Yun Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yun Wang has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Dapeng Oliver Wu
City University of Hong Kong
6 shared publications - Yulan Guo
Sun Yat-sen University
6 shared publications - Longguang Wang
Sun Yat-sen University
5 shared publications - Zhanjie Zhang
Zhejiang University
3 shared publications - Jiahao Zheng
Department of Computer Science, City University of Hong Kong, Hong Kong SAR, China
2 shared publications - Junjie Hu
School of Artificial Intelligence, The Chinese University of Hong Kong, Shenzhen, China
2 shared publications
Latest Publications
- 3 October 2026
- Published by Springer Nature
- p. 111-127
InnoText: A Unified Model for Visual Text Generation and Editing
- 3 October 2026
- Published by Springer Nature
- p. 111-127
InnoText: A Unified Model for Visual Text Generation and Editing
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (5) , 3586-3601
SageIQ: Scene-Graph-Guided Blind Image Quality Assessment
Conventional Blind Image Quality Assessment (BIQA) methods typically assess the entire image quality, which is suboptimal for tasks like autonomous driving that concern specific Task-Aligned Region (TAR). Moreover, we observe that advanced Multimodal Large Language Model (MLLM)-based BIQA models exhibit bias when evaluating small regions, leading to inaccurate perceptual judgments. To address these issues, we propose SageIQ (Scene-graph-guided Evaluation for Image Quality), a pipeline SageIQ-P for TAR localization, an approach consisting of an MLLM-based BIQA model SageIQ-M, and a dataset SageIQ-D. SageIQ-P is designed to automatically identify and evaluate TARs, with the advantages of being training-free and allowing plug-in integration of off-the-shelf BIQA models without retraining. It operates in three stages: scenegraphbased triplet construction, LLMdriven triplet analysis, and integration of weighted BIQA scores into a final assessment. Since SageIQ-P can produce small-sized TAR crops that may encounter small-region scoring bias in existing BIQA models, we propose SageIQ-M to alleviate this bias by injecting scale information through scale-aware images and size-prompted cues, achieving size awareness across both visual and textual modalities. In addition, we develop a fully automated approach to construct a region-level test set SageIQ-D, significantly reducing the human effort needed. Experimental results demonstrate that our methods achieve superior BIQA performance.
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (5) , 3586-3601
SageIQ: Scene-Graph-Guided Blind Image Quality Assessment
Conventional Blind Image Quality Assessment (BIQA) methods typically assess the entire image quality, which is suboptimal for tasks like autonomous driving that concern specific Task-Aligned Region (TAR). Moreover, we observe that advanced Multimodal Large Language Model (MLLM)-based BIQA models exhibit bias when evaluating small regions, leading to inaccurate perceptual judgments. To address these issues, we propose SageIQ (Scene-graph-guided Evaluation for Image Quality), a pipeline SageIQ-P for TAR localization, an approach consisting of an MLLM-based BIQA model SageIQ-M, and a dataset SageIQ-D. SageIQ-P is designed to automatically identify and evaluate TARs, with the advantages of being training-free and allowing plug-in integration of off-the-shelf BIQA models without retraining. It operates in three stages: scenegraphbased triplet construction, LLMdriven triplet analysis, and integration of weighted BIQA scores into a final assessment. Since SageIQ-P can produce small-sized TAR crops that may encounter small-region scoring bias in existing BIQA models, we propose SageIQ-M to alleviate this bias by injecting scale information through scale-aware images and size-prompted cues, achieving size awareness across both visual and textual modalities. In addition, we develop a fully automated approach to construct a region-level test set SageIQ-D, significantly reducing the human effort needed. Experimental results demonstrate that our methods achieve superior BIQA performance.
- 27 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 4494-4508
SMFormer: Empowering Self-Supervised Stereo Matching via Foundation Models and Data Augmentation
Recent self-supervised stereo matching methods have made significant progress. They typically rely on the photometric consistency assumption, which presumes corresponding points across views share the same appearance. However, this assumption could be compromised by real-world disturbances, resulting in invalid supervisory signals and a significant accuracy gap compared to supervised methods. To address this issue, we propose SMFormer, a framework integrating more reliable self-supervision guided by the Vision Foundation Model (VFM) and data augmentation. We first incorporate the VFM with the Feature Pyramid Network (FPN), providing a discriminative and robust feature representation against disturbance in various scenarios. We then devise an effective data augmentation mechanism that ensures robustness to various transformations. The data augmentation mechanism explicitly enforces consistency between learned features and those influenced by illumination variations. Additionally, it regularizes the output consistency between disparity predictions of strong augmented samples and those generated from standard samples. Experiments on multiple mainstream benchmarks demonstrate that our SMFormer achieves state-of-the-art (SOTA) performance among self-supervised methods and even competes on par with supervised ones. Remarkably, in the challenging Booster benchmark, SMFormer even outperforms some SOTA supervised methods, such as CFNet.
