Parallel Dilated CNN for Detecting and Classifying Defects in Surface Steel Strips in Real-Time

Abstract

To improve the quality of steel industry, automatic defects inspection and classification is of great importance. This paper proposed and developed DSTEELNet convolution neural network (CNN) architecture to improve detection accuracy and the required time to detect defects in surface steel strips. DSTEELNet includes three parallel stacks of convolution blocks. Each convolution block used dilated convolution that expands the receptive fields and increase the feature resolutions. The experimental results indicate significant improvements in accuracy and illustrate that the DSTEELNet achieves 97% mAP to detect defects in surface steel strips on NEU dataset and able to detect defect in single image in 22 ms.