sketch_to_image

Using CGAN we have generated Realistic Faces from sketches, facades to buildings etc.

How to Use

Get datasets
Copy the train-model.py in the directory of the data-set that you want to train.
Run the train-model.py after installing all the required dependencies.
Copy the final .h5 files generated in the webapp/models directory, named using the following nomenclature -
- facades2buildings_discriminator.h5
- facades2buildings_generator.h5
- maps2buildings_discriminator.h5
- maps2buildings_generator.h5
- scapes2city_discriminator.h5
- scapes2city_generator.h5
- sketch2faces_discriminator.h5
- sketch2faces_generator.h5
Run the webapp/app.py file to start the flask app to run your models and provide input and generate output using a GUI.

Description of the included files

sketch.py was used to convert images to sketches to augment the training data.
merge.py was used to merge the ground truth and the input image, as expected by the Neural Network.

Example Output