Learning-Deep-Learning

Pseudo-LiDAR++: Accurate Depth for 3D Object Detection in Autonomous Driving

June 2019

tl;dr: Improve depth estimation of pseudo-lidar with stereo depth network (SDN) and sparse depth measurements on “landmark” pixels with few-line lidars.

Overall impression

This is exactly what I wanted to do when I read pseudo-lidar. However we could still explore the idea with radar data.

Pseudo-lidar bridged half of the gap between RGB-based and lidar based 3D object detection but does not perform well for far-away object. Pseudo-lidar++ uses sparse 3D measurement to de-bias the depth estimation.

The Uber ATG group also publishes several papers (ContFuse, MMF) on this idea, although not as explicit as the pseudo-lidar paper or this one.

Key ideas

Technical details

Notes