Changxing Ding

Jiehong Lin, Zewei Wei, Changxing Ding et al.

It is difficult to precisely annotate object instances and their semantics in 3D space, and as such, synthetic data are extensively used for these tasks, e.g., category-level 6D object pose and size estimation. However, the easy annotations in synthetic domains bring the downside effect of synthetic-to-real (Sim2Real) domain gap. In this work, we aim to address this issue in the task setting of Sim2Real, unsupervised domain adaptation for category-level 6D object pose and size estimation. We propose a method that is built upon a novel Deep Prior Deformation Network, shortened as DPDN. DPDN learns to deform features of categorical shape priors to match those of object observations, and is thus able to establish deep correspondence in the feature space for direct regression of object poses and sizes. To reduce the Sim2Real domain gap, we formulate a novel self-supervised objective upon DPDN via consistency learning; more specifically, we apply two rigid transformations to each object observation in parallel, and feed them into DPDN respectively to yield dual sets of predictions; on top of the parallel learning, an inter-consistency term is employed to keep cross consistency between dual predictions for improving the sensitivity of DPDN to pose changes, while individual intra-consistency ones are used to enforce self-adaptation within each learning itself. We train DPDN on both training sets of the synthetic CAMERA25 and real-world REAL275 datasets; our results outperform the existing methods on REAL275 test set under both the unsupervised and supervised settings. Ablation studies also verify the efficacy of our designs. Our code is released publicly at https://github.com/JiehongLin/Self-DPDN.

Zhiyin Shao, Xinyu Zhang, Meng Fang et al.

Text-to-image person re-identification (ReID) aims to search for pedestrian images of an interested identity via textual descriptions. It is challenging due to both rich intra-modal variations and significant inter-modal gaps. Existing works usually ignore the difference in feature granularity between the two modalities, i.e., the visual features are usually fine-grained while textual features are coarse, which is mainly responsible for the large inter-modal gaps. In this paper, we propose an end-to-end framework based on transformers to learn granularity-unified representations for both modalities, denoted as LGUR. LGUR framework contains two modules: a Dictionary-based Granularity Alignment (DGA) module and a Prototype-based Granularity Unification (PGU) module. In DGA, in order to align the granularities of two modalities, we introduce a Multi-modality Shared Dictionary (MSD) to reconstruct both visual and textual features. Besides, DGA has two important factors, i.e., the cross-modality guidance and the foreground-centric reconstruction, to facilitate the optimization of MSD. In PGU, we adopt a set of shared and learnable prototypes as the queries to extract diverse and semantically aligned features for both modalities in the granularity-unified feature space, which further promotes the ReID performance. Comprehensive experiments show that our LGUR consistently outperforms state-of-the-arts by large margins on both CUHK-PEDES and ICFG-PEDES datasets. Code will be released at https://github.com/ZhiyinShao-H/LGUR.

Wentao Tan, Changxing Ding, Pengfei Wang et al.

Domain generalization (DG) for person re-identification (ReID) is a challenging problem, as access to target domain data is not permitted during the training process. Most existing DG ReID methods update the feature extractor and classifier parameters based on the same features. This common practice causes the model to overfit to existing feature styles in the source domain, resulting in sub-optimal generalization ability on target domains. To solve this problem, we propose a novel style interleaved learning (IL) framework. Unlike conventional learning strategies, IL incorporates two forward propagations and one backward propagation for each iteration. We employ the features of interleaved styles to update the feature extractor and classifiers using different forward propagations, which helps to prevent the model from overfitting to certain domain styles. To generate interleaved feature styles, we further propose a new feature stylization approach. It produces a wide range of meaningful styles that are both different and independent from the original styles in the source domain, which caters to the IL methodology. Extensive experimental results show that our model not only consistently outperforms state-of-the-art methods on large-scale benchmarks for DG ReID, but also has clear advantages in computational efficiency. The code is available at https://github.com/WentaoTan/Interleaved-Learning.

Xin Lin, Changxing Ding, Yibing Zhan et al.

Scene graph generation (SGG) aims to detect objects and predict their pairwise relationships within an image. Current SGG methods typically utilize graph neural networks (GNNs) to acquire context information between objects/relationships. Despite their effectiveness, however, current SGG methods only assume scene graph homophily while ignoring heterophily. Accordingly, in this paper, we propose a novel Heterophily Learning Network (HL-Net) to comprehensively explore the homophily and heterophily between objects/relationships in scene graphs. More specifically, HL-Net comprises the following 1) an adaptive reweighting transformer module, which adaptively integrates the information from different layers to exploit both the heterophily and homophily in objects; 2) a relationship feature propagation module that efficiently explores the connections between relationships by considering heterophily in order to refine the relationship representation; 3) a heterophily-aware message-passing scheme to further distinguish the heterophily and homophily between objects/relationships, thereby facilitating improved message passing in graphs. We conducted extensive experiments on two public datasets: Visual Genome (VG) and Open Images (OI). The experimental results demonstrate the superiority of our proposed HL-Net over existing state-of-the-art approaches. In more detail, HL-Net outperforms the second-best competitors by 2.1$\%$ on the VG dataset for scene graph classification and 1.2$\%$ on the IO dataset for the final score. Code is available at https://github.com/siml3/HL-Net.

Zhihao Liang, Zhangjin Huang, Changxing Ding et al.

Recovery of an underlying scene geometry from multiview images stands as a long-time challenge in computer vision research. The recent promise leverages neural implicit surface learning and differentiable volume rendering, and achieves both the recovery of scene geometry and synthesis of novel views, where deep priors of neural models are used as an inductive smoothness bias. While promising for object-level surfaces, these methods suffer when coping with complex scene surfaces. In the meanwhile, traditional multi-view stereo can recover the geometry of scenes with rich textures, by globally optimizing the local, pixel-wise correspondences across multiple views. We are thus motivated to make use of the complementary benefits from the two strategies, and propose a method termed Helix-shaped neural implicit Surface learning or HelixSurf; HelixSurf uses the intermediate prediction from one strategy as the guidance to regularize the learning of the other one, and conducts such intertwined regularization iteratively during the learning process. We also propose an efficient scheme for differentiable volume rendering in HelixSurf. Experiments on surface reconstruction of indoor scenes show that our method compares favorably with existing methods and is orders of magnitude faster, even when some of existing methods are assisted with auxiliary training data. The source code is available at https://github.com/Gorilla-Lab-SCUT/HelixSurf.

Xin Lin, Changxing Ding, Jing Zhang et al.

Scene graph generation (SGG) aims to detect objects and predict the relationships between each pair of objects. Existing SGG methods usually suffer from several issues, including 1) ambiguous object representations, as graph neural network-based message passing (GMP) modules are typically sensitive to spurious inter-node correlations, and 2) low diversity in relationship predictions due to severe class imbalance and a large number of missing annotations. To address both problems, in this paper, we propose a regularized unrolling network (RU-Net). We first study the relation between GMP and graph Laplacian denoising (GLD) from the perspective of the unrolling technique, determining that GMP can be formulated as a solver for GLD. Based on this observation, we propose an unrolled message passing module and introduce an $\ell_p$-based graph regularization to suppress spurious connections between nodes. Second, we propose a group diversity enhancement module that promotes the prediction diversity of relationships via rank maximization. Systematic experiments demonstrate that RU-Net is effective under a variety of settings and metrics. Furthermore, RU-Net achieves new state-of-the-arts on three popular databases: VG, VRD, and OI. Code is available at https://github.com/siml3/RU-Net.

Fei Wang, Liang Ding, Jun Rao et al.

