Enhancing Deepfake Detection through Hybrid MobileNet-LSTM Model with Real-Time Image and Video Analysis

Chinna Venkataswamy, Pabbathi Jacob Vijaya Kumar, Bavanam Yeswanth Kumar Reddy, Mangali Suri Babu, Neelam Venkatesh, Vuluvala Pavasubralali Reddy

2025

Abstract

Deepfake technology is spreading false information and posing a threat to digital security. A Hybrid MobileNetLSTM Model for real-time deepfake detection in videos and images is presented in this project. MobileNet, a lightweight CNN, extracts spatial features, and LSTM records temporal dependencies in video sequences, both of which guarantee high detection accuracy while utilizing minimal computation. Transfer learning is used to train the model on large datasets of real and fake media for better generalization. Heatmaps and probability scores that can be interpreted are provided by the system, which is integrated with OpenCV and TensorFlow. TensorFlow Lite is used to optimize real-time inference, making it possible to use it on mobile and edge devices. On live video feeds, realtime performance is validated, and experimental evaluations demonstrate superior accuracy to traditional CNN-based approaches. This scalable detection system provides support for media forensics, social media verification, and digital security. GAN-based adversarial training, which will bridge the gap between high accuracy and real-time deployment, will be one of the future enhancements. This will make the system more resistant to changing deepfake techniques.

Paper Citation

in Harvard Style

Venkataswamy C., Kumar P., Reddy B., Babu M., Venkatesh N. and Reddy V. (2025). Enhancing Deepfake Detection through Hybrid MobileNet-LSTM Model with Real-Time Image and Video Analysis. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 394-399. DOI: 10.5220/0013930400004919

in Bibtex Style

@conference{icrdicct`2525,
author={Chinna Venkataswamy and Pabbathi Kumar and Bavanam Reddy and Mangali Babu and Neelam Venkatesh and Vuluvala Reddy},
title={Enhancing Deepfake Detection through Hybrid MobileNet-LSTM Model with Real-Time Image and Video Analysis},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={394-399},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013930400004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Enhancing Deepfake Detection through Hybrid MobileNet-LSTM Model with Real-Time Image and Video Analysis
SN - 978-989-758-777-1
AU - Venkataswamy C.
AU - Kumar P.
AU - Reddy B.
AU - Babu M.
AU - Venkatesh N.
AU - Reddy V.
PY - 2025
SP - 394
EP - 399
DO - 10.5220/0013930400004919
PB - SciTePress