Wen Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Wen Wang has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Qingyong Li
Beijing Jiaotong University
13 shared publications Qingyong Li
7 shared publications- Wenju Sun
Beijing Jiaotong University
5 shared publications - Yang Liu
Key Laboratory of Big Data and Artificial Intelligence in Transportation, Beijing Jiaotong University, Beijing, China
5 shared publications - Yangli-Ao Geng
Key Laboratory of Big Data & Artificial Intelligence in Transportation (Ministry of Education), Beijing Jiaotong University, Beijing 100044, China
4 shared publications Fengjiao Liang
Key Laboratory of Big Data and Artificial Intelligence in Transportation, Beijing Jiaotong University, Beijing, China
2 shared publications
Latest Publications
- 9 October 2026
- Published by Elsevier in Pattern Recognition
- Vol. 183
MaskCL: Mask-driven Cross-Modal Collaborative Learning for multimodal anomaly detection
- 9 October 2026
- Published by Elsevier in Pattern Recognition
- Vol. 183
MaskCL: Mask-driven Cross-Modal Collaborative Learning for multimodal anomaly detection
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (10) , 9257-9267
CKM: Continual Anomaly Detection Based on Compact Knowledge Memory
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (10) , 9257-9267
CKM: Continual Anomaly Detection Based on Compact Knowledge Memory
- 18 September 2026
- Published by Springer Nature
- p. 131-145
CSA-3D: Cross-View Semantic Alignment for Unsupervised Point Cloud Segmentation
- 18 September 2026
- Published by Springer Nature
- p. 131-145
CSA-3D: Cross-View Semantic Alignment for Unsupervised Point Cloud Segmentation
- 13 September 2026
- Published by Springer Nature
- p. 573-591
Uncertainty-Driven Gaussian Sphere Propagation for 3D Semantic Segmentation
- 13 September 2026
- Published by Springer Nature
- p. 573-591
Uncertainty-Driven Gaussian Sphere Propagation for 3D Semantic Segmentation
- Open Access
- 23 June 2026
- Published by Elsevier in SSRN Electronic Journal
MaskCL: Mask-driven Cross-Modal Collaborative Learning for Multimodal Anomaly Detection
Multimodal industrial anomaly detection has recently advanced by leveraging the complementary nature of 2D textures and 3D geometries. Among existing strategies, mapping-based paradigms have emerged as a prominent solution for establishing cross-modal alignment. However, such methods often suffer from a mapping shortcut, where the model gravitates toward trivial point-wise numerical fitting instead of capturing global structural correspondences. To address this issue, we propose Mask-driven Cross-Modal Collaborative Learning (MaskCL), a novel multimodal anomaly detection framework that shifts from passive feature mapping to active structural reasoning under information constraints. By intentionally imposing an information bottleneck through a Foreground-aware Random Masking (FRM) module, we obstruct the trivial identity mapping and compel the framework to perform active semantic inference by leveraging the inherent physical consistency between modalities. To effectively process incomplete cross-modal feature representations induced by masking, we further develop a Collaborative Feature Learning (CFL) architecture that reconciles intra-modal structural reconstruction with cross-modal semantic alignment. Extensive experiments on the MVTec 3D-AD and Eyecandies benchmarks demonstrate that MaskCL achieves state-of-the-art performance, offering superior anomaly localization and robustness even in few-shot scenarios. Code will be available at: https://github.com/lilifai422/MaskCL-Mask-driven-Collaborative-Cross-Modal-Learning
- 23 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMaskCL: Mask-driven Cross-Modal Collaborative Learning for Multimodal Anomaly Detection
Multimodal industrial anomaly detection has recently advanced by leveraging the complementary nature of 2D textures and 3D geometries. Among existing strategies, mapping-based paradigms have emerged as a prominent solution for establishing cross-modal alignment. However, such methods often suffer from a mapping shortcut, where the model gravitates toward trivial point-wise numerical fitting instead of capturing global structural correspondences. To address this issue, we propose Mask-driven Cross-Modal Collaborative Learning (MaskCL), a novel multimodal anomaly detection framework that shifts from passive feature mapping to active structural reasoning under information constraints. By intentionally imposing an information bottleneck through a Foreground-aware Random Masking (FRM) module, we obstruct the trivial identity mapping and compel the framework to perform active semantic inference by leveraging the inherent physical consistency between modalities. To effectively process incomplete cross-modal feature representations induced by masking, we further develop a Collaborative Feature Learning (CFL) architecture that reconciles intra-modal structural reconstruction with cross-modal semantic alignment. Extensive experiments on the MVTec 3D-AD and Eyecandies benchmarks demonstrate that MaskCL achieves state-of-the-art performance, offering superior anomaly localization and robustness even in few-shot scenarios. Code will be available at: https://github.com/lilifai422/MaskCL-Mask-driven-Collaborative-Cross-Modal-Learning
- 20 December 2025
- Published by Elsevier in Pattern Recognition
- Vol. 174
Towards plastic and stable incremental learning: A dual-learner framework with cumulative parameter averaging
- 20 December 2025
- Published by Elsevier in Pattern Recognition
- Vol. 174
Towards plastic and stable incremental learning: A dual-learner framework with cumulative parameter averaging
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 5178-5187
Task Arithmetic in Trust Region: A Training-Free Model Merging Approach to Navigate Knowledge Conflicts
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 5178-5187
Open AccessTask Arithmetic in Trust Region: A Training-Free Model Merging Approach to Navigate Knowledge Conflicts
- 5 August 2025
- Published by Elsevier in Neurocomputing
- Vol. 653
Decoupled likelihood modeling: A scalable approach for incremental generalized category discovery
- 5 August 2025
- Published by Elsevier in Neurocomputing
- Vol. 653
Decoupled likelihood modeling: A scalable approach for incremental generalized category discovery
- 19 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
LECLIP: Boosting Zero-Shot Anomaly Detection With Local Enhanced CLIP