The multimedia community has shown a significant interest in perceiving and representing the physical world with multimodal pretrained neural network models, and among them, the visual-language pertaining (VLP) is, currently, the most captivating topic. However, there have been few endeavors dedicated to the exploration of 1) whether essential linguistic knowledge (e.g., semantics and syntax) can be extracted during VLP, and 2) how such linguistic knowledge impact or enhance the multimodal alignment. In response, here we aim to elucidate the impact of comprehensive linguistic knowledge, including semantic expression and syntactic structure, on multimodal alignment. Specifically, we design and release the SNARE, the first large-scale multimodal alignment probing benchmark, to detect the vital linguistic components, e.g., lexical, semantic, and syntax knowledge, containing four tasks: Semantic structure, Negation logic, Attribute ownership, and Relationship composition. Based on our proposed probing benchmarks, our holistic analyses of five advanced VLP models illustrate that the VLP model: i) shows insensitivity towards complex syntax structures and relies on content words for sentence comprehension; ii) demonstrates limited comprehension of combinations between sentences and negations; iii) faces challenges in determining the presence of actions or spatial relationships within visual information and struggles with verifying the correctness of triple combinations. We make our benchmark and code available at \url{https://github.com/WangFei-2019/SNARE/}.

Xubin Zhong, Changxing Ding, Zijian Li et al.

Human-Object Interaction (HOI) detection is a core task for high-level image understanding. Recently, Detection Transformer (DETR)-based HOI detectors have become popular due to their superior performance and efficient structure. However, these approaches typically adopt fixed HOI queries for all testing images, which is vulnerable to the location change of objects in one specific image. Accordingly, in this paper, we propose to enhance DETR's robustness by mining hard-positive queries, which are forced to make correct predictions using partial visual cues. First, we explicitly compose hard-positive queries according to the ground-truth (GT) position of labeled human-object pairs for each training image. Specifically, we shift the GT bounding boxes of each labeled human-object pair so that the shifted boxes cover only a certain portion of the GT ones. We encode the coordinates of the shifted boxes for each labeled human-object pair into an HOI query. Second, we implicitly construct another set of hard-positive queries by masking the top scores in cross-attention maps of the decoder layers. The masked attention maps then only cover partial important cues for HOI predictions. Finally, an alternate strategy is proposed that efficiently combines both types of hard queries. In each iteration, both DETR's learnable queries and one selected type of hard-positive queries are adopted for loss computation. Experimental results show that our proposed approach can be widely applied to existing DETR-based HOI detectors. Moreover, we consistently achieve state-of-the-art performance on three benchmarks: HICO-DET, V-COCO, and HOI-A. Code is available at https://github.com/MuchHair/HQM.

Wenhao Xu, Zhiyin Shao, Changxing Ding

Text-based person re-identification (ReID) aims to identify images of the targeted person from a large-scale person image database according to a given textual description. However, due to significant inter-modal gaps, text-based person ReID remains a challenging problem. Most existing methods generally rely heavily on the similarity contributed by matched word-region pairs, while neglecting mismatched word-region pairs which may play a decisive role. Accordingly, we propose to mine false positive examples (MFPE) via a jointly optimized multi-branch architecture to handle this problem. MFPE contains three branches including a false positive mining (FPM) branch to highlight the role of mismatched word-region pairs. Besides, MFPE delicately designs a cross-relu loss to increase the gap of similarity scores between matched and mismatched word-region pairs. Extensive experiments on CUHK-PEDES demonstrate the superior effectiveness of MFPE. Our code is released at https://github.com/xx-adeline/MFPE.

Changyong Shu, Hemao Wu, Hang Zhou et al.

The head swapping task aims at flawlessly placing a source head onto a target body, which is of great importance to various entertainment scenarios. While face swapping has drawn much attention, the task of head swapping has rarely been explored, particularly under the few-shot setting. It is inherently challenging due to its unique needs in head modeling and background blending. In this paper, we present the Head Swapper (HeSer), which achieves few-shot head swapping in the wild through two delicately designed modules. Firstly, a Head2Head Aligner is devised to holistically migrate pose and expression information from the target to the source head by examining multi-scale information. Secondly, to tackle the challenges of skin color variations and head-background mismatches in the swapping procedure, a Head2Scene Blender is introduced to simultaneously modify facial skin color and fill mismatched gaps in the background around the head. Particularly, seamless blending is achieved with the help of a Semantic-Guided Color Reference Creation procedure and a Blending UNet. Extensive experiments demonstrate that the proposed method produces superior head swapping results in a variety of scenes.

Kan Wang, Changxing Ding, Jianxin Pang et al.

Video-based person re-identification (ReID) is challenging due to the presence of various interferences in video frames. Recent approaches handle this problem using temporal aggregation strategies. In this work, we propose a novel Context Sensing Attention Network (CSA-Net), which improves both the frame feature extraction and temporal aggregation steps. First, we introduce the Context Sensing Channel Attention (CSCA) module, which emphasizes responses from informative channels for each frame. These informative channels are identified with reference not only to each individual frame, but also to the content of the entire sequence. Therefore, CSCA explores both the individuality of each frame and the global context of the sequence. Second, we propose the Contrastive Feature Aggregation (CFA) module, which predicts frame weights for temporal aggregation. Here, the weight for each frame is determined in a contrastive manner: i.e., not only by the quality of each individual frame, but also by the average quality of the other frames in a sequence. Therefore, it effectively promotes the contribution of relatively good frames. Extensive experimental results on four datasets show that CSA-Net consistently achieves state-of-the-art performance.

Chang Sun, Dongliang Liao, Changxing Ding

Open-vocabulary human-object interaction (HOI) detection aims to localize and recognize all human-object interactions in an image, including those unseen during training. Existing approaches usually rely on the collaboration between a conventional HOI detector and a Vision-Language Model (VLM) to recognize unseen HOI categories. However, feature fusion in this paradigm is challenging due to significant gaps in cross-model representations. To address this issue, we introduce SL-HOI, a StreamLined open-vocabulary HOI detection framework based solely on the powerful DINOv3 model. Our design leverages the complementary strengths of DINOv3's components: its backbone for fine-grained localization and its text-aligned vision head for open-vocabulary interaction classification. Moreover, to facilitate smooth cross-attention between the interaction queries and the vision head's output, we propose first feeding both the interaction queries and the backbone image tokens into the vision head, effectively bridging their representation gaps. All DINOv3 parameters in our approach are frozen, with only a small number of learnable parameters added, allowing a fast adaptation to the HOI detection task. Extensive experiments show that SL-HOI achieves state-of-the-art performance on both the SWiG-HOI and HICO-DET benchmarks, demonstrating the effectiveness of our streamlined model architecture. Code is available at https://github.com/MPI-Lab/SL-HOI.

Yiyang Chen, Shanshan Zhao, Changxing Ding et al.

In recent years, cross-modal domain adaptation has been studied on the paired 2D image and 3D LiDAR data to ease the labeling costs for 3D LiDAR semantic segmentation (3DLSS) in the target domain. However, in such a setting the paired 2D and 3D data in the source domain are still collected with additional effort. Since the 2D-3D projections can enable the 3D model to learn semantic information from the 2D counterpart, we ask whether we could further remove the need of source 3D data and only rely on the source 2D images. To answer it, this paper studies a new 3DLSS setting where a 2D dataset (source) with semantic annotations and a paired but unannotated 2D image and 3D LiDAR data (target) are available. To achieve 3DLSS in this scenario, we propose Cross-Modal and Cross-Domain Learning (CoMoDaL). Specifically, our CoMoDaL aims at modeling 1) inter-modal cross-domain distillation between the unpaired source 2D image and target 3D LiDAR data, and 2) the intra-domain cross-modal guidance between the target 2D image and 3D LiDAR data pair. In CoMoDaL, we propose to apply several constraints, such as point-to-pixel and prototype-to-pixel alignments, to associate the semantics in different modalities and domains by constructing mixed samples in two modalities. The experimental results on several datasets show that in the proposed setting, the developed CoMoDaL can achieve segmentation without the supervision of labeled LiDAR data. Ablations are also conducted to provide more analysis. Code will be available publicly.

