Skip to content

KITTI-3D Benchmarks.md

Latest commit

 

History

History
180 lines (128 loc) · 11.8 KB

KITTI-3D Benchmarks.md

File metadata and controls

180 lines (128 loc) · 11.8 KB

KITTI-3D Benchmarks

This benchmark follows the nice work: 3dodi-survey and Awesome-Monocular-3D-detection, thanks a lot! !

Standard Setting

Paper Code Venue Easy Moderate* Hard
MonoATT None CVPR'23 24.72 17.37 15.00
MonoPGC None ICRA'23 24.68 17.17 14.14
MonoDDE None CVPR'22 24.93 17.14 15.10
MonoDETR Pytorch arxiv'22 25.00 16.47 13.58
MonoCon Pytorch AAAI'22 22.50 16.46 13.95
IDMS None arxiv'22 22.50 16.19 13.49
MonoJSG Pytorch CVPR'22 22.69 16.14 13.64
DAE None CVPR'22 23.62 16.10 13.41
OBMO Pytorch arxiv'22 22.71 15.70 13.23
GPENet None arxiv'22 22.41 15.44 12.84
HOMO None CVPR'22 21.75 14.94 13.07
MonoEdge None WACV'23 21.08 14.47 12.73
DEVIANT Pytorch ECCV'22 21.88 14.46 11.89
MonoGround Pytorch CVPR'22 21.37 14.36 12.67
GUPNet Pytorch ICCV'21 20.11 14.20 11.77
MonoFlex Pytorch CVPR'21 19.94 13.89 12.07
MonoRCNN Pytorch ICCV'21 18.36 12.65 10.03
MonoDLE Pytorch CVPR'21 17.23 12.26 10.29
MonoPair None CVPR'20 13.04 9.99 8.65
SMOKE Pytorch CVPRW'20 14.17 9.88 8.63
MonoGRNet Pytorch AAAI'19 9.61 5.74 4.25

LiDAR Signal [training]

Code Venue Easy Moderate* Hard
CMKD Pytorch ECCV'22 25.09 16.99 15.30
Shape-Aware None arxiv'22 23.84 16.52 13.88
DID-M3D Pytorch ECCV'22 24.40 16.29 13.75
MonoDistill Pytorch ICLR'22 22.97 16.03 13.60
MonoDTR Pytorch CVPR'22 21.99 15.39 12.73
CaDDN Pytorch CVPR'21 19.17 13.41 11.46
MonoRUn Pytorch CVPR'21 19.65 12.30 10.58
Ouyang et al. None ACCV'20 11.52 8.26 6.97
MonoPSR None CVPR'19 10.76 7.25 5.85

KD Method & LiDAR Signal [training]

Code Venue Easy Moderate* Hard
Mix-Teaching Pytorch arXiv'22 26.89 18.54 15.79
CMKD Pytorch ECCV'22 25.09 16.99 15.30
ADD None arXiv'22 25.61 16.81 13.79
MonoDistill Pytorch ICLR'22 22.97 16.03 13.60

Weak Supervision

Code Venue Easy Moderate* Hard
WeakMono3D Pytorch CVPR'23 6.98 4.85 4.45
WeakM3D Pytorch ICLR'22 5.03 2.26 1.63

Semi Supervised Active Learning

Code Venue Easy Moderate* Hard
MonoLiG None arxiv'23 - - -

Temporal Sequence [training&testing]

Venue Easy Moderate* Hard
DfM ECCV'22 22.94 16.82 14.65
Kinematic3D ECCV'20 19.07 12.72 9.17
QD-3DT arXiv'21 12.91 9.33 7.86

Occupancy

Code Venue Easy Moderate* Hard
OccupancyM3D Pytorch arxiv'23 26.87 19.96 17.15

CAD Model [training]

Venue Easy Moderate* Hard
DCD ECCV'22 23.81 15.09 13.21
AutoShape ICCV'21 22.47 14.17 11.36

Stereo Pair [training]

Venue Easy Moderate* Hard
SGM3D arXiv'21 22.46 14.65 12.97
DLE BMVC'21 24.23 14.33 10.30
Ground-Aware RA-L'21 21.65 13.25 9.91
RTM3D ECCV'20 14.41 10.34 8.77

External Data [training]

Aux. Data Venue Easy Moderate* Hard
DD3D DDAD-15M ICCV'21 23.22 16.34 14.20
Lian et al. COCO CVPR'22 23.41 15.26 12.80
MonoEF KITTI-Raw CVPR'21 21.29 13.87 11.71
DFRNet KITTI-Depth ICCV‘21 19.40 13.63 10.35
PCT KITTI-Depth NeurIPS'21 21.00 13.37 11.31
DDMP-3D KITTI-Depth CVPR'21 19.71 12.78 9.80
Demystifying KITTI-Depth ICCV'21 22.40 12.53 10.64
IAFA COCO ACCV'20 17.81 12.01 10.61
D4LCN KITTI-Depth CVPR'20 16.65 11.72 9.51
PatchNet KITTI-Depth ECCV'20 15.68 11.12 10.17
AM3D KITTI-Depth ICCV'19 16.50 10.74 9.52
Neighbor-Vote KITTI-Depth ACM MM'21 15.57 9.90 8.89
MonoCInIS Instance Mask ICCVW'20 11.08 7.02 5.63
Decoupled-3D KITTI-Depth AAAI'20 11.08 7.02 5.63
MonoFENet KITTI-Depth T-IP'19 8.35 5.14 4.10

CAD Model [training], External Dataset [training]

Aux. Dataset Venue Easy Moderate* Hard
ROI-10D KITTI-Depth CVPR'19 4.32 2.02 1.46

LiDAR Signal [training], External Dataset [training]

Aux. Dataset Venue Easy Moderate* Hard
pseudo-Stereo KITTI-Depth CVPR'22 23.74 17.74 15.14
DA-3Ddet KITTI-Depth ECCV'20 16.77 11.50 8.93

Stereo Pair [training&testing]

Venue Easy Moderate* Hard
RTS3D AAAI'21 58.51 37.38 31.12
Stereo CenterNet Neurocomputing'22 49.94 31.30 25.62
SIDE WACV'19 47.69 30.82 25.68
Stereo R-CNN CVPR'19 47.58 30.23 23.72
TLNet CVPR'19 7.64 4.37 3.74

CAD Model [training], Stereo Pair [training&testing]

Venue Easy Moderate* Hard
Disp R-CNN T-PAMI'20 67.02 43.27 36.43
Disp R-CNN CVPR'20 58.53 37.91 31.93

LiDAR Signal [training], Stereo Pair [training&testing]

Venue Easy Moderate* Hard
LIGA Stereo ICCV'21 81.39 64.66 57.22
CDN NeurIPS'20 74.52 54.22 46.36
DSGN CVPR'20 73.50 52.18 45.14
YOLOStereo3D ICRA'21 65.68 41.25 30.42

External Dataset [training], Stereo Pair [training&testing]

Aux. Dataset Venue Easy Moderate* Hard
RT3DStereo SceneFlow ITSC'19 29.90 23.28 18.96