Composite insulator real-time segmentation method and system based on DeepLabV < 3 + >

Abstract

The invention relates to a composite insulator real-time segmentation method and system based on DeepLabV < 3 + >, and the method comprises the following steps: S1, obtaining a power equipment infrared image which is shot in a power inspection process and contains a composite insulator, and constructing an original data set; s2, performing data amplification on the original data set to obtain a training data set; s3, constructing an improved DeepLabV < 3 + > network: replacing a backbone network of the DeepLabV < 3 + > with a lightweight neural network MobileNetV2 to improve real-time performance, introducing a lightweight efficient channel attention module ECA to realize local cross-channel interaction without dimension reduction, and then adding a Point fine segmentation module at an output end of the DeepLabV < 3 + > for post-processing to further improve a semantic segmentation result; training the improved DeepLabV < 3 + > network through the training data set to obtain a trained improved DeepLabV < 3 + > network; and S4, processing the shot infrared image of the power equipment through the trained improved DeepLabV3 + network so as to segment the composite insulator in real time. According to the method and the system, the real-time performance and the accuracy of composite insulator segmentation can be improved.

Description

technical field [0001] The invention belongs to the field of power transmission lines, and in particular relates to a method and system for real-time segmentation of composite insulators based on DeepLabV3+, which can be applied to the intelligent inspection and detection of UAVs in the operation and maintenance of power transmission lines. Background technique [0002] With the increase in the scale of use and service life, composite insulators have frequently experienced defects such as abnormal heating in recent years, and further development of defects may lead to serious failures such as internal insulation breakdown and broken strings. Using infrared detection technology to detect defects in insulators is a relatively mature live detection technology. The infrared images of composite insulators can directly reflect the overheating defects of insulators, so as to detect defects in advance and prevent their deterioration. Therefore, in order to detect overheating defects...

AI Technical Summary

Problems solved by technology

This patent achieves the segmentation of insulators in a simple background, but because this traditional method relies too much on image preprocessing, preprocessing directly affects the segmentation effect, and image preprocessing is largely done manually The selection of shape feature operators, therefore, the feature selection process is complicated, and for the complex and changeable background in actual inspection, the segmentation accuracy is not good

Images