Fig. 7
From: A road surface reconstruction dataset for autonomous driving
Inference results by monocular depth estimation methods. From up to down: input RGB images, surface normal maps, and colored point clouds. For better visualization, we show the surface normal maps calculated from the depth maps.
