On June 26, 2026, Dr. Yulin Zhu from the Department of Computer Science at Hong Kong Chu Hai College, along with MSAAI student Ying Wang, presented their latest research findings at 2026 IEEE 7th International Conference on Artificial Intelligence and Electromechanical Automation (AIEA2026).
- Ying Wang, Wai Lun Lo, Tai Chiu Hsung, Sik Ho Tsang, Xiaoxing Yang, Zhen Zhang, Yuni Lai, Xiaoyu Xue, Kai Zhou, Yulin Zhu, “Self-Supervised Contrastive Learning Enhanced UAV Thermal-Imaging-Based PV Defect Detection”, 2026 IEEE 7th International Conference on Artificial Intelligence and Electromechanical Automation (IEEE-AIEA 2026), Shenzhen, China, Jun 2026.
To address the challenges of insufficient detection accuracy, missed detections of small targets, and low-contrast defects in UAV-based thermal imaging inspection of photovoltaic
plants, this paper proposes SSLYOLO, an enhanced detection model built upon YOLO v8s that integrates the self-supervised contrastive pretraining with focal loss. A systematic ablation study and comprehensive comparison with mainstream detectors are conducted. The core innovation of this paper lies in proposing a two-stage training framework that combines the self-supervised contrastive pre-training with focal loss to address the label scarcity and class imbalance issues simultaneously. Experimental results demonstrate that our model outperforms all compared models. The proposed method meets the requirements of edge deployment and real-time inference, and can be directly applied to UAV-based intelligent inspection platforms for photovoltaic plants.
The Hong Kong Chu Hai College remains committed to supporting students’ research endeavors, encouraging publication in international conferences and journals. More are comi
ng!
