Research

Paper

AI LLM February 27, 2026

NAU-QMUL: Utilizing BERT and CLIP for Multi-modal AI-Generated Image Detection

Authors

Xiaoyu Guo, Arkaitz Zubiaga

Abstract

With the aim of detecting AI-generated images and identifying the specific models responsible for their generation, we propose a multi-modal multi-task model. The model leverages pre-trained BERT and CLIP Vision encoders for text and image feature extraction, respectively, and employs cross-modal feature fusion with a tailored multi-task loss function. Additionally, a pseudo-labeling-based data augmentation strategy was utilized to expand the training dataset with high-confidence samples. The model achieved fifth place in both Tasks A and B of the `CT2: AI-Generated Image Detection' competition, with F1 scores of 83.16\% and 48.88\%, respectively. These findings highlight the effectiveness of the proposed architecture and its potential for advancing AI-generated content detection in real-world scenarios. The source code for our method is published on https://github.com/xxxxxxxxy/AIGeneratedImageDetection.

Metadata

arXiv ID: 2602.23863
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-02-27
Fetched: 2026-03-02 06:04

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.23863v1</id>\n    <title>NAU-QMUL: Utilizing BERT and CLIP for Multi-modal AI-Generated Image Detection</title>\n    <updated>2026-02-27T10:03:54Z</updated>\n    <link href='https://arxiv.org/abs/2602.23863v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.23863v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>With the aim of detecting AI-generated images and identifying the specific models responsible for their generation, we propose a multi-modal multi-task model. The model leverages pre-trained BERT and CLIP Vision encoders for text and image feature extraction, respectively, and employs cross-modal feature fusion with a tailored multi-task loss function. Additionally, a pseudo-labeling-based data augmentation strategy was utilized to expand the training dataset with high-confidence samples. The model achieved fifth place in both Tasks A and B of the `CT2: AI-Generated Image Detection' competition, with F1 scores of 83.16\\% and 48.88\\%, respectively. These findings highlight the effectiveness of the proposed architecture and its potential for advancing AI-generated content detection in real-world scenarios. The source code for our method is published on https://github.com/xxxxxxxxy/AIGeneratedImageDetection.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-02-27T10:03:54Z</published>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Xiaoyu Guo</name>\n    </author>\n    <author>\n      <name>Arkaitz Zubiaga</name>\n    </author>\n  </entry>"
}