Application of Convolutional Neural Network (CNN) to Optimize Quality Inspection System for Filamentary Materials

In order to improve the quality inspection accuracy and efficiency of filamentary materials, an automated quality inspection system is designed based on convolutional neural network (CNN). Through image preprocessing, feature extraction and quality evaluation modules, the system is able to efficiently recognize normal materials and materials with different degrees of defects. The experimental results show that the accuracy of the model is 90.2% on slightly defective materials and 94.5% on severely defective materials, and the overall accuracy reaches 92.7% with the error rate reduced to 7.1%. The system significantly improves detection speed and accuracy and facilitates more intelligent quality control of threaded materials.