Training free Cross-Modal Alignment for Multilingual Image Retrieval

Usman Shahid; Manan Jain; Natalie Parde

Training free Cross-Modal Alignment for Multilingual Image Retrieval

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

We investigate whether multilingual image-text retrieval can be made practical without training a multi-modal model, using frozen multilingual uni-modal encoders and inducing cross-modal alignment through relative representations computed against a large external anchor set. We use English image-caption pairs as anchors and evaluate caption-to-image retrieval on 3,600 images with captions in 36 languages (some low-resource). Across multiple encoder pairings and hyperparameter settings, the method yields usable performance, positioning training-free relative representation as a baseline for multilingual cross-modal alignment when target-language multi-modal training data is unavailable.

Anthology ID:: 2026.findings-aacl.203
Volume:: Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026
Month:: November
Year:: 2026
Address:: Hengqin, China
Editors:: Yue Zhang, Ruifeng Xu, Ayu Purwarianti, Tong Rong, Jun Suzuki
Venue:: Findings
SIG:
Publisher:: Association for Computational Linguistics
Note:
Pages:: 4039–4049
Language:
URL:: https://aclanthology.org/2026.findings-aacl.203/
DOI:
Bibkey:
Cite (ACL):: Usman Shahid, Manan Jain, and Natalie Parde. 2026. Training free Cross-Modal Alignment for Multilingual Image Retrieval. In Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026, pages 4039–4049, Hengqin, China. Association for Computational Linguistics.
Cite (Informal):: Training free Cross-Modal Alignment for Multilingual Image Retrieval (Shahid et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-aacl.203.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{shahid-etal-2026-training,
    title = "Training free Cross-Modal Alignment for Multilingual Image Retrieval",
    author = "Shahid, Usman  and
      Jain, Manan  and
      Parde, Natalie",
    editor = "Zhang, Yue  and
      Xu, Ruifeng  and
      Purwarianti, Ayu  and
      Rong, Tong  and
      Suzuki, Jun",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {IJCNLP}-{AACL} 2026",
    month = nov,
    year = "2026",
    address = "Hengqin, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-aacl.203/",
    pages = "4039--4049",
    ISBN = "979-8-89176-444-6",
    abstract = "We investigate whether multilingual image-text retrieval can be made practical without training a multi-modal model, using frozen multilingual uni-modal encoders and inducing cross-modal alignment through relative representations computed against a large external anchor set. We use English image-caption pairs as anchors and evaluate caption-to-image retrieval on 3,600 images with captions in 36 languages (some low-resource). Across multiple encoder pairings and hyperparameter settings, the method yields usable performance, positioning training-free relative representation as a baseline for multilingual cross-modal alignment when target-language multi-modal training data is unavailable."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shahid-etal-2026-training">
    <titleInfo>
        <title>Training free Cross-Modal Alignment for Multilingual Image Retrieval</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Usman</namePart>
        <namePart type="family">Shahid</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Manan</namePart>
        <namePart type="family">Jain</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Natalie</namePart>
        <namePart type="family">Parde</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Yue</namePart>
            <namePart type="family">Zhang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ruifeng</namePart>
            <namePart type="family">Xu</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ayu</namePart>
            <namePart type="family">Purwarianti</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tong</namePart>
            <namePart type="family">Rong</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jun</namePart>
            <namePart type="family">Suzuki</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Hengqin, China</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-444-6</identifier>
    </relatedItem>
    <abstract>We investigate whether multilingual image-text retrieval can be made practical without training a multi-modal model, using frozen multilingual uni-modal encoders and inducing cross-modal alignment through relative representations computed against a large external anchor set. We use English image-caption pairs as anchors and evaluate caption-to-image retrieval on 3,600 images with captions in 36 languages (some low-resource). Across multiple encoder pairings and hyperparameter settings, the method yields usable performance, positioning training-free relative representation as a baseline for multilingual cross-modal alignment when target-language multi-modal training data is unavailable.</abstract>
    <identifier type="citekey">shahid-etal-2026-training</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-aacl.203/</url>
    </location>
    <part>
        <date>2026-11</date>
        <extent unit="page">
            <start>4039</start>
            <end>4049</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Training free Cross-Modal Alignment for Multilingual Image Retrieval
%A Shahid, Usman
%A Jain, Manan
%A Parde, Natalie
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-444-6
%F shahid-etal-2026-training
%X We investigate whether multilingual image-text retrieval can be made practical without training a multi-modal model, using frozen multilingual uni-modal encoders and inducing cross-modal alignment through relative representations computed against a large external anchor set. We use English image-caption pairs as anchors and evaluate caption-to-image retrieval on 3,600 images with captions in 36 languages (some low-resource). Across multiple encoder pairings and hyperparameter settings, the method yields usable performance, positioning training-free relative representation as a baseline for multilingual cross-modal alignment when target-language multi-modal training data is unavailable.
%U https://aclanthology.org/2026.findings-aacl.203/
%P 4039-4049

Download as File

Markdown (Informal)

[Training free Cross-Modal Alignment for Multilingual Image Retrieval](https://aclanthology.org/2026.findings-aacl.203/) (Shahid et al., Findings 2026)

Training free Cross-Modal Alignment for Multilingual Image Retrieval (Shahid et al., Findings 2026)

ACL

Usman Shahid, Manan Jain, and Natalie Parde. 2026. Training free Cross-Modal Alignment for Multilingual Image Retrieval. In Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026, pages 4039–4049, Hengqin, China. Association for Computational Linguistics.