- 27 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 4494-4508
SMFormer: Empowering Self-Supervised Stereo Matching via Foundation Models and Data Augmentation
Recent self-supervised stereo matching methods have made significant progress. They typically rely on the photometric consistency assumption, which presumes corresponding points across views share the same appearance. However, this assumption could be compromised by real-world disturbances, resulting in invalid supervisory signals and a significant accuracy gap compared to supervised methods. To address this issue, we propose SMFormer, a framework integrating more reliable self-supervision guided by the Vision Foundation Model (VFM) and data augmentation. We first incorporate the VFM with the Feature Pyramid Network (FPN), providing a discriminative and robust feature representation against disturbance in various scenarios. We then devise an effective data augmentation mechanism that ensures robustness to various transformations. The data augmentation mechanism explicitly enforces consistency between learned features and those influenced by illumination variations. Additionally, it regularizes the output consistency between disparity predictions of strong augmented samples and those generated from standard samples. Experiments on multiple mainstream benchmarks demonstrate that our SMFormer achieves state-of-the-art (SOTA) performance among self-supervised methods and even competes on par with supervised ones. Remarkably, in the challenging Booster benchmark, SMFormer even outperforms some SOTA supervised methods, such as CFNet.
- 2 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 34 (5) , 1430-1444
General Diffusion Transformer for Arbitrary Medical Image Translation
Medical image translation plays a crucial role in assisting clinical diagnosis by enabling cross-modal synthesis (e.g., Computed Tomography to Magnetic Resonance Imaging) and super-resolution, effectively addressing clinical challenges such as radiation exposure, prolonged scan times, and allergic reactions to contrast agents. However, existing approaches primarily focus on developing specialized models for specific tasks, limiting their adaptability across different applications. Developing a general model capable of handling arbitrary medical image translation tasks not only enhances cross-domain generalization but also aligns with the broader trend of artificial intelligence evolving from specialized to general-purpose solutions. Achieving such one-for-all model, however, presents three key challenges: (1) varying task complexity, (2) modality discrepancies, and (3) structural variations across anatomical regions within the same modality. To tackle the first challenge, we utilize an advanced diffusion-based training paradigm to endow the denoising model with extensive pattern coverage capabilities, thereby handling tasks of varying difficulty levels. Subsequently, a General Diffusion Transformer incorporating a Fuzzy Mixture-of-Experts (FMoE) module and an Entropy-guided Attention Soft Prompt (EASP) module is proposed. The FMoE module, equipped with nonlinear modeling capabilities, is designed to address modality discrepancies, while the EASP module is employed to enhance the model's perception of structural variations in images. Extensive qualitative and quantitative experiments demonstrate the effectiveness of the proposed model in the arbitrary medical image translation task.
- 2 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 34 (5) , 1430-1444
General Diffusion Transformer for Arbitrary Medical Image Translation
Medical image translation plays a crucial role in assisting clinical diagnosis by enabling cross-modal synthesis (e.g., Computed Tomography to Magnetic Resonance Imaging) and super-resolution, effectively addressing clinical challenges such as radiation exposure, prolonged scan times, and allergic reactions to contrast agents. However, existing approaches primarily focus on developing specialized models for specific tasks, limiting their adaptability across different applications. Developing a general model capable of handling arbitrary medical image translation tasks not only enhances cross-domain generalization but also aligns with the broader trend of artificial intelligence evolving from specialized to general-purpose solutions. Achieving such one-for-all model, however, presents three key challenges: (1) varying task complexity, (2) modality discrepancies, and (3) structural variations across anatomical regions within the same modality. To tackle the first challenge, we utilize an advanced diffusion-based training paradigm to endow the denoising model with extensive pattern coverage capabilities, thereby handling tasks of varying difficulty levels. Subsequently, a General Diffusion Transformer incorporating a Fuzzy Mixture-of-Experts (FMoE) module and an Entropy-guided Attention Soft Prompt (EASP) module is proposed. The FMoE module, equipped with nonlinear modeling capabilities, is designed to address modality discrepancies, while the EASP module is employed to enhance the model's perception of structural variations in images. Extensive qualitative and quantitative experiments demonstrate the effectiveness of the proposed model in the arbitrary medical image translation task.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35, 9184-9197
CCD: Continual Consistency Diffusion for Lifelong Generative Modeling
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35, 9184-9197
CCD: Continual Consistency Diffusion for Lifelong Generative Modeling
- 16 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 4123-4137