Zeyu Jiang, Changxing Ding

In this report, we describe the technical details of our submission to the EPIC-Kitchens Action Anticipation Challenge 2022. In this competition, we develop the following two approaches. 1) Anticipation Time Knowledge Distillation using the soft labels learned by the teacher model as knowledge to guide the student network to learn the information of anticipation time; 2) Verb-Noun Relation Module for building the relationship between verbs and nouns. Our method achieves state-of-the-art results on the testing set of EPIC-Kitchens Action Anticipation Challenge 2022.

Yifei Liu, Qiong Cao, Yandong Wen et al.

This paper addresses the problem of generating lifelike holistic co-speech motions for 3D avatars, focusing on two key aspects: variability and coordination. Variability allows the avatar to exhibit a wide range of motions even with similar speech content, while coordination ensures a harmonious alignment among facial expressions, hand gestures, and body poses. We aim to achieve both with ProbTalk, a unified probabilistic framework designed to jointly model facial, hand, and body movements in speech. ProbTalk builds on the variational autoencoder (VAE) architecture and incorporates three core designs. First, we introduce product quantization (PQ) to the VAE, which enriches the representation of complex holistic motion. Second, we devise a novel non-autoregressive model that embeds 2D positional encoding into the product-quantized representation, thereby preserving essential structure information of the PQ codes. Last, we employ a secondary stage to refine the preliminary prediction, further sharpening the high-frequency details. Coupling these three designs enables ProbTalk to generate natural and diverse holistic co-speech motions, outperforming several state-of-the-art methods in qualitative and quantitative evaluations, particularly in terms of realism. Our code and model will be released for research purposes at https://feifeifeiliu.github.io/probtalk/.

Fei Wang, Li Shen, Liang Ding et al.

Zeroth-Order optimization presents a promising memory-efficient paradigm for fine-tuning Large Language Models by relying solely on forward passes. However, its practical adoption is severely constrained by slow wall-clock convergence and high estimation variance. In this work, we dissect the runtime characteristics of ZO algorithms and identify a critical system bottleneck where the generation of perturbations and parameter updates accounts for over 40% of the training latency. We argue that the standard uniform exploration strategy is fundamentally flawed as it fails to account for the heterogeneous sensitivity of layers in deep networks, resulting in computationally wasteful blind searches. To address this structural mismatch, we propose AdaLeZO, an Adaptive Layer-wise ZO optimization framework. By formulating the layer selection process as a non-stationary Multi-Armed Bandit problem, AdaLeZO dynamically allocates the limited perturbation budget to the most sensitive parameters. We further introduce an Inverse Probability Weighting mechanism based on sampling with replacement, which guarantees unbiased gradient estimation while effectively acting as a temporal denoiser to reduce variance. Extensive experiments on LLaMA and OPT models ranging from 6.7B to 30B parameters demonstrate that AdaLeZO achieves 1.7x to 3.0x wall-clock acceleration compared to state-of-the-art methods. Crucially, AdaLeZO functions as a universal plug-and-play module that seamlessly enhances the efficiency of existing ZO optimizers without incurring additional memory overhead.

Zhuolong Li, Xingao Li, Changxing Ding et al.

Detecting human-object interaction (HOI) has long been limited by the amount of supervised data available. Recent approaches address this issue by pre-training according to pseudo-labels, which align object regions with HOI triplets parsed from image captions. However, pseudo-labeling is tricky and noisy, making HOI pre-training a complex process. Therefore, we propose an efficient disentangled pre-training method for HOI detection (DP-HOI) to address this problem. First, DP-HOI utilizes object detection and action recognition datasets to pre-train the detection and interaction decoder layers, respectively. Then, we arrange these decoder layers so that the pre-training architecture is consistent with the downstream HOI detection task. This facilitates efficient knowledge transfer. Specifically, the detection decoder identifies reliable human instances in each action recognition dataset image, generates one corresponding query, and feeds it into the interaction decoder for verb classification. Next, we combine the human instance verb predictions in the same image and impose image-level supervision. The DP-HOI structure can be easily adapted to the HOI detection task, enabling effective model parameter initialization. Therefore, it significantly enhances the performance of existing HOI detection models on a broad range of rare categories. The code and pre-trained weight are available at https://github.com/xingaoli/DP-HOI.

Yiyang Chen, Lunhao Duan, Shanshan Zhao et al.

Rotation invariance is an important requirement for point shape analysis. To achieve this, current state-of-the-art methods attempt to construct the local rotation-invariant representation through learning or defining the local reference frame (LRF). Although efficient, these LRF-based methods suffer from perturbation of local geometric relations, resulting in suboptimal local rotation invariance. To alleviate this issue, we propose a Local-consistent Transformation (LocoTrans) learning strategy. Specifically, we first construct the local-consistent reference frame (LCRF) by considering the symmetry of the two axes in LRF. In comparison with previous LRFs, our LCRF is able to preserve local geometric relationships better through performing local-consistent transformation. However, as the consistency only exists in local regions, the relative pose information is still lost in the intermediate layers of the network. We mitigate such a relative pose issue by developing a relative pose recovery (RPR) module. RPR aims to restore the relative pose between adjacent transformed patches. Equipped with LCRF and RPR, our LocoTrans is capable of learning local-consistent transformation and preserving local geometry, which benefits rotation invariance learning. Competitive performance under arbitrary rotations on both shape classification and part segmentation tasks and ablations can demonstrate the effectiveness of our method. Code will be available publicly at https://github.com/wdttt/LocoTrans.

Deli Cai, Haoyang Ma, Changxing Ding

Trajectory-controlled human motion generation aims to synthesize realistic human motions conditioned on both textual descriptions and spatial trajectories. However, existing methods suffer from two critical limitations: first, the conflict between text and trajectory conditions disrupts the denoising process, resulting in compromised motion quality or inaccurate trajectory following; second, the use of redundant motion representations introduces inconsistencies between motion components, leading to instability during trajectory control. To address these challenges, we propose CMC, a decoupled framework that effectively coordinates text and trajectory conditions through a divide-and-conquer strategy. CMC follows a divide-and-conquer paradigm, comprising two cascaded stages: Trajectory Control and Motion Completion. In the first stage, a diffusion model generates a simplified representation of the controlled joints under trajectory guidance, based on the given trajectories, ensuring accurate and stable trajectory following. In the second stage, a text-conditioned diffusion inpainting model generates full-body motions using the simplified representation from the first stage as partial observations. To mitigate overfitting caused by limited inpainting training data, we further introduce the Selective Inpainting Mechanism (SIM), which alternates between text-to-motion generation and motion inpainting tasks during training. Experiments on HumanML3D and KIT datasets demonstrate that CMC achieves state-of-the-art performance in control accuracy and motion quality, demonstrating its effectiveness in coordinating multimodal conditions and representations.

Songjin Cai, Linjie Zhong, Ling Guo et al.

