cameras. While showing impressive performance, existing methods rely on ideal conditions with the availability of uniform and high-quality event sequences and accurate camera poses, and mainly focus on the object level reconstruction, thus limiting their practical applications. In this work, we propose AE-NeRF to address the challenges of learning event-based NeRF from non-ideal conditions, including non-uniform event sequences, noisy poses, and various scales of scenes. Our method exploits the density of event streams and jointly learn a pose correction module with an event-based NeRF (e-NeRF) framework for robust 3D reconstruction from inaccurate camera poses. To generalize to larger scenes, we propose hierarchical event distillation with a proposal e-NeRF network and a vanilla e-NeRF network to resample and refine the reconstruction process. We further propose an event reconstruction loss and a temporal loss to improve the view consistency of the reconstructed scene. We established a comprehensive benchmark that includes large-scale scenes to simulate practical non-ideal conditions, incorporating both synthetic and challenging real-world event datasets. The experimental results show that our method achieves a new state-of-the-art in event-based 3D reconstruction.

Xiang Fang, Wanlong Fang, Changshuo Wang, Daizong Liu, Keke Tang, Jianfeng Dong, Pan Zhou, Beibei Li

Given some video-query pairs with untrimmed videos and sentence queries, temporal sentence grounding (TSG) aims to locate query-relevant segments in these videos. Although previous respectable TSG methods have achieved remarkable success, they train each video-query pair separately and ignore the relationship between different pairs. To this end, in this paper, we pose a brand-new setting: Multi-Pair TSG, which aims to co-train these pairs. We propose a novel video-query co-training approach, Multi-Thread Knowledge Transfer Network, to locate a variety of video-query pairs effectively and efficiently. Firstly, we mine the spatial and temporal semantics across different queries to cooperate with each other. To learn intra- and inter-modal representations simultaneously, we design a cross-modal contrast module to explore the semantic consistency by a self-supervised strategy. To fully align visual and textual representations between different pairs, we design a prototype alignment strategy to 1) match object prototypes and phrase prototypes for spatial alignment, and 2) align activity prototypes and sentence prototypes for temporal alignment. Finally, we develop an adaptive negative selection module to adaptively generate a threshold for cross-modal matching. Extensive experiments show the effectiveness and efficiency of our proposed method.

Wenxuan Fang, Junkai Fan, Yu Zheng, Jiangwei Weng, Ying Tai, Jun Li

Image dehazing, particularly with learning-based methods, has gained significant attention due to its importance in real-world applications. However, relying solely on the RGB color space often fall short, frequently leaving residual haze. This arises from two main issues: the difficulty in obtaining clear textural features from hazy RGB images and the complexity of acquiring real haze/clean image pairs outside controlled environments like smoke-filled scenes. To address these issues, we first propose a novel Structure Guided Dehazing Network (SGDN) that leverages the superior structural properties of YCbCr features over RGB. It comprises two key modules: Bi-Color Guidance Bridge (BGB) and Color Enhancement Module (CEM). BGB integrates a phase integration module and an interactive attention module, utilizing the rich texture features of the YCbCr space to guide the RGB space, thereby recovering clearer features in both frequency and spatial domains. To maintain tonal consistency, CEM further enhances the color perception of RGB features by aggregating YCbCr channel information. Furthermore, for effective supervised learning, we introduce a Real-World Well-Aligned Haze dataset, which includes a diverse range of scenes from various geographical regions and climate conditions. Experimental results demonstrate that our method surpasses existing state-of-the-art methods across multiple real-world smoke/haze datasets.

