search

FVPLab / Argus-3D

92 stars 6 forks source link

How to process my own datasets? #13

Open LaFeuilleMorte opened 1 month ago

LaFeuilleMorte commented 1 month ago

Great work! I'd like to use my own datasets for training. Do you have any scripts that allow me to convert my own datasets (some .obj/ .glb/ .stl format meshes) to the same format as in the Objaverse-Mix. Many thanks ! Besides, I've some questions regarding this dataset:

  1. I've taken a deep look into this datasets. However I found some extreme low quality data inside this dataset. Is there anyway we can filtered these out?
  2. How to filter subdataset by category? Is there any meta data I can use?
AImind commented 4 weeks ago

The conversion process is somewhat complex. We rendered multi-view images following the guidelines from https://github.com/allenai/objaverse-rendering, generated text captions using BLIP2, generated occupancy data based on https://github.com/autonomousvision/occupancy_networks, sampled point cloud in Blender using Poisson-disk sampling.

A1: Our filtering process was rather basic. For more comprehensive filtering, you can refer to https://github.com/ashawkey/objaverse_filter A2: The Objaverse dataset lacks category annotations. If you need category-based filtering, you might explore the Objaverse-LVIS dataset, which is a subset of Objaverse with LVIS annotations.