Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs
This addresses the problem of unreliable visual grounding in multimodal AI systems for users relying on accurate image understanding, though it is incremental as it builds on existing CLIP-based methods.
The paper reveals systematic visual shortcomings in multimodal LLMs (MLLMs) by identifying 'CLIP-blind pairs' where CLIP perceives visually different images as similar, and constructs the MMVP benchmark showing state-of-the-art systems like GPT-4V struggle with basic visual patterns, often providing incorrect answers and hallucinations. It proposes a Mixture of Features (MoF) approach that integrates vision self-supervised learning features with MLLMs, significantly enhancing visual grounding capabilities.
Is vision good enough for language? Recent advancements in multimodal models primarily stem from the powerful reasoning abilities of large language models (LLMs). However, the visual component typically depends only on the instance-level contrastive language-image pre-training (CLIP). Our research reveals that the visual capabilities in recent multimodal LLMs (MLLMs) still exhibit systematic shortcomings. To understand the roots of these errors, we explore the gap between the visual embedding space of CLIP and vision-only self-supervised learning. We identify ''CLIP-blind pairs'' - images that CLIP perceives as similar despite their clear visual differences. With these pairs, we construct the Multimodal Visual Patterns (MMVP) benchmark. MMVP exposes areas where state-of-the-art systems, including GPT-4V, struggle with straightforward questions across nine basic visual patterns, often providing incorrect answers and hallucinated explanations. We further evaluate various CLIP-based vision-and-language models and found a notable correlation between visual patterns that challenge CLIP models and those problematic for multimodal LLMs. As an initial effort to address these issues, we propose a Mixture of Features (MoF) approach, demonstrating that integrating vision self-supervised learning features with MLLMs can significantly enhance their visual grounding capabilities. Together, our research suggests visual representation learning remains an open challenge, and accurate visual grounding is crucial for future successful multimodal systems.