Shijie Fang, Hongping Gan

Deep Unfolding Networks (DUNs), with their outstanding performance and partial interpretability, have revitalized the field of pan-sharpening. However, the current DUNs for pan-sharpening rely entirely on implicit deep priors, ignoring the intrinsic physical prior knowledge of multispectral image (MS) and panchromatic image (PAN) to guide the reconstruction process. Moreover, these methods often depend on single-scale prior features, failing to adequately capture multiscale information, resulting in spatial and spectral distortions in detail. In this paper, we introduce a spatial-spectral prior-aware framework for pan-sharpening, called SSPF, which formulates a constrained minimization problem integrating MS and PAN prior knowledge based on spatial and spectral domains. We further develop SSPF into a lightweight deep unfolding network, called SSUN-Net, which provides more efficient prior feature extraction and requires less computational cost. Additionally, we augment SSUN-Net's capabilities by integrating a customized multi-scale prior structure (MPS). MPS imposes constraints on the solution space at various scales through regularization, which markedly enhances the reconstruction of intricate details. Extensive experiments demonstrate the significant advantages of our proposed SSUN-Net over the current SOTA methods.

Han Fang, Kejiang Chen, Zijin Yang, Bosen Cui, Weiming Zhang, Ee-Chien Chang

Generative image watermarking inserts secret watermarks into generated images and plays an important role in tracing the usages of generative models. For watermarking of diffusion models, inversion-based framework emerges as an effective approach. Such framework employs a robust mechanism to embed the watermark into the starting latent before ``forward sampling'', thereby generating images with the implicit watermark. During watermark detection, inversion techniques are employed to reverse the process and obtain the watermarked latent, followed by further extraction. The robustness of this technique hinges primarily on the embedding mechanism and inversion accuracy. Previous methods predominantly focused on enhancing the robustness of the embedding mechanism but overlooked the reduction of the inversion errors. However, our results show that inversion error will significantly affect the overall robustness. Therefore, in this paper, we delve into the inversion error aspect and propose CoSDA, a compensation sampling and drift alignment-based approach. The inversion error primarily accumulated during two stages: the internal error incurred by the algorithm, and the inevitable external noise. We observe that the main source of internal error comes from the mismatch in conditions (e.g. prompt, guidance scale) between forward and backward sampling processes. Therefore, we propose a compensation-based forward sampling, compensating for certain mismatch conditions and reducing the inversion error caused by the mismatch. Addressing external error caused by inevitable image distortions (e.g. JPEG compression), we introduce a drift-alignment approach, where a neural network is trained adversarially to restore the original watermarked latent from the distorted counterpart. Experimental results show that CoSDA effectively enhances watermark robustness while maintaining the visual quality of generated images.

Zhen Fan, Peng Dai, Zhuo Su, Xu Gao, Zheng Lv, Jiarui Zhang, Tianyuan Du, Guidong Wang, Yang Zhang

Egocentric human pose estimation (HPE) using wearable sensors is essential for VR/AR applications. Most methods rely solely on either egocentric-view images or sparse Inertial Measurement Unit (IMU) signals, leading to inaccuracies due to self-occlusion in images or the sparseness and drift of inertial sensors. Most importantly, the lack of real-world datasets containing both modalities is a major obstacle to progress in this field. To overcome the barrier, we propose EMHI, a multimodal Egocentric human Motion dataset with Head-Mounted Display (HMD) and body-worn IMUs, with all data collected under the real VR product suite. Specifically, EMHI provides synchronized stereo images from downward-sloping cameras on the headset and IMU data from body-worn sensors, along with pose annotations in SMPL format. This dataset consists of 885 sequences captured by 58 subjects performing 39 actions, totaling about 28.5 hours of recording. We evaluate the annotations by comparing them with optical marker-based SMPL fitting results. To substantiate the reliability of our dataset, we introduce MEPoser, a new baseline method for multimodal egocentric HPE, which employs a multimodal fusion encoder, temporal feature encoder, and MLP-based regression heads. The experiments on EMHI show that MEPoser outperforms existing single-modal methods and demonstrates the value of our dataset in solving the problem of egocentric HPE. We believe the release of EMHI and the method could advance the research of egocentric HPE and expedite the practical implementation of this technology in VR/AR products.

