Benchmark RadarAI BENCHMARK PROFILE
SportMV-Bench
SportMV-Bench evaluates multi-view sports video understanding with 1022 multi-view bundles and 3015 QA pairs across 10 sports, covering perception, rule-aware event interpretation, and adjudicative reasoning.
- Released
- 2026-07-13
- Readiness
- Paper only
- Primary field
- Cybersecurity
Why it matters
This benchmark addresses the lack of multi-view sports video evaluation, revealing that MLLMs struggle with fine-grained visual perception and view selection, guiding future development in video reasoning.
Motivation
Recent Multimodal Large Language Models (MLLMs) achieve strong performance on single-view video understanding benchmarks.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.