S2VT_keras this is a keras version of paper S2VT: Sequence to Sequence: Video to Text the feature extraction code is from https://github.com/KangSooHan/S2VT_tensorflow, thanks him:) already tested, can work.