Movie Frame Prediction Using Convolutional Long Short Term Memory
Prathmesh Sambrekar, Satyadhyan Chickerur · 2019
Video frame prediction involving predicting the future frames given some input frames has been in the limelight in the recent years. However, robust models have not been developed and tested which could handle the videos involving sequence of frames having characters and objects having high temporal variations. And more significantly there are no models which embrace the sequence of frames which are extracted from movies (Example: Cartoon movies).The focus of this paper is to explore a model which could learn the spatio-temporal consistency amongst the frames which are sampled from a cartoon movie. To this end, we propose a Convolutional Long Short Term Memory based architecture wherein the model harnesses the ability of conventional Long Short Term Memory (LSTM) and convolutional neural network (CNN) to learn the time series and spatial features of the given video frame sequences.