search

ZzZZCHS / Chat-Scene

Code for "Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifiers" (NeurIPS 2024)
MIT License
113 stars 8 forks source link

Point cloud data feature #28

Closed KimWu1994 closed 7 months ago

KimWu1994 commented 7 months ago

Hello, in preprocess/extract_3dfeat.py and feature file 'data/annotations/scannet_pointgroup_uni3d_feats.pt'. For example, one of the data is 'scene0000_00_02'. I think the first '0000' regards the scene number. But what does the second '00' regard? What does the third '02' regard?

ZzZZCHS commented 7 months ago

For 'scene0000_00_02', 'scene0000_00' is the scene id, '02' is the object id.

KimWu1994 commented 7 months ago

How do I know which object the object id corresponds to? Is it sorted by the name of the object?

ZzZZCHS commented 7 months ago

For 3D instance segmentation models like PointGroup and Mask3D, their predicted result for each scene is a list of instances with their ID, predicted mask, and confidence score. We directly use the IDs in the predicted results. Since the predicted results (especially the masks) are too large to be released, I recommend you referring to their repos (PointGroup, Mask3D) to run the instance segmentation with their pretrained weights. We postprocess the predicted results into sets of (ID, bbox, class name), which can be further visualized in the 3D scene. We will release the processing scripts within this week.

KimWu1994 commented 7 months ago

Thanks so much !!! I get it.