We collect images using a forward-looking camera mounted on the vehicle. The images are collected at different times under different weather condition. We process the images to obtain lane markings through a customized image segmentation method. We project the lane marking points in the high-definition map into the image. The files are named according to the unix epoch timestamp when the vehicle received the images. 
We utilize colors to differentiate our output:
White: the lane marking pixels extracted from the image;
Green: the lane marking points are consistent with the map;
Yellow: the lane marking points are undetermined;
Purple: the lane marking points are inconsistent with the map;