Generative-Jamdani

Based on pix2pix-tensorflow which is tensorflow port of pix2pix by Isola et al. Visit Jamdani Artist to see the live demonstration.

Interactive Demo

Sample output generated by our Generative-Jamdani from input sketches, like these examples from the original paper:

This is an implementation of pix2pix-tensorflow on Jamdani Noksha dataset. It is meant to be a faithful implementation of the original work and so does not add anything.

Setup

Prerequisites

• Tensorflow 1.4.1

Recommended

• Linux with Tensorflow GPU edition + cuDNN

Getting Started

# clone this repo
git clone https://github.com/raihan-tanvir/generative-jamdani.git
cd generative-jamdani

python pix2pix.py \
  --mode train \
  --output_dir model/ \
  --max_epochs 150 \
  --input_dir dataset/train \
  --which_direction BtoA
# test the model
python pix2pix.py \
  --mode test \
  --output_dir output/ \
  --input_dir dataset/test \
  --checkpoint model/

The test run will output an HTML file at output/index.html that shows input/output/target image sets.

Datasets

The data format used by this program is the same as the original pix2pix format, which consists of images of input and desired output side by side like:

For example:

Datasets Link

dataset	example
Boundary Version 1116 images
Enhanched Resolution Version 1983 images
Skeleton Version 7932 images
Reduced Version 913 images
Sketch Version 910 images

Or download the entire Jamdani Noksha dataset from here

Citation

@INPROCEEDINGS{9392654,
  author={M. T. R. {Shawon} and R. {Tanvir} and H. F. {Shifa} and S. {Kar} and M. I. {Jubair}},
  booktitle={2020 23rd International Conference on Computer and Information Technology (ICCIT)}, 
  title={Jamdani Motif Generation using Conditional GAN}, 
  year={2020},
  volume={},
  number={},
  pages={1-6},
  doi={10.1109/ICCIT51783.2020.9392654}}

Acknowledgments

This is a implementation of pix2pix-tensorflow on Jamdani Noksha dataset. Thanks to the Tensorflow Affinelayer for making such a wonderful port! And special thanks to Phillip Isola for the original pix2pix.