RoSe: Robust Self-Supervised Stereo Matching Under Adverse Weather Conditions
Recent self-supervised stereo matching methods have made significant progress, but their performance significantly degrades under adverse weather conditions such as night, rain, and fog. We identify two primary weaknesses contributing to this performance degradation. First, adverse weather introduces noise and reduces visibility, making CNN-based feature extractors struggle with degraded regions like reflective and textureless areas. Second, these degraded regions can disrupt accurate pixel correspondences, leading to ineffective supervision based on the photometric consistency assumption. To address these challenges, we propose injecting robust priors derived from the visual foundation model into the CNN-based feature extractor to improve feature representation under adverse weather conditions. We then introduce scene correspondence priors to construct robust supervisory signals rather than relying solely on the photometric consistency assumption. Specifically, we create synthetic stereo datasets with realistic weather degradations. These datasets feature clear and adverse image pairs that maintain the same semantic context and disparity, preserving the scene correspondence property. With this knowledge, we propose a robust self-supervised training paradigm, consisting of two key steps: robust self-supervised scene correspondence learning and adverse weather distillation. Both steps aim to align underlying scene results from clean and adverse image pairs, thus improving model disparity estimation under adverse weather effects. Extensive experiments demonstrate the effectiveness and versatility of our proposed solution, which outperforms existing state-of-the-art self-supervised methods. Codes are available at https://github.com/cocowy1/RoSe-Robust-Self-supervised-Stereo-Matching-under-Adverse-Weather-Conditions.
- 16 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 4123-4137
RoSe: Robust Self-Supervised Stereo Matching Under Adverse Weather Conditions
Recent self-supervised stereo matching methods have made significant progress, but their performance significantly degrades under adverse weather conditions such as night, rain, and fog. We identify two primary weaknesses contributing to this performance degradation. First, adverse weather introduces noise and reduces visibility, making CNN-based feature extractors struggle with degraded regions like reflective and textureless areas. Second, these degraded regions can disrupt accurate pixel correspondences, leading to ineffective supervision based on the photometric consistency assumption. To address these challenges, we propose injecting robust priors derived from the visual foundation model into the CNN-based feature extractor to improve feature representation under adverse weather conditions. We then introduce scene correspondence priors to construct robust supervisory signals rather than relying solely on the photometric consistency assumption. Specifically, we create synthetic stereo datasets with realistic weather degradations. These datasets feature clear and adverse image pairs that maintain the same semantic context and disparity, preserving the scene correspondence property. With this knowledge, we propose a robust self-supervised training paradigm, consisting of two key steps: robust self-supervised scene correspondence learning and adverse weather distillation. Both steps aim to align underlying scene results from clean and adverse image pairs, thus improving model disparity estimation under adverse weather effects. Extensive experiments demonstrate the effectiveness and versatility of our proposed solution, which outperforms existing state-of-the-art self-supervised methods. Codes are available at https://github.com/cocowy1/RoSe-Robust-Self-supervised-Stereo-Matching-under-Adverse-Weather-Conditions.
- 10 May 2025
- Published by Elsevier in Neural Networks
- Vol. 189, 107556
SPAST: Arbitrary style transfer with style priors via pre-trained large-scale model
- 10 May 2025
- Published by Elsevier in Neural Networks
- Vol. 189, 107556
SPAST: Arbitrary style transfer with style priors via pre-trained large-scale model
- 14 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 34 (10577149) , 1204-1218
ADStereo: Efficient Stereo Matching With Adaptive Downsampling and Disparity Alignment
The balance between accuracy and computational efficiency is crucial for the applications of deep learning-based stereo matching algorithms in real-world scenarios. Since matching cost aggregation is usually the most computationally expensive component, a common practice is to construct cost volumes at a low resolution for aggregation and then directly regress a high-resolution disparity map. However, current solutions often suffer from limitations such as the loss of discriminative features caused by downsampling operations that treat all pixels equally, and spatial misalignment resulting from repeated downsampling and upsampling. To overcome these challenges, this paper presents two sampling strategies: the Adaptive Downsampling Module (ADM) and the Disparity Alignment Module (DAM), to prioritize real-time inference while ensuring accuracy. The ADM leverages local features to learn adaptive weights, enabling more effective downsampling while preserving crucial structure information. On the other hand, the DAM employs a learnable interpolation strategy to predict transformation offsets of pixels, thereby mitigating the spatial misalignment issue. Building upon these modules, we introduce ADStereo, a real-time yet accurate network that achieves highly competitive performance on multiple public benchmarks. Specifically, our ADStereo runs over$5\times $ faster than the current state-of-the-art CREStereo (0.054s vs.$0.29{s}$ ) under the same hardware while achieving comparable accuracy (1.82% vs. 1.69%) on the KITTI stereo 2015 benchmark. The codes are available at: https://github.com/cocowy1/ADStereo.- 14 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 34 (10577149) , 1204-1218
