Deep-Physics is a compact TensorFlow model trained to recreate intuitive physics. Deep-Physics allows for a simple generation of dynamic environments without the use of any classical physics formulas.
Deep-Physics can be applied to any environment. We chose to provide users with a simple javascript implementation using TensorFlow.js and P5.js. Simply clone the repository and run the index.html.
We have included our unpublished research paper which explains the theory behind our research.
Research project made by Alexandre Lavoie and James Kary from Vanier College, QC, Canada. We want to thank Dr. Ivan Ivanov for his guidance.