RSRDDPS

Fine-grained road reconstruction recovers the road surface profile up to the macro-texture level. It provides the possibility of predicting the vehicle response in advance, thus benefiting decision making and control of autonomous vehicles to improve the driving safety and comfort.The RSRD provides high-resolution, high-accuracy, and multi-condition stereo images and point clouds. It contains 2800 data pairs with dense point cloud labels and 13K pairs with sparse labels. This dataset can act as a benchmark for monocular depth estimation, multi-view stereo, binocular stereo matching, structure from motion, or direct point cloud processing.