Photo Coloring Using End2end CNN based Model!
Colouring photos could bring you a lot of fun to your daily life. Just imagine if you coloured your grandma’s old photo and showed it to your family, how surprised and happy they would be.
Colouring a picture with Photoshop is a so exhausting process, so we implemented this Photo coloring tool, which is based on Let there be Color, it's Deep Learning based on CNN, it can color any grayscale picture with any color scale and picture resolution.
📖 Documentation
How to Run
Install the requirements:
pip3 install -r requirements.txt Running the web app locally
python3 WebApp/app.pyThen the endpoint will be running on http://127.0.0.1:5000/
How to use the Pre-Trained model
- Download the Pre-trained model from here: https://www.dropbox.com/s/j9sehdm7fs65wax/model.zip?dl=0
- Unzip the file into
algorithm/Models/ - Run the Run the Endpoint again
Results
The results are not bad at all! a lot of test cases gonna be so realistic, but the model still needs more training
| Original Image (grayscale) | Resultant Image (colored) |
|---|---|
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Paper
This project is an implementation of the Let there be Color, published 2016.
Dataset
- Dataset used is MIT Places Scence Dataset. Sample of the data used
- The pre-trained model is trained only on Landscapes, seas and gardens pictures.
Model Used
Experiments
References & Tutorials
Useful materials
- Introduction to the problem: http://tinyclouds.org/colorize/
- More details about the solution: http://hi.cs.waseda.ac.jp/~iizuka/projects/colorization/en/
Other related projects
- https://github.com/pavelgonchar/colornet
- https://github.com/satoshiiizuka/siggraph2016_colorization
- https://github.com/OmarSayedMostafa/Deep-learning-Colorization-for-visual-media
Tutorials
- CNN: http://cs231n.github.io/convolutional-networks/
- CNN: https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
- CNN: https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/
- CNN: https://www.youtube.com/watch?v=Gu0MkmynWkw
- CNN: https://www.youtube.com/watch?v=2-Ol7ZB0MmU
- CNN: https://www.youtube.com/watch?v=s716QXfApa0
- Image Colorization: https://www.youtube.com/watch?v=2IWhh1gd_p0
- IMage Colorization: https://www.youtube.com/watch?v=S1HE9SpZl8A
- Image Colorization: https://blog.floydhub.com/colorizing-b-w-photos-with-neural-networks/
- Image Colorization: https://www.youtube.com/watch?v=tknFQULAAs0
- Image Colorization: https://www.youtube.com/watch?v=KO7W0Qq8yUE
- Deep learning: https://www.youtube.com/watch?v=He4t7Zekob0
Shortcuts
- YUV: https://en.wikipedia.org/wiki/YUV
- CNN: https://en.wikipedia.org/wiki/Convolutional_neural_network
CNN practice
- https://becominghuman.ai/building-an-image-classifier-using-deep-learning-in-python-totally-from-a-beginners-perspective-be8dbaf22dd8
- http://cv-tricks.com/tensorflow-tutorial/training-convolutional-neural-network-for-image-classification/
- http://richzhang.github.io/colorization/
- https://github.com/microic/niy/tree/master/examples/colorizing_photos
- https://www.kaggle.com/preslavrachev/wip-photo-colorization-using-keras/notebook
- https://blog.floydhub.com/colorizing-b-w-photos-with-neural-networks/
- https://gist.github.com/standarderror/43582e9a15038806da8a846903438ebe
- https://github.com/coolioasjulio/Cat-Dog-CNN-Classifier
- https://gist.github.com/standarderror/43582e9a15038806da8a846903438ebe
- https://becominghuman.ai/building-an-image-classifier-using-deep-learning-in-python-totally-from-a-beginners-perspective-be8dbaf22dd8 https://medium.com/@parthvadhadiya424/hello-world-program-in-keras-with-cnn-dog-vs-cat-classification-efc6f0da3cc5
Other related papers
Useful Articles
- https://hackernoon.com/colorising-black-white-photos-using-deep-learning-4da22a05f531
- http://whatimade.today/two-weeks-of-colorizebot-conclusions-and-statistics/
Video colorization
Future Work
- Training, Training and more Training
- Train on more data clusters
- Extend the project to be able to colorize Videos