-
Re: Has Human Officiating in the NBA jumped the shark?
 Originally Posted by ralph_i_el
Line judge in tennis is incredibly simple compared to reffing basketball, but I get what you're saying.
I was reading today what detection tech the Waymo robotaxi has: lidar, radar, ultrasonic sensors (like a bat's echolocation), thermal imaging. It costs $110K per car. You can 'see' in every sense- heat, sound, reflection. Using cameras, radar, laser.
In my earlier post, I point out how the NBA is already using computer vision to evaluate human refs and their mistakes.
According to AI, the tech for AI refs is 5-10 years away. The hawkeye tech from tennis is already there.
"These could be automated right now with near-perfect accuracy:
Out-of-bounds calls (computer vision is excellent at line detection)
Goaltending / basket interference
Shot clock & game clock violations
Foot on the line (2 vs 3 points)
Defensive 3-seconds
Backcourt violations
Whether a ball was blocked or tipped
2. What’s close but not perfect yet (~3–5 years)?
These require better body-tracking, collision mapping, and intent estimation:
Off-ball contact (pushes, grabs)
Charge/block decisions where players collide at angles
Travel detection
Illegal screens
AI can track bodies extremely well today, but the NBA standard would require:
millisecond-level joint tracking
reliable force estimation
context understanding (e.g., “was this enough to be a foul?”)
Readiness level: ~70%.
Expected maturity: 3–5 years.
✅ 3. What’s the hardest and furthest away (~5–10+ years)?
Calls requiring intent, degree, and context, like:
Whether contact “affected the play”
Flopping vs legitimate reaction
Whether a swipe was on the ball or on the arm
“Marginal vs illegal” contact
Continuation calls ("and-one" judgment)
Flagrant-hood (reckless vs unnecessary vs excessive)
These require semantic judgment, not just detection.
Today’s AI is improving fast at physical reasoning, but to reach NBA consistency:
needs highly accurate force estimation from visual data
needs deep behavior modeling of players
needs to understand basketball context, not just physics
Readiness level: ~40%.
Expected maturity: 5–10 years for raw technology to be good enough."
Posting Permissions
- You may not post new threads
- You may not post replies
- You may not post attachments
- You may not edit your posts
-
Forum Rules
|