DeeCLIP: A Robust and Generalizable Transformer-Based Framework for Detecting AI-Generated Images Article Swipe
Mamadou Keïta
,
Wassim Hamidouche
,
Hessen Bougueffa Eutamene
,
Abdelmalik Taleb‐Ahmed
,
Abdenour Hadid
·
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2504.19876
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2504.19876
This paper introduces DeeCLIP, a novel framework for detecting AI-generated images using CLIP-ViT and fusion learning. Despite significant advancements in generative models capable of creating highly photorealistic images, existing detection methods often struggle to generalize across different models and are highly sensitive to minor perturbations. To address these challenges, DeeCLIP incorporates DeeFuser, a fusion module that combines high-level and low-level features, improving robustness against degradations such as compression and blurring. Additionally, we apply triplet loss to refine the embedding space, enhancing the model's ability to distinguish between real and synthetic content. To further enable lightweight adaptation while preserving pre-trained knowledge, we adopt parameter-efficient fine-tuning using low-rank adaptation (LoRA) within the CLIP-ViT backbone. This approach supports effective zero-shot learning without sacrificing generalization. Trained exclusively on 4-class ProGAN data, DeeCLIP achieves an average accuracy of 89.00% on 19 test subsets composed of generative adversarial network (GAN) and diffusion models. Despite having fewer trainable parameters, DeeCLIP outperforms existing methods, demonstrating superior robustness against various generative models and real-world distortions. The code is publicly available at https://github.com/Mamadou-Keita/DeeCLIP for research purposes.
Related Topics
Concepts
No concepts available.
Metadata
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2504.19876
- https://arxiv.org/pdf/2504.19876
- OA Status
- green
- OpenAlex ID
- https://openalex.org/W4414746156
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4414746156Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2504.19876Digital Object Identifier
- Title
-
DeeCLIP: A Robust and Generalizable Transformer-Based Framework for Detecting AI-Generated ImagesWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2025Year of publication
- Publication date
-
2025-04-28Full publication date if available
- Authors
-
Mamadou Keïta, Wassim Hamidouche, Hessen Bougueffa Eutamene, Abdelmalik Taleb‐Ahmed, Abdenour HadidList of authors in order
- Landing page
-
https://arxiv.org/abs/2504.19876Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2504.19876Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2504.19876Direct OA link when available
- Cited by
-
0Total citation count in OpenAlex
Full payload
| id | https://openalex.org/W4414746156 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2504.19876 |
| ids.doi | https://doi.org/10.48550/arxiv.2504.19876 |
| ids.openalex | https://openalex.org/W4414746156 |
| fwci | |
| type | preprint |
| title | DeeCLIP: A Robust and Generalizable Transformer-Based Framework for Detecting AI-Generated Images |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T12859 |
| topics[0].field.id | https://openalex.org/fields/13 |
| topics[0].field.display_name | Biochemistry, Genetics and Molecular Biology |
| topics[0].score | 0.8159999847412109 |
| topics[0].domain.id | https://openalex.org/domains/1 |
| topics[0].domain.display_name | Life Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1304 |
| topics[0].subfield.display_name | Biophysics |
| topics[0].display_name | Cell Image Analysis Techniques |
| topics[1].id | https://openalex.org/T12357 |
| topics[1].field.id | https://openalex.org/fields/17 |
| topics[1].field.display_name | Computer Science |
| topics[1].score | 0.7925000190734863 |
| topics[1].domain.id | https://openalex.org/domains/3 |
| topics[1].domain.display_name | Physical Sciences |
| topics[1].subfield.id | https://openalex.org/subfields/1707 |
| topics[1].subfield.display_name | Computer Vision and Pattern Recognition |
| topics[1].display_name | Digital Media Forensic Detection |
| topics[2].id | https://openalex.org/T11689 |
| topics[2].field.id | https://openalex.org/fields/17 |
| topics[2].field.display_name | Computer Science |
