Power equipment identification method and device and terminal equipment

Abstract

The invention is suitable for the technical field of image recognition, and discloses a power equipment recognition method and device and terminal equipment. The electrical equipment identification method comprises the following steps: acquiring an electrical equipment image, preprocessing the electrical equipment image, and establishing an electrical equipment image data set; based on the power equipment image data set, training to obtain a power equipment identification model; and inputting the to-be-identified electrical equipment picture into the electrical equipment identification model to obtain an identification result of the electrical equipment. A power equipment image data set is preprocessed, the problem that the operation speed is low due to the fact that the size difference of power energy equipment is large is solved, a power equipment recognition model is established according to the power equipment image data set, the model is improved in the three aspects of feature extraction, feature fusion and feature matching, and the recognition accuracy is improved. While the model operation efficiency is improved, the identification precision of the power equipment is improved.

Description

technical field [0001] The present application belongs to the technical field of image recognition, and in particular relates to a power equipment recognition method, device and terminal equipment. Background technique [0002] In recent years, the application of deep learning convolutional neural network in the field of image feature information extraction has achieved remarkable results. However, most of the existing neural network algorithms for image recognition cannot well solve the problem of feature information extraction of images of different scales. To solve the problem of multi-scale image recognition, feature pyramid methods, including SSD, STDN and other methods, are Scale image features are integrated into small-scale object recognition to improve recognition accuracy. [0003] At present, most object recognition methods use the method of enumerating anchor frames for feature extraction. Faster RCNN uses manual selection of anchor frames, and Yolo uses statist...

AI Technical Summary

Problems solved by technology

However, the feature pyramid method realizes the integration of large-scale object features into small-scale object recognition, which improves the recognition accuracy of small-scale objects while ignoring the recognition of large-scale objects, resulting in the problem that the recognition accuracy of large-scale objects is much lower than that of small-scale target objects.

Images