Generating realistic and physically plausible 3D Human-Object Interactions (HOI) remains a key challenge in motion generation. One primary reason is that describing these physical constraints with words alone is difficult. To address this limitation, we propose a new paradigm: extracting rich interaction priors from easily accessible 2D images. Specifically, we introduce ViHOI, a novel framework that enables diffusion-based generative models to leverage rich, task-specific priors from 2D images to enhance generation quality. We utilize a large Vision-Language Model (VLM) as a powerful prior-extraction engine and adopt a layer-decoupled strategy to obtain visual and textual priors. Concurrently, we design a Q-Former-based adapter that compresses the VLM's high-dimensional features into compact prior tokens, which significantly facilitates the conditional training of our diffusion model. Our framework is trained on motion-rendered images from the dataset to ensure strict semantic alignment between visual inputs and motion sequences. During inference, it leverages reference images synthesized by a text-to-image generation model to improve generalization to unseen objects and interaction categories. Experimental results demonstrate that ViHOI achieves state-of-the-art performance, outperforming existing methods across multiple benchmarks and demonstrating superior generalization.

Wentao Tan, Qiong Cao, Chao Xue et al.

The chart-to-code generation task requires MLLMs to convert chart images into executable code. This task faces two main challenges: limited data diversity and the difficulty of maintaining visual consistency between generated charts and the original ones. Existing datasets mainly rely on synthetic seed data to prompt GPT models for code generation, resulting in homogeneous samples that limit model generalization to real-world chart styles. To address this, we propose ReChartPrompt, leveraging real-world, human-designed charts extracted from arXiv papers as prompts. By harnessing the rich content and diverse visual styles of arXiv charts, we construct ReChartPrompt-240K, a large-scale and highly diverse dataset that better reflects realistic chart variations. For the second challenge, although SFT improves code understanding by optimizing next-token prediction, it does not provide direct supervision on visual features. As a result, it often fails to guarantee that the generated charts visually match the original ones. To address this, we propose ChartSimRL, a GRPO-based reinforcement learning algorithm guided by a novel chart similarity reward. This reward consists of two components: attribute similarity, which measures the overlap of chart attributes like layout and color between the generated and original charts, and visual similarity, which evaluates overall visual features, including texture, using convolutional neural networks. Unlike traditional text-based rewards, our reward accounts for the multimodal nature of the chart-to-code generation task, significantly enhancing the model's ability to accurately reproduce charts. Integrating ReChartPrompt and ChartSimRL, we develop the ChartMaster model, achieving SOTA results among 7B-parameter models and rivaling GPT-4o on various chart-to-code benchmarks. All resources are available at https://github.com/WentaoTan/ChartMaster.

Wentao Tan, Qiong Cao, Yibing Zhan et al.

Achieving human-like reasoning capabilities in Multimodal Large Language Models (MLLMs) has long been a goal. Current methods primarily focus on synthesizing positive rationales, typically relying on manual annotations or complex systems. Moreover, they often overlook negative reasoning, which limits the model's generalization ability and robustness in multimodal inference. To address this gap, we propose a novel framework: \textbf{S}elf-Aligning \textbf{M}ultimodal Reasoning with \textbf{A}nswer-O\textbf{r}iented Chain-of-\textbf{T}hought (SMART). SMART employs an answer-oriented chain-of-thought (AoT) prompt to automatically construct high-quality data. Drawing inspiration from human proof-based strategies, AoT leverages both correct and incorrect answers to extract key visual information that links questions and answers. When provided with correct answers, the model produces strong positive rationales. Conversely, when correct answers are replaced with incorrect alternatives, the model generates an erroneous yet compelling reasoning path, serving as a form of discriminative negative rationale. Models trained with AoT-generated data outperform those trained on manually annotated datasets, demonstrating superior reasoning capabilities. Consequently, SMART establishes an iterative generation-optimization method that continually enhances the model's reasoning skills. Experiments indicate that the SMART framework significantly improves various MLLMs, regardless of model architecture, parameter size, or pre-training dataset. The code is available at https://github.com/WentaoTan/SMART.

Fei Wang, Li Shen, Liang Ding et al.

Large Language Models excel at natural language processing tasks, but their massive size leads to high computational and storage demands. Recent works have sought to reduce their model size through layer-wise structured pruning. However, they tend to ignore retaining the capabilities in the pruned part. In this work, we re-examine structured pruning paradigms and uncover several key limitations: 1) notable performance degradation due to direct layer removal, 2) incompetent linear weight layer aggregation, and 3) the lack of effective post-training recovery mechanisms. To address these limitations, we propose CoMe, including a progressive layer pruning framework with a Concatenation-based Merging technology and a hierarchical distillation post-training process. Specifically, we introduce a channel sensitivity metric that utilizes activation intensity and weight norms for fine-grained channel selection. Subsequently, we employ a concatenation-based layer merging method to fuse the most critical channels across adjacent layers, enabling progressive model size reduction. Finally, we propose a hierarchical distillation protocol that leverages the correspondences between the original and pruned model layers established during pruning, thereby enabling efficient knowledge transfer. Experiments on seven benchmarks show that CoMe achieves state-of-the-art performance; when pruning 30% of LLaMA-2-7b's parameters, the pruned model retains 83% of its original average accuracy. Our code is available at https://github.com/MPI-Lab/CoMe.

Yiyang Chen, Shanshan Zhao, Lunhao Duan et al.

Diffusion-based models, widely used in text-to-image generation, have proven effective in 2D representation learning. Recently, this framework has been extended to 3D self-supervised learning by constructing a conditional point generator for enhancing 3D representations. However, its performance remains constrained by the 3D diffusion model, which is trained on the available 3D datasets with limited size. We hypothesize that the robust capabilities of text-to-image diffusion models, particularly Stable Diffusion (SD), which is trained on large-scale datasets, can help overcome these limitations. To investigate this hypothesis, we propose PointSD, a framework that leverages the SD model for 3D self-supervised learning. By replacing the SD model's text encoder with a 3D encoder, we train a point-to-image diffusion model that allows point clouds to guide the denoising of rendered noisy images. With the trained point-to-image diffusion model, we use noise-free images as the input and point clouds as the condition to extract SD features. Next, we train a 3D backbone by aligning its features with these SD features, thereby facilitating direct semantic learning. Comprehensive experiments on downstream point cloud tasks and ablation studies demonstrate that the SD model can enhance point cloud self-supervised learning. Code is publicly available at https://github.com/wdttt/PointSD.

Zhiyin Shao, Xinyu Zhang, Changxing Ding et al.

The pre-training task is indispensable for the text-to-image person re-identification (T2I-ReID) task. However, there are two underlying inconsistencies between these two tasks that may impact the performance; i) Data inconsistency. A large domain gap exists between the generic images/texts used in public pre-trained models and the specific person data in the T2I-ReID task. This gap is especially severe for texts, as general textual data are usually unable to describe specific people in fine-grained detail. ii) Training inconsistency. The processes of pre-training of images and texts are independent, despite cross-modality learning being critical to T2I-ReID. To address the above issues, we present a new unified pre-training pipeline (UniPT) designed specifically for the T2I-ReID task. We first build a large-scale text-labeled person dataset "LUPerson-T", in which pseudo-textual descriptions of images are automatically generated by the CLIP paradigm using a divide-conquer-combine strategy. Benefiting from this dataset, we then utilize a simple vision-and-language pre-training framework to explicitly align the feature space of the image and text modalities during pre-training. In this way, the pre-training task and the T2I-ReID task are made consistent with each other on both data and training levels. Without the need for any bells and whistles, our UniPT achieves competitive Rank-1 accuracy of, ie, 68.50%, 60.09%, and 51.85% on CUHK-PEDES, ICFG-PEDES and RSTPReid, respectively. Both the LUPerson-T dataset and code are available at https;//github.com/ZhiyinShao-H/UniPT.

