search

OpenRobotLab / PointLLM

[ECCV 2024] PointLLM: Empowering Large Language Models to Understand Point Clouds
https://runsenxu.com/projects/PointLLM
450 stars 22 forks source link

PointBERT re-training Code #26

Closed fattho0919 closed 2 months ago

fattho0919 commented 2 months ago

Hello,

Thanks for the fantastic work.

I want to re-train the PointBERT in scene level, you mention that you re-train the PointBERT with color information so could you please provide the pretraining code to me, thanks a lot!!!

RunsenXu commented 2 months ago

Hi, Thanks for your interest in our work. You may refer to the original ULIP-2 codebase.

KzZheng commented 2 months ago

Hi Runsen @RunsenXu,

Do you know the features of your PointBERT aligns with which open-clip model? In the ULIP-2 paper, they mentioned using 'ViT-g-14', but its feature dimension is 1024, which does not match with 768 in the PointBERT. I wonder whether you know the answer. Thanks!

Best, Kz

RunsenXu commented 2 months ago

OpenCLIP ViT-L/14, datacomp xl s13b b90

Details are in the supplementary of the paper.