Defect detection method of steel strip based on Faster-RCNN
KOU Xu-peng1,2,3, LIU Shuai-jun1,2,3, MA Zhi-run1,2,3
1. School of Big Data, Yunnan Agricultural University, Kunming 650201, Yunnan, China; 2. Agricultural Big Data Engineering Research Center of Yunnan Province, Kunming 650201, Yunnan, China; 3. Green Agricultural Product Big Data Intelligent Information Processing Engineering Research Center, Kunming 650201, Yunnan, China
Abstract:At present, there are some problems in the steel strip defect detection task in the actual industrial production, such as difficult data collection and poor defect recognition. Therefore, a kind of steel strip defect detection algorithm based on Faster-RCNN—FRDNet is proposed. The anchor box parameters are obtained by k-means clustering, which makes the generated box more in line with the proportion of target defect categories and improves the accuracy of defect detection. At the same time, the network structure is fine-tuned by model migration, so that the steel strip defect detection model can better adapt to the target defect task, effectively solve the problem of less target data on the surface defects of the target steel strip, and enhance the generalization of the model. Experimental results show that the mAP of the model on the GC10-DET steel strip defect data set reaches 67.6%, which is 4.9% higher than that of the original model results, and the detection speed is 27.2FPS, meeting the requirements of the detection task.
寇旭鹏, 刘帅君, 麻之润. 基于Faster-RCNN的钢带缺陷检测方法[J]. 中国冶金, 2021, 31(4): 77-83.
KOU Xu-peng, LIU Shuai-jun, MA Zhi-run. Defect detection method of steel strip based on Faster-RCNN[J]. China Metallurgy, 2021, 31(4): 77-83.
Ren S, He K, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(6): 1137.
[11]
Liu W,Anguelov D,Erhan D,et al. SSD: Single shot multibox detector[C]//European Conference on Computer Vision. Amsterdam: The University of Amsterdam Springer,Cham,2016:21.
[12]
Redmon J,Divvala S,Girshick R,et al. You only look once: Unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, Nevada: Institute of Electrical and Electronics Engineers, 2016:779.
Arora P,Varshney S. Analysis of k-means and k-medoids algorithm for big data[J]. Procedia Computer Science,2016,78:507.
[16]
HE Y,SONG K,MENG Q,et al. An end-to-end steel surface defect detection approach via fusing multiple hierarchical features[J]. IEEE Transactions on Instrumentation and Measurement,2019,69(4):1493.