Shengcong Chen, Changxing Ding, Dacheng Tao et al.

Nucleus segmentation is usually the first step in pathological image analysis tasks. Generalizable nucleus segmentation refers to the problem of training a segmentation model that is robust to domain gaps between the source and target domains. The domain gaps are usually believed to be caused by the varied image acquisition conditions, e.g., different scanners, tissues, or staining protocols. In this paper, we argue that domain gaps can also be caused by different foreground (nucleus)-background ratios, as this ratio significantly affects feature statistics that are critical to normalization layers. We propose a Distribution-Aware Re-Coloring (DARC) model that handles the above challenges from two perspectives. First, we introduce a re-coloring method that relieves dramatic image color variations between different domains. Second, we propose a new instance normalization method that is robust to the variation in foreground-background ratios. We evaluate the proposed methods on two H$\&$E stained image datasets, named CoNSeP and CPM17, and two IHC stained image datasets, called DeepLIIF and BC-DeepLIIF. Extensive experimental results justify the effectiveness of our proposed DARC model. Codes are available at \url{https://github.com/csccsccsccsc/DARC

Pengfei Wang, Changxing Ding, Wentao Tan et al.

Unsupervised Domain Adaptive (UDA) object re-identification (Re-ID) aims at adapting a model trained on a labeled source domain to an unlabeled target domain. State-of-the-art object Re-ID approaches adopt clustering algorithms to generate pseudo-labels for the unlabeled target domain. However, the inevitable label noise caused by the clustering procedure significantly degrades the discriminative power of Re-ID model. To address this problem, we propose an uncertainty-aware clustering framework (UCF) for UDA tasks. First, a novel hierarchical clustering scheme is proposed to promote clustering quality. Second, an uncertainty-aware collaborative instance selection method is introduced to select images with reliable labels for model training. Combining both techniques effectively reduces the impact of noisy labels. In addition, we introduce a strong baseline that features a compact contrastive loss. Our UCF method consistently achieves state-of-the-art performance in multiple UDA tasks for object Re-ID, and significantly reduces the gap between unsupervised and supervised Re-ID performance. In particular, the performance of our unsupervised UCF method in the MSMT17$\to$Market1501 task is better than that of the fully supervised setting on Market1501. The code of UCF is available at https://github.com/Wang-pengfei/UCF.

Zefeng Ding, Changxing Ding, Zhiyin Shao et al.

Text-to-image person re-identification (ReID) aims to search for images containing a person of interest using textual descriptions. However, due to the significant modality gap and the large intra-class variance in textual descriptions, text-to-image ReID remains a challenging problem. Accordingly, in this paper, we propose a Semantically Self-Aligned Network (SSAN) to handle the above problems. First, we propose a novel method that automatically extracts semantically aligned part-level features from the two modalities. Second, we design a multi-view non-local network that captures the relationships between body parts, thereby establishing better correspondences between body parts and noun phrases. Third, we introduce a Compound Ranking (CR) loss that makes use of textual descriptions for other images of the same identity to provide extra supervision, thereby effectively reducing the intra-class variance in textual features. Finally, to expedite future research in text-to-image ReID, we build a new database named ICFG-PEDES. Extensive experiments demonstrate that SSAN outperforms state-of-the-art approaches by significant margins. Both the new ICFG-PEDES database and the SSAN code are available at https://github.com/zifyloo/SSAN.

Junyang Huang, Changxing Ding

The past few years have witnessed great progress in the domain of face recognition thanks to advances in deep learning. However, cross pose face recognition remains a significant challenge. It is difficult for many deep learning algorithms to narrow the performance gap caused by pose variations; the main reasons for this relate to the intra-class discrepancy between face images in different poses and the pose imbalances of training datasets. Learning pose-robust features by traversing to the feature space of frontal faces provides an effective and cheap way to alleviate this problem. In this paper, we present a method for progressively transforming profile face representations to the canonical pose with an attentive pair-wise loss. Firstly, to reduce the difficulty of directly transforming the profile face features into a frontal pose, we propose to learn the feature residual between the source pose and its nearby pose in a block-byblock fashion, and thus traversing to the feature space of a smaller pose by adding the learned residual. Secondly, we propose an attentive pair-wise loss to guide the feature transformation progressing in the most effective direction. Finally, our proposed progressive module and attentive pair-wise loss are light-weight and easy to implement, adding only about 7:5% extra parameters. Evaluations on the CFP and CPLFW datasets demonstrate the superiority of our proposed method. Code is available at https://github.com/hjy1312/AGPM.

Xubin Zhong, Xian Qu, Changxing Ding et al.

Modern human-object interaction (HOI) detection approaches can be divided into one-stage methods and twostage ones. One-stage models are more efficient due to their straightforward architectures, but the two-stage models are still advantageous in accuracy. Existing one-stage models usually begin by detecting predefined interaction areas or points, and then attend to these areas only for interaction prediction; therefore, they lack reasoning steps that dynamically search for discriminative cues. In this paper, we propose a novel one-stage method, namely Glance and Gaze Network (GGNet), which adaptively models a set of actionaware points (ActPoints) via glance and gaze steps. The glance step quickly determines whether each pixel in the feature maps is an interaction point. The gaze step leverages feature maps produced by the glance step to adaptively infer ActPoints around each pixel in a progressive manner. Features of the refined ActPoints are aggregated for interaction prediction. Moreover, we design an actionaware approach that effectively matches each detected interaction with its associated human-object pair, along with a novel hard negative attentive loss to improve the optimization of GGNet. All the above operations are conducted simultaneously and efficiently for all pixels in the feature maps. Finally, GGNet outperforms state-of-the-art methods by significant margins on both V-COCO and HICODET benchmarks. Code of GGNet is available at https: //github.com/SherlockHolmes221/GGNet.

Shengcong Chen, Changxing Ding, Minfeng Liu et al.

Nucleus segmentation is a challenging task due to the crowded distribution and blurry boundaries of nuclei. Recent approaches represent nuclei by means of polygons to differentiate between touching and overlapping nuclei and have accordingly achieved promising performance. Each polygon is represented by a set of centroid-to-boundary distances, which are in turn predicted by features of the centroid pixel for a single nucleus. However, using the centroid pixel alone does not provide sufficient contextual information for robust prediction and thus degrades the segmentation accuracy. To handle this problem, we propose a Context-aware Polygon Proposal Network (CPP-Net) for nucleus segmentation. First, we sample a point set rather than one single pixel within each cell for distance prediction. This strategy substantially enhances contextual information and thereby improves the robustness of the prediction. Second, we propose a Confidence-based Weighting Module, which adaptively fuses the predictions from the sampled point set. Third, we introduce a novel Shape-Aware Perceptual (SAP) loss that constrains the shape of the predicted polygons. Here, the SAP loss is based on an additional network that is pre-trained by means of mapping the centroid probability map and the pixel-to-boundary distance maps to a different nucleus representation. Extensive experiments justify the effectiveness of each component in the proposed CPP-Net. Finally, CPP-Net is found to achieve state-of-the-art performance on three publicly available databases, namely DSB2018, BBBC06, and PanNuke. Code of this paper is available at \url{https://github.com/csccsccsccsc/cpp-net

Xubin Zhong, Changxing Ding, Xian Qu et al.

