Classification detection method and device based on lightweight classification model

Abstract

The invention provides a classification detection method and device based on a lightweight classification model. The method comprises the steps of obtaining a training sample image, and determining the category of the training sample image; performing classification marking on the training sample image according to the category to obtain a sample marking image; inputting the sample marking image into the lightweight classification model for training to obtain a trained lightweight classification model; and inputting a to-be-evaluated sample image into the trained lightweight classification model, and judging and outputting the category of the to-be-evaluated sample image. According to the lightweight classification model, the model basic units in the trunk adopt a simple double-branch structure, transmission of to-be-evaluated sample image data is reserved while the number of the model basic units is increased, the parameter space can be lightened by using the convolution layer with the number of output channels equal to the number of convolution groups, the initialization width of the model is small, the model is lighter, so that the speed of processing the to-be-evaluated sample image data by the model is higher, the detection effect is better, and the detection efficiency is higher.

Description

technical field [0001] The present application relates to the technical field of deep learning models, in particular to a classification detection method and device based on a lightweight classification model. Background technique [0002] In the field of industrial inspection, due to the variety of surface defects of products. When classifying and detecting product surface defect categories, traditional recognition algorithms can be used to extract features from product images, and then surface defects and their categories can be determined according to the results of feature extraction. However, it is difficult to extract all effective features in product images by using traditional recognition algorithms such as grayscale extraction algorithms, area extraction algorithms, and contour extraction algorithms, and the detection effect on product images is poor. [0003] Since the deep learning model can automatically learn product image features, the product image is detecte...

AI Technical Summary

Problems solved by technology

[0005] This application provides a classification detection method and device based on a lightweight classification model to solve the problem that the detection method using a deep learning model in the prior art has low detection efficiency for product images, resulting in low production efficiency of products

Images