Zero-shot anomaly detection (ZSAD) is a critical task that detects anomalies without any training samples from the target application, which is crucial for applications in diverse fields such as industrial quality control and medical imaging analysis. Recent advances have seen the application of contrastive language-image pretraining (CLIP) in ZSAD, exploiting its robust visual-linguistic alignment and zero-shot learning capabilities. However, CLIP is primarily designed for natural image classification, emphasizing global visual embeddings, while anomaly detection (AD) requires a more accurate representation of anomalous regions and more precise local visual embeddings. To overcome these limitations, this article proposes the local enhanced CLIP (LECLIP) framework for ZSAD. LECLIP incorporates a local alignment (LA) module that divides images into blocks and aligns them with learnable text embeddings, ensuring precise relevance expression. Furthermore, a training-free echo-attention (EA) is proposed to complement the traditional QKV attention, enabling the model to capture both global and local image details effectively, thus providing a more accurate and detailed image representation. Experimental results show that LECLIP achieves superior performance on 15 challenging datasets, including six industrial datasets and nine medical datasets. Code is available at https://github.com/lyy70/LECLIP
- 19 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
LECLIP: Boosting Zero-Shot Anomaly Detection With Local Enhanced CLIP
Zero-shot anomaly detection (ZSAD) is a critical task that detects anomalies without any training samples from the target application, which is crucial for applications in diverse fields such as industrial quality control and medical imaging analysis. Recent advances have seen the application of contrastive language-image pretraining (CLIP) in ZSAD, exploiting its robust visual-linguistic alignment and zero-shot learning capabilities. However, CLIP is primarily designed for natural image classification, emphasizing global visual embeddings, while anomaly detection (AD) requires a more accurate representation of anomalous regions and more precise local visual embeddings. To overcome these limitations, this article proposes the local enhanced CLIP (LECLIP) framework for ZSAD. LECLIP incorporates a local alignment (LA) module that divides images into blocks and aligns them with learnable text embeddings, ensuring precise relevance expression. Furthermore, a training-free echo-attention (EA) is proposed to complement the traditional QKV attention, enabling the model to capture both global and local image details effectively, thus providing a more accurate and detailed image representation. Experimental results show that LECLIP achieves superior performance on 15 challenging datasets, including six industrial datasets and nine medical datasets. Code is available at https://github.com/lyy70/LECLIP
- Open Access
CrackCLIP: Adapting Vision-Language Models for Weakly Supervised Crack Segmentation
Weakly supervised crack segmentation aims to create pixel-level crack masks with minimal human annotation, which often only differentiate between crack and normal no-crack patches. This task is crucial for assessing structural integrity and safety in real-world industrial applications, where manually labeling the location of cracks at the pixel level is both labor-intensive and impractical. Addressing the challenges of labeling uncertainty, this paper presents CrackCLIP, a novel approach that leverages language prompts to augment the semantic context and employs the Contrastive Language–Image Pre-Training (CLIP) model to enhance weakly supervised crack segmentation. Initially, a gradient-based class activation map is used to generate pixel-level coarse pseudo-labels from a trained crack patch classifier. The estimated coarse pseudo-labels are utilized to fine-tune additional linear adapters, which are integrated into the frozen image encoders of CLIP to adapt the CLIP model to the specialized task of crack segmentation. Moreover, specific textual prompts are crafted for crack characteristics, which are input into the frozen text encoder of CLIP to extract features encapsulating the semantic essence of the cracks. The final crack segmentation is determined by comparing the similarity between text prompt features and visual patch token features. Comparative experiments on the Crack500, CFD, and DeepCrack datasets demonstrate that the proposed framework outperforms existing weakly supervised crack segmentation methods, and the pre-trained vision-language model exhibits strong potential for crack feature learning, thereby enhancing the overall performance and generalization capabilities of the proposed framework.
Open AccessCrackCLIP: Adapting Vision-Language Models for Weakly Supervised Crack Segmentation
Weakly supervised crack segmentation aims to create pixel-level crack masks with minimal human annotation, which often only differentiate between crack and normal no-crack patches. This task is crucial for assessing structural integrity and safety in real-world industrial applications, where manually labeling the location of cracks at the pixel level is both labor-intensive and impractical. Addressing the challenges of labeling uncertainty, this paper presents CrackCLIP, a novel approach that leverages language prompts to augment the semantic context and employs the Contrastive Language–Image Pre-Training (CLIP) model to enhance weakly supervised crack segmentation. Initially, a gradient-based class activation map is used to generate pixel-level coarse pseudo-labels from a trained crack patch classifier. The estimated coarse pseudo-labels are utilized to fine-tune additional linear adapters, which are integrated into the frozen image encoders of CLIP to adapt the CLIP model to the specialized task of crack segmentation. Moreover, specific textual prompts are crafted for crack characteristics, which are input into the frozen text encoder of CLIP to extract features encapsulating the semantic essence of the cracks. The final crack segmentation is determined by comparing the similarity between text prompt features and visual patch token features. Comparative experiments on the Crack500, CFD, and DeepCrack datasets demonstrate that the proposed framework outperforms existing weakly supervised crack segmentation methods, and the pre-trained vision-language model exhibits strong potential for crack feature learning, thereby enhancing the overall performance and generalization capabilities of the proposed framework.