Lane line multi-task learning detection method based on road segmentation

Abstract

The invention relates to the technical field of road traffic image recognition, and discloses a lane line multi-task learning detection method based on road segmentation, and the method comprises thesteps: constructing a multi-task learning network, carrying out the recognition and processing of an inputted road image, and outputting road segmentation data and lane line detection data; constructing a link coding structure, and connecting the two sub-networks; and performing alternate fine adjustment training on the two sub-networks, correcting and improving the lane line detection data precision by adopting a cross entropy loss function, and finally outputting corrected lane line detection data. The detection scale is diversified, the robustness is improved, the detection precision in a complex scene is improved, and the speed is high. The two sub-networks are connected through the link coding structure, and the information amount obtained by the feature map is increased through parameter hard sharing of the two sub-networks.

Description

technical field [0001] The invention relates to the technical field of image recognition of road traffic, in particular to a lane line multi-task learning detection method based on road segmentation. Background technique [0002] At present, the methods of lane line detection are mainly divided into traditional algorithms and detection algorithms based on deep learning. [0003] The traditional algorithm adopts the method of "feature extraction and expression + feature matching" for detection, which can be divided into edge feature method and color feature method. in: [0004] The edge feature method counts the global gradient angle through the edge distribution function, and determines the position of the lane line according to the symmetry of the lane line. The advantage is that it has good robustness to the shape of the lane line, and it can still detect the lane line reliably in the case of strong interference; the disadvantage is that the extraction of feature informa...

AI Technical Summary

Problems solved by technology

In order to overcome the problem that the detection accuracy of the existing lane line detection network drops significantly in complex scenes, the road segmentation network DeepLab v3+ is fused with the above-mentioned SCNN into a multi-task learning network, and the two sub-networks are connected by link coding structure

Images