| topics[2].score | 0.7002999782562256 |
| topics[2].domain.id | https://openalex.org/domains/3 |
| topics[2].domain.display_name | Physical Sciences |
| topics[2].subfield.id | https://openalex.org/subfields/1702 |
| topics[2].subfield.display_name | Artificial Intelligence |
| topics[2].display_name | Adversarial Robustness in Machine Learning |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2504.19876 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | cc-by |
| locations[0].pdf_url | https://arxiv.org/pdf/2504.19876 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | https://openalex.org/licenses/cc-by |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2504.19876 |
| locations[1].id | pmh:oai:HAL:hal-05086705v1 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306402512 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | False |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | HAL (Le Centre pour la Communication Scientifique Directe) |
| locations[1].source.host_organization | https://openalex.org/I1294671590 |
| locations[1].source.host_organization_name | Centre National de la Recherche Scientifique |
| locations[1].source.host_organization_lineage | https://openalex.org/I1294671590 |
| locations[1].license | cc-by |
| locations[1].pdf_url | |
| locations[1].version | submittedVersion |
| locations[1].raw_type | Preprints, Working Papers, ... |
| locations[1].license_id | https://openalex.org/licenses/cc-by |
| locations[1].is_accepted | False |
| locations[1].is_published | False |
| locations[1].raw_source_name | 2025 |
| locations[1].landing_page_url | https://hal.science/hal-05086705 |
| locations[2].id | doi:10.48550/arxiv.2504.19876 |
| locations[2].is_oa | True |
| locations[2].source.id | https://openalex.org/S4306400194 |
| locations[2].source.issn | |
| locations[2].source.type | repository |
| locations[2].source.is_oa | True |
| locations[2].source.issn_l | |
| locations[2].source.is_core | False |
| locations[2].source.is_in_doaj | False |
| locations[2].source.display_name | arXiv (Cornell University) |
| locations[2].source.host_organization | https://openalex.org/I205783295 |
| locations[2].source.host_organization_name | Cornell University |
| locations[2].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[2].license | cc-by |
| locations[2].pdf_url | |
| locations[2].version | |
| locations[2].raw_type | article |
| locations[2].license_id | https://openalex.org/licenses/cc-by |
| locations[2].is_accepted | False |
| locations[2].is_published | |
| locations[2].raw_source_name | |
| locations[2].landing_page_url | https://doi.org/10.48550/arxiv.2504.19876 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5031314564 |
| authorships[0].author.orcid | https://orcid.org/0009-0009-7618-9253 |
| authorships[0].author.display_name | Mamadou Keïta |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Mamadou Keita |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5084205574 |
| authorships[1].author.orcid | https://orcid.org/0000-0002-0143-1756 |
| authorships[1].author.display_name | Wassim Hamidouche |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Wassim Hamidouche |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5094583687 |
| authorships[2].author.orcid | https://orcid.org/0009-0009-0556-9996 |
| authorships[2].author.display_name | Hessen Bougueffa Eutamene |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Hessen Bougueffa Eutamene |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5088214859 |
| authorships[3].author.orcid | https://orcid.org/0000-0001-7218-3799 |
| authorships[3].author.display_name | Abdelmalik Taleb‐Ahmed |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Abdelmalik Taleb-Ahmed |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5013928164 |
| authorships[4].author.orcid | https://orcid.org/0000-0001-9092-735X |
| authorships[4].author.display_name | Abdenour Hadid |
| authorships[4].countries | AE |
| authorships[4].affiliations[0].raw_affiliation_string | SUAD_SAFIR (United Arab Emirates) |
| authorships[4].affiliations[1].institution_ids | https://openalex.org/I4210141120 |
| authorships[4].affiliations[1].raw_affiliation_string | SUAD - Sorbonne University Abu Dhabi (Ile d’Al Reem Abu Dhabi Émirats Arabes Unis PO Box 38044 - United Arab Emirates) |
