Benchmark RadarAI BENCHMARK PROFILE
BasketballBench
Evaluates basketball understanding through 7,980 questions over text, image, and video, covering ten tasks from event recognition to structured game knowledge.
- Released
- 2026-08-24
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Measures integrated multimodal reasoning in a domain where separate capabilities must work together, revealing gaps in current MLLMs.
Motivation
Understanding a basketball game requires recognizing events, localizing actions, identifying players, and relating these to structured game knowledge.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.