GitHub - kapilgarg2004/VisionGuard: A deepfake detection system using CNN, ResNet50, VGG19, and VGG16+LSTM was tested, with the hybrid model performing best (96.79% training accuracy, 95.96% validation accuracy, 0.0854 loss). By combining spatial and temporal features, it outperforms existing models, enhancing digital media authenticity.

kapilgarg2004 / VisionGuard Public

Notifications You must be signed in to change notification settings
Fork 0
Star 0

A deepfake detection system using CNN, ResNet50, VGG19, and VGG16+LSTM was tested, with the hybrid model performing best (96.79% training accuracy, 95.96% validation accuracy, 0.0854 loss). By combining spatial and temporal features, it outperforms existing models, enhancing digital media authenticity.

0 stars 0 forks Branches Tags Activity

Star

Notifications

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
__pycache__		__pycache__
Deepfake.py		Deepfake.py
api.py		api.py
app.py		app.py
dataset-LENOVO_IDEAPAD.py		dataset-LENOVO_IDEAPAD.py
dataset.py		dataset.py
dataset_torch.py		dataset_torch.py
dataset_torch_new.py		dataset_torch_new.py
dataset_torch_new.py.bak		dataset_torch_new.py.bak
evaluate-LENOVO_IDEAPAD.py		evaluate-LENOVO_IDEAPAD.py
evaluate.py		evaluate.py
evaluate_torch.py		evaluate_torch.py
metrics.py		metrics.py
model_architecture.py		model_architecture.py
model_torch.py		model_torch.py
plot_learning_curves.py		plot_learning_curves.py
prediction.py		prediction.py
requirements.txt		requirements.txt
roc_curve.py		roc_curve.py
server.py		server.py
temp.py		temp.py
train-LENOVO_IDEAPAD.py		train-LENOVO_IDEAPAD.py
train.py		train.py
train_torch.py		train_torch.py
training_log.csv		training_log.csv