How to prepare training data ？

[MagicRedZero]

Data Preparation

Place your vertices data (.npy files) and audio data (.wav files) in /vertices_npy and /wav folders, respectively.

Save the templates of all subjects to a templates.pkl file and put it in , as done for BIWI and vocaset dataset. Export an arbitary template to .ply format and put it in /.

Ask about Data Preparation of Play with Your Own Data .npy templates.pkl .ply How are these documents prepared?

[MMMDY]

the same question! Have you solved it?

[Doubiiu]

Hi, Since I didn't try it on the customized dataset either, the best way I think is to inspect the corresponding files of vocaset and BIWI: for .npy, it may be in <frame_num $\times$ vertex_num $\times$ 3>, and the value are just 3d positions of vertices. The templates.pkl is the dictionary of neutral face (3d position of vertices), e.g. {'a': <vertex_num $\times$ 3>, 'b': ....}, and .ply is the topology for rendering purpose, the index of vertices to form a triangle facet, e.g. [[1, 2, 3], [2,3,7], ....].

[DarlingZzzz]

Hi, Since I didn't try it on the customized dataset either, the best way I think is to inspect the corresponding files of vocaset and BIWI: for .npy, it may be in <frame_num × vertex_num × 3>, and the value are just 3d positions of vertices. The templates.pkl is the dictionary of neutral face (3d position of vertices), e.g. {'a': <vertex_num × 3>, 'b': ....}, and .ply is the topology for rendering purpose, the index of vertices to form a triangle facet, e.g. [[1, 2, 3], [2,3,7], ....].

Hi, I don't know how to get the BIWI data set, can you tell me how you got it