A reweighting offset bin classification network for surface defect detection and location of metal components. (January 2022)

• Record Type: Journal Article
• Authors: Gu, Caohan; Lu, Yi; Chen, Mingzhi; Sun, Guifang; Ni, Zhonghua
• Is Part Of: Measurement. Volume 187(2022)
• Publication Date: 2022-01
• Subjects: Computer vision -- Offset bin classification network -- High quality defect location -- Deep learning; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2021.110166
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A new method of locating defect for in situ laser cladding repair is proposed. An optimized offset discretization method for accurate location is designed. A new training strategy is proposed to balance the weight norm of offset bins. The experiment results show that the location performance of our method is better. Abstract: The detection and accurate location of metal components surface defects are the basis of automatic laser cladding/laser additive manufacturing repair. Most deep-learning methods utilize regression networks trained with non-maximum suppression and loss function to correct the offset. However, these methods are unable to implement sufficient punishment for accidental offset, resulting in an inaccurate offset prediction. In this investigation, we proposed a reweighting offset bin classification network to solve this issue. The network discretizes the offset values into multiple offset bins by its probability distribution and corrects the offset of the bounding box by predicting the probability of these bins. In addition, the offset bins are classified into different balance groups for parallel training to ensure that all bins have relatively balanced weight norms. Comparative experiments conducted on the advanced detection network and defect datasets demonstrate that our method can locate defects accurately.