Zhongyu Li

Bowen Yin, Xuying Zhang, Zhongyu Li et al.

We present DFormer, a novel RGB-D pretraining framework to learn transferable representations for RGB-D segmentation tasks. DFormer has two new key innovations: 1) Unlike previous works that encode RGB-D information with RGB pretrained backbone, we pretrain the backbone using image-depth pairs from ImageNet-1K, and hence the DFormer is endowed with the capacity to encode RGB-D representations; 2) DFormer comprises a sequence of RGB-D blocks, which are tailored for encoding both RGB and depth information through a novel building block design. DFormer avoids the mismatched encoding of the 3D geometry relationships in depth maps by RGB pretrained backbones, which widely lies in existing methods but has not been resolved. We finetune the pretrained DFormer on two popular RGB-D tasks, i.e., RGB-D semantic segmentation and RGB-D salient object detection, with a lightweight decoder head. Experimental results show that our DFormer achieves new state-of-the-art performance on these two tasks with less than half of the computational cost of the current best methods on two RGB-D semantic segmentation datasets and five RGB-D salient object detection datasets. Our code is available at: https://github.com/VCIP-RGBD/DFormer.

Zhong-Yu Li, Shanghua Gao, Ming-Ming Cheng

Learning representations with self-supervision for convolutional networks (CNN) has been validated to be effective for vision tasks. As an alternative to CNN, vision transformers (ViT) have strong representation ability with spatial self-attention and channel-level feedforward networks. Recent works reveal that self-supervised learning helps unleash the great potential of ViT. Still, most works follow self-supervised strategies designed for CNN, e.g., instance-level discrimination of samples, but they ignore the properties of ViT. We observe that relational modeling on spatial and channel dimensions distinguishes ViT from other networks. To enforce this property, we explore the feature SElf-RElation (SERE) for training self-supervised ViT. Specifically, instead of conducting self-supervised learning solely on feature embeddings from multiple views, we utilize the feature self-relations, i.e., spatial/channel self-relations, for self-supervised learning. Self-relation based learning further enhances the relation modeling ability of ViT, resulting in stronger representations that stably improve performance on multiple downstream tasks. Our source code is publicly available at: https://github.com/MCG-NKU/SERE.

Zhong-Yu Li, Bo-Wen Yin, Yongxiang Liu et al.

Incorporating heterogeneous representations from different architectures has facilitated various vision tasks, e.g., some hybrid networks combine transformers and convolutions. However, complementarity between such heterogeneous architectures has not been well exploited in self-supervised learning. Thus, we propose Heterogeneous Self-Supervised Learning (HSSL), which enforces a base model to learn from an auxiliary head whose architecture is heterogeneous from the base model. In this process, HSSL endows the base model with new characteristics in a representation learning way without structural changes. To comprehensively understand the HSSL, we conduct experiments on various heterogeneous pairs containing a base model and an auxiliary head. We discover that the representation quality of the base model moves up as their architecture discrepancy grows. This observation motivates us to propose a search strategy that quickly determines the most suitable auxiliary head for a specific base model to learn and several simple but effective methods to enlarge the model discrepancy. The HSSL is compatible with various self-supervised methods, achieving superior performances on various downstream tasks, including image classification, semantic segmentation, instance segmentation, and object detection. The codes are available at https://github.com/NK-JittorCV/Self-Supervised/.

Shanghua Gao, Zhong-Yu Li, Qi Han et al.

Temporal/spatial receptive fields of models play an important role in sequential/spatial tasks. Large receptive fields facilitate long-term relations, while small receptive fields help to capture the local details. Existing methods construct models with hand-designed receptive fields in layers. Can we effectively search for receptive field combinations to replace hand-designed patterns? To answer this question, we propose to find better receptive field combinations through a global-to-local search scheme. Our search scheme exploits both global search to find the coarse combinations and local search to get the refined receptive field combinations further. The global search finds possible coarse combinations other than human-designed patterns. On top of the global search, we propose an expectation-guided iterative local search scheme to refine combinations effectively. Our RF-Next models, plugging receptive field search to various models, boost the performance on many tasks, e.g., temporal action segmentation, object detection, instance segmentation, and speech synthesis. The source code is publicly available on http://mmcheng.net/rfnext.

Zhong-Yu Li, Xin Jin, Boyuan Sun et al.

