search

gabeur / mmt

Multi-Modal Transformer for Video Retrieval
http://thoth.inrialpes.fr/research/MMT/
Apache License 2.0
259 stars 41 forks source link

H5 files with video features #10

Closed qjyyyy closed 3 years ago

qjyyyy commented 3 years ago

Thank you for your generous sharing. I want to know the difference between ‘features.audio’ and 'features_t.audio' in the H5 file.

qjyyyy commented 3 years ago

I see that all H5 files have the same 'features_t.audio', so I want to know what it stands for. and others,like 'features_t.face','features_t.flow',etc

gabeur commented 3 years ago

Thanks for your interest in our work. features_t.flow are used to encode the time in the video [start_time, end_time] from which the features were extracted. Similarly for the other modalities. But we actually use the get_feature_timings function to calculate those times, so please disregard the features_t data.