spatial_temporal_view_decomposition not used in evaluate_a_set_of_videos.py

[turian]

Why isn't spatial_temporal_view_decomposition used in evaluate_a_set_of_videos.py?

Honestly, you should have an evaluate.py module so you don't duplicate so much code. It also makes it easier for others to build predictors, like my repo where I have to disentangle gnarly differences between the two implementations.

[vdlad]

This is terribly late but perhaps helps others,
ViewDecompositionDataset class which internally calls spatial_temporal_view_decomposition through its __getitem__ method