| _version_ | 1866901533177675776 |
|---|---|
| author | Patil, U. S. Sontakke, Omkar Ravindra Jadhav, Kaustubh Dattatray Khot, Sujal Subhash Shinde, Soham Santosh Kumbhar, Pranav Bajirao |
| author_facet | Patil, U. S. Sontakke, Omkar Ravindra Jadhav, Kaustubh Dattatray Khot, Sujal Subhash Shinde, Soham Santosh Kumbhar, Pranav Bajirao |
| contents | <p><strong>Abstract —</strong><br>This study presents a robust framework for deepfake detection using advanced machine learning techniques. The proposed framework leverages a hybrid model that integrates convolutional neural networks and vision transformers to accurately distinguish between authentic and synthetic media. The methodology employs systematic preprocessing and feature extraction techniques to enhance detection accuracy, which is crucial due to the increasing sophistication of deepfake generation methods. Experimental evaluations using benchmark datasets such as FaceForensics++, Celeb-DF, and the DeepFake Detection Challenge dataset demonstrate promising results, achieving up to 97.2% accuracy. The framework is designed to be computationally efficient and scalable for real-world applications in cybersecurity and media forensics. This study contributes to ongoing research by evaluating hybrid classifier architectures and establishing a baseline for future exploration in deepfake detection.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_17660017 |
| institution | Zenodo |
| language | eng |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | DEEPFAKE DETECTION USING MACHINE LEARNING Patil, U. S. Sontakke, Omkar Ravindra Jadhav, Kaustubh Dattatray Khot, Sujal Subhash Shinde, Soham Santosh Kumbhar, Pranav Bajirao Deepfake Detection Machine Learning CNN Vision Transformer Digital Forensics <p><strong>Abstract —</strong><br>This study presents a robust framework for deepfake detection using advanced machine learning techniques. The proposed framework leverages a hybrid model that integrates convolutional neural networks and vision transformers to accurately distinguish between authentic and synthetic media. The methodology employs systematic preprocessing and feature extraction techniques to enhance detection accuracy, which is crucial due to the increasing sophistication of deepfake generation methods. Experimental evaluations using benchmark datasets such as FaceForensics++, Celeb-DF, and the DeepFake Detection Challenge dataset demonstrate promising results, achieving up to 97.2% accuracy. The framework is designed to be computationally efficient and scalable for real-world applications in cybersecurity and media forensics. This study contributes to ongoing research by evaluating hybrid classifier architectures and establishing a baseline for future exploration in deepfake detection.</p> |
| title | DEEPFAKE DETECTION USING MACHINE LEARNING |
| topic | Deepfake Detection Machine Learning CNN Vision Transformer Digital Forensics |
| url | https://doi.org/10.5281/zenodo.17660017 |