Wenxiao Fan, Kan Li

Noisy labels can negatively impact the performance of deep neural networks. One common solution is label refurbishment, which involves reconstructing noisy labels through predictions and distributions. However, these methods may introduce problematic semantic associations, a phenomenon that we identify as Semantic Contamination. Through an analysis of Robust LR, a representative label refurbishment method, we found that utilizing the logits of views for refurbishment does not adequately balance the semantic information of individual classes. Conversely, using the logits of models fails to maintain consistent semantic relationships across models, which explains why label refurbishment methods frequently encounter issues related to Semantic Contamination. To address this issue, we propose a novel method called Collaborative Cross Learning, which utilizes semi-supervised learning on refurbished labels to extract appropriate semantic associations from embeddings across views and models. Experimental results show that our method outperforms existing approaches on both synthetic and real-world noisy datasets, effectively mitigating the impact of label noise and Semantic Contamination.

Rui Fan, Weidong Hao, Juntao Guan, Lai Rui, Lin Gu, Tong Wu, Fanhong Zeng, Zhangming Zhu

Event Cameras offer appealing advantages, including power efficiency and ultra-low latency, driving forward advancements in edge applications. In order to leverage mature frame-based algorithms, most approaches typically compute dense, image-like representations from sparse, asynchronous events. However, they are often unable to capture comprehensive information or are computationally intensive, which hinders the edge deployment of event-based vision. Meanwhile, pillar-based paradigms have been proven to be efficient and well established for dense representations of sparse data. Hence, from a novel pillar-based perspective, we present EventPillars, an efficient, comprehensive framework for dense event representations. To summarize, it (i) incorporates the Temporal Event Range to describe an intact temporal distribution, (ii) Activates the Event Polarities to explicitly record the scene dynamics, (iii) enhances the target awareness by a spatial attention prior from Normalized Event Density, (iv) can be plug-and-played into different downstream tasks. Extensive experiments show that our EventPillars records a new state-of-the-art precision on object recognition and detection datasets with surprisingly 9.2× and 4.5× lower computation and storage consumption. This brings a new insight into dense event representations and is promising to boost the edge deployment of event-based vision.

Junkai Fan, Kun Wang, Zhiqiang Yan, Xiang Chen, Shangbing Gao, Jun Li, Jian Yang

In this paper, we study the challenging problem of simultaneously removing haze and estimating depth from real monocular hazy videos. These tasks are inherently complementary: enhanced depth estimation improves dehazing via the atmospheric scattering model (ASM), while superior dehazing contributes to more accurate depth estimation through the brightness consistency constraint (BCC). To tackle these intertwined tasks, we propose a novel depth-centric learning framework that integrates the ASM model with the BCC constraint. Our key idea is that both ASM and BCC rely on a shared depth estimation network. This network simultaneously exploits adjacent dehazed frames to enhance depth estimation via BCC and uses the refined depth cues to more effectively remove haze through ASM. Additionally, we leverage a non-aligned clear video and its estimated depth to independently regularize the dehazing and depth estimation networks. This is achieved by designing two discriminator networks: D_MFIR enhances high-frequency details in dehazed videos, and D_MDR reduces the occurrence of black holes in low-texture regions. Extensive experiments demonstrate that the proposed method outperforms current state-of-the-art techniques in both video dehazing and depth estimation tasks, especially in real-world hazy scenes.

Haozhi Fan, Yuan Cao

Cross-modal retrieval, as an emerging field within multimedia research, has gained significant attention in recent years. Unsupervised cross-modal hashing methods are attractive due to their ability to capture latent relationships within the data without label supervision and to produce compact hash codes for high search efficiency. However, the text modality exhibits worse representation ability compared with the image modality, leading to weak guidance to construct the joint similarity matrix. Moreover, most unsupervised cross-modal hashing methods are based on pairwise similarities for training, resulting in non-aggregating data distribution in the hash space. In this paper, we propose a novel Vision-guided Text Mining for Unsupervised Cross-modal Hashing via Community Similarity Quantization, termed VTM-UCH. Specifically, we first find the one-to-one correspondence between each word and each vision (image or object) based on the Contrastive Language-Image Pre-training (CLIP) model and compute the text similarities according to the clustering of their corresponding visions. Then, we define the fine-grained object-level image similarities and design the joint similarity matrix based on the text and image similarities. Accordingly, we construct an undirected graph to compute the communities as the pseudo-centers and adjust the pairwise similarities to improve the hash codes distribution. The experimental results on two common datasets verify the accuracy improvements in comparison with state-of-the-art baselines.

