Nest Detection Using Coarse-to-Fine Searching Strategy
Nest on pylon is a great threat to the safety and function of electric power system. Thus detection of nest has been considered as a vital task when checking transmission line through intelligent surveillance system. Traditional detection methods for nests are basically based on artificial identification, resulting a lot of time consumption and human waste. In addition, it’s impractical to apply this method to the situation with massive pending data. In this paper, we propose an automatic framework for nest detection. This work firstly locates the pylon, then detects nest on it through color analysis. To boost the speed of pylon location, our work combines the technology of detection and tracking after extracting HOG features of pylons. In addition, building on the observation that most pylons are standing out from background for the sake of their sharp outlines and unique color patterns, a number of filtering operations in this work are designed to significantly decrease the number of candidate boxes. Experiments have been conducted to show that this framework could achieve promisingly precise and robust detection for nest in complex environment.
KeywordsNest detection Tracking Computer vision
The work was supported by State Key Research and Development Program (2016YFB1001003). This work was also supported by NSFC (U1611461, 61502301, 61527804, 61671298) and STCSM17511105401, China’s Thousand Youth Talents Plan, the 111 Project under Grant B07022, and the Shanghai Key Laboratory of Digital Media Processing and Transmissions.
- 2.Dalal, N., Triggs, B.: Histograms of oriented gradients for human detection. In: 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005, vol. 1, pp. 886–893. IEEE (2005)Google Scholar
- 3.Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., Fei-Fei, L.: Imagenet: a large-scale hierarchical image database. In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2009, pp. 248–255. IEEE (2009)Google Scholar
- 8.Kégl, B.: The return of AdaBoost.MH: multi-class hamming trees. arXiv preprint arXiv:1312.6086 (2013)
- 9.Liu, Q., Qu, Y.: Hog and color based adaboost pedestrian detection. In: 2011 Seventh International Conference on Natural Computation (ICNC), vol. 1, pp. 584–587. IEEE (2011)Google Scholar
- 10.Wang, Z.-R., Jia, Y.-L., Huang, H., Tang, S.-M.: Pedestrian detection using boosted hog features. In: 2008 11th International IEEE Conference on Intelligent Transportation Systems, ITSC 2008, pp. 1155–1160. IEEE (2008)Google Scholar
- 12.Wu, Y., Lim, J., Yang, M.-H.: Online object tracking: a benchmark. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2411–2418 (2013)Google Scholar
- 13.Zhu, C., Wang, T.: Analysis and prevention of bird damage in transmission line. Cent. China Electr. Power 22(4), 46–49 (2009)Google Scholar