Abstract
Aiming at the problems of low accuracy of traditional railway catenary insulator identification methods, long time period and poor multiscale target recognition effect, an improved faster R-CNN railway catenary insulator identification algorithm is proposed. Firstly, by introducing feature pyramid networks (FPN) to fuse deep feature images with shallow feature images, a feature image with strong semantic information and high resolution is obtained, which solves the problem of multi-scale recognition. Then, the traditional nonmaximum suppression (NMS) algorithm is optimized by Gaussian weight reduction function, and online hard example mining (OHEM) technology is added in the training process to improve the recall rate and accuracy of insulator identification. Experimental results on the test set demonstrated that the improved faster R-CNN algorithm achieved a precision rate of 98.3%, a recall rate of 96.43%, an accuracy rate of 94.85% and F1 score of 97.35%, which were increased by 2.3 percentage points, 4.35 percentage points, 6.17 percentage points and 3.36 percentage points, respectively, compared with the original faster R-CNN algorithm. It can effectively identify targets of different scales in the image of railway catenary insulators, has a higher accuracy rate, and can more accurately identify insulators.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
REFERENCES
Yan, L., Wei, N., and Zuo, X., 4C analysis method for key OCS equipment of high speed railway and its application, Electrified Railway, 2019, vol. 30, no. 4, pp. 81–83.
Zhang, G., Liu, Zh., Han, Ye, et al., A fast fuzzy matching method of fault detection for rod insulators of high-speed railways, J. China Railway Soc., 2013, vol. 35, no. 5, pp. 27–33.
Yao, X., Liu, Li, and Li, Zh., Identification method of catenary insulator based on canny edge feature point, Insul. Surge Arresters, 2020, vol. 24, no. 1, pp. 142–148.
Xu, J., Liu, Zh., Han, Zh., et al., Application of point cloud registration in 3D reconstruction of catenary parts based on SIFT and LBP, J. China Railway Soc., 2017, vol. 39, no. 10, pp. 76–81.
Zhang, Guinan., Liu, Zhigang., and Han, Ye., Automatic recognition for catenary insulators of high-speed railway based on contourlet transform and Chan–Vese mode, Optik, 2016, vol. 127, no. 1, pp. 215–221. https://doi.org/10.1016/j.ijleo.2015.10.049
Han, Ye, Liu, Zh., Geng, X., et al., Fracture detection of ear pieces in catenary support devices of high-speed railway based on HOG features and two-dimensional Gabor transform, J. China Railway Soc., 2017, vol. 39, no. 2, pp. 52–57.
Wu, T., Wang, W., Yu, Li, et al., Insulator defect detection method for lightweight YOLOV3, Comput. Eng., 2019, vol. 45, no. 8, pp. 275–280.
Peng, X., Liu, Ya., Wang, Ke, Zhang, B., Qian, J., Chen, Chi, and Yang, B., An automatically locating method for insulator object based on CNNs, Geomatics Inf. Sci. Wuhan Univ., 2019, vol. 44, no. 4, pp. 563–569. https://doi.org/10.13203/j.whugis20170175
Chen, Ch., Yang, X., Zhou, Yu., et al., Research on insulator target recognition method based on convolutional neural network, J. Shenyang Agric. Univ., 2019, vol. 50, no. 4, pp. 501–506.
Chen, Zh. and Min, F., Detection method of catenary insulator based on improved YOLO V3, J. Wuhan Inst. Technol., 2020, vol. 42, no. 4, pp. 462–466.
Du, P., Chen, M., and Su, T., Deep Learning and Target Detection, Beijing: Publ. House of Electronics Industry, 2020.
Zeiler, M.D. and Fergus, R., Visualizing and understanding convolutional networks, Computer Vision–ECCV 2014, Fleet, D., Pajdla, T., Schiele, B., and Tuytelaars, T., Eds., Lecture Notes in Computer Science, vol. 8689, Cham: Springer, 2014, pp. 818–833. https://doi.org/10.1007/978-3-319-10590-1_53
Simonyan, K. and Zisserman, A., Very deep convolutional networks for large-scale image recognition, Int. Conf. Learning Representations, 2015. arXiv:1409.1556 [cs.CV]
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Zh., Karpathy, A., Khosla, A., Bernstein, M., Berg, A.C, and Fei-Fei, Li, ImageNet large scale visual recognition challenge, Int. J. Comput. Vision, 2015, vol. 115, no. 3, pp. 211–252. https://doi.org/10.1007/s11263-015-0816-y
Uijlings, J.R.R., van de Sande, K.E.A., Gevers, T., and Smeulders, A.W.M., Selective search for object recognition, Int. J. Comput. Vision, 2013, vol. 104, no. 2, pp. 154–171. https://doi.org/10.1007/s11263-013-0620-5
Ren, S., He, K., Girshick, R., and Sun, J., Faster R-CNN: Towards real-time object detection with region proposal networks, Adv. Neural Inf. Process. Syst., 2015, vol. 28.
Lin, T.-Yi, Dollár, P., Girshick, R., He, K., Hariharan, B., and Belongie, S., Feature pyramid networks for object detection, 2017 IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Honolulu, Hawaii, 2017, IEEE, 2017, pp. 936–944. https://doi.org/10.1109/CVPR.2017.106
Xi, Rui., Jiang, K., and Zhang, W., Lv, Z.Q., and Hou, J.L., Recognition method for potato buds based on improved faster R-CNN, Trans. Chin. Soc. Agric. Mach., 2020, vol. 51, no. 4, pp. 216–223.
Shrivastava, A., Gupta, A., and Girshick, R., Training region-based object detectors with online hard example mining, Proc. 2016 IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Las Vegas, 2016, IEEE, 2016, pp. 761–769. https://doi.org/10.1109/CVPR.2016.89
Funding
Project supported by the National Natural Science Foundation of China (no. 51767015).
Author information
Authors and Affiliations
Corresponding authors
Ethics declarations
The authors declare that they have no conflicts of interest.
About this article
Cite this article
Jiang Xiang Ju, Du Xiao Liang Railway Catenary Insulator Recognition Based on Improved Faster R-CNN. Aut. Control Comp. Sci. 56, 553–563 (2022). https://doi.org/10.3103/S0146411622060074
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.3103/S0146411622060074