Image segmentation method, related device, equipment and storage medium

Abstract

The invention discloses an image segmentation method related to an artificial intelligence technology. The method comprises the following steps: obtaining original image features of an original image; a first image feature is obtained through an encoder in the image segmentation model, the local vision module comprises a spatial enhancement module, the spatial enhancement module generates a spatial scaling factor based on the input image feature, and the spatial scaling factor can correct the spatial feature of the input image feature; obtaining a second image feature through a decoder in the image segmentation model; splicing the second image feature and a third image feature output by a first local vision module in the encoder to obtain a target image feature; and generating a target segmented image according to the target image features. The invention further provides a device, equipment and a medium. According to the method, the spatial scaling factor is used for adjusting the spatial feature of the image feature, and the spatial scaling factor is adaptively generated based on the input image feature, so that the difficulty of model configuration is reduced.

Description

technical field [0001] The present application relates to the technical field of computer vision, and in particular, to an image segmentation method, related apparatus, equipment, and storage medium. Background technique [0002] With the continuous development of computer technology and artificial intelligence (Artificial Intelligence, AI) technology, AI technology has achieved remarkable results in tasks such as image segmentation and image recognition. With the emergence of more and more efficient network structures, computer vision and natural language processing also tend to converge. Using Transformers to complete visual tasks has become a new research direction to reduce the complexity of the structure and explore possible Scalability and training efficiency. [0003] At present, a visual Transformer that can solve computer vision problems is proposed. In order to enhance the feature representation capability of the Visual Transformer, in the existing scheme, additi...

AI Technical Summary

Problems solved by technology

[0004] However, the inventors found that there are at least the following problems in the existing solutions. It is difficult to decide in advance how many branches to add to the visual Transformer and which linear transformation to use, and it often takes multiple attempts to know the optimal configuration.

As a result, the model configuration process is cumbersome and reduces the efficiency of model configuration

Images