PREMIER Demonstrator

This demonstrator implements four video forensic tools denoted as PremierNet1, PremierNet2, and PremierNet3 (see details), which provide an authenticity prediction and an attention map for each frame.
By clicking on one of the buttons below, you can choose which type of videos you want to analyze:

Short videos: a grid will appear depicting 2-second long videos, which can be analyzed in real time with the preferred tools.
Long videos: two grids will appear depicting real and fake videos with longer duration (above 10 seconds). Due to the computational time required for those clips, pre-computed results obtained with both tools are displayed.
Custum video: you will be able to upload your own video, which can be analyzed in real time with the preferred tools.

References

[PremierNet1]	N. Bonettini, E. D. Cannas, S. Mandelli, L. Bondi, P. Bestagini and S. Tubaro, "Video Face Manipulation Detection Through Ensemble of CNNs," IEEE International Conference on Pattern Recognition (ICPR), 2021, pp. 5012-5019, doi: 10.1109/ICPR48806.2021.9412711.
[PremierNet2]	A. Rössler, et al. "Faceforensics++: Learning to detect manipulated facial images." Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019. Please refer to this repository
[PremierNet3]	The method is an ensamble of four EfficientNetB7, trained on FaceForensics++, and the avereging of Class Activation Maps (CAMs) is provided.
[Videos]	A. Rössler, et al. "Faceforensics++: Learning to detect manipulated facial images." Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019. Please refer to this repository B. Dolhansky, et al. "The deepfake detection challenge (dfdc) preview dataset." arXiv preprint arXiv:1910.08854 (2019).