3D data augmentation and dual-branch model for robust face forgery detection
We propose Dual-Branch Network (DBNet), a novel deepfake detection framework that addresses key limitations of existing works by jointly modeling 3D-temporal and fine-grained texture representations. Specifically, we aim to investigate how to (1) capture dynamic properties and spatial details in a u...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Graphical Models |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1524070325000025 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832087760958128128 |
|---|---|
| author | Changshuang Zhou Frederick W.B. Li Chao Song Dong Zheng Bailin Yang |
| author_facet | Changshuang Zhou Frederick W.B. Li Chao Song Dong Zheng Bailin Yang |
| author_sort | Changshuang Zhou |
| collection | DOAJ |
| description | We propose Dual-Branch Network (DBNet), a novel deepfake detection framework that addresses key limitations of existing works by jointly modeling 3D-temporal and fine-grained texture representations. Specifically, we aim to investigate how to (1) capture dynamic properties and spatial details in a unified model and (2) identify subtle inconsistencies beyond localized artifacts through temporally consistent modeling. To this end, DBNet extracts 3D landmarks from videos to construct temporal sequences for an RNN branch, while a Vision Transformer analyzes local patches. A Temporal Consistency-aware Loss is introduced to explicitly supervise the RNN. Additionally, a 3D generative model augments training data. Extensive experiments demonstrate our method achieves state-of-the-art performance on benchmarks, and ablation studies validate its effectiveness in generalizing to unseen data under various manipulations and compression. |
| format | Article |
| id | doaj-art-7d176fc9555d4a70b98b1905124fecbe |
| institution | Kabale University |
| issn | 1524-0703 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Graphical Models |
| spelling | doaj-art-7d176fc9555d4a70b98b1905124fecbe2025-02-06T05:11:17ZengElsevierGraphical Models1524-07032025-03-011381012553D data augmentation and dual-branch model for robust face forgery detectionChangshuang Zhou0Frederick W.B. Li1Chao Song2Dong Zheng3Bailin Yang4Zhejiang Gongshang University, ChinaUniversity of Durham, United KingdomZhejiang Gongshang University, ChinaZheng Dong Universal Ubiquitous Technology Co., LTD, ChinaZhejiang Gongshang University, China; Corresponding author.We propose Dual-Branch Network (DBNet), a novel deepfake detection framework that addresses key limitations of existing works by jointly modeling 3D-temporal and fine-grained texture representations. Specifically, we aim to investigate how to (1) capture dynamic properties and spatial details in a unified model and (2) identify subtle inconsistencies beyond localized artifacts through temporally consistent modeling. To this end, DBNet extracts 3D landmarks from videos to construct temporal sequences for an RNN branch, while a Vision Transformer analyzes local patches. A Temporal Consistency-aware Loss is introduced to explicitly supervise the RNN. Additionally, a 3D generative model augments training data. Extensive experiments demonstrate our method achieves state-of-the-art performance on benchmarks, and ablation studies validate its effectiveness in generalizing to unseen data under various manipulations and compression.http://www.sciencedirect.com/science/article/pii/S1524070325000025Dual-branch network3D data augmentationDeepfake detection |
| spellingShingle | Changshuang Zhou Frederick W.B. Li Chao Song Dong Zheng Bailin Yang 3D data augmentation and dual-branch model for robust face forgery detection Graphical Models Dual-branch network 3D data augmentation Deepfake detection |
| title | 3D data augmentation and dual-branch model for robust face forgery detection |
| title_full | 3D data augmentation and dual-branch model for robust face forgery detection |
| title_fullStr | 3D data augmentation and dual-branch model for robust face forgery detection |
| title_full_unstemmed | 3D data augmentation and dual-branch model for robust face forgery detection |
| title_short | 3D data augmentation and dual-branch model for robust face forgery detection |
| title_sort | 3d data augmentation and dual branch model for robust face forgery detection |
| topic | Dual-branch network 3D data augmentation Deepfake detection |
| url | http://www.sciencedirect.com/science/article/pii/S1524070325000025 |
| work_keys_str_mv | AT changshuangzhou 3ddataaugmentationanddualbranchmodelforrobustfaceforgerydetection AT frederickwbli 3ddataaugmentationanddualbranchmodelforrobustfaceforgerydetection AT chaosong 3ddataaugmentationanddualbranchmodelforrobustfaceforgerydetection AT dongzheng 3ddataaugmentationanddualbranchmodelforrobustfaceforgerydetection AT bailinyang 3ddataaugmentationanddualbranchmodelforrobustfaceforgerydetection |