search

nv-nguyen / gigapose

[CVPR 2024] PyTorch implementation of GigaPose: Fast and Robust Novel Object Pose Estimation via One Correspondence
https://nv-nguyen.github.io/gigaPose/
MIT License
144 stars 12 forks source link

Inference with BOP Challenge 2024 HOPE Dataset - Missing keys_to_shard.json #15

Open Seunggyu-Lee opened 3 months ago

Seunggyu-Lee commented 3 months ago

Hi,

Thank you for your great work.

I was trying to inference your model with the BOP Challenge 2024 dataset "HOPE" but it does not seem to have the keys_to_shard.json file (possibly due to dataset version changes?).

As a result, the inference code does not work.

Is there any way or suggestions to use your model with the 2024 HOPE dataset?

If I've misunderstood something or if there are additional steps I need to follow, your guidance would be greatly appreciated.

Thank you for your help!

Seunggyu-Lee commented 3 months ago

I think even lmo dataset does not have keys_to_shard.json file too

sp9103 commented 3 months ago

Hi, Seunggyu

Could you check how the dataset was obtained? If it was downloaded directly through Hugging Face or the BOP website, it’s normal for it not to be included.

When you download it using the script below, as mentioned in the ReadME, it calls two conversion scripts, and one of them, convert_imagewise_to_webdataset.py, generates the key_to_shard.json file.

python -m src.scripts.download_test_bop23