Existing object detection methods often consider sRGB input, which was compressed from RAW data using ISP originally designed for visualization. However, such compression might lose crucial information for detection, especially under complex light and weather conditions. We introduce the AODRaw dataset, which offers 7,785 high-resolution real RAW images with 135,601 annotated instances spanning 62 categories, capturing a broad range of indoor and outdoor scenes under 9 distinct light and weather conditions. Based on AODRaw that supports RAW and sRGB object detection, we provide a comprehensive benchmark for evaluating current detection methods. We find that sRGB pre-training constrains the potential of RAW object detection due to the domain gap between sRGB and RAW, prompting us to directly pre-train on the RAW domain. However, it is harder for RAW pre-training to learn rich representations than sRGB pre-training due to the camera noise. To assist RAW pre-training, we distill the knowledge from an off-the-shelf model pre-trained on the sRGB domain. As a result, we achieve substantial improvements under diverse and adverse conditions without relying on extra pre-processing modules. Code and dataset are available at https://github.com/lzyhha/AODRaw.

Feng Chen, Zhuxiu Xu, Tianzhe Chu et al.

Data scarcity remains a fundamental bottleneck for embodied intelligence. Existing approaches use large language models (LLMs) to automate gripper-based simulation generation, but they transfer poorly to dexterous manipulation, which demands more specialized environment design. Meanwhile, dexterous manipulation tasks are inherently more difficult due to their higher degrees of freedom. Massively generating feasible and trainable dexterous hand tasks remains an open challenge. To this end, we present GenDexHand, a generative simulation pipeline that autonomously produces diverse robotic tasks and environments for dexterous manipulation. GenDexHand introduces a closed-loop refinement process that adjusts object placements and scales based on vision-language model (VLM) feedback, substantially improving the average quality of generated environments. Each task is further decomposed into sub-tasks to enable sequential reinforcement learning, reducing training time and increasing success rates. Our work provides a viable path toward scalable training of diverse dexterous hand behaviors in embodied intelligence by offering a simulation-based solution to synthetic data generation. Our website: https://winniechen2002.github.io/GenDexHand/.

Yunheng Li, ZhongYu Li, Quansheng Zeng et al.

Pre-trained vision-language models, e.g., CLIP, have been successfully applied to zero-shot semantic segmentation. Existing CLIP-based approaches primarily utilize visual features from the last layer to align with text embeddings, while they neglect the crucial information in intermediate layers that contain rich object details. However, we find that directly aggregating the multi-level visual features weakens the zero-shot ability for novel classes. The large differences between the visual features from different layers make these features hard to align well with the text embeddings. We resolve this problem by introducing a series of independent decoders to align the multi-level visual features with the text embeddings in a cascaded way, forming a novel but simple framework named Cascade-CLIP. Our Cascade-CLIP is flexible and can be easily applied to existing zero-shot semantic segmentation methods. Experimental results show that our simple Cascade-CLIP achieves superior zero-shot performance on segmentation benchmarks, like COCO-Stuff, Pascal-VOC, and Pascal-Context. Our code is available at: https://github.com/HVision-NKU/Cascade-CLIP

Jinqian Chen, Jihua Zhu, Qinghai Zheng et al.

Federated learning encounters substantial challenges with heterogeneous data, leading to performance degradation and convergence issues. While considerable progress has been achieved in mitigating such an impact, the reliability aspect of federated models has been largely disregarded. In this study, we conduct extensive experiments to investigate the reliability of both generic and personalized federated models. Our exploration uncovers a significant finding: \textbf{federated models exhibit unreliability when faced with heterogeneous data}, demonstrating poor calibration on in-distribution test data and low uncertainty levels on out-of-distribution data. This unreliability is primarily attributed to the presence of biased projection heads, which introduce miscalibration into the federated models. Inspired by this observation, we propose the "Assembled Projection Heads" (APH) method for enhancing the reliability of federated models. By treating the existing projection head parameters as priors, APH randomly samples multiple initialized parameters of projection heads from the prior and further performs targeted fine-tuning on locally available data under varying learning rates. Such a head ensemble introduces parameter diversity into the deterministic model, eliminating the bias and producing reliable predictions via head averaging. We evaluate the effectiveness of the proposed APH method across three prominent federated benchmarks. Experimental results validate the efficacy of APH in model calibration and uncertainty estimation. Notably, APH can be seamlessly integrated into various federated approaches but only requires less than 30\% additional computation cost for 100$\times$ inferences within large models.

