Automatic Detection of Beginner's Welding Joint
- Resource Type
- Conference
- Authors
- Kato, Shigeru; Hino, Takanori; Kumeno, Hironori; Kagawa, Tomomichi; Nobuhara, Hajime
- Source
- 2020 Joint 11th International Conference on Soft Computing and Intelligent Systems and 21st International Symposium on Advanced Intelligent Systems (SCIS-ISIS) Soft Computing and Intelligent Systems and 21st International Symposium on Advanced Intelligent Systems (SCIS-ISIS), 2020 Joint 11th International Conference on. :1-3 Dec, 2020
- Subject
- Communication, Networking and Broadcast Technologies
Computing and Processing
Robotics and Control Systems
Signal Processing and Analysis
Welding
Training
Image edge detection
Feature extraction
Convolutional neural networks
Support vector machines
Steel
CNN
R-CNN
Machine Learning
- Language
This paper describes the construction of a system for the automatic evaluation of stainless steel plates welded by beginners. As a subgoal for that purpose, we constructed RCNN that automatically detects welded joints. In the experiment, fifty welded plate pictures were used for training RCNN. When several pictures of welded plates not used for training were inputted to trained RCNN, it was confirmed that the welded joint part could be detected almost properly.