Image semantic segmentation model training method for traffic road scene

Abstract

The invention provides an image semantic segmentation model training method for a traffic road scene. The method comprises the steps of constructing a semantic segmentation basic model, adjusting the structure of a basic network to form a semantic segmentation initial model, and training the semantic segmentation initial model by using a sample image training set of the traffic road scene to obtain an image semantic segmentation model. In the prior art, an up-sampling operator of an up-sampling module of an image semantic segmentation model uses a nearest neighbor interpolation mode for calculation; the problem that the accuracy of a final image semantic segmentation result is poor due to the fact that a feature map, output by an up-sampling module, of an image semantic segmentation model loses a large amount of pixel information compared with an original input image is solved.

Description

technical field [0001] The invention relates to the technical field of computer vision image processing, in particular to an image semantic segmentation model training method for traffic road scenes. Background technology [0002] Image semantic segmentation is one of the core research problems in the field of computer vision. The goal of image semantic segmentation is to assign labels to each pixel of the input image, that is, to realize the object classification task at the pixel level. The pixels of the input image are predicted and classified through the image semantic segmentation model to generate semantic labels, and finally the image is divided into several pixel regions with a specific semantic meaning. [0003] In the traffic road scene, image semantic segmentation technology is widely used. Image semantic segmentation technology provides the possibility of information perception in the traffic road scene by accurately analyzing and distinguishing the targets such as dr...

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to provide a kind of image semantic segmentation model training method for traffic road scenes, to solve the problem in the prior art that the upsampling operator of the upsampling module of the image semantics segmentation model uses the nearest neighbor interpolation mode calculation; Compared with the original input image, the feature map output by the upsampling module of the image semantic segmentation model loses a large amount of pixel information, which affects the semantic segmentation performance of the image semantic segmentation model, resulting in poor accuracy of the final image semantic segmentation result. question

Images