Research on the Algorithm for Detecting Appearance Defects of Automotive Precision Parts Based on Computer Vision

Various defects are inevitably generated during the manufacturing process of parts, such as size deviation, scratches, cracks, scratches, surface pollution, etc. These defects are important factors leading to system operation failures, and are more serious in high-precision systems, such as some precision parts in automotive engines. Appearance defect detection is an indispensable part. Traditional detection relies on manual means, with low detection accuracy and multiple influencing factors. Therefore, improving the accuracy of precision component detection is imperative. This article introduces a deep learning-based method to identify and classify defects in automotive precision parts, which is significantly superior to traditional machine learning methods. Our deep learning model has an accuracy of 92.5%, an accuracy of 89.6%, a recall rate of 91.2%, an $F 1$ score of $\mathbf{9 0. 4 \%}$, and a significant average detection time of $\mathbf{0. 5}$ seconds. The implementation of this method can greatly improve manufacturing efficiency and product quality, ensuring higher safety standards and customer satisfaction.