Multi-scale space joint model and visual detection method thereof

Abstract

The invention discloses a multi-scale space joint model and a visual detection method thereof, and belongs to the technical field of machine visual defect detection, the multi-scale space joint model comprises an FCOS basic network, the FCOS basic network is connected with a classification fusion pyramid and a regression fusion pyramid, and the classification fusion pyramid is connected to the regression fusion pyramid through a class space regression similarity module. The multi-scale class space regression quality information and the multi-scale regression space feature information of the image data can be extracted through the FCOS basic network, and the multi-scale class space regression quality information and the multi-scale regression space feature information are subjected to data fusion processing through the classification fusion pyramid and the regression fusion pyramid. Therefore, defect features are enhanced, and the accuracy of defect detection is improved. According to the method, the category feature information is input into the regression fusion pyramid to guide information fusion of the regression space, so that the condition that classification and regression results are inconsistent in a defect detection stage when the model is used for model training can be reduced, and the defect detection accuracy is ensured.

Description

technical field [0001] The invention relates to the technical field of machine vision defect detection, in particular to a multi-scale space joint model and a visual detection method thereof. Background technique [0002] In the process of industrial production, due to the manufacturing process, use intensity, working environment and other reasons, product defects are often unavoidable, and inspection of the product surface is one of the important links to ensure the quality of the finished product. Manual inspection is one of the traditional inspection solutions. Although this method does not have high technical requirements for relevant personnel, it needs to train workers to identify complex surface defects. Compared with machine identification, manual inspection is time-consuming. , The efficiency is low, and it is easily affected by the external environment and subjective judgment in the detection process, resulting in unstable detection accuracy, prone to missed detect...

AI Technical Summary

Problems solved by technology

[0009] The purpose of the present invention is to overcome the problems of low detection accuracy and positioning accuracy in the prior art, and provide a multi-scale spatial joint model and its visual detection method

Images