Abstract: Aiming at the commonly used computer vision-based PV panel defect recognition model to extract features difficult, slow inference speed and other problems, this paper proposes a YOLOv8 improvement algorithm that incorporates a two-layer routing attention mechanism visual converter Biformer as the backbone network, which improved model adaptability by dynamically adjusting attention span; designing a feature extraction network incorporating the SE attention mechanism, which improves the sensitivity of the network to key features, thus improving the feature characterization; Better Student semi-supervised learning method(BSLM) is proposed, which improves the model accuracy while reducing the amount of data labeling. Experimentally, it is proved that the mAP@0.5 of the algorithm proposed in this paper reaches 83.9%, which is 3.1% higher than that of YOLOv8n algorithm, far exceeding other one-stage algorithms, and it can better meet the accuracy requirements of defect detection of PV panels. The FPS reaches 101.01, which meets the real-time requirements of defect detection of PV panels. After using the BSLM proposed in this paper, the mAP@0.5 reaches 90.7%, which further improves the model performance.