Magnetic shoe defect segmentation method based on deep learning

Abstract

The invention relates to a magnetic shoe defect segmentation method based on deep learning. The method comprises the steps of S1, preprocessing marked data, dividing the data into a training set and a test set according to a proportion, and performing training of an improved MobileNetV3 semantic divider to obtain a semantic divider for magnetic shoe defect detection; and S2, inputting a to-be-detected magnetic shoe defect image, adjusting the to-be-detected magnetic shoe defect image to a uniform size, and then performing magnetic shoe surface defect detection by using a semantic segmentation device for magnetic shoe defect detection. The method has the characteristics of high automation degree, no environmental influence, low cost and the like. Machine recognition is utilized, manual intervention recognition is not needed, and the detection precision and speed meet the industrial requirements.

Description

technical field [0001] The present invention relates to the fields of deep learning and defect segmentation, in particular to a method for segmenting magnetic tile defects based on deep learning. Background technique [0002] Semantic segmentation is one of the key problems in the field of computer vision today. Semantic segmentation is a classification at the pixel level, and pixels belonging to the same category must be classified into one category, so semantic segmentation understands images from the pixel level. Note that semantic segmentation is different from instance segmentation. For example, if there are multiple people in a photo, for semantic segmentation, it is only necessary to classify the pixels of all the people into one category, but instance segmentation also divides the pixels of different people. categorized into different categories. That is to say, instance segmentation is a step further than semantic segmentation. Semantic segmentation has a wide ra...

AI Technical Summary

Problems solved by technology

[0003] At present, most of the magnetic tile defect detection is still using the human eye detection method, which consumes a lot of human experience and physical strength, high labor costs, low efficiency, and slow detection speed.

There are also traditional methods to detect defects, but the traditional methods have higher requirements for lighting and shooting angles, and more parameters need to be adjusted, which is more cumbersome

Obviously, the existing magnetic tile defect detection methods are inefficient and have high labor costs, while the recognition accuracy and speed of the semantic segmentation algorithm provide strong conditions for magnetic tile defect detection

Images