Mobile phone surface defect accurate grading method based on mixed attention deformation convolutional neural network

Abstract

The invention provides a waste mobile phone surface defect accurate grading method based on a deformation convolutional neural network, and aims to solve the problem that surface defects are difficult to accurately grade in a waste mobile phone recovery process. According to the method, the mixed attention mechanism model is designed, the performance of the model can be optimized, the recognition model based on the deformation convolutional neural network is established, and accurate grading of the surface defects of the waste mobile phone is achieved. According to the method, good rapidity and accuracy are kept for mobile phone surface defect grading in different scenes, and the waste mobile phone recycling efficiency and the economic benefits of recycling enterprises can be improved.

Description

technical field [0001] The invention utilizes the mixed attention mechanism deformed convolutional neural network method for identifying surface defects of used mobile phones to realize rapid and accurate identification of surface defects of mobile phones during the recycling process of used mobile phones. In the process of recycling used mobile phones, the defects of different degrees of mobile phones are one of the important evaluation criteria that affect the quality of recycling. However, the surface defects of mobile phones are affected by light and shooting angle, which can easily lead to unclear defects, and the defects have different shapes and sizes. However, it is difficult for the convolutional network model to effectively capture the details of local features, which seriously affects the accurate identification of surface defects of mobile phones. The surface defect identification method of used mobile phones based on the deformed convolutional neural network of mi...

AI Technical Summary

Problems solved by technology

In the process of recycling mobile phones, the images of used mobile phones are collected by relying on the photos taken by the inspectors. However, due to the influence of the inspection photo angle, equipment, light source and other conditions, the image clarity of the surface defects of the mobile phone is affected, which increases the accuracy of the model. Recognition difficulty

Moreover, due to the different sizes and shapes of defects, it is easy for the model to ignore some feature details in the process of feature extraction, which leads to poor recognition accuracy of traditional convolutional network models when detecting surface defects of mobile phones

Images