Closed streckus closed 5 months ago
Hi @streckus,
Let me check, I'll get back to you shortly.
Best,
Maximilian
Hi @streckus,
I've noticed the real-world data was missing object poses. Thank you for pointing this out. I've now uploaded them to the same folder as the MGNV2-Real dataset. Please merge these with the existing scene folders in dataset_real/ before proceeding.
To check the poses, you can run (git pull
):
python3 Scripts/colored_pcl.py --visualize_pose --scene 46 --viewpt 3 --data_root ./dataset_real
Before starting, please ensure you've reviewed difficult_objects.txt
and difficult_scenes.txt
to confirm the data meets your needs.
Best, Maximilian
Hi @maximiliangilles ,
thank you for updating the dataset with the missing pose information. This resolves the issue :)
Best, Michael
Hi, thanks for the great dataset. I wasn't able to find the ground truth poses for the real-world data. I understand these annotations are only provided for a subset of frames. For which frames are the ground truth poses provided and where do I find the pose annotations? Best regards, Michael