Method and device for carrying out image segmentation processing by utilizing deep transfer learning

Abstract

The invention discloses a method and a device for carrying out image segmentation processing by utilizing deep transfer learning. The method comprises the following steps: acquiring metal microstructure images at a first temperature; training a segmentation model by using each image of the first training set, wherein, the training is verified by using the verification set after each round of training by using each image, and an evaluation index of the segmentation model after the round of training is obtained; selecting an optimal model according to the evaluation index of each round, and obtaining model parameters corresponding to the optimal model; constructing a new segmentation model by using the model parameters; adjusting the new segmentation model by using metal microstructure images at a second temperature as a second training set; and performing prediction by using the adjusted new segmentation model to obtain a prediction result. Through the invention, the problem of poor model prediction effect caused by few training set images in the prior art is solved, so that the prediction effect of models is improved.

Description

technical field [0001] This application relates to the field of image processing, in particular, to a method and device for image segmentation processing using deep transfer learning. Background technique [0002] Microstructure is the key to the relationship between material composition, process, organization and performance, which can reveal the synergy between material composition and process, and then adjust the performance of materials. Therefore, how to use deep learning technology to improve the accuracy of microstructure identification and effectively extract microstructure image features is of great significance for further understanding the relationship between microstructure and material properties. [0003] In the field of metallic structural materials, microstructural images are characterized by the presence of different interlaced precipitated phases within small areas. figure 2 is a comparison chart between the model prediction results and the real labels obt...

AI Technical Summary

Problems solved by technology

[0004] The embodiment of the present application provides a method and device for image segmentation processing using deep transfer learning to at least solve the problem in the prior art that the model prediction effect is poor due to the lack of images in the training set

Images