1 paper · 1 filter
Yuan Wang, Shujian Gao, Songtao Jiang +2
Existing medical video benchmarks primarily evaluate whether a model produces the correct answer, but rarely assess whether it answers at the right time. In real clinical settings,…