A Fabric Defect Detection Algorithm Based on YOLOv8

In response to the practical problems faced by manual defect detection in the textile industry, this paper studies an automatic fabric defect detection algorithm based on machine vision. Starting from the research background of deep learning based fabric defect detection algorithms, this article introduces the latest YOLOv8 algorithm model released by Ultratics. Based on the complex and diverse textures of textiles, a coordinate attention mechanism is added to the YOLOv8n model to enhance the network's ability to extract features. The experimental results show that the fabric defect detection algorithm designed in this article has high detection accuracy, mAP up to 79.20%, fast speed, and FPS up to 95.4, which can meet the requirements of online real-time detection with low computational resource requirements and can be deployed on edge intelligent computing devices. The algorithm designed in this article has good comprehensive performance and can promote the deployment of textile defect automatic detection technology in the industry.