About this Abstract |
Meeting |
Advances in Welding and Additive Manufacturing Research 2022
|
Symposium
|
Advances in Welding and Additive Manufacturing Research 2022
|
Presentation Title |
Welding Defects Classification by Weakly Supervised Semantic Segmentation |
Author(s) |
baoxin zhang, xiaopeng wang, jinhan cui, xu wang, yongzhe fa, xinghua yu |
On-Site Speaker (Planned) |
baoxin zhang |
Abstract Scope |
Radiographic non-destructive evaluation (NDE) is an important technique to understand defects in welds. These radiographs require certified workers to interpret them to identify the presence of defects. Recent developed deep learning techniques, especially semantic segmentation, could be used to help welding defect detection and classification. The use of image segmentation technology to obtain performance evaluations of the presence, location, and size of defects can improve the stability of defect evaluations while saving a great deal of time. However, supervised instance segmentation requires a large number of manually implemented pixel-level annotations, which greatly increases the difficulty and cost of achieving nondestructive evaluations. In our work, the weakly supervised semantic segmentation based on Cut-Cascade RCNN model is used for classification of defects. The cascade RCNN is used to obtain the region of interest (ROI) and classification information. In the ROI, adaptive threshold segmentation of the defects is implemented and the image is filtered to obtain the mask information. The accuracy of using Cut-Cascade RCNN model in our own x-ray dataset size can reach up to 90 %. |
Proceedings Inclusion? |
Undecided |