Research

Paper

TESTING March 17, 2026

MLLM-based Textual Explanations for Face Comparison

Authors

Redwan Sony, Anil K Jain, Ross Arun

Abstract

Multimodal Large Language Models (MLLMs) have recently been proposed as a means to generate natural-language explanations for face recognition decisions. While such explanations facilitate human interpretability, their reliability on unconstrained face images remains underexplored. In this work, we systematically analyze MLLM-generated explanations for the unconstrained face verification task on the challenging IJB-S dataset, with a particular focus on extreme pose variation and surveillance imagery. Our results show that even when MLLMs produce correct verification decisions, the accompanying explanations frequently rely on non-verifiable or hallucinated facial attributes that are not supported by visual evidence. We further study the effect of incorporating information from traditional face recognition systems, viz., scores and decisions, alongside the input images. Although such information improves categorical verification performance, it does not consistently lead to faithful explanations. To evaluate the explanations beyond decision accuracy, we introduce a likelihood-ratio-based framework that measures the evidential strength of textual explanations. Our findings highlight fundamental limitations of current MLLMs for explainable face recognition and underscore the need for a principled evaluation of reliable and trustworthy explanations in biometric applications. Code is available at https://github.com/redwankarimsony/LR-MLLMFR-Explainability.

Metadata

arXiv ID: 2603.16629
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-17
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.16629v1</id>\n    <title>MLLM-based Textual Explanations for Face Comparison</title>\n    <updated>2026-03-17T15:01:00Z</updated>\n    <link href='https://arxiv.org/abs/2603.16629v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.16629v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Multimodal Large Language Models (MLLMs) have recently been proposed as a means to generate natural-language explanations for face recognition decisions. While such explanations facilitate human interpretability, their reliability on unconstrained face images remains underexplored. In this work, we systematically analyze MLLM-generated explanations for the unconstrained face verification task on the challenging IJB-S dataset, with a particular focus on extreme pose variation and surveillance imagery. Our results show that even when MLLMs produce correct verification decisions, the accompanying explanations frequently rely on non-verifiable or hallucinated facial attributes that are not supported by visual evidence. We further study the effect of incorporating information from traditional face recognition systems, viz., scores and decisions, alongside the input images. Although such information improves categorical verification performance, it does not consistently lead to faithful explanations. To evaluate the explanations beyond decision accuracy, we introduce a likelihood-ratio-based framework that measures the evidential strength of textual explanations. Our findings highlight fundamental limitations of current MLLMs for explainable face recognition and underscore the need for a principled evaluation of reliable and trustworthy explanations in biometric applications. Code is available at https://github.com/redwankarimsony/LR-MLLMFR-Explainability.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-17T15:01:00Z</published>\n    <arxiv:comment>Accepted at 14th International Workshop on Biometrics and Forensics (IWBF)</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Redwan Sony</name>\n    </author>\n    <author>\n      <name>Anil K Jain</name>\n    </author>\n    <author>\n      <name>Ross Arun</name>\n    </author>\n  </entry>"
}