Guodong Fan, Zishu Yao, Guang-Yong Chen, Jian-Nan Su, Min Gan

Retinex-based methods have become a general approach for solving low-light image enhancement (LLIE). However, traditional methods require post-processing of illumination (e.g., gamma correction), which lacks adaptability and disrupts the illumination structure. Retinex-based deep networks typically follow a ‘decomposition-adjustment-exposure control’ process, which is redundant and lacks robustness. One major issue is the inaccuracy in estimating and decomposing the initial illumination. Accurate initial illumination can prevent further post-processing instability. We propose IniRetinex, rethinking the Retinex-based LLIE method from the perspective of initialization. By using neural networks to provide reasonable initial illumination and solving for smooth illumination through optimization, higher performance LLIE is achieved. We construct a two-layer convolutional neural network to capture the low-frequency structure of the image, adaptively compensating for classical initial illumination and avoiding additional post-processing. The network requires no pre-training and can be implemented in an unsupervised manner with just a few iterations, making it highly efficient. Additionally, we propose a new illumination optimization strategy by introducing an additional proximal penalty term, improving illumination in areas with varying levels and enhancing image details. Extensive experiments on various low-light image datasets demonstrate that our method achieves state-of-the-art (SOTA) results on multiple benchmarks, offering higher stability and inference efficiency compared to current advanced methods.

Zheng-Peng Duan, Jiawei Zhang, Zheng Lin, Xin Jin, XunDong Wang, Dongqing Zou, Chun-Le Guo, Chongyi Li

Image retouching aims to enhance the visual quality of photos. Considering the different aesthetic preferences of users, the target of retouching is subjective. However, current retouching methods mostly adopt deterministic models, which not only neglects the style diversity in the expert-retouched results and tends to learn an average style during training, but also lacks sample diversity during inference. In this paper, we propose a diffusion-based method, named DiffRetouch. Thanks to the excellent distribution modeling ability of diffusion, our method can capture the complex fine-retouched distribution covering various visual-pleasing styles in the training data. Moreover, four image attributes are made adjustable to provide a user-friendly editing mechanism. By adjusting these attributes in specified ranges, users are allowed to customize preferred styles within the learned fine-retouched distribution. Additionally, the affine bilateral grid and contrastive learning scheme are introduced to handle the problem of texture distortion and control insensitivity respectively. Extensive experiments have demonstrated the superior performance of our method on visually appealing and sample diversity.

Zheng-Peng Duan, Jiawei Zhang, Siyu Liu, Zheng Lin, Chun-Le Guo, Dongqing Zou, Jimmy Ren, Chongyi Li

Seamlessly moving objects within a scene is a common requirement for image editing, but it is still a challenge for existing editing methods. Especially for real-world images, the occlusion situation further increases the difficulty. The main difficulty is that the occluded portion needs to be completed before movement can proceed. To leverage the real-world knowledge embedded in the pre-trained diffusion models, we propose a Diffusion-based framework specifically designed for Occluded Object Movement, named DiffOOM. The proposed DiffOOM consists of two parallel branches that perform object de-occlusion and movement simultaneously. The de-occlusion branch utilizes a background color-fill strategy and a continuously updated object mask to focus the diffusion process on completing the obscured portion of the target object. Concurrently, the movement branch employs latent optimization to place the completed object in the target location and adopts local text-conditioned guidance to integrate the object into new surroundings appropriately. Extensive evaluations across various metrics demonstrate the superior performance of our method, which is further validated by a comprehensive user study.

