Efficient Cable Surface Defect Detection with Deep Learning

Abstract

Efficient detection of cable surface defects can prevent and reduce the potential dangers in the process of high voltage transmission. In order to achieve efficient detection of cable surface defects and solve the problem of low detection accuracy of small and unobvious defects on cable surface, we propose an efficient cable surface defect detection model with deep learning. Firstly, the lightweight backbone feature extraction network is used to extract the preliminary defect features. Secondly, the parallel convolution module and serial convolution module are designed to obtain abundant defect features and reduce the number of model parameters. Then, the feature fusion module is designed to fuse the shallow features with deep features to enhance the features of small and unobvious defects. Finally, the obtained features are put into the corresponding detection head to get the final prediction results. The experimental results on local cable dataset show that our method achieves favorable trade-off between the accuracy, speed and model size of the cable surface defect detection, which meets the requirements of high accuracy, high speed and small model in industrial application.