Ball tracking in Padel Videos using Convolutional Neural Networks

The goal of the project is the development and evaluation of a deep-learning application for the reconstruction of the ball trajectory in a padel match, based on a video captured from a camera angle similar to the de-facto standard in padel streaming. The project will explore and evaluate different approaches to detect the ball in individual frames of the video, to discard false negatives, to reconstruct free flight segments, and finally to merge reconstructed flight segments into trajectories.