Deep Learning requires tremendous amounts of data, while acquiring real-world data is tedious. A state-of-the-art approach is to augment data that aids in training AI. This project investigates the opportunity to generate improved data using videos.
The goal is to extract IMU data from videos using Deep Learning. Real-world data comprising IMU data and videos from animals such as horses, goats, and sheep can be used to evaluate the approach.
20% Theory, 60% Simulations, 20%Writing
Viet Duc Le, email@example.com, room ZI