Chip defect image classification method based on ResNet network

Abstract

The invention relates to the technical field of deep learning, in particular to a chip defect image classification method based on a ResNet network, classification is performed through the ResNet network, and the method comprises the following steps: dividing obtained sample data into a training set, a verification set and a test set; preprocessing the sample; using a training set and a verification set in the processed sample image for training an established network model; using the trained network model as a test model, using the remaining test set in a test network, and finally outputting a classification result through an activation function. Through preprocessing, the situation that a large amount of calculation work and time need to be consumed in order to process the whole image due to the fact that the original image size of a sample is large is avoided, through data enhancement, the possibility of over-fitting is prevented, and the classification performance can be improved by increasing more feature information; the problem of gradient disappearance or gradient explosion and the problem of learning efficiency degradation are solved through the residual block.

Description

technical field [0001] The invention relates to the technical field of deep learning, in particular to a chip defect image classification method based on a ResNet network. Background technique [0002] An integrated circuit chip is a circuit with a specific function that integrates a certain number of electronic components together through a semiconductor process. Since its invention, integrated circuit chips are almost everywhere. Modern computing, communication, manufacturing, Internet and transportation systems all rely on the existence of integrated circuit chips. However, defects may be caused in every link of the integrated circuit chip manufacturing process, and the existence of defects will directly affect the lifespan and reliability. Therefore, the detection of integrated circuit chips is particularly important. [0003] In traditional production, manual detection is used for detection. Since there are many variable factors in manual detection and cannot be quan...

AI Technical Summary

Problems solved by technology

Since there are many variable factors in manual detection and cannot be described quantitatively, which affects the accuracy and reliability of detection, manual detection has the disadvantages of large subjective factors, poor real-time performance, high labor intensity, and high false detection rate.

In the process of product production, with the increase of production speed and stricter quality requirements, traditional manual inspection methods can no longer complete the production tasks with quality and quantity

Images