Analyzing the postures of athletes in training videos is a time-consuming and repetitive task, often requiring biomechanical engineers to manually sift through hours of footage.
We present an innovative application of computer vision and artificial intelligence that sped this task up by a factor of 10 for canoe athletes preparing for the Olympic Games.
We show in detail the bottom-up approach of applying computer vision and AI on real world problems and give an in-depth view on pitfalls and lessons learned.
One of the main lessons being that using AI still requires a deep understanding of the domain and the data at hand.
From the talk you will get a better understanding of Computer Vision techniques, how to apply them to real-world challenges, and how to avoid common pitfalls.