主要研究方向为计算机视觉语义理解、图像内容安全、异常检测、知识蒸馏、大数据分析等，具体涉及对图像进行深度解析与推理，实现场景理解、物体检测与识别、属性分析、行为识别等，结合语义分割与关系建模，使机器能够“看懂”图像内容及其逻辑关联，对违规、敏感或有害视觉内容进行识别与过滤，对视频或图像序列中的非常规事件或行为，通过时序建模与弱监督蒸馏学习，在安防监控、工业质检等场景中实时发现异常模式等。
相关技术应用于分布式光伏智慧运维系统、知识图谱构建与推演系统、平台服务能力仿真工具、通信数据业务信息库等信息化服务平台研发，相关成果在国家电网、航天一院、中国船舶集团、中国信息通信研究院等多家重点企事业单位应用。
团队长期招收实习生、硕博研究生，欢迎投递简历至
[email protected]。
[1] Jianyuan Wang, Liang She*, Wei Wang, et al. Distilled Center and Scale Prediction: Distilling anchor-free pedestrian detector for edge computing[J]. Internet of Things, 2025, (29), 101444
[2] Jianyuan Wang, Heng Du, Yangyan Zeng*. PDFormer: Efficient Vision Transformer for Photovoltaic Defect Detection[J]. IEEE Transactions on Consumer Electronics, 2025, 71, (2), 6602 - 6611
[3] Jianyuan Wang, Kexin Liu*, Yucheng Zhang, et al. Recent Advances of Few-shot Learning Methods and Applications[J]. SCIENCE CHINA Technological Sciences, 2023, 66, 920–944
[4] Xinyue Liu, Jianyuan Wang*, Biao Leng, et al. Dual-Modeling Decouple Distillation for Unsupervised Anomaly Detection[C]. in Proceedings of the 32nd ACM International Conference on Multimedia, 2024, 10, pp. 5035-5044
[5] Xinyue Liu, Jianyuan Wang*, Biao Leng, et al. Unlocking the Potential of Reverse Distillation for Anomaly Detection[C]. in Proceedings of the 39th AAAI Conference on Artificial Intelligence, 2025, 4, pp. 5640-5648