ADStereo: Efficient Stereo Matching With Adaptive Downsampling and Disparity Alignment
The balance between accuracy and computational efficiency is crucial for the applications of deep learning-based stereo matching algorithms in real-world scenarios. Since matching cost aggregation is usually the most computationally expensive component, a common practice is to construct cost volumes at a low resolution for aggregation and then directly regress a high-resolution disparity map. However, current solutions often suffer from limitations such as the loss of discriminative features caused by downsampling operations that treat all pixels equally, and spatial misalignment resulting from repeated downsampling and upsampling. To overcome these challenges, this paper presents two sampling strategies: the Adaptive Downsampling Module (ADM) and the Disparity Alignment Module (DAM), to prioritize real-time inference while ensuring accuracy. The ADM leverages local features to learn adaptive weights, enabling more effective downsampling while preserving crucial structure information. On the other hand, the DAM employs a learnable interpolation strategy to predict transformation offsets of pixels, thereby mitigating the spatial misalignment issue. Building upon these modules, we introduce ADStereo, a real-time yet accurate network that achieves highly competitive performance on multiple public benchmarks. Specifically, our ADStereo runs over$5\times $ faster than the current state-of-the-art CREStereo (0.054s vs.$0.29{s}$ ) under the same hardware while achieving comparable accuracy (1.82% vs. 1.69%) on the KITTI stereo 2015 benchmark. The codes are available at: https://github.com/cocowy1/ADStereo.- 11 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 33 (10577149) , 6425-6438
Cost Volume Aggregation in Stereo Matching Revisited: A Disparity Classification Perspective
Cost aggregation plays a critical role in existing stereo matching methods. In this paper, we revisit cost aggregation in stereo matching from disparity classification and propose a generic yet efficient Disparity Context Aggregation (DCA) module to improve the performance of CNN-based methods. Our approach is based on an insight that a coarse disparity class prior is beneficial to disparity regression. To obtain such a prior, we first classify pixels in an image into several disparity classes and treat pixels within the same class as homogeneous regions. We then generate homogeneous region representations and incorporate these representations into the cost volume to suppress irrelevant information while enhancing the matching ability for cost aggregation. With the help of homogeneous region representations, efficient and informative cost aggregation can be achieved with only a shallow 3D CNN. Our DCA module is fully-differentiable and well-compatible with different network architectures, which can be seamlessly plugged into existing networks to improve performance with small additional overheads. It is demonstrated that our DCA module can effectively exploit disparity class priors to improve the performance of cost aggregation. Based on our DCA, we design a highly accurate network named DCANet, which achieves state-of-the-art performance on several benchmarks.
- 11 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 33 (10577149) , 6425-6438
Cost Volume Aggregation in Stereo Matching Revisited: A Disparity Classification Perspective
Cost aggregation plays a critical role in existing stereo matching methods. In this paper, we revisit cost aggregation in stereo matching from disparity classification and propose a generic yet efficient Disparity Context Aggregation (DCA) module to improve the performance of CNN-based methods. Our approach is based on an insight that a coarse disparity class prior is beneficial to disparity regression. To obtain such a prior, we first classify pixels in an image into several disparity classes and treat pixels within the same class as homogeneous regions. We then generate homogeneous region representations and incorporate these representations into the cost volume to suppress irrelevant information while enhancing the matching ability for cost aggregation. With the help of homogeneous region representations, efficient and informative cost aggregation can be achieved with only a shallow 3D CNN. Our DCA module is fully-differentiable and well-compatible with different network architectures, which can be seamlessly plugged into existing networks to improve performance with small additional overheads. It is demonstrated that our DCA module can effectively exploit disparity class priors to improve the performance of cost aggregation. Based on our DCA, we design a highly accurate network named DCANet, which achieves state-of-the-art performance on several benchmarks.
- 2 October 2024
- Published by Springer Nature
- p. 146-162
Learning Representations from Foundation Models for Domain Generalized Stereo Matching
- 2 October 2024
- Published by Springer Nature
- p. 146-162
Learning Representations from Foundation Models for Domain Generalized Stereo Matching