6D Object Pose Estimation on T-LESS Primesense SIXD BOP 2018 (test)
29.5Object Recall (errvsd < 0.3)Pix2Pose
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Pix2PoseInput=RGB only2019.08 | 29.5 | — | |
| KehlInput=RGB-D2019.08 | 24.6 | — | |
| ImplicitInput=RGB-D2019.08 | 18.4 | — | |
| BrachmannInput=RGB-D2019.08 | 17.8 | — | |
| AAEInput Modality=Depth, Refinement=ICP, Bounding Box Source=Ground Truth 2D2019.02 | 0.8405 | 0.33 | |
| AAEInput Modality=Depth, Refinement=ICP2019.02 | 0.6857 | 0.4 | |
| Drost et al.Input Modality=Depth, Refinement=edge2019.02 | 0.675 | 21.5 | |
| Vidal et al.Input Modality=Depth, Refinement=ICP2019.02 | 0.6651 | 4.7 | |
| AAETraining Modality=RGB‡ (untextured CAD models), Input Modality=RGB, Bounding Box Source=Ground Truth 2D2019.02 | 0.3834 | 0.006 | |
| RetinaNetTraining Modality=RGB‡ (untextured CAD models), Input Modality=RGB2019.02 | 0.2679 | 0.077 | |
| Kehl et al.Input Modality=RGB-D, Refinement=ICP2019.02 | 0.246 | 1.8 | |
| RetinaNetTraining Modality=RGB† (3D reconstructions), Input Modality=RGB2019.02 | 0.1926 | 0.077 | |
| Brachmann et al.Input Modality=RGB-D2019.02 | 0.1784 | 13.5 | |
| SSDTraining Modality=RGB† (3D reconstructions), Input Modality=RGB2019.02 | 0.1467 | 0.024 |