A Deep Learning Based Surface Defect Detection Method with Coarse Granularity for Corrugated Fuel Tubes

Surface defect detection is a crucial task to achieve the automation goal in industrial production process. Nowadays, owing to the rapid development of computer vision, lots of detection methods based on deep learning have been proposed and applied in industrial manufacturing process. However, different from other deep learning tasks, the rarity of labeled defect training samples makes it a great challenge to attain expected performance for the surface defect detection of corrugated fuel tubes. To address this issue, this paper proposes a network with rougher detection granularity to obtain higher accuracy and recall rate. The boundary of detected defects will be less precise, which, however, is much less important than the gain of detection accuracy and is a tolerable cost. Experiments on two different datasets are carried out to demonstrate the effectiveness of this proposed method.