Human-Object Interaction (HOI) detection is important to human-centric scene understanding tasks. Existing works tend to assume that the same verb has similar visual characteristics in different HOI categories, an approach that ignores the diverse semantic meanings of the verb. To address this issue, in this paper, we propose a novel Polysemy Deciphering Network (PD-Net) that decodes the visual polysemy of verbs for HOI detection in three distinct ways. First, we refine features for HOI detection to be polysemyaware through the use of two novel modules: namely, Language Prior-guided Channel Attention (LPCA) and Language Prior-based Feature Augmentation (LPFA). LPCA highlights important elements in human and object appearance features for each HOI category to be identified; moreover, LPFA augments human pose and spatial features for HOI detection using language priors, enabling the verb classifiers to receive language hints that reduce intra-class variation for the same verb. Second, we introduce a novel Polysemy-Aware Modal Fusion module (PAMF), which guides PD-Net to make decisions based on feature types deemed more important according to the language priors. Third, we propose to relieve the verb polysemy problem through sharing verb classifiers for semantically similar HOI categories. Furthermore, to expedite research on the verb polysemy problem, we build a new benchmark dataset named HOI-VerbPolysemy (HOIVP), which includes common verbs (predicates) that have diverse semantic meanings in the real world. Finally, through deciphering the visual polysemy of verbs, our approach is demonstrated to outperform state-of-the-art methods by significant margins on the HICO-DET, V-COCO, and HOI-VP databases. Code and data in this paper are available at https://github.com/MuchHair/PD-Net.

Shengcong Chen, Changxing Ding, Dacheng Tao

Nucleus segmentation is an important task in medical image analysis. However, machine learning models cannot perform well because there are large amount of clusters of crowded nuclei. To handle this problem, existing approaches typically resort to sophisticated hand-crafted post-processing strategies; therefore, they are vulnerable to the variation of post-processing hyper-parameters. Accordingly, in this paper, we devise a Boundary-assisted Region Proposal Network (BRP-Net) that achieves robust instance-level nucleus segmentation. First, we propose a novel Task-aware Feature Encoding (TAFE) network that efficiently extracts respective high-quality features for semantic segmentation and instance boundary detection tasks. This is achieved by carefully considering the correlation and differences between the two tasks. Second, coarse nucleus proposals are generated based on the predictions of the above two tasks. Third, these proposals are fed into instance segmentation networks for more accurate prediction. Experimental results demonstrate that the performance of BRP-Net is robust to the variation of post-processing hyper-parameters. Furthermore, BRP-Net achieves state-of-the-art performances on both the Kumar and CPM17 datasets. The code of BRP-Net will be released at https://github.com/csccsccsccsc/brpnet.

Xin Lin, Changxing Ding, Jinquan Zeng et al.

Scene graph generation (SGG) aims to detect objects in an image along with their pairwise relationships. There are three key properties of scene graph that have been underexplored in recent works: namely, the edge direction information, the difference in priority between nodes, and the long-tailed distribution of relationships. Accordingly, in this paper, we propose a Graph Property Sensing Network (GPS-Net) that fully explores these three properties for SGG. First, we propose a novel message passing module that augments the node feature with node-specific contextual information and encodes the edge direction information via a tri-linear model. Second, we introduce a node priority sensitive loss to reflect the difference in priority between nodes during training. This is achieved by designing a mapping function that adjusts the focusing parameter in the focal loss. Third, since the frequency of relationships is affected by the long-tailed distribution problem, we mitigate this issue by first softening the distribution and then enabling it to be adjusted for each subject-object pair according to their visual appearance. Systematic experiments demonstrate the effectiveness of the proposed techniques. Moreover, GPS-Net achieves state-of-the-art performance on three popular databases: VG, OI, and VRD by significant gains under various settings and metrics. The code and models are available at \url{https://github.com/taksau/GPS-Net}.

Changxing Ding, Kan Wang, Pengfei Wang et al.

Part-level representations are important for robust person re-identification (ReID), but in practice feature quality suffers due to the body part misalignment problem. In this paper, we present a robust, compact, and easy-to-use method called the Multi-task Part-aware Network (MPN), which is designed to extract semantically aligned part-level features from pedestrian images. MPN solves the body part misalignment problem via multi-task learning (MTL) in the training stage. More specifically, it builds one main task (MT) and one auxiliary task (AT) for each body part on the top of the same backbone model. The ATs are equipped with a coarse prior of the body part locations for training images. ATs then transfer the concept of the body parts to the MTs via optimizing the MT parameters to identify part-relevant channels from the backbone model. Concept transfer is accomplished by means of two novel alignment strategies: namely, parameter space alignment via hard parameter sharing and feature space alignment in a class-wise manner. With the aid of the learned high-quality parameters, MTs can independently extract semantically aligned part-level features from relevant channels in the testing stage. MPN has three key advantages: 1) it does not need to conduct body part detection in the inference stage; 2) its model is very compact and efficient for both training and testing; 3) in the training stage, it requires only coarse priors of body part locations, which are easy to obtain. Systematic experiments on four large-scale ReID databases demonstrate that MPN consistently outperforms state-of-the-art approaches by significant margins. Code is available at https://github.com/WangKan0128/MPN.

Chenhong Zhou, Changxing Ding, Xinchao Wang et al.

Class imbalance has emerged as one of the major challenges for medical image segmentation. The model cascade (MC) strategy significantly alleviates the class imbalance issue via running a set of individual deep models for coarse-to-fine segmentation. Despite its outstanding performance, however, this method leads to undesired system complexity and also ignores the correlation among the models. To handle these flaws, we propose a light-weight deep model, i.e., the One-pass Multi-task Network (OM-Net) to solve class imbalance better than MC does, while requiring only one-pass computation. First, OM-Net integrates the separate segmentation tasks into one deep model, which consists of shared parameters to learn joint features, as well as task-specific parameters to learn discriminative features. Second, to more effectively optimize OM-Net, we take advantage of the correlation among tasks to design both an online training data transfer strategy and a curriculum learning-based training strategy. Third, we further propose sharing prediction results between tasks and design a cross-task guided attention (CGA) module which can adaptively recalibrate channel-wise feature responses based on the category-specific statistics. Finally, a simple yet effective post-processing method is introduced to refine the segmentation results. Extensive experiments are conducted to demonstrate the effectiveness of the proposed techniques. Most impressively, we achieve state-of-the-art performance on the BraTS 2015 testing set and BraTS 2017 online validation set. Using these proposed approaches, we also won joint third place in the BraTS 2018 challenge among 64 participating teams. The code is publicly available at https://github.com/chenhong-zhou/OM-Net.

Yifei Liu, Changxing Ding, Ling Guo et al.

Diffusion models have seen widespread adoption for text-driven human motion generation and related tasks due to their impressive generative capabilities and flexibility. However, current motion diffusion models face two major limitations: a representational gap caused by pre-trained text encoders that lack motion-specific information, and error propagation during the iterative denoising process. This paper introduces Reconstruction-Anchored Diffusion Model (RAM) to address these challenges. First, RAM leverages a motion latent space as intermediate supervision for text-to-motion generation. To this end, RAM co-trains a motion reconstruction branch with two key objective functions: self-regularization to enhance the discrimination of the motion space and motion-centric latent alignment to enable accurate mapping from text to the motion latent space. Second, we propose Reconstructive Error Guidance (REG), a testing-stage guidance mechanism that exploits the diffusion model's inherent self-correction ability to mitigate error propagation. At each denoising step, REG uses the motion reconstruction branch to reconstruct the previous estimate, reproducing the prior error patterns. By amplifying the residual between the current prediction and the reconstructed estimate, REG highlights the improvements in the current prediction. Extensive experiments demonstrate that RAM achieves significant improvements and state-of-the-art performance. Our code will be released.

