Fabric Defect Detection Using Customized Deep Convolutional Neural Network for Circular Knitting Fabrics

Visual inspection is a main stage of quality assurance process in many applications. In this paper, we propose new network architecture for detecting the fabric defects based on convolutional neural network. Four different pre-trained and customized model architectures have compared terms performance. Results has been evaluated defect dataset 13.800 images. Among existing Inception V3, MobileNetV2, Xception ResNet50 methods, InceptionV3 achieved 78% classification success. Our designed deep could achieve 97% The experimental works show that effective defects.