search

NVlabs / FoundationPose

[CVPR 2024 Highlight] FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects
https://nvlabs.github.io/FoundationPose/
Other
1.01k stars 112 forks source link

Reason your own model data #42

Closed wsq1010 closed 1 month ago

wsq1010 commented 1 month ago

Hello I have a question to ask, in addition to the model file and camera calibration parameters and rgb images are also necessary in the reasoning

wenbowen123 commented 1 month ago

can you clarify your question?

wsq1010 commented 1 month ago

Deduce which files are needed in FoundationPose

wsq1010 commented 1 month ago

Can you tell me how to generate the file under the mask_cosypose path，What are the files in the mask_cosypose path.

wenbowen123 commented 1 month ago

The files you mentioned are all needed. Refer to run_demo.py for what data are used.

wsq1010 commented 1 month ago

Hi, I found a problem when extrapolating my own data.It seems that only the results of the first frame of data inference are normal.There are some problems in the subsequent inference graph, may I ask why this is the case? 1 2 8 12

wenbowen123 commented 1 month ago

Your data is likely not in the right format. Please check https://github.com/NVlabs/FoundationPose/issues/44#issuecomment-2048141043