Image segmentation method and system based on edge auxiliary calculation and mask attention

Abstract

The invention discloses an image segmentation method and system based on edge auxiliary calculation and mask attention, and the method comprises the steps: building a feature encoder constructed by a multi-stage cascaded residual module, obtaining an edge feature map through fusing three superficial layer feature maps, obtaining an edge prediction image after feature dimension reduction, enhancing the representation capability of the first three layers of feature encoders, and obtaining an edge prediction image; the last-order residual error module sequentially passes through a plurality of feature decoders and a mask attention module, the mask attention module is used for improving the focus attention of the feature decoder of each level on a local area, and a segmentation result image predicted by a corresponding scale is output at each level. And fusing the output feature map of the feature decoder and the edge feature map of the first three-order residual error module, and predicting a final segmentation result image through feature dimension reduction. Compared with an existing image segmentation method, the method can provide more accurate segmentation edge prediction, is suitable for image segmentation in various complex scenes, and is higher in generalization performance and better in segmentation effect.

Description

technical field [0001] The invention belongs to the field of computer vision, and relates to an image segmentation method and system based on edge auxiliary calculation and mask attention. Background technique [0002] At present, the image segmentation technology based on deep learning is an important research direction in the field of computer vision, and has been widely used. The existing image segmentation method is to use the deep learning model to classify each pixel in the image, and finally get each pixel. Semantic categories of pixels. However, the existing methods still have the following problems. The model does not accurately segment the edge of the target in the image, cannot make full use of multi-scale context information, there is too much information loss in the prediction process, and the target loss function optimized by the model is too large. Single, unable to effectively model, the above problems will eventually affect the segmentation effect of the mo...

AI Technical Summary

Problems solved by technology

However, the existing methods still have the following problems. The model does not accurately segment the edge of the target in the image, cannot make full use of multi-scale context information, there is too much information loss in the prediction process, and the target loss function optimized by the model is too large. Single, unable to effectively model, the above problems will eventually affect the segmentation effect of the model

Images