The advent of Vision Language Models (VLM) has allowed researchers to investigate the visual understanding of a neural network using natural language. Beyond object classification and detection, VLMs are capable of visual comprehension and common-sense reasoning. This naturally led to the question: How do VLMs respond when the image itself is inherently unreasonable? To this end, we present IllusionVQA: a diverse dataset of challenging optical illusions and hard-to-interpret scenes to test the capability of VLMs in two distinct multiple-choice VQA tasks - comprehension and soft localization. GPT4V, the best-performing VLM, achieves 62.99% accuracy (4-shot) on the comprehension task and 49.7% on the localization task (4-shot and Chain-of-Thought). Human evaluation reveals that humans achieve 91.03% and 100% accuracy in comprehension and localization. We discover that In-Context Learning (ICL) and Chain-of-Thought reasoning substantially degrade the performance of GeminiPro on the localization task. Tangentially, we discover a potential weakness in the ICL capabilities of VLMs: they fail to locate optical illusions even when the correct answer is in the context window as a few-shot example.

翻译：视觉语言模型（VLM）的出现使研究者能够通过自然语言探究神经网络对图像的理解。除目标分类与检测外，VLM已具备视觉理解与常识推理能力。这自然引出一个问题：当图像本身具有内在不合理性时，VLM将如何应对？为此，我们提出IllusionVQA：一个包含挑战性光学错觉与难以解读场景的多样化数据集，用于测试VLM在两项多选视觉问答（VQA）任务——理解与软定位——中的能力。表现最佳的VLM模型GPT4V在理解任务中达到62.99%（4-shot准确率），在定位任务中达到49.7%（4-shot与思维链）。人工评估显示，人类在理解与定位任务中分别达到91.03%与100%准确率。我们发现，上下文学习（ICL）与思维链推理会显著降低GeminiPro在定位任务中的表现。此外，我们意外发现VLM在ICL能力上的潜在弱点：即使正确答案已作为少样本示例存在于上下文窗口中，它们仍无法定位光学错觉目标。