Research on Engineering Vehicle Target Detection in Aerial Photography Environment based on YOLOX
Mingjiang Zhang, Chengyuan Wang, Jungang Yang, Kouquan Zheng · 2021
The aerial target detection of engineering vehicles such as excavators is the key technology for UAV inspection of optical cable lines and petroleum pipelines. It is proposed to apply the deep learning YOLOX (You Only Look Once version X) target detection algorithm to the target detection of engineering vehicles in aerial images. Based on the Pytorch deep learning framework, through the production of engineering vehicle aerial photography data sets, simulation has realized the target detection of engineering vehicles, and its target recognition [email protected]:0.95 value reaches 59.38%. Based on the same data set and training conditions, its detection accuracy and speed surpass classic algorithms such as YOLOv4 and YOLOv5. The simulation results can provide a certain reference for the research of aerial inspection and maintenance of underground pipeline facilities.