search

ZCMax / LLaVA-3D

A Simple yet Effective Pathway to Empowering LLaVA to Understand and Interact with 3D World
169 stars 4 forks source link

About Training Data #15

Open LightManxx opened 1 week ago

LightManxx commented 1 week ago

Hello, I am trying to make the dataset in the paper by myself. I found that the data in SceneVerse are all in point cloud format, without image, depth, etc. Does the original dataset contain these data or have you done other processing?

ZCMax commented 1 week ago

Thanks for your question~ We only use the language annotation from the SceneVerse dataset, which only provides the point cloud data, you can find the image, depth, etc in EmbodiedScan.