Transmission line image defect diagnosis based on YOLOv3 and SNIPER cascade lifting algorithm
Authors:
Open Access
- 1 October 2019
- conference paper
- Published by IOP Publishing in IOP Conference Series: Earth and Environmental Science
- Vol. 332 (4) , 042032
Abstract
Deep learning classifies information and combines the underlying features to form more abstract high-level features. Its ability of autonomous learning also provides favorable conditions for industrial application. Aiming at the problem of insufficient performance of current small target detection, this paper proposes a deep learning algorithm using One-stage fusion two-stage. The influence of background noise on network training is reduced. The algorithm combines multi-scale training methods and achieves sample equilibrium adjustment based on sample distribution and stratified sampling. The high efficiency of YOLOv3 method for large target detection is well used, and the multi-scale training advantages of SNIPER algorithm are also played out, which makes the intelligent detection effect of pin defect of transmission line get some performance improvement.Keywords
This publication has 5 references indexed in Scilit:
- Mask R-CNN2017
- Feature Pyramid Networks for Object Detection2017
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2016
- Selective Search for Object RecognitionInternational Journal of Computer Vision, 2013
- Infrared Image Edge Detection of High Voltage Insulator ChainApplied Mechanics and Materials, 2011