This project is a comparative study on the performance of multiple transfer learning models in deepfake detection. The dataset used in this project is split into a training set of 100k images, a test set of 20k images, and a validation set of 20k images.
The following transfer learning models were tried for deepfake detection:
- InceptionResNetV2
- InceptionV3
- MobileNet
- MobileNetV2
- VGG16
- VGG19
- Xception
- EfficientNetB0
- EfficientNetB1
- EfficientNetB2
- EfficientNetB3
- EfficientNetB4
- EfficientNetB5
- EfficientNetB6
- EfficientNetB7
- ResNet50
- ResNet101
- ResNet152
- ResNet50V2
- ResNet101V2
- ResNet152V2
- EfficientNetV2B0
- EfficientNetV2B1
- EfficientNetV2B2
- EfficientNetV2B3
- DenseNet121
- DenseNet169
- DenseNet201
Here are the results
The project requires Python 3.5 or later and the following libraries:
- NumPy
- Tensorflow