NVlabs / FoundationPose

[CVPR 2024 Highlight] FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects
https://nvlabs.github.io/FoundationPose/
Other
1.48k stars 201 forks source link

Regarding model-free related files #167

Closed dengyufuqin closed 2 months ago

dengyufuqin commented 4 months ago

First of all, thank you for your contribution and sharing.

I have a question. In the method in model-free, I have a few questions I would like to ask.

  1. What data should be in cam_in_ob?
  2. What data is in depth_enhanced?
  3. What is the meaning of select_frames.yml?

Thank you very much

wenbowen123 commented 4 months ago

1)camera pose wrt. object 2)it's just depth image 3)You dont need this for your custom data. This was used to sample few-shot frames on the public benchmark.

8866lt commented 3 months ago

Thank you for your contribution and sharing!!!

I hope this message finds you well. I am currently working on a project involving 6D pose estimation and tracking of novel objects and am highly interested in your FoundationPose framework.

Could you please provide detailed guidance on how to create and train a custom dataset in a model-free setting using FoundationPose? Specifically, I am looking for information on:

  1. How to Prepar and structure the dataset.
  2. Required annotations and formats.
  3. Training pipeline and key parameters to adjust.

Thank you for your time and assistance. I look forward to your response.