| Submission name | TF6D (CNOS) + Megapose (5 hypothesis) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Submission time (UTC) | Nov. 29, 2024, 12:40 p.m. | ||||||||||
| User | tf6d | ||||||||||
| Task | Model-based 6D localization of unseen objects | ||||||||||
| Dataset | LM-O | ||||||||||
| Description | 1) Submitted to: BOP Challenge 2024 2) Training-free method: Our method does not require any task-specific training 3) Pose refinement: Yes, the pose refinement network is from MegaPose with 5 hypotheses 4) Onboarding data: 40 rendered templates 5) Used 3D models: Default, CAD 6) Testing data: RGB 7) Segmentation method: Default, CNOS | ||||||||||
| Evaluation scores |
|
| User | tf6d |
|---|---|
| Publication | |
| Implementation | |
| Training image modalities | None |
| Test image modalities | RGB |
| Description | Submitted to: BOP Challenge 2024 Training data: None, our method does not require any task-specific training Onboarding data: Model-based rendering, 40 rendered templates Used 3D models: Default, CAD Notes: 1) Testing data: RGB 2) Segmentation method: Default, CNOS 3) Our method does not introduce any additional or task-specific training 4) Pose refinement: Yes, the pose refinement network is from MegaPose with 5 hypotheses 5) Key idea: Our method leverages 3D-consistent features derived from 3D foundation models for training-free 6D pose estimation, rather than relying on 2D foundation models (e.g., DINO). |
| Computer specifications | NVIDIA A100 80GB PCIe + Intel(R) Xeon(R) Gold 6342 CPU @ 2.80GHz |