A Comprehensive Review on the Prediction of Video Frame and Motion Contents

The capability of predicting, or analyzing the future outcomes of motion pictures using intelligent systems is the video frame prediction process. Video frame prediction is a significant and challenging aspect of Computer Vision (CV) applications. Video prediction plays a vital role in terms of robotics and Artificial Intelligence (AI) applications in detecting the future frames of a video, depending on which, the decision-making process can be carried out. The existing video frame prediction processes include pixel law based and motion-based predictions, which have been proven to be ambiguous. This leads to inaccurate predictions and indirectly deteriorates the performance of the real time applications. This paper performs a detailed review of the existing methodologies, their merits, pitfalls and challenges that create havoc in the video frame prediction process. In addition, this paper performs a detailed analysis on the existing datasets that assist in the video prediction process. This review summarizes various methodologies and datasets along with their performance analysis, which will open research possibilities in the field of video frame prediction.