Industrial image region-of-interest segmentation algorithm based on double-branch network

Abstract

The invention relates to an industrial image region-of-interest segmentation algorithm based on a double-branch network, and the algorithm comprises the following steps: S1, collecting an image data set of an industrial part; s2, performing region-of-interest labeling on the data set according to different quality inspection requirements; s3, preprocessing and amplifying the data set and dividing the data set; s4, constructing a double-branch network, wherein the double-branch network comprises a semantic information extraction branch and a spatial information extraction branch; according to the optimization target, training the network through the region-of-interest labeling data set to obtain a trained model; and S5, inputting a to-be-detected industrial image into the trained model to obtain a corresponding region-of-interest segmentation image. According to the method, a double-branch structure is adopted, the industrial image is divided into a plurality of regions of interest in real time according to the quality detection requirement of the industrial part, subsequent detection is facilitated, and the method has the advantages that the segmentation speed is high, the segmentation precision is high, manual extraction of the regions of interest can be replaced, the product quality evaluation difficulty is reduced, and the efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of image visual detection, in particular to an industrial image region-of-interest segmentation algorithm based on a dual-branch network. Background technique [0002] With the development of science and technology, industrial production has gradually entered the era of intelligence. The smart devices and products produced by industrial intelligence often require detector integrity and functionality. In the past, quality assessment of industrial parts was often performed manually, which not only required a lot of labor, but also contained a lot of subjectivity that affected its objective results. The quality assessment of the same industrial part has to go through many kinds of procedures, and the area of ​​interest that needs to be inspected corresponding to each procedure is not very different. Therefore, how to quickly and accurately segment an industrial part image into ROI images required by different...

AI Technical Summary

Problems solved by technology

Although these real-time segmentation networks can achieve a leap in inference speed, due to the sudden reduction in the number of parameters, although the speed requirements are met, a lot of accuracy is lost.

The main reason is that most of them choose to lose the low-level spatial information for segmentation, and the performance at the edge of the details is very poor, which will lose the low-level spatial information of the image, which greatly affects the accuracy of the network.

Images