Zhong-Yu Li, Yu-Song Hu, Bo-Wen Yin et al.

Vision representation learning, especially self-supervised learning, is pivotal for various vision applications. Ensemble learning has also succeeded in enhancing the performance and robustness of the vision models. However, traditional ensemble strategies are impractical for representation learning, especially self-supervised representation learning that requires large-scale datasets and long schedules. This is because they require k times more training and inference computation costs for an ensemble of k models. Differently, we introduce Multi-Token Enhancing (MTE) that extracts multiple auxiliary tokens simultaneously from a single model to enhance representation learning, while incurring minimal additional training costs and no additional inference costs. These auxiliary tokens, including auxiliary CLS tokens and adaptively pooled tokens, capture complementary information due to their differences. Meanwhile, to address the increase in inference costs, we distill the knowledge acquired by the auxiliary tokens into a global token during pre-training. Consequently, we can discard the auxiliary tokens during inference without incurring additional costs. Our MTE is compatible with various self-supervised loss functions and architectures, consistently improving performances across different downstream tasks. Our source code will be made publicly available.

Zuqi Huang, Mengxin Tian, Huan Liu et al.

Accurate cell counting in immunohistochemistry (IHC) images is critical for quantifying protein expression and aiding cancer diagnosis. However, the task remains challenging due to the chromogen overlap, variable biomarker staining, and diverse cellular morphologies. Regression-based counting methods offer advantages over detection-based ones in handling overlapped cells, yet rarely support end-to-end multi-class counting. Moreover, the potential of foundation models remains largely underexplored in this paradigm. To address these limitations, we propose a rank-aware agglomeration framework that selectively distills knowledge from multiple strong foundation models, leveraging their complementary representations to handle IHC heterogeneity and obtain a compact yet effective student model, CountIHC. Unlike prior task-agnostic agglomeration strategies that either treat all teachers equally or rely on feature similarity, we design a Rank-Aware Teacher Selecting (RATS) strategy that models global-to-local patch rankings to assess each teacher's inherent counting capacity and enable sample-wise teacher selection. For multi-class cell counting, we introduce a fine-tuning stage that reformulates the task as vision-language alignment. Discrete semantic anchors derived from structured text prompts encode both category and quantity information, guiding the regression of class-specific density maps and improving counting for overlapping cells. Extensive experiments demonstrate that CountIHC surpasses state-of-the-art methods across 12 IHC biomarkers and 5 tissue types, while exhibiting high agreement with pathologists' assessments. Its effectiveness on H&E-stained data further confirms the scalability of the proposed method.

Jialong Mai, Xiaofen Xing, Yawei Li et al.

Recent research has focused on applying speech large language model (SLLM) to improve speech emotion recognition (SER). However, the inherently high frame rate in speech modality severely limits the signal processing and understanding capabilities of SLLM. For example, a SLLM with a 4K context window can only process 80 seconds of audio at 50Hz feature sampling rate before reaching its capacity limit. Input token compression methods used in SLLM overlook the continuity and inertia of emotions across multiple conversation turns. This paper proposes a Dynamic Parameter Memory (DPM) mechanism with contextual semantics and sentence-level emotion encoding, enabling processing of unlimited-length audio with limited context windows in SLLM. Specifically, DPM progressively encodes sentence-level information and emotions into a temporary LoRA module during inference to effectively "memorize" the contextual information. We trained an emotion SLLM as a backbone and incorporated our DPM into inference for emotion recognition in conversation (ERC). Experimental results on the IEMOCAP dataset show that DPM significantly improves the emotion recognition capabilities of SLLM when processing long audio sequences, achieving state-of-the-art performance.

Zhong-Yu Li, Yunheng Li, Deng-Ping Fan et al.

Masked image modeling has achieved great success in learning representations but is limited by the huge computational costs. One cost-saving strategy makes the decoder reconstruct only a subset of masked tokens and throw the others, and we refer to this method as partial reconstruction. However, it also degrades the representation quality. Previous methods mitigate this issue by throwing tokens with minimal information using temporal redundancy inaccessible for static images or attention maps that incur extra costs and complexity. To address these limitations, we propose a progressive reconstruction strategy and a furthest sampling strategy to reconstruct those thrown tokens in an extremely lightweight way instead of completely abandoning them. This approach involves all masked tokens in supervision to ensure adequate pre-training, while maintaining the cost-reduction benefits of partial reconstruction. We validate the effectiveness of the proposed method across various existing frameworks. For example, when throwing 50% patches, we can achieve lossless performance of the ViT-B/16 while saving 28% FLOPs and 36% memory usage compared to standard MAE. Our source code will be made publicly available

Yunheng Li, Zhongyu Li, Shanghua Gao et al.

