Surface Defect Detection Algorithm for Aluminum Profiles based on Deep Learning

Abstract

The surface quality of aluminum profiles directly affects the performance and safety of the final product. Efficient and accurate surface defect detection has become particularly important for ensuring product quality. To solve the low efficiency and low accuracy of traditional detection methods, the You Only Look Once version 5 is used to detect surface defects on aluminum profiles. Improvements are made to the anchor frame mechanism, data augmentation method, and attention mechanism, adjusting the loss function to more accurately identify small defects. The research results indicated that the average accuracy, recall, accuracy, and F1 score of the research method for detecting multiple categories were 0.95, 0.90, 0.94, and 0.91, respectively. Compared with before improvement, the research method had the lowest loss value and tended to stabilize faster. After 300 iterations, the loss value, the first improvement, and before the improvement were 0.0173, 0.0204, and 0.0288. The average accuracy value of detection for multiple categories stabilized faster. Simulation analysis showed that the highest detection accuracy, false detection rate, and missed detection rate of the research method for 10 types were 99.2%, 1.3%, and 1.4%, respectively. The successful application of this method can provide reference for the surface defect detection in other materials, which has broad promotion value.

翻译

搜索

复制