Method and system for detecting construction machinery in power transmission line area

Abstract

The invention discloses a method and system for detecting construction machinery in a power transmission line area. The detection method comprises the steps of: firstly, generating a codebook background model of a to-be-detected area based on video data; matching each pixel point in a to-be-detected image with the codebook background model to obtain a pixel point coordinate set of a moving targetin the to-be-detected image; according to the pixel point coordinate set of the moving target, determining the maximum bounding rectangle coordinate of each moving target in the to-be-detected area, and establishing a Box set; and inputting the to-be-detected image and the Box set into a trained Fast R-CNN model for monitoring. According to the method, moving target detection and deep learning target detection are combined, so that Fast R-CNN only pays attention to foreground targets in image information. Therefore, the redundant calculation of the algorithm is greatly reduced, and meanwhile,by utilizing the moving target detection effect of the codebook, the Fast R-CNN can achieve a relatively good detection effect by virtue of a small number of image samples. And the detection precisionand the detection speed of the construction machinery in the power transmission line are improved.

Description

technical field [0001] The invention relates to the technical field of transmission line detection, in particular to a detection method and system for construction machinery in the transmission line area. Background technique [0002] Through statistics on the intrusion behavior of transmission lines, it is found that large-scale machinery touching the line is the main factor of external force intrusion on transmission lines. Therefore, automatic identification of large-scale machinery through transmission line monitoring information can greatly reduce the workload of inspectors. Detect the intrusion of large machinery in time to ensure the safe operation of transmission lines. [0003] At present, there are few researches on large-scale mechanical intrusion detection based on transmission line monitoring video. The current intrusion detection methods used for images are one type of intrusion detection method based on moving objects. Typical methods include frame difference ...

AI Technical Summary

Problems solved by technology

However, the intrusion detection method based on moving object detection is prone to false detection due to the interference of the field environment. At the same time, due to the limitations of the algorithm itself, it cannot provide the category information of the intrusion object (such as people, vehicles, etc.), which is not conducive to the investigation in practical applications. work, there are a lot of redundant calculations

Although the method based on image classification has strong adaptability to the background and can provide category information of intruding objects, it requires a large number of labeled image samples, and a small number of image samples cannot achieve good results.

Images