|
|
NFL Scouts, Coaches Go Deep With AI for Evaluating Draft Prospects
|
|
|
|
|
|
Welcome back. Computer vision and machine learning are set to upend the NFL draft, where scouts and coaches still rely on physical tryouts that are likely to become problematic in the age of Covid-19.
Also: The latest on the coronavirus pandemic and AI.
|
|
|
|
|
Scouts still time players with hand-held stopwatches during workouts. PHOTO: JULIO CORTEZ/ASSOCIATED PRESS
|
|
|
The old college try. NFL scouts and coaches have a poor record of evaluating draft prospects, an analog process that relies on hand-held stopwatches, 40-yard dashes, and college-game press clippings, much as it did a half-century ago. Slants, a software startup, is developing digital tools that can apply computer vision to footage from college games and track players’ movements on the field, The Wall Street Journal reports.
The data is then processed with machine learning models to provide real-world metrics to evaluate players.
Better stats. The startup’s software offers the kind of game-day precision NFL teams have about their current players, offering an apples-to-apples comparison. For instance, it can determine the top speed of LSU wide receiver Ja’Marr Chase by pulling up video of the 2020 national, when he caught a 52-yard touchdown pass from Joe Burrow, while mapping his exact route on the play.
[continues below]
|
|
|
|
|
Ja'Marr Chase of the LSU Tigers catches the ball for a 52-yard touchdown against the Clemson Tigers in the National Championship game. PHOTO: JONATHAN BACHMAN/GETTY IMAGES
|
|
|
Remote tryouts. Currently in beta, the technology is likely to catch the league’s eye as the coronavirus outbreak persists, making teams unable to put prospects through their paces in person and limited to watching footage of private workouts and video-conference interviews.
Accelerating change. While data has revolutionized sports, the college draft has remained stuck in the past. Startups like Slants are developing technology that can give NFL teams more advanced information on draft picks than they’ve ever had before.
|
|
|
|
A person was loaded into an ambulance last month at the Life Care Center in Kirkland, Wash. Tracking based on 911 calls could help spot clusters of cases at nursing homes. PHOTO: TED S. WARREN/ASSOCIATED PRESS
|
|
|
Coronavirus pulls 911 data into tracking tool. Seattle’s fire department is testing an artificial intelligence tool from Copenhagen-based Corti that collects data on 911 callers’ reported symptoms as well as their breathing patterns and coughs, The Wall Street Journal’s Sarah Krouse reports. The city uses that information to map locations with large numbers of suspected infections, which in turn help determine where to dispatch first responders to encourage more social-distancing measures.
The system has alerted the city to nursing homes and assisted-living facilities with clusters of cases, and the department has dispatched officers to ensure that staff are maintaining spacing and wearing proper protective gear. “I like to visually see it in a clean shot to understand how big the problem is,” said Fire Chief Harold Scoggins.
|
|
|
|
AI used to monitor coronavirus patient. A clinical team in Boston used an AI tool to remotely monitor breathing, movement and sleep patterns for a Covid-19 patient, VentureBeat reported. The wireless tool, known as Emerald, was developed by MIT’s Computer Science and Artificial Intelligence Laboratory. It's been deployed in nearly 200 hospitals, homes, and assistive-care facilities over the past 18 months, but this marks the first time it’s been used to track the progression of a patient with the novel coronavirus.
|
|
|
AI startups raise $6.9 billion in the first quarter. U.S. artificial-intelligence startups raised $6.9 billion in the first three months of this year, according to VentureBeat. It based its report on data from the National Venture Capital Association. However, the coronavirus pandemic will likely damp funding in all sectors for the remainder of the year, according to the report.
|
|
|
|
Skip the AI proof-of-concept stage. Companies that deployed AI projects without conducting a proof-of-concept phase achieved returns on investment that were three times greater than companies that did, Accenture analysts wrote in Harvard Business Review. Based on a 2019 survey of 1,500 C-suite executives, the analysts said the companies that abandoned proof-of-concepts and went straight to scaling succeeded in their initiatives twice as often and completed scaling projects more quickly.
Piloting as an alternative. One of the drawbacks to conducting proof of concepts for AI projects is that companies often don’t consider issues such as model risk, data bias and privacy until it’s time for production, according to the report. One substitute is to embrace piloting, which takes a fully baked tool and launches it directly into the real world on a small scale. That allows the company to accurately see how the new technology will be received by customers, the article said.
|
|
|
Governors asserted their authority to reopen their states’ economies, in a counter to the president’s argument that he alone has the power to end the lockdown. (WSJ)
JPMorgan Chase and Wells Fargo set aside billions of dollars to get ready for a flood of customers to default on their loans as the pandemic pummels the economy. (WSJ)
|
|
|
|