A new benchmark shows multimodal AI models consistently underperform humans at perceiving occluded objects, and the gap is not closed by model scaling or extended reasoning.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond the Visible: Benchmarking Occlusion Perception in Multimodal Large Language Models
A new benchmark shows multimodal AI models consistently underperform humans at perceiving occluded objects, and the gap is not closed by model scaling or extended reasoning.