<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hasanebrahimi-etal-2026-viser">
    <titleInfo>
        <title>VisER: Visual Evidence and Reliance for Object Hallucination Detection in LVLMs</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Afsaneh</namePart>
        <namePart type="family">Hasanebrahimi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hanxun</namePart>
        <namePart type="family">Huang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Christopher</namePart>
        <namePart type="family">Leckie</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sarah</namePart>
        <namePart type="family">Erfani</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>Object hallucination remains a persistent reliability issue in large vision-language models, where generated object mentions may sound plausible but lack visual grounding. Recent training-free detectors use internal signals such as token likelihood, attention, visual confidence, or image-text similarity to identify hallucinated objects. These signals are useful, but they are often source-confounded. They measure how strongly an object is supported inside the model without distinguishing whether that support comes from object-specific visual evidence or the generated text prefix. In difficult cases, a hallucinated object can still receive high internal support because it fits the scene, is associated with nearby visual cues, or follows naturally from the generated text prefix. We propose VisER, a training-free two-sided metric for object-level hallucination detection. VisER evaluates each generated object mention from two complementary views. Visual Evidence measures whether object-context compatibility is backed by object-specific evidence from image tokens. Visual Reliance measures whether the object is supported more by the image than by the generated prefix. Combining these views gives a more source-aware grounding score, while avoiding additional object-level verification generations. Across multiple LVLMs and benchmarks, VisER improves AUROC and AUPR over a range of baselines.</abstract>
    <identifier type="citekey">hasanebrahimi-etal-2026-viser</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.2645/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>60186</start>
            <end>60202</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T VisER: Visual Evidence and Reliance for Object Hallucination Detection in LVLMs
%A Hasanebrahimi, Afsaneh
%A Huang, Hanxun
%A Leckie, Christopher
%A Erfani, Sarah
%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 hasanebrahimi-etal-2026-viser
%X Object hallucination remains a persistent reliability issue in large vision-language models, where generated object mentions may sound plausible but lack visual grounding. Recent training-free detectors use internal signals such as token likelihood, attention, visual confidence, or image-text similarity to identify hallucinated objects. These signals are useful, but they are often source-confounded. They measure how strongly an object is supported inside the model without distinguishing whether that support comes from object-specific visual evidence or the generated text prefix. In difficult cases, a hallucinated object can still receive high internal support because it fits the scene, is associated with nearby visual cues, or follows naturally from the generated text prefix. We propose VisER, a training-free two-sided metric for object-level hallucination detection. VisER evaluates each generated object mention from two complementary views. Visual Evidence measures whether object-context compatibility is backed by object-specific evidence from image tokens. Visual Reliance measures whether the object is supported more by the image than by the generated prefix. Combining these views gives a more source-aware grounding score, while avoiding additional object-level verification generations. Across multiple LVLMs and benchmarks, VisER improves AUROC and AUPR over a range of baselines.
%U https://aclanthology.org/2026.emnlp-main.2645/
%P 60186-60202

Markdown (Informal)

[VisER: Visual Evidence and Reliance for Object Hallucination Detection in LVLMs](https://aclanthology.org/2026.emnlp-main.2645/) (Hasanebrahimi et al., EMNLP 2026)

• VisER: Visual Evidence and Reliance for Object Hallucination Detection in LVLMs (Hasanebrahimi et al., EMNLP 2026)

ACL

• Afsaneh Hasanebrahimi, Hanxun Huang, Christopher Leckie, and Sarah Erfani. 2026. VisER: Visual Evidence and Reliance for Object Hallucination Detection in LVLMs. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 60186–60202, Budapest, Hungary. Association for Computational Linguistics.