Movement prediction is a new field of deep-learning system that can do the very human task of interpreting what is happening in a photo or video and predicting the likely future movement of objects in the image. The system can express the most plausible futures by adding animations to still images, such as cars that would ultimately crash, people who might move in a street, or a train that might roll forward on its tracks. Current generations of computer vision focus on Tracking
past movement or identifying objects in the present. Movement prediction provides a new direction for exploring computer vision by giving machines the ability to understand how objects move in the real world in order to predict future action.