Towards a better classification of deepfake videos

• Matheus P. R. Vieira UFOP
• Arthur Negrão de F. M. C. UFOP
• Guilherme A. L. Silva UFOP
• Eduardo Luz UFOP
• Pedro Silva UFOP

DOI: https://doi.org/10.5753/eniac.2025.13951

Abstract

The increasing accessibility and realism of deepfake technology pose serious threats to information integrity, privacy, and public trust. In this work, we propose a deep learning model for detecting deepfake videos based on short video segments. The architecture combines a TimeDistributed wrapper over an Xception backbone with global average pooling and a dense classification layer. To enhance performance, a Bayesian hyperparameter optimization strategy was employed, tuning both architectural and training parameters. The proposed model was evaluated on the Celeb-DF-V2 dataset, using only 24 frames per video, approximately one second of content. Despite this constraint, the model achieved a competitive accuracy of 97.30% and an AUC of 0.9957, outperforming several existing approaches that rely on longer video sequences. These results demonstrate the feasibility of detecting deepfakes efficiently using shorter clips, suggesting a viable direction for real-time and resource-constrained applications.

References

Alanazi, S. and Asif, S. (2024). Exploring deepfake technology: creation, consequences and countermeasures. Human-Intelligent Systems Integration, 6(1):49–60.

Chen, H. and Magramo, K. (2024). Finance worker pays out $25 million after video call with deepfake chief financial officer.

Chen, H.-S., Rouhsedaghat, M., Ghani, H., Hu, S., You, S., and Kuo, C. C. J. (2021). Defakehop: A light-weight high-performance deepfake detector.

de Lima, O., Franklin, S., Basu, S., Karwoski, B., and George, A. (2020). Deepfake detection using spatiotemporal convolutional networks.

Frazier, P. I. (2018). A tutorial on bayesian optimization.

Heidari, A., Jafari Navimipour, N., Dag, H., and Unal, M. (2024). Deepfake detection using deep learning methods: A systematic and comprehensive review. WIREs Data Mining and Knowledge Discovery, 14(2):e1520.

Heo, Y.-J., Yeo, W.-H., and Kim, B.-G. (2023). Deepfake detection algorithm based on improved vision transformer. Applied Intelligence, 53(7):7512–7527.

Hernandez-Ortega, J., Tolosana, R., Fierrez, J., and Morales, A. (2022). DeepFakes Detection Based on Heart Rate Estimation: Single-and Multi-frame, pages 255–273. Springer International Publishing, Cham.

Khormali, A. and Yuan, J.-S. (2021). Add: Attention-based deepfake detection approach. Big Data and Cognitive Computing, 5(4).

Kirby, P. (2023). Muharrem ince: Turkish candidate dramatically pulls out before election.

Le, T.-N., Nguyen, H. H., Yamagishi, J., and Echizen, I. (2021). Openforensics: Large-scale challenging dataset for multi-face forgery detection and segmentation in-the-wild.

Li, Y., Yang, X., Sun, P., Qi, H., and Lyu, S. (2020). Celeb-df: A large-scale challenging dataset for deepfake forensics.

Masud, U., Sadiq, M., Masood, S., Ahmad, M., and El-Latif, A. A. A. (2023). Lw-deepfakenet: a lightweight time distributed cnn-lstm network for real-time deepfake video detection. Signal, Image and Video Processing, 17(8):4029–4037.

Mirsky, Y. and Lee, W. (2021). The creation and detection of deepfakes: A survey. ACM Computing Surveys, 54(1):1–41.

O’Malley, T., Bursztein, E., Long, J., Chollet, F., Jin, H., Invernizzi, L., et al. (2019). Kerastuner. [link].

Reis, P. M. G. I. and Ribeiro, R. O. (2024). A forensic evaluation method for deepfake detection using dcnn-based facial similarity scores. Forensic Science International, 358:111747.

Serengil, S. I. and Ozpinar, A. (2020). Lightface: A hybrid deep face recognition framework. In 2020 Innovations in Intelligent Systems and Applications Conference (ASYU), pages 23–27. IEEE.

Singh, A., Saimbhi, A. S., Singh, N., and Mittal, M. (2020). Deepfake video detection: A time-distributed approach. SN Computer Science, 1(4):212.

Wilcox, L. M., Allison, R. S., Helliker, J., Dunk, B., and Anthony, R. C. (2015). Evidence that viewers prefer higher frame-rate film. 12(4).

Yang, K., Liu, L., and Wen, Y. (2024). The impact of bayesian optimization on feature selection. Scientific Reports, 14(1):3948.

Yermakov, A., Cech, J., and Matas, J. (2025). Unlocking the hidden potential of clip in generalizable deepfake detection.

Zhang, A., Lipton, Z. C., Li, M., and Smola, A. J. (2023). Dive into Deep Learning. Cambridge University Press. [link].

Łabuz, M. and Nehring, C. (2024). On the way to deep fake democracy? deep fakes in election campaigns in 2023. European Political Science.