This project is no longer accepting applications. Subscribe to our newsletter to be notified of new projects!
Develop machine learning models to classify player actions from tracking data and automatically generate game highlights.
In the fast-paced world of sports streaming, automated highlight generation is becoming increasingly crucial for engaging fans and providing quick insights. In this Build Project, you'll step into the role of a Machine Learning Engineer to develop an end-to-end system for automated sports action recognition and highlight generation. You will develop a fully functional automated sports highlight generation system, including a trained action classification model, a video processing pipeline, and a highlight generation algorithm.
Under the guidance of an experienced industry expert, you'll build machine learning models to classify player actions from tracking data, create a video processing pipeline, and implement highlight generation algorithms. You'll gain hands-on experience with cutting-edge technologies like computer vision, deep learning, and large language models, simulating the workflow of a real data science team in the sports tech industry.
Introduction to the project goals and timeline. You'll set up your development environment, explore the player tracking dataset, and create initial visualizations to gain insights.
Learn techniques for data preprocessing and feature engineering specific to sports tracking data. You'll prepare the dataset for model training and identify key features for action classification.
Experiment with different machine learning models for action classification. You'll train models, evaluate their performance, and select the best one for your system.
Optimize your chosen model and prepare it for deployment. You'll also learn best practices for documenting your model development process.
Introduction to video processing using Python libraries. You'll implement basic video manipulation tasks crucial for highlight generation.
Develop a pipeline that integrates your action classification model with video processing to generate highlights automatically. Begin refining the system to improve highlight quality.
Continue refining your highlight generation system, focusing on improving the quality and relevance of the generated highlights. Implement techniques to enhance the storytelling aspect, potentially using large language models for summarization.
Prepare and deliver your final presentation, showcasing your automated highlight generation system. You'll demonstrate the key features, discuss challenges overcome, and reflect on your learning experience throughout the project.
Get access to all of our Build projects, including this one, by creating your Build account!
Get started by submitting your application.
We'll notify you when projects reopen. In the meantime, you can explore our resources and learn more about our Fellows.
Paul Kefer is a machine learning engineer specializing in autonomous sports broadcasting. He leverages his expertise in artificial intelligence and computer vision to develop cutting-edge systems that revolutionize how sports events are captured and broadcast. In his free time, Paul is an enthusiast of immersive technologies, exploring the realms of augmented and virtual reality. He seeks thrills through indoor skydiving and can often be found cruising along the beach on his electric longboard, enjoying the perfect blend of technology and outdoor adventure.