Identifies that MLLMs fail to perceive visual illusions due to a high-frequency attention bias and provides a plug-and-play fix that boosts accuracy from 13% to 84%.
March 25, 2026
Original Paper
SMSP: A Plug-and-Play Strategy of Multi-Scale Perception for MLLMs to Perceive Visual Illusions
arXiv · 2603.23118
The Takeaway
This reveals a fundamental misalignment between human and model vision that can lead to safety risks. The proposed SMSP framework allows existing MLLMs to 'see' hidden patterns and semantic content they previously ignored, improving their robustness in complex visual environments.
From the abstract
Recent works have shown that Multimodal Large Language Models (MLLMs) are highly vulnerable to hidden-pattern visual illusions, where the hidden content is imperceptible to models but obvious to humans. This deficiency highlights a perceptual misalignment between current MLLMs and humans, and also introduces potential safety concerns. To systematically investigate this failure, we introduce IlluChar, a comprehensive and challenging illusion dataset, and uncover a key underlying mechanism for the