Skip to content

Latest commit

 

History

History
90 lines (46 loc) · 2.78 KB

appro.md

File metadata and controls

90 lines (46 loc) · 2.78 KB

3D Object Detection/ Approaches

Type 1: monocular (M), stereo (S), LiDAR 64 beams (L), LiDAR 4 beams (L4), RADAR (R)
Type 2: supervised (sup), unsupervised (unsup), semi-supervised, (semi-sup), self-supervised (self-sup)

Monocular

Supervised

ICCV 19 Pseudo-LiDAR e2e ()[Notes]

CVPR 19 Pseudo-LiDAR (convert depth map into pseudo 3d pcl) [Notes]

Unsupervised

CVPR 17 Unsupervised Monocular Depth Estimation with Left-Right Consistency (end-to-end/ novel loss that enforces left-right depth/ An effective evaluation) [TOREAD]

Semi-supervised

Self-supervised

ECCV 20 Pseudo RGB-D for Self-Improving MonocularSLAM and Depth Prediction (outperform state-of-the-art self-supervised monocular and stereo depth prediction networks (e.g., Monodepth2)) [TOREAD]

Binocular

Supervised

arXiv 20 CG-Stereo

CVPR 20 Pseudo-LiDAR V3 E2E

ICRL 20 Pseudo-LiDAR ++ () [Notes]

CVPR 19 Pseudo-LiDAR [Notes]

Unsupervised

Semi-supervised

Self-supervised

LiDAR only

Supervised

CVPR 19 PointRCNN () [Notes]

Unsupervised

Semi-supervised

Self-supervised

Fusion

Supervised

Unsupervised

Semi-supervised

Self-supervised

Downsample LiDAR

Supervised

ICRL 20 Pseudo-LiDAR ++ () [Notes]

Unsupervised

Semi-supervised

Self-supervised