By Charles Carter, 20/10/22
Machine learning engineers at Danish startup Jabbr have developed an AI called DeepStrike that automatically analyses the performance of boxing opponents using camera feeds.
The innovation could be used to replace boxing judges and provide athletes with accurate stats on their sparring training sessions to help them identify areas for improvement.
How does it work?
DeepStrike uses deep learning to measure 50 metrics including punch types, punches landed, quality, footwork, aggression, pressure, and more.
It can produce metrics from footage taken on 1, 2 or 3 amateur smartphones and professionally recorded footage.
As well as performance stats, it automatically edits and mixes the footage to produce round highlights with zoom, view-shifts and tracking.
The company is currently carrying out beta testing with potential customers who can register on their website to get access to the application.
What are the potential benefits?
Boxing scoring can be prone to human error as well as deliberate manipulation by corrupt or biased judges.
Corruption is a significant problem in boxing. A 2021 independent investigation into the 2016 Rio Olympics boxing tournament found evidence of bribes, favours and judges throwing matches.
DeepStrike could help to eradicate corruption and human error from judging, ensuring fair results.
Questions for you. Comment below
- First thought that comes into your head?
- Pros and cons according to you?
- Other applications of this approach?
- What could this be combined with?
Links
https://www.theguardian.com/sport/2021/sep/30/judges-signals-fix-olympic-boxing-bouts-mclaren-report
A french startup ST37 (https://www.st37.fr/) is developping similar approach baed on AI; they started on fencing but expand the application beyond.