大型視覺語言模型如何識別圖像中的文字？揭示OCR頭部的獨特作用

How Do Large Vision-Language Models See Text in Image? Unveiling the Distinctive Role of OCR Heads

May 21, 2025

作者: Ingeol Baek, Hwan Chang, Sunghyun Ryu, Hwanhee Lee

cs.AI

摘要

儘管大型視覺語言模型（LVLMs）已取得顯著進展，但在其可解釋性以及如何定位和解釋圖像中的文本信息方面仍存在差距。本文探討了多種LVLMs，以識別負責從圖像中識別文本的特定頭部，我們稱之為光學字符識別頭（OCR Head）。關於這些頭部的發現如下：（1）稀疏性較低：與先前的檢索頭不同，大量頭部被激活以從圖像中提取文本信息。（2）質性差異：OCR頭部具有與一般檢索頭部顯著不同的特性，其特徵相似度較低。（3）靜態激活：這些頭部的激活頻率與其OCR分數高度一致。我們在下游任務中驗證了這些發現，通過將思維鏈（CoT）應用於OCR和傳統檢索頭部，並對這些頭部進行遮罩。我們還展示了在OCR頭部內重新分配匯聚標記值可以提升性能。這些見解提供了對LVLMs處理圖像中嵌入文本信息的內部機制的深入理解。

English

Despite significant advancements in Large Vision Language Models (LVLMs), a gap remains, particularly regarding their interpretability and how they locate and interpret textual information within images. In this paper, we explore various LVLMs to identify the specific heads responsible for recognizing text from images, which we term the Optical Character Recognition Head (OCR Head). Our findings regarding these heads are as follows: (1) Less Sparse: Unlike previous retrieval heads, a large number of heads are activated to extract textual information from images. (2) Qualitatively Distinct: OCR heads possess properties that differ significantly from general retrieval heads, exhibiting low similarity in their characteristics. (3) Statically Activated: The frequency of activation for these heads closely aligns with their OCR scores. We validate our findings in downstream tasks by applying Chain-of-Thought (CoT) to both OCR and conventional retrieval heads and by masking these heads. We also demonstrate that redistributing sink-token values within the OCR heads improves performance. These insights provide a deeper understanding of the internal mechanisms LVLMs employ in processing embedded textual information in images.