Preparing The Dataset

[ardianumam]

Hi,
Thanks for the interesting work! I've been reading the paper and would like to get the method details and the data flow by digging the code using a common debug tool (step over, step into, etc). Could you give some directions on how to prepare the minimum dataset for that? I break it down as follows.

1. There are some download items on the d-faust's page (see Image 2), what will be the minimum to be downloaded since downloading them all is very big?
2. For the config items below, (i) what will be the filetype and directory structure in path L1? (ii) basically csv file in L2, we can define by ourselves to split train-test data? (iii) what data should we put in folder L3 - L6?

L1. dataset_directory: "/media/paschalidoud/MyData/onet_data/Humans/D-FAUST/"
L2. splits_file: "../config/d-faust_splits.csv"
L3. mesh_folder: "mesh_seq"
L4. points_folder: "points_seq"
L5. surface_points_folder: "surface_points_seq"
L6. renderings_folder: "renderings-downsampled"

3. Or for the specific purpose mentioned in the beginning, is it easier if ShapeNet dataset is used instead?

Many thanks!

Best,
Ardian

Image 2