Abstract
Steel Surface defect detection is widely used in product quality control system in factories. But the traditional algorithm has the problem of slow running speed and low accuracy. In this paper, we propose a method based on transfer learning and image segmentation. First, data augmentation is used to extend dataset. Next, we use EfficientNet as the backbone to get the feature map. A new feature map with five different scales is obtained from the repeated BiFPN module. Then, deconvolution is used to fuse the five feature maps together, and a layer of convolution is used to get the mask maps of various defects. After the final training, we use the batch normalization (BN) and convolution layer fusion method to reduce the algorithm run time. Experimental results show that our proposed algorithm can obtain the mean dice coefficient of 0.912. Better than other deep learning algorithms for steel surface defect detection.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
