A method for real-time detection of aerial foreign body image based on depth learn

Abstract

A method for real-time detection of aerial foreign body image base on depth learning include such steps as 1, establishing four aerial foreign body image databases according to difference of foreign body types,. 2, respectively make data sets accord to four aerial photographed foreign object image databases; 3, construct and training a foreign body model of that transmission line; 4, construct andtraining a foreign body model of that shock-proof hammer; 5, construct and training a foreign body model of that voltage equalize loop clamp; 6, construct and training a pole and tow foreign body model; Step S7: Adopt fine-tune to carry out fine tuning on the four aerial foreign object image database models; Step S8: solidifying the four aerial foreign body image library models after fine-tuning;Step S9: Input the image to be tested into four solidified foreign body detection models for detection, and obtain the coordinates and confidence levels of the target rectangular frames of the four detection results after the forward propagation of the network. Compared with the traditional end-to-end model, the invention has higher accuracy.

Description

technical field [0001] The invention relates to a real-time detection method for aerial foreign object images based on deep learning. Background technique [0002] In recent years, unmanned aerial vehicle inspection has gradually become one of the main means of inspection and maintenance of transmission lines. The necessity of automatic detection of foreign objects in aerial images and removal of foreign objects has become increasingly prominent: foreign objects staying on the line for a long time will affect the operation and maintenance of the line. Management leaves a security risk. After the marked foreign matter is detected in the image, targeted foreign matter removal can be carried out and eliminated in time. Therefore, how to automatically detect and exclude foreign objects in aerial images is a technical problem that needs to be solved at present. Contents of the invention [0003] In view of this, the object of the present invention is to provide a real-time de...

AI Technical Summary

Problems solved by technology

[0002] In recent years, unmanned aerial vehicle inspection has gradually become one of the main means of inspection and maintenance of transmission lines. The necessity of automatic detection of foreign objects in aerial images and removal of foreign objects has become increasingly prominent: foreign objects staying on the line for a long time will affect the operation and maintenance of the line. Management leaves safety hazards

Images