Wentao Tan, Changxing Ding, Jiayu Jiang et al.

Text-to-image person re-identification (ReID) retrieves pedestrian images according to textual descriptions. Manually annotating textual descriptions is time-consuming, restricting the scale of existing datasets and therefore the generalization ability of ReID models. As a result, we study the transferable text-to-image ReID problem, where we train a model on our proposed large-scale database and directly deploy it to various datasets for evaluation. We obtain substantial training data via Multi-modal Large Language Models (MLLMs). Moreover, we identify and address two key challenges in utilizing the obtained textual descriptions. First, an MLLM tends to generate descriptions with similar structures, causing the model to overfit specific sentence patterns. Thus, we propose a novel method that uses MLLMs to caption images according to various templates. These templates are obtained using a multi-turn dialogue with a Large Language Model (LLM). Therefore, we can build a large-scale dataset with diverse textual descriptions. Second, an MLLM may produce incorrect descriptions. Hence, we introduce a novel method that automatically identifies words in a description that do not correspond with the image. This method is based on the similarity between one text and all patch token embeddings in the image. Then, we mask these words with a larger probability in the subsequent training epoch, alleviating the impact of noisy textual descriptions. The experimental results demonstrate that our methods significantly boost the direct transfer text-to-image ReID performance. Benefiting from the pre-trained model weights, we also achieve state-of-the-art performance in the traditional evaluation settings.

Jianbin Zheng, Minghui Hu, Zhongyi Fan et al.

Latent Consistency Model (LCM) extends the Consistency Model to the latent space and leverages the guided consistency distillation technique to achieve impressive performance in accelerating text-to-image synthesis. However, we observed that LCM struggles to generate images with both clarity and detailed intricacy. Consequently, we introduce Trajectory Consistency Distillation (TCD), which encompasses trajectory consistency function and strategic stochastic sampling. The trajectory consistency function diminishes the parameterisation and distillation errors by broadening the scope of the self-consistency boundary condition with trajectory mapping and endowing the TCD with the ability to accurately trace the entire trajectory of the Probability Flow ODE in semi-linear form with an Exponential Integrator. Additionally, strategic stochastic sampling provides explicit control of stochastic and circumvents the accumulated errors inherent in multi-step consistency sampling. Experiments demonstrate that TCD not only significantly enhances image quality at low NFEs but also yields more detailed results compared to the teacher model at high NFEs.

Xu Yang, Changxing Ding, Zhibin Hong et al.

Image-based virtual try-on is an increasingly important task for online shopping. It aims to synthesize images of a specific person wearing a specified garment. Diffusion model-based approaches have recently become popular, as they are excellent at image synthesis tasks. However, these approaches usually employ additional image encoders and rely on the cross-attention mechanism for texture transfer from the garment to the person image, which affects the try-on's efficiency and fidelity. To address these issues, we propose an Texture-Preserving Diffusion (TPD) model for virtual try-on, which enhances the fidelity of the results and introduces no additional image encoders. Accordingly, we make contributions from two aspects. First, we propose to concatenate the masked person and reference garment images along the spatial dimension and utilize the resulting image as the input for the diffusion model's denoising UNet. This enables the original self-attention layers contained in the diffusion model to achieve efficient and accurate texture transfer. Second, we propose a novel diffusion-based method that predicts a precise inpainting mask based on the person and reference garment images, further enhancing the reliability of the try-on results. In addition, we integrate mask prediction and image synthesis into a single compact model. The experimental results show that our approach can be applied to various try-on tasks, e.g., garment-to-person and person-to-person try-ons, and significantly outperforms state-of-the-art methods on popular VITON, VITON-HD databases.

Wentao Tan, Qiong Cao, Yibing Zhan et al.

Human preference alignment can greatly enhance Multimodal Large Language Models (MLLMs), but collecting high-quality preference data is costly. A promising solution is the self-evolution strategy, where models are iteratively trained on data they generate. However, current techniques still rely on human- or GPT-annotated data and sometimes require additional models or ground truth answers. To address these issues, we propose a novel multimodal self-evolution framework that enables the model to autonomously generate high-quality questions and answers using only unannotated images. First, we implement an image-driven self-questioning mechanism, allowing the model to create and evaluate questions based on image content, regenerating them if they are irrelevant or unanswerable. This sets a strong foundation for answer generation. Second, we introduce an answer self-enhancement technique, starting with image captioning to improve answer quality. We also use corrupted images to generate rejected answers, forming distinct preference pairs for optimization. Finally, we incorporate an image content alignment loss function alongside Direct Preference Optimization (DPO) loss to reduce hallucinations, ensuring the model focuses on image content. Experiments show that our framework performs competitively with methods using external information, offering a more efficient and scalable approach to MLLMs.

Guowei Wang, Changxing Ding, Wentao Tan et al.

Test-time adaptation (TTA) is a task that continually adapts a pre-trained source model to the target domain during inference. One popular approach involves fine-tuning model with cross-entropy loss according to estimated pseudo-labels. However, its performance is significantly affected by noisy pseudo-labels. This study reveals that minimizing the classification error of each sample causes the cross-entropy loss's vulnerability to label noise. To address this issue, we propose a novel Decoupled Prototype Learning (DPL) method that features prototype-centric loss computation. First, we decouple the optimization of class prototypes. For each class prototype, we reduce its distance with positive samples and enlarge its distance with negative samples in a contrastive manner. This strategy prevents the model from overfitting to noisy pseudo-labels. Second, we propose a memory-based strategy to enhance DPL's robustness for the small batch sizes often encountered in TTA. We update each class's pseudo-feature from a memory in a momentum manner and insert an additional DPL loss. Finally, we introduce a consistency regularization-based approach to leverage samples with unconfident pseudo-labels. This approach transfers feature styles of samples with unconfident pseudo-labels to those with confident pseudo-labels. Thus, more reliable samples for TTA are created. The experimental results demonstrate that our methods achieve state-of-the-art performance on domain generalization benchmarks, and reliably improve the performance of self-training-based methods on image corruption benchmarks. The code will be released.

Mengqing Xue, Yifei Liu, Ling Guo et al.

Human-object interaction (HOI) synthesis is crucial for creating immersive and realistic experiences for applications such as virtual reality. Existing methods often rely on simplified object representations, such as the object's centroid or the nearest point to a human, to achieve physically plausible motions. However, these approaches may overlook geometric complexity, resulting in suboptimal interaction fidelity. To address this limitation, we introduce ROG, a novel diffusion-based framework that models the spatiotemporal relationships inherent in HOIs with rich geometric detail. For efficient object representation, we select boundary-focused and fine-detail key points from the object mesh, ensuring a comprehensive depiction of the object's geometry. This representation is used to construct an interactive distance field (IDF), capturing the robust HOI dynamics. Furthermore, we develop a diffusion-based relation model that integrates spatial and temporal attention mechanisms, enabling a better understanding of intricate HOI relationships. This relation model refines the generated motion's IDF, guiding the motion generation process to produce relation-aware and semantically aligned movements. Experimental evaluations demonstrate that ROG significantly outperforms state-of-the-art methods in the realism and semantic accuracy of synthesized HOIs.

Fei Wang, Li Shen, Liang Ding et al.

