FPS for RGB, Flow and Object Features

fpv-iplab / rulstm

Code for the Paper: Antonino Furnari and Giovanni Maria Farinella. What Would You Expect? Anticipating Egocentric Actions with Rolling-Unrolling LSTMs and Modality Attention. International Conference on Computer Vision, 2019.

http://iplab.dmi.unict.it/rulstm

132 stars 33 forks source link

FPS for RGB, Flow and Object Features #6

Closed guanglun95 closed 4 years ago

guanglun95 commented 4 years ago

Thank you very much for your excellent work. I wonder whether the rgb, flow and object features come from the same frames sampled at the fps of 4 (as you mentioned in readme)?

antoninofurnari commented 4 years ago

We are providing two sets of pre-extracted features, which can be downloaded with two scripts:

download_data.sh: downloads the "small" set, which includes features for frames subsampled at 4fps;
download_data_full.sh: downloads the "large" set, which includes features for all videos at 30fps (the videos have been previously re-encoded at 30fps)