Defect Detection of Metal Sheets Based on Improved YOLOX Algorithm
- Resource Type
- Conference
- Authors
- Luo, Bing; Wang, Hongwei; Jia, Jingkun; Qin, Na; Du, Yuanfu; Xie, Linzi
- Source
- 2023 CAA Symposium on Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS) Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS), 2023 CAA Symposium on. :1-6 Sep, 2023
- Subject
- Aerospace
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Shape
Metals
Sheet metal processing
Production
Feature extraction
Product design
Safety
metal sheets
YOLOX
defect detection
loss function
- Language
Surface defect detection of metal sheets is an important link in industrial production. Identifying and locating defects on the surface of metal sheets is important to ensure safely production, improve product quality and save costs for sheet metal processing center. This paper proposes an improved exceeding yolo series(YOLOX) algorithm to complete the identification and location of defects. The novelty of this method is that according to the different situations of various types of defects. The loss function is improved by optimizing the balance parameters and increasing the weight factor. Leting the network strengthen the learning ability of defects with a small number of samples, and reasonably distribute the weight of each defect to ensure sample balance. Experiments show that the mAP of improved YOLOX on the metal sheets test dataset is 91.54%. It has a greater improvement compared with unimproved and other classic deep learning network accuracy. The improved YOLOX algorithm has practical engineering significance for the detection of metal sheets surface defects.