Engineering Review : International journal for publishing of original researches from the aspect of structural analysis, materials and new technologies in the field of mechanical engineering, shipbuilding, fundamental engineering sciences, computer sciences, electrical engin, Vol. 44 No. 4 - SI 2024, 2024.
Original scientific paper
https://doi.org/10.30765/er.2583
Advanced deepfake detection leveraging swin transformer technology
Soumya Ranjan Mishra
; Computer Science and Engineering, KIIT (Deemed to be) University, Bhubaneswar, India
Hitesh Mohapatra
; Computer Science and Engineering, KIIT (Deemed to be) University, Bhubaneswar, India
*
Seyed Ahmad Edalatpanah
; Department of Applied Mathematics, Ayandegan Institute of Higher Education, Tonekabon, Iran
Mahendra Kumar Gourisaria
; Computer Science and Engineering, KIIT (Deemed to be) University, Bhubaneswar, India
* Corresponding author.
Abstract
The widespread use of deepfake technology in recent years has made it extremely difficult to differentiate between real and fake images, usually AI-generated images. Effective detection techniques are desperately needed because one can generate fake images and spread them with ease. This research paper examines how effective the SWIN Transformer, a new transformer-based architecture, is for detecting deep fake images. The foundation of the suggested detection framework is an architecture made up of bottleneck, encoder, and decoder parts which is a type of SWIN transformer. It uses various self-attention mechanisms and advanced features to analyse the images closely whether it is a real image or a deepfake one. It relies on the concept of shifted windows during the processing of the images and is considered more effective than the traditional CNN methods. Our test results show how well the SWIN Transformer-based method performs in precisely recognizing deep fake images. The accuracy is found to be 97.91\% for CelebDF dataset and 95.715\% for FF++ dataset. The AUC for the newly modelled SWIN transformer is 0.99 and 0.9625 for CelebDF and FF++ datasets respectively. The Log Loss was calculated to be 0.034 for CelebDF dataset and 0.1573 for FF++ dataset. The proposed methodology not only enhances the accuracy of detecting manipulated images but also offers potential for scalable and efficient deployment in real-world scenarios where the proliferation of deepfakes presents significant challenges to maintaining trust and authenticity in visual media.
Keywords
deepfake; image classification; SWIN trans- formers; fake image generation; image detection; hierarchical representation; transformer block; quadratic complexity; SWIN transformer blocks; object detection
Hrčak ID:
324599
URI
Publication date:
23.12.2024.
Visits: 0 *