https://github.com/gsurma/image_generator

DCGAN image generator 🖼️.
https://github.com/gsurma/image_generator

dcgan dcgan-tensorflow deep-learning gan generative-adversarial-network jupyter machine-learning neural-network python python3 simpsons simpsons-dataset tensorflow

Last synced: about 3 hours ago
JSON representation

DCGAN image generator 🖼️.

• Host: GitHub
• URL: https://github.com/gsurma/image_generator
• Owner: gsurma
• License: mit
• Created: 2019-01-20T02:48:34.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2021-07-09T08:52:13.000Z (over 3 years ago)
• Last Synced: 2023-10-20T20:07:11.925Z (about 1 year ago)
• Topics: dcgan, dcgan-tensorflow, deep-learning, gan, generative-adversarial-network, jupyter, machine-learning, neural-network, python, python3, simpsons, simpsons-dataset, tensorflow
• Language: Jupyter Notebook
• Homepage: https://gsurma.github.io
• Size: 17.5 MB
• Stars: 214
• Watchers: 7
• Forks: 82
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • Funding: .github/FUNDING.yml
  • License: LICENSE

Awesome Lists containing this project

README

        


  



# Image Generator

Python notebook containing TensorFlow DCGAN implementation. It was trained on a [Simpsons Faces](https://www.kaggle.com/kostastokis/simpsons-faces) dataset.




Check out corresponding Kaggle kernel: [Image Generator](https://www.kaggle.com/greg115/image-generator-dcgan-the-simpsons-dataset).

Check out corresponding Medium article:

[Image Generator - Drawing Cartoons with Generative Adversarial Networks](https://towardsdatascience.com/image-generator-drawing-cartoons-with-generative-adversarial-networks-45e814ca9b6b)



  



## DCGAN

Network architecture by [Radford et al., 2015](https://arxiv.org/abs/1511.06434).



## Training

Visualization of training with the following hyperparameteres.



	IMAGE_SIZE = 128

	NOISE_SIZE = 100

	LR_D = 0.00004

	LR_G = 0.0004

	BATCH_SIZE = 64

	EPOCHS = 300

	BETA1 = 0.5

	WEIGHT_INIT_STDDEV = 0.02

	EPSILON = 0.00005

## Results

Cherry-picked generated samples.



As expected, there were some funny-looking malformed faces as well.



## Author

**Greg (Grzegorz) Surma**

[**PORTFOLIO**](https://gsurma.github.io)

[**GITHUB**](https://github.com/gsurma)

[**BLOG**](https://medium.com/@gsurma)