search

Pointcept / PointTransformerV3

[CVPR'24 Oral] Official repository of Point Transformer V3 (PTv3)
MIT License
758 stars 44 forks source link

Model Checkpoints #101

Open little-aztl opened 18 hours ago

little-aztl commented 18 hours ago

Fantastic work! Could you please tell me when the new model and checkpoints will be released? My collaborators and I are looking forward to taking advantages of this great work :) Thank you very much if you can release them before long.

Gofinge commented 18 hours ago

One month after CVPR DDL! With a super powerful self-supervised model weight.

Gofinge commented 16 hours ago

Fantastic work! Could you please tell me when the new model and checkpoints will be released? My collaborators and I are looking forward to taking advantages of this great work :) Thank you very much if you can release them before long.

BTW, which model weight to your team want to use? Maybe I can check my experiments records.

little-aztl commented 13 hours ago

Fantastic work! Could you please tell me when the new model and checkpoints will be released? My collaborators and I are looking forward to taking advantages of this great work :) Thank you very much if you can release them before long.

BTW, which model weight to your team want to use? Maybe I can check my experiments records.

Thanks for your reply! Our team want to use a pretrained model to obtain the local and global feature of an indoor scene. So we may be interested in model trained on an indoor dataset.

Gofinge commented 2 hours ago

In this case you can first use pre-trained PTv3 with multi-dataset joint training here(https://huggingface.co/Pointcept/PointTransformerV3/tree/main).

And after cvpr, we will have a big update on the codebase, and released self-supervised pre-trained PTv3, this version is more flexible and powerful.