Urban road remote sensing image segmentation method and device, electronic equipment and storage medium

Abstract

The invention discloses an urban road remote sensing image segmentation method. The method comprises the following steps: acquiring an urban road remote sensing image to be processed; inputting a pre-trained segmentation model, and performing feature extraction by using the trunk feature extraction network to obtain low-level features and high-level features; extracting multi-scale features from the advanced features by using a multi-scale feature extraction network, and performing convolution operation on multi-scale fusion features obtained by feature fusion to obtain first features; performing convolution on the low-level features by using a feature fusion network, performing up-sampling operation on the first feature, performing feature fusion on two operation results, and performing convolution and up-sampling on a feature fusion result to obtain a segmentation result of the urban road remote sensing image; creating the multi-scale feature extraction network as a pooling structure of a softpool spatial pyramid with holes based on pooling operation; segmentation model training uses a focus loss function. Information loss caused by pooling can be reduced, and the problem of unbalance of positive and negative samples is solved.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a method, device, electronic equipment and storage medium for segmenting urban road remote sensing images. Background technique [0002] In recent years, image segmentation technology has become a hot research direction in the field of computer vision, and its research results can be effectively applied to transportation, surveying and mapping, face recognition and other fields. At the same time, with the continuous development of remote sensing technology, the resolution of remote sensing images has been greatly improved, and high-resolution remote sensing images have become an important data source for digital image processing. Because high-resolution remote sensing images can provide rich feature information of ground features, and at the same time update quickly and with high precision, they have played an important role in urban planning, traffic management, vehicl...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the embodiments of the present invention is to provide a method, device, electronic equipment and storage medium for urban road remote sensing image segmentation, so as to reduce the information loss caused by pooling as much as possible while maintaining the function of the pooling layer, and solve the problem of positive and negative The problem of sample imbalance

Images