Research on Surface Defect Detection System of Chip Inductors Based on Machine Vision

Artificial inspection of surface defects in chip inductors faces issues such as low efficiency and poor accuracy. To enhance production efficiency, increase intelligence, and reduce production costs, this paper proposes the use of machine vision technology for chip inductor surface defect detection. Specifically, the paper builds upon the DETR model, improving its feature extraction network and attention mechanism. The approach involves transferring the pre-trained detection model to generalize it for chip inductor surface defect datasets. Consequently, the improved DETR model is applied to chip inductor surface defect detection. Experimental results demonstrate that the enhanced DETR model successfully detects chip inductor surface defects, improving the feature extraction and object localization capabilities of the network while reducing training time. The application of machine vision in chip inductor surface defect detection enhances efficiency, addresses the issue of lengthy DETR model training and poor small object detection performance, achieves classification and localization of chip inductor surface defects, and validates the feasibility of the detection method.