| Submission name | |||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Submission time (UTC) | Oct. 8, 2025, 4:06 a.m. | ||||||||||||||||||||||||||
| User | jduke | ||||||||||||||||||||||||||
| Task | Model-free 2D detection of unseen objects | ||||||||||||||||||||||||||
| Dataset | HOT3D | ||||||||||||||||||||||||||
| Description | updated to using Hot3d masks that have been made available in https://huggingface.co/datasets/bop-benchmark/hot3d/commits/main 30fe9674782f32e1e5edba98476b6ff4300132c5. See also https://github.com/Vision-Kek/cnos25/commits/main/ 641bd08f5c0d116c0561e32e05973b995ff29ce3 | ||||||||||||||||||||||||||
| Evaluation scores |
|
| User | jduke |
|---|---|
| Publication | None |
| Implementation | https://github.com/Vision-Kek/cnos25 |
| Training image modalities | None |
| Test image modalities | RGB |
| Description | Submitted to: BOP Challenge 2025 Training data: No training data used. Onboarding data: Static onboarding. Method: The CNOS pipeline. with 2025 models. YOLOE is used for proposals, dinov3 for descriptors. |
| Computer specifications | RTX4090 |