In collaboration with an industry partner, we have access to a number of recorded sports events performed with multiple cameras, providing a 360 degree view of the match. The initial project is centered around detecting and tracking the ball as it moves around. Once the tracking is working as expected, we would like to apply machine learning to predict the ball movement. As we have the actual ball movements, we can use the next frames to validate during training. The solution should eventually be able to run in real-time, being able to predict ball movement based on previous ball direction and cues from the frames.