HookNet: Efficient Multiscale Context Aggregation for High-Accuracy Detection of Fabric Defects

Automatic defect detection is a critical stage of quality control in the textile industry. Because defects vary in texture structure, size, and spatial distribution, existing defect detection methods have difficulty in achieving a perfect tradeoff between detection efficiency, accuracy, and generalizability. This article presents an advanced fabric defect detection model based on a deep learning algorithm. A novel parallel dilated attention module (PDAM) was designed: this is nested in the deep layers of the neural network backbone to establish global channel dependencies and capture multiscale contextual information. Furthermore, a hook-shaped feature pyramid network (HookFPN) was developed for multiscale context aggregation: this directs the network to focus on lower level features and is helpful for further improving detection efficiency without sacrificing detection accuracy. In addition, the Alpha-GIoU loss is used to improve the accuracy of bounding box regression because of its modulated power parameter $\alpha $ . The advantages and effectiveness of the proposed method were statistically analyzed using data from several public datasets. The code is available at https://github.com/aabb605/HookNet.

Automated summary

This article presents an advanced fabric defect detection model based on deep learning algorithm. The model incorporates a parallel dilated attention module (PDAM) and a hook-shaped feature pyramid network (HookFPN), which leads to improved detection efficiency without sacrificing accuracy.