Chen Duan, Qianyi Jiang, Pei Fu, Jiamin Chen, Shengxi Li, Zining Wang, Shan Guo, Junfeng Luo

In the field of scene text spotting, previous OCR methods primarily relied on image encoders and pre-trained text information, but they often overlooked the advantages of incorporating human language instructions. To address this gap, we propose InstructOCR, an innovative instruction-based scene text spotting model that leverages human language instructions to enhance the understanding of text within images. Our framework employs both text and image encoders during training and inference, along with instructions meticulously designed based on text attributes. This approach enables the model to interpret text more accurately and flexibly. Extensive experiments demonstrate the effectiveness of our model and we achieve state-of-the-art results on widely used benchmarks. Furthermore, the proposed framework can be seamlessly applied to scene text VQA tasks. By leveraging instruction strategies during pre-training, the performance on downstream VQA tasks can be significantly improved, with a 2.6% increase on the TextVQA dataset and a 2.1% increase on the ST-VQA dataset. These experimental results provide insights into the benefits of incorporating human language instructions for OCR-related tasks.

Yongkun Du, Zhineng Chen, Caiyan Jia, Xieping Gao, Yu-Gang Jiang

Scene Text Recognition (STR) methods have demonstrated robust performance in word-level text recognition. However, in real applications the text image is sometimes long due to detected with multiple horizontal words. It triggers the requirement to build long text recognition models from readily available short (i.e., word-level) text datasets, which has been less studied previously. In this paper, we term this task Out of Length (OOL) text recognition. We establish the first Long Text Benchmark (LTB) to facilitate the assessment of different methods in long text recognition. Meanwhile, we propose a novel method called OOL Text Recognition with sub-String Matching (SMTR). SMTR comprises two cross-attention-based modules: one encodes a sub-string containing multiple characters into next and previous queries, and the other employs the queries to attend to the image features, matching the sub-string and simultaneously recognizing its next and previous character. SMTR can recognize text of arbitrary length by iterating the process above. To avoid being trapped in recognizing highly similar sub-strings, we introduce a regularization training to compel SMTR to effectively discover subtle differences between similar sub-strings for precise matching. In addition, we propose an inference augmentation strategy to alleviate confusion caused by identical sub-strings in the same text and improve the overall recognition efficiency. Extensive experimental results reveal that SMTR, even when trained exclusively on short text, outperforms existing methods in public short text benchmarks and exhibits a clear advantage on LTB.

Keyu Du, Hao Xu, Haipeng Li, Hong Qu, Chi-Wing Fu, Shuaicheng Liu

Scene-level point cloud registration is very challenging when considering dynamic foregrounds. Existing indoor datasets mostly assume rigid motions, so the trained models cannot robustly handle scenes with non-rigid motions. On the other hand, non-rigid datasets are mainly object-level, so the trained models cannot generalize well to complex scenes. This paper presents HybridReg, a new approach to 3D point cloud registration, learning uncertainty mask to account for hybrid motions: rigid for backgrounds and non-rigid/rigid for instance-level foregrounds. First, we build a scene-level 3D registration dataset, namely HybridMatch, designed specifically with strategies to arrange diverse deforming foregrounds in a controllable manner. Second, we account for different motion types and formulate a mask-learning module to alleviate the interference of deforming outliers. Third, we exploit a simple yet effective negative log-likelihood loss to adopt uncertainty to guide the feature extraction and correlation computation. To our best knowledge, HybridReg is the first work that exploits hybrid motions for robust point cloud registration. Extensive experiments show HybridReg's strengths, leading it to achieve state-of-the-art performance on both widely-used indoor and outdoor datasets.

Chenghu Du, Junyin Wang, Feng Yu, Shengwu Xiong

Efficiently applying fully supervised learning to virtual try-on tasks is challenging due to the lack of paired ground truth in available training samples. Recent works have achieved virtual try-ons by employing self-supervised learning-based inpainting paradigms. However, this approach is heavily dependent on the constraints of inpainting masks. An incorrect mask can mislead the generated results, while overly large mask areas can lose essential original information, thereby hindering the synthesis of high-quality results. To address these problems, we propose a latent diffusion model-based virtual try-on network that achieves fully supervised learning using the concept of cycle consistency and knowledge distillation. Specifically, we divide our approach into pretext and downstream tasks. In the pretext task, we generate a pseudo-label (pseudo-person image) to form paired training samples, which enables the downstream task to achieve fully supervised learning. To prevent the unreliable pseudo-person image from introducing irresponsible prior knowledge, we propose a noise-covering strategy, which aims at fully optimizing the pseudo-label to eliminate the impact of the incorrect inpainting mask as much as possible. Additionally, we propose a skin refinement loss to further enhance the generation of details in the skin region. Extended experiments demonstrate that our proposed method is superior to state-of-the-art methods.

