Joint weakly and fully supervised learning for surface defect segmentation from images

作者：

Highlights：

• We propose a novel weakly supervised semantic segmentation method for small defects.

• An elaborate Siamese network is proposed for the hybrid supervised dataset.

• Our hybrid supervised method obtains excellent results and saves the labeling cost.

摘要

•We propose a novel weakly supervised semantic segmentation method for small defects.•An elaborate Siamese network is proposed for the hybrid supervised dataset.•Our hybrid supervised method obtains excellent results and saves the labeling cost.

论文关键词：Weakly supervised learning,Defect detection,Semantic segmentation

论文评审过程：Received 2 August 2021, Revised 9 May 2022, Accepted 24 June 2022, Available online 28 June 2022, Version of Record 7 July 2022.

论文官网地址：https://doi.org/10.1016/j.image.2022.116807