宿勇

性别: 男

职称: 工学博士，讲师，硕士生导师

毕业院校: 天津大学

毕业时间: 2019年6月

研究方向: 计算机视觉、边缘计算、智能机器人、智慧供应链

联系方式：Email:suyong@tju.edu.cn

个人教学

主讲课程:

《机器学习》

《智能科学与技术课程设计》

教学成果:

指导本科生发表多篇论文

指导多个本科生获得天津师范大学优秀本科生毕业论文

多人保研与出国留学

研究生招生名额：学硕（3人）/年；专硕（1人）/年

研究方向与成果

研究方向：

2025-至今：医学图像处理、基于强化学习的智能决策、智能控制理论

及应用

2023-2024：异常检测、姿态估计、零样本识别、动作预测、OOD

研究成果:

近几年在计算机视觉、多媒体和信号处理等相关领域重要刊物发表国际期刊会议论文40余篇，其中发表的中科院一区TOP期刊和CCF-A类顶级会议论文如下：

(1)(第一作者) Coming Out of the Dark: Human Pose Estimation in Low-Light Conditions.Conference:International Joint Conference on Artificial Intelligence(IJCAI), 2025.CCF-A Top Conference

(2)(第一作者) Semantic-Driven Dual Consistency Learning for Weakly Supervised Video Anomaly Detection.Journal:Pattern Recognition, 2025.中科院1区top

(3)(第一作者) Federated Weakly-Supervised Video Anomaly Detection with Mixture of Local-to-Global Experts.Journal:Information Fusion, 2025.中科院1区top

(4)(第一作者共同) Dual-Detector Re-optimization for Federated Weakly Supervised Video Anomaly Detection via Adaptive Dynamic Recursive Mapping.Journal:IEEE Transactions on Industrial Informatics, 2025.中科院1区top

(5)(第一作者) Anomalies Cannot Materialize or Vanish Out of Thin Air: A Hierarchical Multiple Instance Learning with Position-Scale Awareness for Video Anomaly Detection.Journal:Expert Systems with Applications, 2024.中科院1区top

(6)(第一作者) VPE-WSVAD: Visual Prompt Exemplars for Weakly-Supervised Video Anomaly Detection.Journal:Knowledge-Based Systems, 2024.中科院1区top

(7)(通讯作者) Unbiased Spatial-Temporal Atomic Fusion-Based Zero-Shot Action Recognition.Journal:Information Fusion, 2025.中科院1区top

(8)(第一作者) Prime: Privacy-Preserving Video Anomaly Detection via Motion Exemplar Guidance.Journal:Knowledge-Based Systems, 2023.中科院1区top

(9)(通讯作者) Energy-Based Temporal Summarized Attentive Network for Zero-Shot Action Recognition.Journal:IEEE Transactions on Multimedia, 2023.中科院1区top

(10)(通讯作者) Learning by Erasing: Conditional Entropy Based Transferable Out-of-Distribution Detection.Conference:AAAI Conference on Artificial Intelligence(AAAI), 2024.CCF-A Top Conference

(11)(通讯作者) Ventral & Dorsal Stream Theory Based Zero-Shot Action Recognition.Journal:Pattern Recognition, 2022.中科院1区top

(12)（第四作者）Parametric T-Spline Face Morphable Model for Detailed Fitting in Shape Subspace.Conference:IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2017.CCF-A Top Conference

(13)(第四作者) MeshPAD: Payload-Aware Mesh Distortion for 3D Steganography Based on Geometric Deep Learning.Journal:Expert Systems with Applications, 2024.中科院1区top