lmm077 / Video_Inbetweening

Tensorflow implementation of paper "From Here to There: Video Inbetweening Using Direct 3D Convolutions" https://arxiv.org/pdf/1905.10240.pdf

Video_Inbetweening

Unofficially Tensorflow implementation of From Here to There: Video Inbetweening Using Direct 3D Convolutions (https://arxiv.org/pdf/1905.10240.pdf)

update

Still training...

Requirement

Tensorflow

Model description

T = 16 and D = 128

input: (x_s, x_e), Gaussian noise vector u ∈ R^D

output:(x_s, xˆ1, . . . , xˆT −2, x_e)

0. Download Dataset KTH

./data/KTH/download.sh

1. Train

python3 train_kth_multigpu.py --gpu 0 --batch_size 32 --lr 0.0001

Thanks

Some codes are from here.