Submission: MegaPose-CNOS_fastSAM+CoarseBest/HB

Download submission
Submission name
Submission time (UTC) Nov. 2, 2023, 7:10 a.m.
User agimus-happypose
Task Model-based 6D localization of unseen objects
Dataset HB
Training model type Default
Training image type Synthetic (custom)
Description
Evaluation scores
AR:0.251
AR_MSPD:0.397
AR_MSSD:0.222
AR_VSD:0.133
average_time_per_image:15.188

Method: MegaPose-CNOS_fastSAM+CoarseBest

User agimus-happypose
Publication Labbé et al.: MegaPose: 6D Pose Estimation of Novel Objects via Render & Compare, CoRL 2022
Implementation https://github.com/agimus-project/happypose
Training image modalities RGB
Test image modalities RGB
Description

This submission was prepared by Mederic Fourmy, Elliot Maître, Lucas Manuelli, Yann Labbé

In this submission, CNOS_fastSAM [A] detections (default detections for Task 4) are used as input to the MegaPose pose estimation method [B]. Detections are selected in each frame strictly according to the provided the object ids and number of instances provided for the challenge in the test datasets

For each detection, we run the MegaPose coarse network and take the best scoring as a prediction.

The following improvement has been made over the original MegaPose paper [B]:
- The orientation of the coarse hypotheses are obtained my discretizing SO(3) in 576 values using [C].

Instructions to reproduce these results will be made available on https://github.com/agimus-project/happypose.

[A] Nguyen et al.: CNOS: A Strong Baseline for CAD-based Novel Object Segmentation, arXiv 2023 [B] Labbé et al.: MegaPose: 6D Pose Estimation of Novel Objects via Render & Compare, CoRL 2022 [C] Yershova et al.: Generating Uniform Incremental Grids on SO(3) Using the Hopf Fibration, IJRR 2009, reference implementation: http://lavalle.pl/software/so3/so3.html

This work was granted access to the HPC resources of IDRIS under the allocation 011014301 made by GENCI

Computer specifications NVIDIA Tesla V100 32Go