Identification Method for Cone Yarn Based on the Improved Faster R-CNN Model

To solve the problems of high labor intensity, low efficiency, and frequent errors in manual identification cone yarn types, this study five kinds were taken as research objects, an method for based on improved Faster R-CNN model was proposed. In total, 2750 images collected samples real textile industry environments, then data enhancement performed after marking targets. The ResNet50 with strong representation ability used feature network to replace VGG16 backbone original extract features dataset. Training a stochastic gradient descent approach obtain optimally weighted file predict categories yarn. Using same training environmental settings, we compared proposed paper two mainstream target detection algorithms, YOLOv3 + DarkNet-53 VGG16. results showed that algorithm had highest mean average precision rate types at 99.95%, 2.24% higher 1.19% higher. Regarding defects, shielding, wear, can correctly identify these issues without misdetection occurring, greater than 99.91%.