A Visual Detection Method of Defective Products Based on Improved DeepLabv3+ Segmentation Network

In the traditional electronic product production line, the products need to be manually inspected, and the production efficiency is low. After the application of machine vision defect detection system, the instant detection and rapid classification of products are realized. The overall efficiency of the production line will be improved. A visual detection method for defective products based on an improved DeepLabv3+ segmentation network is proposed in this paper. By training a deep learning model, combining high-resolution cameras and advanced image processing algorithms, the improved DeepLabv3+ segmentation network can realize real-time detection and classification of tiny defects. With the hardware platform of 360 ring image imaging system and reflection suppression imaging system, the accurate recognition and real-time detection of diverse workpieces on complex production lines are realized. The experimental results show that the proposed method can improve the overall detection rate and accuracy of intelligent detection equipment. The method improves the applicability and generalization ability of the model in the same type of tasks.