Identifying the GANs generated image.
Programming Language:
- Python 3.6
Libraries needed to Run this project:
- tensorlow-1.13.0
- opencv-python
- keras-2.3.0
- matplotlib
- os
- mtcnn
- pillow
Required model architecture and weights files:
you can download the weights and model architecture file from here: https://drive.google.com/open?id=1KujeIcVzoRWHgBlO-BzpClxhujUbeDmV
Dataset:
Dataset for training the model are available here: https://drive.google.com/file/d/1smby8vBB0g8bNtUsQ10OdjF-4FK9k_GS/view
Instructions to train the model:
- Dataset which is provided in the link is already preprocessed and it contains images which has only faces i.e. MTCNN is already applied.
- You need to download the dataset and just run the model training section since face detection part is already done.
- give the path of folders into the ImageDataGenerator and run the model it will start training.
Insutuctions to test the model:
- You will to download the model.json and model.h5 file and will have to give the path of these files in the testing of the notebook.
- Then choose the image on which you want to make the prediction give that imaeg path in the testing section and predictions will be generated.
References:
https://medium.com/@gouravbais08/fake-face-image-classification-91a6225e708d
<script src="https://gist.github.com/gouravsinghbais/656938bc2ae8818c10cc39982405fecd.js?file=gist.md"></script>