A glass insulator sheet localization method based on end-to-end key point detection

Abstract

The invention discloses a method for locating glass insulator sheets based on end-to-end key point detection, including: 1) construction and labeling of an instance segmentation data set of glass insulators for power inspection; 2) data expansion by using a data enhancement algorithm; 3) The instance segmentation model is trained and the minimum circumscribed polygon image of the area where the glass insulator is located is cut off as a data set for key point detection in the next step; 4) Label the key point detection data set and do data expansion; 5) Design an end-to-end key point The point detection model is continuously optimized for training; 6) The trained instance segmentation model and the key point detection model are connected in series: input the image of the glass insulator to be detected into the trained instance segmentation model, and cut out the region of the segmentation result for input To the trained key point detection model, the coordinate values ​​of the key points where each glass insulator sheet is located in the figure are obtained. The invention can improve the positioning speed and accuracy of the glass insulator sheet.

Description

technical field [0001] The invention relates to the technical fields of image pattern recognition and computer vision, in particular to a method for locating glass insulator sheets based on end-to-end key point detection. Background technique [0002] Insulators are very important and common components in the transmission network. They play the role of insulation and support. Once a fault occurs, it will cause contact between transmission lines or with towers, resulting in power interruption accidents, which will affect the social economy and people's livelihood. have a big impact. In the current insulator consumption of transmission lines, glass insulators are eliminated due to self-explosion, no need for zero measurement and excellent anti-aging performance, accounting for nearly one-third of the proportion. However, glass insulators still have some troublesome problems in the operation and maintenance process. Problems, such as high self-explosion rate, etc., so correctl...

AI Technical Summary

Problems solved by technology

It is reasonable and feasible to divide the entire algorithm process into three stages, but in the second stage, in order to accurately segment the glass insulators in various scenarios to obtain the position information of each piece of insulators, the instance segmentation is continued. Algorithm, its model complexity is too high, it is difficult to meet the real-time requirements, and it takes a lot of manpower and time to mark the outline of each piece of glass insulator in a large number of image data sets

Images