@inproceedings{shabtay-etal-2026-look,
title = "Look Where It Matters: High-Resolution Crops Retrieval for Efficient {VLM}s",
author = "Shabtay, Nimrod and
Kimhi, Moshe and
Harary, Sivan and
Spector, Artem and
Rivlin, Ehud and
Baskin, Chaim and
Giryes, Raja and
Schwartz, Eli",
editor = "T.Y.S.S., Santosh and
Agarwal, Amit and
Gupta, Vivek and
Kementchedjhieva, Yova and
Garimella, Aparna and
Hou, Yufang",
booktitle = "The Proceedings of Workshop on Document Intelligence and Understanding",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.docinsights-1.25/",
pages = "357--384",
ISBN = "979-8-89176-477-4",
abstract = "Vision-language models (VLMs) typically process images at a native high-resolution, forcing a trade-off between accuracy and computational efficiency: high-resolution inputs capture fine details but incur significant computational costs, while low-resolution inputs advocate for efficiency, they potentially miss critical visual information, specifically OCR related. We present AwaRes, a spatial-on-demand framework that resolves this accuracy{--}efficiency trade-off by operating on a low-resolution global view and using tool-calling to retrieve only high-resolution segment needed for a given query, rather than escalating to the full image. We construct this supervision automatically: a judge compares low- vs. high-resolution answers to label whether cropping is needed, and an oracle grounding model localizes the evidence for the correct answer, which we map to a discrete crop set to form multi-turn tool-use trajectories. We train our framework with SFT followed by multi-turn GRPO with a composite reward that combines semantic answer correctness with explicit crop-cost penalties. AwaRes provides a practical, deployment-friendly path to high-detail VLM reasoning under tight compute budgets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shabtay-etal-2026-look">
<titleInfo>
<title>Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nimrod</namePart>
<namePart type="family">Shabtay</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Moshe</namePart>
<namePart type="family">Kimhi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sivan</namePart>
<namePart type="family">Harary</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Artem</namePart>
<namePart type="family">Spector</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ehud</namePart>
<namePart type="family">Rivlin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chaim</namePart>
<namePart type="family">Baskin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Raja</namePart>
<namePart type="family">Giryes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eli</namePart>
<namePart type="family">Schwartz</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>The Proceedings of Workshop on Document Intelligence and Understanding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Santosh</namePart>
<namePart type="family">T.Y.S.S.</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Amit</namePart>
<namePart type="family">Agarwal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yova</namePart>
<namePart type="family">Kementchedjhieva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aparna</namePart>
<namePart type="family">Garimella</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yufang</namePart>
<namePart type="family">Hou</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">979-8-89176-477-4</identifier>
</relatedItem>
<abstract>Vision-language models (VLMs) typically process images at a native high-resolution, forcing a trade-off between accuracy and computational efficiency: high-resolution inputs capture fine details but incur significant computational costs, while low-resolution inputs advocate for efficiency, they potentially miss critical visual information, specifically OCR related. We present AwaRes, a spatial-on-demand framework that resolves this accuracy–efficiency trade-off by operating on a low-resolution global view and using tool-calling to retrieve only high-resolution segment needed for a given query, rather than escalating to the full image. We construct this supervision automatically: a judge compares low- vs. high-resolution answers to label whether cropping is needed, and an oracle grounding model localizes the evidence for the correct answer, which we map to a discrete crop set to form multi-turn tool-use trajectories. We train our framework with SFT followed by multi-turn GRPO with a composite reward that combines semantic answer correctness with explicit crop-cost penalties. AwaRes provides a practical, deployment-friendly path to high-detail VLM reasoning under tight compute budgets.</abstract>
<identifier type="citekey">shabtay-etal-2026-look</identifier>
<location>
<url>https://aclanthology.org/2026.docinsights-1.25/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>357</start>
<end>384</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs
%A Shabtay, Nimrod
%A Kimhi, Moshe
%A Harary, Sivan
%A Spector, Artem
%A Rivlin, Ehud
%A Baskin, Chaim
%A Giryes, Raja
%A Schwartz, Eli
%Y T.Y.S.S., Santosh
%Y Agarwal, Amit
%Y Gupta, Vivek
%Y Kementchedjhieva, Yova
%Y Garimella, Aparna
%Y Hou, Yufang
%S The Proceedings of Workshop on Document Intelligence and Understanding
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-477-4
%F shabtay-etal-2026-look
%X Vision-language models (VLMs) typically process images at a native high-resolution, forcing a trade-off between accuracy and computational efficiency: high-resolution inputs capture fine details but incur significant computational costs, while low-resolution inputs advocate for efficiency, they potentially miss critical visual information, specifically OCR related. We present AwaRes, a spatial-on-demand framework that resolves this accuracy–efficiency trade-off by operating on a low-resolution global view and using tool-calling to retrieve only high-resolution segment needed for a given query, rather than escalating to the full image. We construct this supervision automatically: a judge compares low- vs. high-resolution answers to label whether cropping is needed, and an oracle grounding model localizes the evidence for the correct answer, which we map to a discrete crop set to form multi-turn tool-use trajectories. We train our framework with SFT followed by multi-turn GRPO with a composite reward that combines semantic answer correctness with explicit crop-cost penalties. AwaRes provides a practical, deployment-friendly path to high-detail VLM reasoning under tight compute budgets.
%U https://aclanthology.org/2026.docinsights-1.25/
%P 357-384
Markdown (Informal)
[Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs](https://aclanthology.org/2026.docinsights-1.25/) (Shabtay et al., DocInsights 2026)
ACL
- Nimrod Shabtay, Moshe Kimhi, Sivan Harary, Artem Spector, Ehud Rivlin, Chaim Baskin, Raja Giryes, and Eli Schwartz. 2026. Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs. In The Proceedings of Workshop on Document Intelligence and Understanding, pages 357–384, Budapest, Hungary. Association for Computational Linguistics.