Chenghu Du, Junyin Wang, Yi Rong, Feng Yu, Shengwu Xiong

A good garment try-on model should learn the transfer between different types of garments while satisfying: 1) high fidelity and 2) low inference speed. Existing methods address either of these two issues, limited processing speed or low generation quality. We directly use a lightweight encoder-decoder, ensuring faster speeds. To tackle the problem of lower image quality typically generated by lighter models, we present GarFast, a simplified, parser-free framework that optimizes the same lightweight network through a two-stage transformation of real data roles (from input to supervision), thereby greatly promoting model convergence. Specifically, first, we propose a correction strategy to prevent the difficulty of convergence caused by the lack of ground truth in the first stage. Second, we propose a fine-grained domain consistency to ensure that the results generated in the unsupervised first stage are highly realistic clothed human images. Finally, we propose a skin-variant refinement loss and a skinMix regularization to amplify texture differences and enhance the realism of skin-variant regions, thereby improving the quality of the generated skin. Extensive experiments thoroughly demonstrate that our method achieves high resolution, near real-time performance, and superior reconstruction quality compared to state-of-the-art approaches, with processing times of less than 0.03 seconds on an Nvidia A100.

Ziheng Ding, Xiaze Zhang, Qi Jing, Ying Cheng, Rui Feng

3D object detection in point clouds is critical in 3D computer vision, autonomous driving, and robotics. Existing point-based detectors, tailored to handle unstructured raw point clouds, often rely on simplistic sampling strategies to select a subset of points for local representation learning and detection. However, the diverse patterns exhibited by multiple types of point cloud data present a significant challenge to the universality of current detectors, particularly those captured by varied sensors (e.g., LiDAR and 4D Imaging Radar). In response to this challenge, we introduce an adaptable point-based single-stage 3D detector, AS-Det, engineered to excel on both LiDAR and 4D Radar point clouds. Specifically, we propose a novel active sampling strategy that actively mines object-related information to achieve efficient sampling and representation across different types of point clouds through end-to-end training. Additionally, we introduce a lightweight multi-scale center feature aggregation module to exploit multi-scale object context for precise and low-cost detection. By integrating the abovementioned modules, AS-Det achieves highly adaptive detection on various point clouds, encompassing different sensors and scales. Experimental results demonstrate the superior performance and adaptability of AS-Det on both LiDAR and 4D Radar point clouds.

Yanbo Ding, Shaobin Zhuang, Kunchang Li, Zhengrong Yue, Yu Qiao, Yali Wang

Despite recent advancements in text-to-image generation, most existing methods struggle to create images with multiple objects and complex spatial relationships in the 3D world. To tackle this limitation, we introduce a generic AI system, namely MUSES, for 3D-controllable image generation from user queries. Specifically, our MUSES develops a progressive workflow with three key components, including (1) Layout Manager for 2D-to-3D layout lifting, (2) Model Engineer for 3D object acquisition and calibration, (3) Image Artist for 3D-to-2D image rendering. By mimicking the collaboration of human professionals, this multi-modal agent pipeline facilitates the effective and automatic creation of images with 3D-controllable objects, through an explainable integration of top-down planning and bottom-up generation. Additionally, existing benchmarks lack detailed descriptions of complex 3D spatial relationships of multiple objects. To fill this gap, we further construct a new benchmark of T2I-3DisBench (3D image scene), which describes diverse 3D image scenes with 50 detailed prompts. Extensive experiments show the state-of-the-art performance of MUSES on both T2I-CompBench and T2I-3DisBench, outperforming recent strong competitors such as DALL-E 3 and Stable Diffusion 3. These results demonstrate a significant step forward for MUSES in bridging natural language, 2D image generation, and 3D world.