@inproceedings{zhang-etal-2026-revico,
title = "{R}e{V}i{C}o: Unveiling the Limitations of {VLM}s in Visual Text Understanding via Error Correction",
author = "Zhang, Bojun and
Liang, Junhong and
Zhai, Feifei and
Ji, Fengxian and
Zhou, Yu",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.845/",
pages = "19606--19622",
ISBN = "979-8-89176-459-0",
abstract = "Vision Language Models (VLMs) have shown great success in general visual tasks, yet they still struggle to deeply understand text within images. In this paper, we introduce **ReViCo** (**Re**al **Vi**sual **Co**rrection), a benchmark designed to evaluate VLM text understanding through a novel task of visual text error correction. ReViCo challenges models with identifying and fixing text errors in real-world images, which requires a deep understanding on visual text. We benchmark various VLMs using two distinct paradigms: prompt-based strategy and targeted model training, both aimed at pushing the limits of current models. Our experiments reveal a striking performance gap between even the best VLMs and human, and further analysis also shows that most models struggle to accurately perceive the visual text, resulting in frequent correction errors. By highlighting these gaps, ReViCo provides a new benchmark foundation for developing more robust and text-aware VLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-revico">
<titleInfo>
<title>ReViCo: Unveiling the Limitations of VLMs in Visual Text Understanding via Error Correction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bojun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junhong</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Feifei</namePart>
<namePart type="family">Zhai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fengxian</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Vision Language Models (VLMs) have shown great success in general visual tasks, yet they still struggle to deeply understand text within images. In this paper, we introduce **ReViCo** (**Re**al **Vi**sual **Co**rrection), a benchmark designed to evaluate VLM text understanding through a novel task of visual text error correction. ReViCo challenges models with identifying and fixing text errors in real-world images, which requires a deep understanding on visual text. We benchmark various VLMs using two distinct paradigms: prompt-based strategy and targeted model training, both aimed at pushing the limits of current models. Our experiments reveal a striking performance gap between even the best VLMs and human, and further analysis also shows that most models struggle to accurately perceive the visual text, resulting in frequent correction errors. By highlighting these gaps, ReViCo provides a new benchmark foundation for developing more robust and text-aware VLMs.</abstract>
<identifier type="citekey">zhang-etal-2026-revico</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.845/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19606</start>
<end>19622</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ReViCo: Unveiling the Limitations of VLMs in Visual Text Understanding via Error Correction
%A Zhang, Bojun
%A Liang, Junhong
%A Zhai, Feifei
%A Ji, Fengxian
%A Zhou, Yu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F zhang-etal-2026-revico
%X Vision Language Models (VLMs) have shown great success in general visual tasks, yet they still struggle to deeply understand text within images. In this paper, we introduce **ReViCo** (**Re**al **Vi**sual **Co**rrection), a benchmark designed to evaluate VLM text understanding through a novel task of visual text error correction. ReViCo challenges models with identifying and fixing text errors in real-world images, which requires a deep understanding on visual text. We benchmark various VLMs using two distinct paradigms: prompt-based strategy and targeted model training, both aimed at pushing the limits of current models. Our experiments reveal a striking performance gap between even the best VLMs and human, and further analysis also shows that most models struggle to accurately perceive the visual text, resulting in frequent correction errors. By highlighting these gaps, ReViCo provides a new benchmark foundation for developing more robust and text-aware VLMs.
%U https://aclanthology.org/2026.emnlp-main.845/
%P 19606-19622
Markdown (Informal)
[ReViCo: Unveiling the Limitations of VLMs in Visual Text Understanding via Error Correction](https://aclanthology.org/2026.emnlp-main.845/) (Zhang et al., EMNLP 2026)
ACL