Soft weighted multi-stage network model applied to semantic segmentation

Abstract

The invention discloses a semantic segmentation network model applied to a complex scene, which adopts a hole transformation feature pyramid module and a stage feature attention module to form a soft weighting multi-stage feature network, void space pyramid pooling and a self-adaptive transformation function are used as transformation functions of deep layer and shallow layer stage features in a module respectively, the whole receptive field of the network is expanded, and the features and global information of a large-scale target are better extracted; and the staged feature attention module is used for extracting different attention degrees on the multi-scale target from the features in different stages, so that the features with identification degrees in different stages are effectively stored. Compared with a feature pyramid module, the hole transformation feature pyramid module can better extract target features with overlarge scale difference from multiple stages. The staged feature attention module can extract different attention degrees on a multi-scale target from features of different stages.

Description

technical field [0001] The invention relates to the field of semantic segmentation, relates to a multi-scale target segmentation network for urban street view images, and in particular to a soft-weighted multi-stage network model applied to semantic segmentation. Background technique [0002] In recent years, with the rapid development of vehicle cameras, surveillance cameras and other equipment, a large number of images of urban street scenes have been generated. Understanding the semantic information contained in these images is an important basis for urban applications such as autonomous driving and intelligent services. [0003] Multi-scale object segmentation in complex scenes such as city streets has always been a challenging problem in semantic segmentation tasks. This kind of image has the problem that the size difference between objects is too large, which makes the network unable to extract the features of objects at various scales while maintaining timeliness. ...

AI Technical Summary

Problems solved by technology

However, these methods not only fail to make full use of the multi-scale information generated by the backbone network, but also fail to alleviate the irreversible loss caused by encoding

Images