Preprint
Jul 2026
HoloCount: A Holistic Visual Counting Benchmark for MLLMs
Through an exhaustive evaluation of over 20 state-of-the-art MLLMs, HoloCount reveals a critical performance gap: even top-tier models degrade significantly as tasks transition from perception to complex analytical reasoning and adverse scenarios.
Jinhong Deng, Limeng Qiao, Guanglu Wan
· 0 citations