Power Transmission Line Image Segmentation Method Based on Binocular Vision and Feature Pyramid Network
- 21 October 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Abstract
Power transmission line image segmentation is an important task, which is the preliminary task of power transmission line defect detection and intrusion detection of obstacles. It can also provide navigation and obstacle avoidance functions for the unmanned aerial vehicle. This paper uses the semantic segmentation method to detect power transmission lines, and an end-to-end semantic segmentation model of power transmission lines based on binocular vision and feature pyramid network is proposed, which is called binocular FPN. The model can detect the power transmission lines and their defects on binocular images. The experiment results show that the segmentation accuracy of binocular FPN is improved compared with the original monocular FPN without significantly increasing the number of parameters.Keywords
This publication has 8 references indexed in Scilit:
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching2020
- Panoptic Feature Pyramid Networks2019
- Squeeze-and-Excitation NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2019
- The Lovasz-Softmax Loss: A Tractable Surrogate for the Optimization of the Intersection-Over-Union Measure in Neural Networks2018
- Feature Pyramid Networks for Object Detection2017
- Deep Residual Learning for Image Recognition2016
- A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation2016
- ImageNet: A large-scale hierarchical image database2009