| authorships[4].institutions[0].id | https://openalex.org/I4210141120 |
| authorships[4].institutions[0].ror | https://ror.org/03e1ymy32 |
| authorships[4].institutions[0].type | education |
| authorships[4].institutions[0].lineage | https://openalex.org/I4210141120 |
| authorships[4].institutions[0].country_code | AE |
| authorships[4].institutions[0].display_name | Sorbonne University Abu Dhabi |
| authorships[4].author_position | last |
| authorships[4].raw_author_name | Abdenour Hadid |
| authorships[4].is_corresponding | False |
| authorships[4].raw_affiliation_strings | SUAD - Sorbonne University Abu Dhabi (Ile d’Al Reem Abu Dhabi Émirats Arabes Unis PO Box 38044 - United Arab Emirates), SUAD_SAFIR (United Arab Emirates) |
| has_content.pdf | True |
| has_content.grobid_xml | True |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2504.19876 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-02T00:00:00 |
| display_name | DeeCLIP: A Robust and Generalizable Transformer-Based Framework for Detecting AI-Generated Images |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T12859 |
| primary_topic.field.id | https://openalex.org/fields/13 |
| primary_topic.field.display_name | Biochemistry, Genetics and Molecular Biology |
| primary_topic.score | 0.8159999847412109 |
| primary_topic.domain.id | https://openalex.org/domains/1 |
| primary_topic.domain.display_name | Life Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1304 |
| primary_topic.subfield.display_name | Biophysics |
| primary_topic.display_name | Cell Image Analysis Techniques |
| cited_by_count | 0 |
| locations_count | 3 |
| best_oa_location.id | pmh:oai:arXiv.org:2504.19876 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | cc-by |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2504.19876 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | https://openalex.org/licenses/cc-by |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2504.19876 |
| primary_location.id | pmh:oai:arXiv.org:2504.19876 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | cc-by |
| primary_location.pdf_url | https://arxiv.org/pdf/2504.19876 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | https://openalex.org/licenses/cc-by |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2504.19876 |
| publication_date | 2025-04-28 |
| publication_year | 2025 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 4, 52 |
| abstract_inverted_index.19 | 135 |
| abstract_inverted_index.To | 45, 91 |
| abstract_inverted_index.an | 129 |
| abstract_inverted_index.as | 66 |
| abstract_inverted_index.at | 171 |
| abstract_inverted_index.in | 19 |
| abstract_inverted_index.is | 168 |
| abstract_inverted_index.of | 23, 132, 139 |
| abstract_inverted_index.on | 123, 134 |
| abstract_inverted_index.to | 33, 42, 75, 84 |
| abstract_inverted_index.we | 71, 100 |
| abstract_inverted_index.The | 166 |
| abstract_inverted_index.and | 13, 38, 58, 68, 88, 144, 163 |
| abstract_inverted_index.are | 39 |
| abstract_inverted_index.for | 7, 173 |
| abstract_inverted_index.the | 77, 81, 109 |
| abstract_inverted_index.This | 0, 112 |
| abstract_inverted_index.code | 167 |
| abstract_inverted_index.loss | 74 |
| abstract_inverted_index.real | 87 |
| abstract_inverted_index.such | 65 |
| abstract_inverted_index.test | 136 |
| abstract_inverted_index.that | 55 |
| abstract_inverted_index.(GAN) | 143 |
| abstract_inverted_index.adopt | 101 |
| abstract_inverted_index.apply | 72 |
| abstract_inverted_index.data, | 126 |
| abstract_inverted_index.fewer | 149 |
| abstract_inverted_index.minor | 43 |
| abstract_inverted_index.novel | 5 |
| abstract_inverted_index.often | 31 |
| abstract_inverted_index.paper | 1 |
| abstract_inverted_index.these | 47 |
| abstract_inverted_index.using | 11, 104 |
| abstract_inverted_index.while | 96 |
| abstract_inverted_index.(LoRA) | 107 |
| abstract_inverted_index.89.00% | 133 |
| abstract_inverted_index.ProGAN | 125 |
| abstract_inverted_index.across | 35 |
| abstract_inverted_index.enable | 93 |
| abstract_inverted_index.fusion | 14, 53 |
| abstract_inverted_index.having | 148 |
| abstract_inverted_index.highly | 25, 40 |
| abstract_inverted_index.images | 10 |
| abstract_inverted_index.models | 21, 37, 162 |
| abstract_inverted_index.module | 54 |
| abstract_inverted_index.refine | 76 |
