A hierarchical training-convolutional neural network with feature alignment for steel surface defect recognition

Steel is a basic material, and vision-based defect recognition is important for quality. Recently, deep learning, especially convolutional neural network (CNN), has become a research hotspot. However, steel defects have poor class separation, which is similar to the background, and different defects show similar textures. This causes some defects unrecognizable and influences production greatly. Thus, current CNNs still need to be improved. With this goal, this paper proposes a hierarchical training-CNN with feature alignment. The proposed method introduces a feature alignment, which maps the unrecognizable defects to the recognizable area, and a hierar-chical training strategy is used to integrate the feature alignment into the training process. With these im-provements, the proposed method achieves improved performance. The recognition results on a public dataset achieve 100%, which outperforms the other CNNs. And it has been developed into a real-world case successfully, which is significantly improved.

Automated summary

This paper proposes a hierarchical training-CNN with feature alignment for vision-based defect recognition in steel. The method achieves improved performance by introducing a feature alignment and a hierarchical training strategy. It outperforms other CNNs in recognition results and has been successfully applied in a real-world case with significant improvement.