This Repository is a Generative Modelling Project.
Author of this Project: Rahul Golder
Project Created as part of the Kaggle Competetion: I'm Something of a Painter Myself.
About the Project: The model ArtistGAN will convert a real-domain image to a monet style painting domain image through a domain transfer feature mapping.
Motivation of Modelling Architecture: CycleGAN Model for Domain Features Transfer.
Getting Started with the Project:
- Run the following code for to import os module and also create a content directory.
import os
os.chdir("/content")
- Run the code to clear the ArtistGAN directory if it exists.
!rm -r /content/ArtistGAN
- Run the code for cloning the Project Repository.
!git clone https://github.com/Goldenprince8420/ArtistGAN.git
- Set Artist Directory as the current directory.
- Run the following script
kaggle.sh
for Kaggle Setup and Data downloads.
!bash kaggle.sh
- Unzip the Training Weights.
! unzip cyclegan-training-weights.zip -d weights
- Run the script file
run.sh
for running the model.
!bash run.sh
The Programme will run with default setup of the model running parameters. It may take 5 to 10 minutes.