Launch AI Platform Notebook
Copy notebook file
Follow the notebook
Analyzing a Tennis Serve with the Video Intelligence API
Already, machine learning plays a role in sports: companies use it to identify players’ unique talents, detect injuries earlier, and broker optimal player trades. Plus, almost every professional sport uses machine learning technology for tracking. The NBA, for example, has deployed a sophisticated vision-based system on courts to track players’ motions, read numbers off their jerseys, analyze how fast they pass the ball, and determin how accurately they shoot under pressure.
In this lab you will learn how to create the core components of a simple vision-based system to analyze a tennis serve using the Video Intelligence API and a pretrained AutoML Vision Object Detection model.
The Video Intelligence API allows developers to use Google video analysis technology as part of their applications. The REST API enables users to annotate videos stored locally or in Cloud Storage with contextual information at the level of the entire video, per segment, per shot, and per frame. Among its many features, the Video Intelligence API offers Person Detection which detects each person in the video and recognizes a large amount of body parts, facial features, and clothing.
Using Person Detection in this lab, you will be able to calculate the position of a tennis player's wrists, knees, elbows, and shoulders over time from a video clip. Then, with an AutoML Vision Object Detection model that's been trained to recognize tennis balls, you will be able to track the position of the ball in the same video and determine its speed. This collective data can then be analyzed to determine many aspects about the tennis serve such as when the serve happened, how powerful the serve was, and what the tennis player's form during the serve was.
This lab is based on the blog post, Can AI Make You a Better Athlete? Using Machine Learning to Analyze Tennis Serves and Penalty Kicks, by Dale Markowitz.
What you'll do
Analyze a tennis serve with the Video Intelligence API
Use Pose Detection to calculate angle of a person's elbows and arms over time
Use a pretrained AutoML Vision Object Detection model to track the speed of the tennis ball
It is recommended, but not required, to have completed the Video Intelligence: Qwik Start lab in order to have a baseline foundation for how the Video Intelligence API is working.
Crea un account Qwiklabs per leggere il resto del lab e tanto altro ancora.
- Acquisisci accesso temporaneo a Google Cloud Console.
- Oltre 200 lab dal livello iniziale a quelli più avanzati.
- Corsi brevi per apprendere secondo i tuoi ritmi.