afislonge/deepfake-detection-project

Train Xception Model after adjustments made to training data (LIME)

Opened this issue · 2 comments

isamia-can-code commented

Training the model after adjustments were made to the training data, using the best model as base model.

isamia-can-code commented

Model 2 generated with the following improvments:

Image

isamia-can-code commented

First model location: /content/drive/My Drive/checkpoints/Xception2/7/model_epoch_14_val_accuracy_0.9115.h5

Second (fine-tuned model) location: /content/drive/My Drive/checkpoints/Xception2/10/model_epoch_11_val_accuracy_0.9414.h5