https://github.com/louishenrifranc/paintforme

Simple Android app to transform picture to similar drawing
https://github.com/louishenrifranc/paintforme

android deeplearning tensorflow vgg19

Last synced: 5 months ago
JSON representation

Simple Android app to transform picture to similar drawing

• Host: GitHub
• URL: https://github.com/louishenrifranc/paintforme
• Owner: louishenrifranc
• Created: 2017-01-22T15:07:57.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2017-02-05T11:44:25.000Z (over 8 years ago)
• Last Synced: 2025-03-31T19:21:19.514Z (7 months ago)
• Topics: android, deeplearning, tensorflow, vgg19
• Language: Java
• Homepage:
• Size: 58.8 MB
• Stars: 7
• Watchers: 3
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
> Do you feel a lack of creativity while drawing sometimes? 

> Stuck at a museum in front of La Joconde, ever wanted to find closely looking alike painting?  

> Just bored about life, make your routine panorama looking alike a painting?  

__Download the PaintForMe app__

# What for

* Take a picture with your phone, or search on on your local drive. 

* Send the picture to the server, and wait for paintings which are similar in the style of the initial sended picture.

http://www.inference.vc/content/images/2016/01/9k-.jpg

# Examples

* Example with a picture of building:  

![](https://github.com/ConUHacks/image_to_painting/blob/master/demos/demo1.gif)

* Example with a picture of Venise:  

![](https://github.com/ConUHacks/image_to_painting/blob/master/demos/demo2.gif)

* Example with a garden picture:  

![](https://github.com/ConUHacks/image_to_painting/blob/master/demos/demo3.gif)

* Example with La Joconde:  

![](https://github.com/ConUHacks/image_to_painting/blob/master/demos/demo4.gif)  

_(still training the accuracy)_

# Usage

## Play with the app

* Get a Firebase Account, setup your 'google-services.json' and build a new project

* or, use the apk provided

## Play with the server

To reproduce the results:  

1. Download the dataset at ![kaggle.com](https://www.kaggle.com/c/painter-by-numbers/data)  

2. Download the neural networks weights at ![google drive](https://drive.google.com/file/d/0Bz7KyqmuGsilZ2RVeVhKY0FyRmc/view). I used this deep neural network architecture [vgg-19](https://arxiv.org/abs/1409.1556)

3. Compute the vector representation for painting using ```script_python/model.py```  

4. Set Firebase config credentials in the ```backend.py```. For more information about the api, I used ![pyrebase](https://github.com/thisbejim/Pyrebase#database)  

5. Launch the server with ```python backend.py```  

# Visualisation of the latent space

Visualisation of a set of pictures using PCA algorithm and tensorboard Embedding

![](https://github.com/ConUHacks/image_to_painting/blob/master/demos/visualisation.gif)