Recent advancements in Chain-of-Thought (CoT) and related rationale-based
works have significantly improved the performance of Large Language Models
(LLMs) in complex reasoning tasks. With the evolution of Multimodal Large
Language Models (MLLMs), enhancing their capability to tackle complex
multimodal reasoning problems is a crucial frontier. However, incorporating
multimodal rationales in CoT has yet to be thoroughly investigated. We propose
the Image-of-Thought (IoT) prompting method, which helps MLLMs to extract
visual rationales step-by-step. Specifically, IoT prompting can automatically
design critical visual information extraction operations based on the input
images and questions. Each step of visual information refinement identifies
specific visual rationales that support answers to complex visual reasoning
questions. Beyond the textual CoT, IoT simultaneously utilizes visual and
textual rationales to help MLLMs understand complex multimodal information. IoT
prompting has improved zero-shot visual reasoning performance across various
visual understanding tasks in different MLLMs. Moreover, the step-by-step
visual feature explanations generated by IoT prompting elucidate the visual
reasoning process, aiding in analyzing the cognitive processes of large
multimodal models

近期在 Chain-of-Thought (CoT) 和相关基于理性的研究中，显着提升了大规模语言模型（LLM）在复杂推理任务中的性能。随着多模态大规模语言模型（MLLM）的发展，增强其解决复杂多模态推理问题的能力是至关重要的。然而，将多模态理性融入 CoT 还没有得到彻底的研究。本文提出了 Image-of-Thought（IoT）提示方法，帮助 MLLM 逐步提取视觉理性。具体而言，IoT 提示可以根据输入图像和问题自动设计关键的视觉信息提取操作。视觉信息的每个步骤都可以确定支持复杂视觉推理问题答案的特定视觉理性。除了文本 CoT，IoT 同时利用视觉和文本合理有助于 MLLM 理解复杂多模态信息。IoT 提示在不同的 MLLM 中改善了各种视觉理解任务的零 - shot 视觉推理性能。此外，IoT 提示生成的逐步视觉特征解释阐明了视觉推理过程，有助于分析大规模多模态模型的认知过程。