Effectively modeling discriminative spatio-temporal information is essential for segmenting activities in long action sequences. However, we observe that existing methods are limited in weak spatio-temporal modeling capability due to two forms of decoupled modeling: (i) cascaded interaction couples spatial and temporal modeling, which over-smooths motion modeling over the long sequence, and (ii) joint-shared temporal modeling adopts shared weights to model each joint, ignoring the distinct motion patterns of different joints. We propose a Decoupled Spatio-Temporal Framework (DeST) to address the above issues. Firstly, we decouple the cascaded spatio-temporal interaction to avoid stacking multiple spatio-temporal blocks, while achieving sufficient spatio-temporal interaction. Specifically, DeST performs once unified spatial modeling and divides the spatial features into different groups of subfeatures, which then adaptively interact with temporal features from different layers. Since the different sub-features contain distinct spatial semantics, the model could learn the optimal interaction pattern at each layer. Meanwhile, inspired by the fact that different joints move at different speeds, we propose joint-decoupled temporal modeling, which employs independent trainable weights to capture distinctive temporal features of each joint. On four large-scale benchmarks of different scenes, DeST significantly outperforms current state-of-the-art methods with less computational complexity.

Shanghua Gao, Zhong-Yu Li, Ming-Hsuan Yang et al.

Empowered by large datasets, e.g., ImageNet, unsupervised learning on large-scale data has enabled significant advances for classification tasks. However, whether the large-scale unsupervised semantic segmentation can be achieved remains unknown. There are two major challenges: i) we need a large-scale benchmark for assessing algorithms; ii) we need to develop methods to simultaneously learn category and shape representation in an unsupervised manner. In this work, we propose a new problem of large-scale unsupervised semantic segmentation (LUSS) with a newly created benchmark dataset to help the research progress. Building on the ImageNet dataset, we propose the ImageNet-S dataset with 1.2 million training images and 50k high-quality semantic segmentation annotations for evaluation. Our benchmark has a high data diversity and a clear task objective. We also present a simple yet effective method that works surprisingly well for LUSS. In addition, we benchmark related un/weakly/fully supervised methods accordingly, identifying the challenges and possible directions of LUSS. The benchmark and source code is publicly available at https://github.com/LUSSeg.

Shang-Hua Gao, Qi Han, Zhong-Yu Li et al.

Temporal receptive fields of models play an important role in action segmentation. Large receptive fields facilitate the long-term relations among video clips while small receptive fields help capture the local details. Existing methods construct models with hand-designed receptive fields in layers. Can we effectively search for receptive field combinations to replace hand-designed patterns? To answer this question, we propose to find better receptive field combinations through a global-to-local search scheme. Our search scheme exploits both global search to find the coarse combinations and local search to get the refined receptive field combination patterns further. The global search finds possible coarse combinations other than human-designed patterns. On top of the global search, we propose an expectation guided iterative local search scheme to refine combinations effectively. Our global-to-local search can be plugged into existing action segmentation methods to achieve state-of-the-art performance.

Qinghai Zheng, Yu Zhang, Jihua Zhu et al.

As a hot research topic, many multi-view clustering approaches are proposed over the past few years. Nevertheless, most existing algorithms merely take the consensus information among different views into consideration for clustering. Actually, it may hinder the multi-view clustering performance in real-life applications, since different views usually contain diverse statistic properties. To address this problem, we propose a novel Tensor-based Intrinsic Subspace Representation Learning (TISRL) for multi-view clustering in this paper. Concretely, the rank preserving decomposition is proposed firstly to effectively deal with the diverse statistic information contained in different views. Then, to achieve the intrinsic subspace representation, the tensor-singular value decomposition based low-rank tensor constraint is also utilized in our method. It can be seen that specific information contained in different views is fully investigated by the rank preserving decomposition, and the high-order correlations of multi-view data are also mined by the low-rank tensor constraint. The objective function can be optimized by an augmented Lagrangian multiplier based alternating direction minimization algorithm. Experimental results on nine common used real-world multi-view datasets illustrate the superiority of TISRL.