Lane line health detection method and device and storage medium

The invention discloses a lane line health detection method and device and a storage medium, and belongs to the technical field of traffic road detection. Firstly, a lane line data set for semantic segmentation is built, an Erfnet segmentation model is built and trained to detect the positions of various lane lines in an incoming frame, then lane fitting, near lane line interception, background filtering and standard RGB pixel replacement are carried out according to a semantic segmentation feature map to obtain a local lane line original image and a local lane line standard image, and the local lane line original image and the local lane line standard image are obtained. Secondly, stretching feature maps obtained after edge convolution is carried out on the original image and the standard image into one-dimensional vectors, carrying out cosine calculation to obtain contour similarity, then respectively transferring pixels of lane line areas of the original image and the standard image from an RGB color space