search

NVlabs / FoundationPose

[CVPR 2024 Highlight] FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects
https://nvlabs.github.io/FoundationPose/
Other
955 stars 99 forks source link

Clear explanation needed of what is possible with this model (given certain inputs) #58

Closed jkim447 closed 2 weeks ago

jkim447 commented 1 month ago

Thank you for the great work. It is not super clear which inputs are absolutely necessary for pose estimation here. If I have RGB (no depth) and a 3D CAD model, can I use your model? Or would I need further input modalities here?

abhishekmonogram commented 1 month ago

It is mentioned in the paper that Foundation Pose is rgbd based method and will require a depth map.

In this paper we propose a unified framework called FoundationPose that performs both pose estimation and tracking for novel objects in both the model-based and model-free setups, using RGBD images

wenbowen123 commented 3 weeks ago

https://github.com/NVlabs/FoundationPose/issues/77