Research on Defect Detection of Electronic Components in Facility Greenhouse Based on Improved YOLOv5

To address challenges in manual detection of electronic component defects in facility greenhouses, this paper presents an electronic component defect detection method using the Improved YOLOv5 recognition algorithm. By introducing the Convolutional Block Attention Module into the YOLOv5 backbone network, the model’s identification and classification of defect types are enhanced, the network’s receptive field is improved, and recognition accuracy is increased. The proposed method also utilizes the
$\alpha $
-CIoU loss function to expedite network regression. The effectiveness of the Improved YOLOv5 model was evaluated on a self-made electronic component dataset. Experimental results revealed an average accuracy of 91.9% and a detection time of 2.1 seconds per frame. Compared to the original YOLOv5 model, the average precision value increased by 1.5%, and the single-image detection speed improved by 0.2 seconds per frame. These improvements meet the accurate and efficient requirements for electronic component defect detection within the greenhouse equipment such as roller shutters and ventilators. This study provides valuable theoretical and technical support for defect detection in electronic components, contributing to the performance optimization of electrical equipment in facility greenhouses. The proposed method shows great potential for further development and application in real-world scenarios.