Power transmission line defect identification method based on saliency map and semantic embedding feature pyramid

Abstract

The invention discloses a power transmission line defect identification method based on a saliency map and a semantic embedding feature pyramid. The method comprises the steps of 1, performing data cleaning and division on a data set; 2, carrying out the super-resolution image generation of the small target of the power transmission line through an EL-ESRGAN super-resolution augmentation algorithm; 3, performing image saliency detection on the data set by constructing a nested U-shaped network; step 4, carrying out data augmentation based on a saliency graph on the data set through a Gridmask and random erasure (Cut Out) algorithm, and generating a classification data set; and 5, carrying out picture classification on the normal set and the defect set by utilizing a ResNet34 classification algorithm through a feature pyramid classification network embedded by deep semantics. According to the method, image saliency detection and data augmentation are combined, the feature pyramid classification network embedded through deep semantics is used as a supplement of ResNet34 classification, the method is used for fault identification in the unmanned aerial vehicle power transmission line inspection image, and the method has high system robustness.

Description

technical field [0001] The invention relates to transmission line inspection image processing, in particular to a transmission line defect identification method based on a saliency map and a semantic embedded feature pyramid. Background technique [0002] In industrial practice, the identification of transmission line defects is more economical and has a faster identification rate. The common method is to rely on the image classification model to classify the output image of the transmission line after the target detection, so as to distinguish whether there is a fault in the component. But in practice, most of the target images of transmission lines are defect-free images, which cannot be used to train supervised learning models for defect identification and analysis. This imbalance of defective and non-defective samples is called the long-tail effect. In order to solve this problem, it can be augmented at the data level, and new samples can be generated for the categories ...

AI Technical Summary

Problems solved by technology

But in practice, most of the transmission line target images are defect-free images, which cannot be used to train supervised learning models for defect recognition and analysis. This imbalance of defective and non-defective samples is called the long tail effect

In order to solve this problem, it can be augmented at the data level, and resampling can be used to generate new samples for categories with insufficient samples. This method will lead to oversampling of a few samples, which will lead to over-fitting of the model and directly reduce the model of feature extraction. performance

Images