Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201446,
        author = {Alwin and Pavitra Premanand and Vijay SP and Kanishka S and Anjana R and N Vishnu Venkatesh},
        title = {REAL-TIME DEEPFAKE DETECTION ON EMBEDDED SYSTEMS: A PORTABLE DEVICE ARCHITECTURE},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {4567-4571},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201446},
        abstract = {The rapid proliferation of hyper-realistic deepfake technology poses severe threats to digital security, public trust, and individual privacy. While high-performance server-side detection models exist, there is a critical gap in localized, real-time verification tools for field investigators. This paper proposes a portable deepfake detection framework optimized for embedded mobile hardware. Utilizing a MobileNetV2 architecture fine-tuned via a dual-phase transfer learning regimen on the Deepfake Detection Challenge (DFDC), Celeb-DF, and FaceForensics++ datasets, the system focuses on identifying high-frequency boundary artifacts and spatial inconsistencies. To ensure efficiency on resource-constrained devices, the model was converted to an INT8-quantized TensorFlow Lite (TFLite) format, reducing the model size from 9MB to 2.3MB and peak RAM usage to 12MB. The resulting system, deployed on a Samsung Galaxy A31, achieves an ROC AUC of 0.8718 with an inference latency of 100–200ms per frame. The framework consists of a native Android application for immediate field verification and a Python-based interface for detailed forensic analysis, effectively bridging the gap between theoretical deep learning and practical digital forensic requirements.},
        keywords = {Deepfake Detection, MobileNetV2, Embedded Systems, INT8 Quantization, Digital Forensics, TensorFlow Lite.},
        month = {May},
        }