PyThia performs video prediction for simple physical systems: given the first few frames in a video, it predicts the images that will appear in the future.
The easiest way to see what PyThia can do is to peruse the demonstration notebook. It is available as an interactive notebook served via Google Colab.
As you will see in that notebook, the PyThia models are trained on sequences of images depicting a round ball bouncing around a closed box.
A trained PyThia model is then capable of predicting the future trajectory. The animation below shows the first half of a video, from which the model predicts the last frame of the video, and then compares the model's prediction with the truth.
The objective of this project is to predict what will happen in the future, more specifically predicting what will happen in the future of a video whose initial frames are given as input. This project is therefore named after a famous oracle purporting to divine the future: in Ancient Greece the Pythia served as oracle at the temple of Apollo, in Delphi.