Skip to content

Latest commit

 

History

History
84 lines (61 loc) · 3.72 KB

README.md

File metadata and controls

84 lines (61 loc) · 3.72 KB

Chainer implementation of "Perceptual Losses for Real-Time Style Transfer and Super-Resolution"

Fast artistic style transfer by using feed forward network.

checkout resize-conv branch which provides better result.

  • input image size: 1024x768
  • process time(CPU): 17.78sec (Core i7-5930K)
  • process time(GPU): 0.994sec (GPU TitanX)

Requirement

$ pip install chainer

Prerequisite

Download VGG16 model and convert it into smaller file so that we use only the convolutional layers which are 10% of the entire model.

sh setup_model.sh

Train

Need to train one image transformation network model per one style target. According to the paper, the models are trained on the Microsoft COCO dataset.

python train.py -s <style_image_path> -d <training_dataset_path> -g <use_gpu ? gpu_id : -1>

Generate

python generate.py <input_image_path> -m <model_path> -o <output_image_path> -g <use_gpu ? gpu_id : -1>

This repo has pretrained models as an example.

  • example:
python generate.py sample_images/tubingen.jpg -m models/composition.model -o sample_images/output.jpg

or

python generate.py sample_images/tubingen.jpg -m models/seurat.model -o sample_images/output.jpg

Transfer only style but not color (--keep_colors option)

python generate.py <input_image_path> -m <model_path> -o <output_image_path> -g <use_gpu ? gpu_id : -1> --keep_colors

A collection of pre-trained models

Fashizzle Dizzle created pre-trained models collection repository, chainer-fast-neuralstyle-models. You can find a variety of models.

Difference from paper

  • Convolution kernel size 4 instead of 3.
  • Training with batchsize(n>=2) causes unstable result.

No Backward Compatibility

Jul. 19, 2016

This version is not compatible with the previous versions. You can't use models trained by the previous implementation. Sorry for the inconvenience!

License

MIT

Reference

Codes written in this repository based on following nice works, thanks to the author.

  • chainer-gogh Chainer implementation of neural-style. I heavily referenced it.
  • chainer-cifar10 Residual block implementation is referred.