Road scene recognition method and system based on multi-task learning neural network

Abstract

The invention provides a road scene recognition method and system based on a multi-task learning neural network, and the method comprises the following steps: obtaining the image information of a roadscene; carrying out semantic segmentation and target detection annotation on the image information to obtain a training data set; performing data enhancement processing on the training data set; constructing a multi-task learning neural network; training the multi-task learning neural network according to the enhanced training data set; and performing target detection and semantic segmentation onthe road scene image according to the trained multi-task learning neural network. According to the method, two tasks of semantic segmentation and target detection can be simultaneously realized by training one neural network, so that the road scene recognition time can be effectively shortened, the road scene recognition efficiency can be improved, and video memory resources can be saved.

Description

technical field [0001] The invention relates to the technical field of road scene recognition, in particular to a road scene recognition method based on a multi-task learning neural network and a road scene recognition system based on a multi-task learning neural network. Background technique [0002] At present, the networks commonly used in road scene recognition can only perform semantic segmentation or target detection alone. Therefore, when actually performing road scene recognition tasks, semantic segmentation and target detection need to be performed separately, and separate training is also required during network training. The semantic segmentation network and the target detection network, thus, tend to cause the recognition process and the training process to occupy more video memory resources and take a long time. Contents of the invention [0003] The present invention aims to solve one of the technical problems in the above-mentioned technologies at least to a...

AI Technical Summary

Problems solved by technology

[0002] At present, the networks commonly used in road scene recognition can only perform semantic segmentation or target detection alone. Therefore, when actually performing road scene recognition tasks, semantic segmentation and target detection need to be performed separately, and separate training is also required during network training. Semantic segmentation network and target detection network, thus, it is easy to cause the recognition process and training process to occupy more video memory resources and take a long time

Images