Machine learning and computer vision are being combined to provide commentary on professional sporting events as they actually occur.
For example, Indian researchers demonstrated that weakly supervised computers could reliably differentiate between what is happening during videos of cricket matches and then provide text-based commentary.
Their research required analyzing many hours of cricket videos, placing them in categories based on already available text descriptions, breaking down the longer videos into smaller scenes to classify each video shot, and then using an algorithm to find matching commentary. Algorithms were subsequently able to accurately label a batsman's cricketing shot using visual-recognition methods for an action that sometimes lasted less than two seconds.
The same research team also analyzed how a computer might be able to deconstruct the action of a tennis match, and presented a study on how machine-learning algorithms could be used to provide text commentary on tennis tournaments.
Machines' ability to process views of the action from many angles simultaneously also could enhance the assessment of game plays, although for now the technology is intended to be used for training and coaching.
The researchers note the video annotation enables computers to search across hundreds of hours of content for specific actions that last only seconds.
From The Washington Post
View Full Article - May Require Free Registration
Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA
No entries found