Freezer surface defect detection based on convolutional neural network

Abstract

The invention provides freezer surface defect detection based on a convolutional neural network. The method aims to detect defects on the surface of the freezer by constructing an appropriate target detection model, mark defect positions and mark classification labels NG on pictures with defects by using a Bounding Box, and only mark classification labels OK on pictures without defects. The methodmainly comprises the following four steps: firstly, selecting an appropriate target detection infrastructure according to task requirements, and further selecting an appropriate backbone network structure; then constructing a target detection network model, expanding a training data set by adopting a data enhancement method and training, and then verifying on a verification data set; further optimizing the model structure and the training strategy according to the verification result, and training and verifying again; and finally, putting the test data into the model to generate a final prediction result, and visually comparing actual results. The method is suitable for freezer surface defect detection based on the convolutional neural network.

Description

technical field [0001] The invention relates to the field of computer vision for image classification and target detection, in particular to a detection of surface defects of a refrigerator based on a convolutional neural network. Background technique [0002] With the rapid development of the field of artificial intelligence, computer vision technology is gradually integrated into people's lives and is changing people's lives, operations, and production methods, providing stronger support for the development of all walks of life. [0003] However, for large-scale manufacturing enterprises, in the context of the era of big data, how to process and make good use of the massive data collected in production activities plays a vital role in reducing costs and improving efficiency in manufacturing. [0004] From the actual production, it can be found that the current manufacturers of freezers still check the surface of the freezer one by one manually when detecting defects on the...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the deficiencies in the prior art, to provide a detection of surface defects of refrigerators based on convolutional neural networks, this method does not rely on manual detection, using computer vision technology, training a target detection model, and then Defects on the surface of the freezer are automatically identified and marked

Images