Peer-reviewed research papers in top-tier journals and conferences.
A novel causal representation consistency learning framework for anomaly detection in surveillance videos. The method leverages causal inference principles to improve robustness and generalization across different scenarios while maintaining real-time performance.
A novel grasp transformer architecture with pixel-wise contrastive learning for improving robotic grasp detection under sparse annotations. The method significantly enhances grasp detection accuracy while reducing annotation requirements.
A comprehensive survey of privacy-preserving techniques for video anomaly detection. The paper covers federated learning, differential privacy, and homomorphic encryption approaches while maintaining detection accuracy and real-time performance.
A novel appearance-motion prototype network for automatic video anomaly detection in industrial systems. The framework combines appearance and motion features through prototype learning to achieve superior anomaly detection performance in real-world industrial scenarios.
A novel causality-inspired representation consistency learning framework for video anomaly detection. The method leverages causal inference principles to learn consistent representations that are robust to spurious correlations and domain shifts.
A robust representation learning framework for transportation activity recognition that addresses distributional and spatial-temporal variations. The method improves recognition accuracy across different transportation modes and environmental conditions.
Research impact and academic recognition over time