Skip to content
line drawing colorization using chainer
Branch: master
Clone or download
Pull request Compare This branch is 8 commits ahead, 186 commits behind pfnet:master.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
cgi-bin/paint_x2_unet fix to work without GPU (-g -1) Jan 28, 2017
log initial commit for open Jan 27, 2017
static
2010_06.png add sample picture Jan 31, 2017
2010_06_correcte_by_changing_save_as_img.jpg add colorized sample picture with correted source code save_as_img() Jan 31, 2017
2010_06_new_BGR2YUV.JPG add colorized sample picture with change COLOR_RGB2YUV to COLOR_BGR2Y… Jan 31, 2017
2010_06_new_changed_color_RGB-2-BGR-by-IrfanView.JPG add colorized sample picture with change colors RGB->BGR using IrfanView Jan 31, 2017
IMG_6731.jpg change edge-detection tool a little Feb 4, 2017
README.md update README Jan 27, 2017
clouds.jpg add files for color test Feb 4, 2017
cv2_demo_cvt.py add files for color test Feb 4, 2017
edge-detection.ipynb change edge-detection tool a little Feb 4, 2017
server.py initial commit for open Jan 27, 2017

README.md

Paints Chainer

Paints Chainer is line drawing colorizer using chainer. Using CNN, you can colorize your scketch automatically / semi-automatically .

image

Requirement

  • Chainer
  • CUDA / cudnn

How to boot UI

UI is html based. using wPaint.js Server side is very basic python server

boot local server python server.py

access to localhost localhost:8000/static/

Learning

main code of colorization is in cgi-bin/paint_x2_unet

to train 1st layer using GPU 0 python train_128.py -g 0 to train 2nd layer using GPU 0 python train_x2.py -g 0

DEMO

http://paintschainer.preferred.tech/

Pre-Trained Models

Download following model files to cgi-bin/paint_x2_unet/models/

http://paintschainer.preferred.tech/downloads/

You can’t perform that action at this time.