Research Appointments
Baiyangdian Ecological IoT Monitoring System
Studied efficient object detection for mining trucks and low-altitude UAV monitoring under practical deployment constraints.
- Developed mining-truck detection using YOLOv7-tiny and Darknet, improving recognition accuracy by 2% with minimal FPS loss.
- Enhanced infrared-visible UAV object detection with YOLOv8, achieving 65% mAP at 30 FPS on a laptop while targeting edge deployment.
Collaborative Research Projects
National Undergraduate Innovation and Entrepreneurship Training
Project lead, 202411104021 (2024), working on UAV perception, automated inspection, and edge inference. Also participated in collaborating student projects 202511104021, 202511104025, and 202611104287.
Multimodal Perception and Emergency Decision Research
Research participant, National Key R&D projects 2024YFC3016800 and 2024YFC3016805. Contributed to H2T-DEIM, TRIM, ManiSeg, visual episodic-memory diagnostics, and EmergUAV-Bench through model implementation, evaluation pipelines, feature analysis, and multimodal benchmarking.