隨著深度學習的快速發展，深度偽造技術也越來越進步了，因此有大量的深度偽造影片在網路上流傳引起熱烈的討論。然而，有些影片的不良用途例如色情，政治誤導，惡作劇都造成許多社會問題並且影響他人的名譽，所以偵測出深度偽造影片成了現今重要的任務。
現今有許多深度偽造偵測模型都面臨著過擬合的問題。因此，本篇論文提出了一個改善模型泛化能力的深度偽造偵測演算法。首先，使用 EfficientNet 作為模型的骨幹，然後加入注意力機制使模型專注於重要信息。最後，遮蓋圖像被設計作為資料擴增加入訓練。所提出方法的訓練集為 FaceForensics++ ，並且測試於 FaceForensics++、Celeb-DF、DFDC 測試集上。實驗結果顯示，本論文提出的方法在比較的方法中擁有更好的泛化能力。

With the rapid development of deep learning, deepfake technology has also improved. Therefore, many deepfake videos are circulating on the Internet and arousing heated discussions. However, some malicious uses of videos, such as pornography, political misleading, and hoaxes, cause many social problems and affect the reputation of others. Therefore, detecting deepfake videos has become an important task nowadays. Nowadays, many deepfake detection models are facing the problem of overfitting. Therefore, this Thesis proposes a deepfake detection algorithm that improves the generalization ability of the model. Firstly, EfficientNet is used as the backbone of the model, then an attention mechanism is added to make the model focus on important information. Finally, the mask image is designed as data augmentation for training. The training set of the proposed method is FaceForensics++, and it is tested on the FaceForensics++, Celeb-DF, and DFDC testing sets. The experimental results show that the proposed method has better generalization ability among the compared methods.

References
[1] Tolosana, R., et al., Deepfakes and beyond: A survey of face manipulation and fake detection. Information Fusion, 2020. 64: p. 131-148.
[2] Li, Y., et al. Celeb-df: A large-scale challenging dataset for deepfake forensics. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.
[3] Rossler, A., et al. Faceforensics++: Learning to detect manipulated facial images. in Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019.
[4] Choi, Y., et al. Stargan: Unified generative adversarial networks for multi-domain image-to-image translation. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
[5] Bondi, L., et al., Training Strategies and Data Augmentations in CNN-based DeepFake Video Detection. arXiv preprint arXiv:2011.07792, 2020.
[6] Bourlard, H. and Y. Kamp, Auto-association by multilayer perceptrons and singular value decomposition. Biological cybernetics, 1988. 59(4): p. 291-294.
[7] Hinton, G.E. and R.R. Salakhutdinov, Reducing the dimensionality of data with neural networks. science, 2006. 313(5786): p. 504-507.
[8] Goodfellow, I.J., et al., Generative adversarial networks. arXiv preprint arXiv:1406.2661, 2014.
[9] Li, L., et al., Faceshifter: Towards high fidelity and occlusion aware face swapping. arXiv preprint arXiv:1912.13457, 2019.
[10] Nirkin, Y., Y. Keller, and T. Hassner. Fsgan: Subject agnostic face swapping and reenactment. in Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019.
[11] Rana, M.S. and A.H. Sung. Deepfakestack: A deep ensemble-based learning technique for deepfake detection. in 2020 7th IEEE International Conference on Cyber Security and Cloud Computing (CSCloud)/2020 6th IEEE International Conference on Edge Computing and Scalable Cloud (EdgeCom). 2020. IEEE.
[12] Kumar, A., A. Bhavsar, and R. Verma. Detecting deepfakes with metric learning. in 2020 8th International Workshop on Biometrics and Forensics (IWBF). 2020. IEEE.
[13] Li, L., et al. Face x-ray for more general face forgery detection. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.
[14] Zhao, H., et al., Multi-attentional deepfake detection. arXiv preprint arXiv:2103.02406, 2021.
[15] Tan, M. and Q. Le. Efficientnet: Rethinking model scaling for convolutional neural networks. in International Conference on Machine Learning. 2019. PMLR.
[16] Tan, M., et al. Mnasnet: Platform-aware neural architecture search for mobile. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019.
[17] Sandler, M., et al. Mobilenetv2: Inverted residuals and linear bottlenecks. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
[18] Hu, J., L. Shen, and G. Sun. Squeeze-and-excitation networks. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
[19] He, K., et al. Deep residual learning for image recognition. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[20] Zhang, K., et al., Joint face detection and alignment using multitask cascaded convolutional networks. IEEE Signal Processing Letters, 2016. 23(10): p. 1499-1503.
[21] Hou, Q., D. Zhou, and J. Feng, Coordinate attention for efficient mobile network design. arXiv preprint arXiv:2103.02907, 2021.
[22] King, D.E., Dlib-ml: A Machine Learning Toolkit. Journal of Machine Learning Research, 2009. 10: p. 1755-1758.
[23] Dolhansky, B., et al., The deepfake detection challenge (dfdc) preview dataset. arXiv preprint arXiv:1910.08854, 2019.
[24] Afchar, D., et al. Mesonet: a compact facial video forgery detection network. in 2018 IEEE International Workshop on Information Forensics and Security (WIFS). 2018. IEEE.
[25] Selvaraju, R.R., et al. Grad-cam: Visual explanations from deep networks via gradient-based localization. in Proceedings of the IEEE international conference on computer vision. 2017.