An industrial part defect detection method, device, equipment and storage medium

Abstract

The application discloses an industrial part defect detection method, device and equipment and a storage medium, applied to the field of defect detection, synthesizes a target sample image based on high and low exposure sample images of a part, and performs defect labeling on the target sample image; the target sample image is input into a defect detection model to obtain a predicted defect position output by a detection head in a detection network and a predicted defect category output by a classification head; a position loss is determined based on the predicted defect position and a labeled defect position, a classification loss is determined based on the predicted defect category and a labeled defect category, model parameters are updated based on the position loss and the classification loss until a trained defect detection model is obtained, and part defect detection is performed based on the trained defect detection model. Image acquisition is performed in a high and low exposure image fusion manner, a classification branch is added based on original detection branches of a defect detection model, an end-to-end multi-task learning framework is constructed, and the accuracy of defect detection and classification is improved.

Images