{"id":15223,"date":"2024-06-27T13:00:20","date_gmt":"2024-06-27T13:00:20","guid":{"rendered":"http:\/\/TheNextWeb=1407862"},"modified":"2024-06-27T13:00:20","modified_gmt":"2024-06-27T13:00:20","slug":"can-ai-cure-the-corruption-and-biases-that-plague-boxing","status":"publish","type":"post","link":"https:\/\/www.londonchiropracter.com\/?p=15223","title":{"rendered":"Can AI cure the corruption and biases that plague boxing?"},"content":{"rendered":"\n

Boxing is a whimsically old-fashioned sport. Feuds are settled in fights, traditions are revered, and ageing faces run the show. <\/span><\/p>\n

Fifty years since their heydays promoting Muhammad Ali, the 92-year-olds Don King and Bob Arum remain leading powerbrokers. They work with overlords who operate in the shadows. In this world, oversight is resisted and new tech greeted with suspicion. <\/span><\/p>\n

At world title fights, judges still fill out scorecards on scraps of paper. They follow four extremely subjective criteria: \u201ceffective aggression,\u201d \u201cring generalship,\u201d \u201cclean punches,\u201d and defence. All these concepts are open to interpretation. Inevitably, they frequently create controversial decisions.<\/span><\/p>\n

The problem extends to the sport\u2019s top analytics tool. Despite the futuristic sheen of the CompuBox<\/a> name, its stats are manually created by two people tapping keys when they see punches. The potential for biases and errors is endless. <\/span><\/p>\n

\n
<\/p>\n

TNW Conference 2025 – Back to NDSM on June 19-20, 2025 – Save the date!<\/p>\n

As we wrapped up our incredible 2024 edition, we’re pleased to announce our return to Amsterdam NDSM in 2025. Registration now!<\/p>\n

<\/a><\/div>\n<\/div>\n

Fans and fighters alike have decried the results for decades. One of them is Allan Svejstrup, a Danish machine learning<\/a> engineer. But Svejstrup (<\/span>pronounced Svar-strop) also had an idea for a solution: computer vision. <\/span><\/p>\n

A boxing aficionado with a PhD in maths, Svejstrup had his brainwave while working at a startup in China. \u201cI also had the CompuBox doubts,\u201d he tells TNW on a Zoom call from the Chinese city of Shenzhen. \u201cI realised I could build a better system myself.\u201d<\/span><\/p>\n

He quickly devised a plan to train an AI<\/a> model on real boxing footage. The model would track not only the number of punches but also their impact, the fight\u2019s flow, and each boxer\u2019s movements. <\/span><\/p>\n

Svejstrup was impressed by the early tests. In 2022, he turned the passion project into a startup<\/a>: <\/span>Jabbr<\/span><\/a>.<\/span><\/p>\n

Two years later, his brainchild had evolved into a fully-grown product. In the new boxing mecca of Saudi Arabia, Jabbr would analyse a fight for the biggest prize of them all: the undisputed heavyweight championship of the world.<\/span><\/p>\n

But before arriving on that grand stage, the company had fights to win outside the ring. <\/span><\/p>\n

Hard knocks<\/h2>\n

Boxing presents big barriers to AI analysis. One emerges when evaluating punches and ringcraft, which are far harder to measure than, say, baseball pitches or golf shots. <\/span><\/p>\n

\u201cI understood why nobody else had done it successfully,\u201d Svejstrup says. \u201cThe standard off-the-shelf tools that machine learning engineers use for soccer or basketball don\u2019t work very well for combat sports.\u201d<\/span><\/p>\n

Another problem is the visual phenomenon of occlusion. In computer vision, occlusion occurs when one object conceals the view of another. It\u2019s rarely an issue when evaluating a tennis shot or home workout (as I <\/span>know from personal experience<\/span><\/a>). These applications can often rely on off-the-shelf products that are fine-tuned on sports data. For boxing, however, they\u2019re ill-fitting tools. <\/span><\/p>\n

Occlusion is particularly troublesome when fighters get close and trade body shots. From one angle, a powerful body shot could be hidden and misclassified as missed. Jabbr\u2019s solution combines multiple cameras with a novel approach to visualisation. Svejstrup named the system DeepStrike.<\/span><\/p>\n