Fine-tuning is powerful for adapting large language models to downstream tasks, but it often results in huge memory usages. A promising approach to mitigate this is using Zeroth-Order (ZO) optimization, which estimates gradients to replace First-Order (FO) gradient calculations, albeit with longer training time due to its stochastic nature. By revisiting the Memory-efficient ZO (MeZO) optimizer, we discover that the full-parameter perturbation and updating processes consume over 50% of its overall fine-tuning time cost. Based on these observations, we introduce a novel layer-wise sparse computation and memory efficient ZO optimizer, named LeZO. LeZO treats layers as fundamental units for sparsification and dynamically perturbs different parameter subsets in each step to achieve full-parameter fine-tuning. LeZO incorporates layer-wise parameter sparsity in the process of simultaneous perturbation stochastic approximation (SPSA) and ZO stochastic gradient descent (ZO-SGD). It achieves accelerated computation during perturbation and updating processes without additional memory overhead. We conduct extensive experiments with the OPT model family on the SuperGLUE benchmark and two generative tasks. The experiments show that LeZO accelerates training without compromising the performance of ZO optimization. Specifically, it achieves over 3x speedup compared to MeZO on the SST-2, BoolQ, and Copa tasks.

Guowei Wang, Changxing Ding

Long-term test-time adaptation (TTA) is a challenging task due to error accumulation. Recent approaches tackle this issue by actively labeling a small proportion of samples in each batch, yet the annotation burden quickly grows as the batch number increases. In this paper, we investigate how to achieve effortless active labeling so that a maximum of one sample is selected for annotation in each batch. First, we annotate the most valuable sample in each batch based on the single-step optimization perspective in the TTA context. In this scenario, the samples that border between the source- and target-domain data distributions are considered the most feasible for the model to learn in one iteration. Then, we introduce an efficient strategy to identify these samples using feature perturbation. Second, we discover that the gradient magnitudes produced by the annotated and unannotated samples have significant variations. Therefore, we propose balancing their impact on model optimization using two dynamic weights. Extensive experiments on the popular ImageNet-C, -R, -K, -A and PACS databases demonstrate that our approach consistently outperforms state-of-the-art methods with significantly lower annotation costs.

Jiayu Jiang, Changxing Ding, Wentao Tan et al.

Text-to-image person re-identification (ReID) aims to retrieve the images of an interested person based on textual descriptions. One main challenge for this task is the high cost in manually annotating large-scale databases, which affects the generalization ability of ReID models. Recent works handle this problem by leveraging Multi-modal Large Language Models (MLLMs) to describe pedestrian images automatically. However, the captions produced by MLLMs lack diversity in description styles. To address this issue, we propose a Human Annotator Modeling (HAM) approach to enable MLLMs to mimic the description styles of thousands of human annotators. Specifically, we first extract style features from human textual descriptions and perform clustering on them. This allows us to group textual descriptions with similar styles into the same cluster. Then, we employ a prompt to represent each of these clusters and apply prompt learning to mimic the description styles of different human annotators. Furthermore, we define a style feature space and perform uniform sampling in this space to obtain more diverse clustering prototypes, which further enriches the diversity of the MLLM-generated captions. Finally, we adopt HAM to automatically annotate a massive-scale database for text-to-image ReID. Extensive experiments on this database demonstrate that it significantly improves the generalization ability of ReID models.

Xu Yang, Shaoli Huang, Shenbo Xie et al.

Co-speech gesture video generation aims to synthesize realistic, audio-aligned videos of speakers, complete with synchronized facial expressions and body gestures. This task presents challenges due to the significant one-to-many mapping between audio and visual content, further complicated by the scarcity of large-scale public datasets and high computational demands. We propose a lightweight framework that utilizes 2D full-body skeletons as an efficient auxiliary condition to bridge audio signals with visual outputs. Our approach introduces a diffusion model conditioned on fine-grained audio segments and a skeleton extracted from the speaker's reference image, predicting skeletal motions through skeleton-audio feature fusion to ensure strict audio coordination and body shape consistency. The generated skeletons are then fed into an off-the-shelf human video generation model with the speaker's reference image to synthesize high-fidelity videos. To democratize research, we present CSG-405-the first public dataset with 405 hours of high-resolution videos across 71 speech types, annotated with 2D skeletons and diverse speaker demographics. Experiments show that our method exceeds state-of-the-art approaches in visual quality and synchronization while generalizing across speakers and contexts. Code, models, and CSG-405 are publicly released at https://mpi-lab.github.io/Democratizing-CSG/

Yupeng Hu, Changxing Ding, Chang Sun et al.

Open vocabulary Human-Object Interaction (HOI) detection is a challenging task that detects all <human, verb, object> triplets of interest in an image, even those that are not pre-defined in the training set. Existing approaches typically rely on output features generated by large Vision-Language Models (VLMs) to enhance the generalization ability of interaction representations. However, the visual features produced by VLMs are holistic and coarse-grained, which contradicts the nature of detection tasks. To address this issue, we propose a novel Bilateral Collaboration framework for open vocabulary HOI detection (BC-HOI). This framework includes an Attention Bias Guidance (ABG) component, which guides the VLM to produce fine-grained instance-level interaction features according to the attention bias provided by the HOI detector. It also includes a Large Language Model (LLM)-based Supervision Guidance (LSG) component, which provides fine-grained token-level supervision for the HOI detector by the LLM component of the VLM. LSG enhances the ability of ABG to generate high-quality attention bias. We conduct extensive experiments on two popular benchmarks: HICO-DET and V-COCO, consistently achieving superior performance in the open vocabulary and closed settings. The code will be released in Github.

Xubin Zhong, Changxing Ding, Yupeng Hu et al.

Human-Object Interaction (HOI) detection is a core task for human-centric image understanding. Recent one-stage methods adopt a transformer decoder to collect image-wide cues that are useful for interaction prediction; however, the interaction representations obtained using this method are entangled and lack interpretability. In contrast, traditional two-stage methods benefit significantly from their ability to compose interaction features in a disentangled and explainable manner. In this paper, we improve the performance of one-stage methods by enabling them to extract disentangled interaction representations. First, we propose Shunted Cross-Attention (SCA) to extract human appearance, object appearance, and global context features using different cross-attention heads. This is achieved by imposing different masks on the cross-attention maps produced by the different heads. Second, we introduce the Interaction-aware Pose Estimation (IPE) task to learn interaction-relevant human pose features using a disentangled decoder. This is achieved with a novel attention module that accurately captures the human keypoints relevant to the current interaction category. Finally, our approach fuses the appearance feature and pose feature via element-wise addition to form the interaction representation. Experimental results show that our approach can be readily applied to existing one-stage HOI detectors. Moreover, we achieve state-of-the-art performance on two benchmarks: HICO-DET and V-COCO.

Jianbin Zheng, Daqing Liu, Chaoyue Wang et al.

Existing multimodal conditional image synthesis (MCIS) methods generate images conditioned on any combinations of various modalities that require all of them must be exactly conformed, hindering the synthesis controllability and leaving the potential of cross-modality under-exploited. To this end, we propose to generate images conditioned on the compositions of multimodal control signals, where modalities are imperfectly complementary, i.e., composed multimodal conditional image synthesis (CMCIS). Specifically, we observe two challenging issues of the proposed CMCIS task, i.e., the modality coordination problem and the modality imbalance problem. To tackle these issues, we introduce a Mixture-of-Modality-Tokens Transformer (MMoT) that adaptively fuses fine-grained multimodal control signals, a multimodal balanced training loss to stabilize the optimization of each modality, and a multimodal sampling guidance to balance the strength of each modality control signal. Comprehensive experimental results demonstrate that MMoT achieves superior performance on both unimodal conditional image synthesis (UCIS) and MCIS tasks with high-quality and faithful image synthesis on complex multimodal conditions. The project website is available at https://jabir-zheng.github.io/MMoT.