Online detection of weld surface defects based on improved incremental learning approach

作者：

Highlights：

• Online detection by applying incremental learning approach to infer weld defects.

• Implementation of a generalized incremental 2DPCA in a recursive form.

• Validation of the detection from the weld dataset and the actual running videos.

• Easy expansion by incorporating more defect classes with different weld widths.

摘要

•Online detection by applying incremental learning approach to infer weld defects.•Implementation of a generalized incremental 2DPCA in a recursive form.•Validation of the detection from the weld dataset and the actual running videos.•Easy expansion by incorporating more defect classes with different weld widths.

论文关键词：Weld surface defects,Online detection,Feature extraction,Principal component analysis,Feedforward neural network

论文评审过程：Received 3 March 2021, Revised 17 October 2021, Accepted 11 December 2021, Available online 14 January 2022, Version of Record 10 February 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116407