Preprocessing read_mp4.py: Capture 20 frame from .mp4 videos and save them in .png formaat Network for vector of size 7 video_recognition_lstm.py: One layer LSTM with 7 hidden units. Get emotion vector of size 7 as input and output one emotion per 20 steps