search

Omni6DPose / GenPose2

[ECCV 2024] GenPose++: A generative category-level 6D object pose estimation and tracking approach proposed in Omni6DPose.
https://jiyao06.github.io/Omni6DPose/
MIT License
26 stars 1 forks source link

Original Dataset #6

Open w-fsachi opened 2 months ago

w-fsachi commented 2 months ago

Thank you for your work!

I'm trying to run GenPose2 on my original dataset for grasping unknown objects by robot arm. But I don't know how to create a ROPE format dataset. How to make own dataset? I'm using Realsense D435 and I can get color, depth and mask data. But I don't know how to get meta data.

Looking forward to your reply.

Jiyao06 commented 1 month ago

Thank you for your interest in GenPose++! For your application, you do not need to create a dataset in ROPE format. A significant part of the metadata in ROPE is used for evaluation purposes, and for running GenPose++, you only need the object masks, corresponding RGBD images, and the camera intrinsic parameters. You can focus on preparing your data according to these requirements for model inference. Additionally, we will soon release an inference version integrated with SAM to enhance usability. Stay tuned to our repository for updates!

Yeager-101 commented 1 month ago

Thank you for your interest in GenPose++! For your application, you do not need to create a dataset in ROPE format. A significant part of the metadata in ROPE is used for evaluation purposes, and for running GenPose++, you only need the object masks, corresponding RGBD images, and the camera intrinsic parameters. You can focus on preparing your data according to these requirements for model inference. Additionally, we will soon release an inference version integrated with SAM to enhance usability. Stay tuned to our repository for updates!

I am also trying to use Realsense D455 to test the algorithm. Is there any related routine? Now I feel a little bit at a loss. Thanks