| abstract_inverted_index.space, | 79 |
| abstract_inverted_index.within | 108 |
| abstract_inverted_index.4-class | 124 |
| abstract_inverted_index.DeeCLIP | 49, 127, 152 |
| abstract_inverted_index.Despite | 16, 147 |
| abstract_inverted_index.Trained | 121 |
| abstract_inverted_index.ability | 83 |
| abstract_inverted_index.address | 46 |
| abstract_inverted_index.against | 63, 159 |
| abstract_inverted_index.average | 130 |
| abstract_inverted_index.between | 86 |
| abstract_inverted_index.capable | 22 |
| abstract_inverted_index.further | 92 |
| abstract_inverted_index.images, | 27 |
| abstract_inverted_index.methods | 30 |
| abstract_inverted_index.model's | 82 |
| abstract_inverted_index.models. | 146 |
| abstract_inverted_index.network | 142 |
| abstract_inverted_index.subsets | 137 |
| abstract_inverted_index.triplet | 73 |
| abstract_inverted_index.various | 160 |
| abstract_inverted_index.without | 118 |
| abstract_inverted_index.CLIP-ViT | 12, 110 |
| abstract_inverted_index.DeeCLIP, | 3 |
| abstract_inverted_index.accuracy | 131 |
| abstract_inverted_index.achieves | 128 |
| abstract_inverted_index.approach | 113 |
| abstract_inverted_index.combines | 56 |
| abstract_inverted_index.composed | 138 |
| abstract_inverted_index.content. | 90 |
| abstract_inverted_index.creating | 24 |
| abstract_inverted_index.existing | 28, 154 |
| abstract_inverted_index.learning | 117 |
| abstract_inverted_index.low-rank | 105 |
| abstract_inverted_index.methods, | 155 |
| abstract_inverted_index.publicly | 169 |
| abstract_inverted_index.research | 174 |
| abstract_inverted_index.struggle | 32 |
| abstract_inverted_index.superior | 157 |
| abstract_inverted_index.supports | 114 |
| abstract_inverted_index.DeeFuser, | 51 |
| abstract_inverted_index.available | 170 |
| abstract_inverted_index.backbone. | 111 |
| abstract_inverted_index.blurring. | 69 |
| abstract_inverted_index.detecting | 8 |
| abstract_inverted_index.detection | 29 |
| abstract_inverted_index.different | 36 |
| abstract_inverted_index.diffusion | 145 |
| abstract_inverted_index.effective | 115 |
| abstract_inverted_index.embedding | 78 |
| abstract_inverted_index.enhancing | 80 |
| abstract_inverted_index.features, | 60 |
| abstract_inverted_index.framework | 6 |
| abstract_inverted_index.improving | 61 |
| abstract_inverted_index.learning. | 15 |
| abstract_inverted_index.low-level | 59 |
| abstract_inverted_index.purposes. | 175 |
| abstract_inverted_index.sensitive | 41 |
| abstract_inverted_index.synthetic | 89 |
| abstract_inverted_index.trainable | 150 |
| abstract_inverted_index.zero-shot | 116 |
| abstract_inverted_index.adaptation | 95, 106 |
| abstract_inverted_index.generalize | 34 |
| abstract_inverted_index.generative | 20, 140, 161 |
| abstract_inverted_index.high-level | 57 |
| abstract_inverted_index.introduces | 2 |
| abstract_inverted_index.knowledge, | 99 |
| abstract_inverted_index.preserving | 97 |
| abstract_inverted_index.real-world | 164 |
| abstract_inverted_index.robustness | 62, 158 |
| abstract_inverted_index.adversarial | 141 |
| abstract_inverted_index.challenges, | 48 |
| abstract_inverted_index.compression | 67 |
| abstract_inverted_index.distinguish | 85 |
| abstract_inverted_index.exclusively | 122 |
| abstract_inverted_index.fine-tuning | 103 |
| abstract_inverted_index.lightweight | 94 |
| abstract_inverted_index.outperforms | 153 |
| abstract_inverted_index.parameters, | 151 |
| abstract_inverted_index.pre-trained | 98 |
| abstract_inverted_index.sacrificing | 119 |
| abstract_inverted_index.significant | 17 |
| abstract_inverted_index.AI-generated | 9 |
| abstract_inverted_index.advancements | 18 |
| abstract_inverted_index.degradations | 64 |
| abstract_inverted_index.distortions. | 165 |
| abstract_inverted_index.incorporates | 50 |
| abstract_inverted_index.Additionally, | 70 |
| abstract_inverted_index.demonstrating | 156 |
| abstract_inverted_index.perturbations. | 44 |
| abstract_inverted_index.photorealistic | 26 |
| abstract_inverted_index.generalization. | 120 |
| abstract_inverted_index.parameter-efficient | 102 |
| abstract_inverted_index.https://github.com/Mamadou-Keita/DeeCLIP | 172 |
| cited_by_percentile_year | |
| countries_distinct_count | 1 |
| institutions_distinct_count | 5 |
| citation_normalized_percentile |