Chainer implementation of "Perceptual Losses for Real-Time Style Transfer and Super-Resolution"
Fast artistic style transfer by using feed forward network.
- input image size: 1024x768
- process time(CPU): 17.78sec (Core i7-5930K)
- process time(GPU): 0.994sec (GPU TitanX)
Differences from original
- default
--image_size
set to 512 (original uses 256) - ability to switch off dataset cropping with
--fullsize
option. Crops by default, preserving aspect ratio - cropping implementation uses
ImageOps.fit
, which always scales and crops, whereas original uses custom solution, which upscales the image if it's smaller than--image_size
, otherwise just crops without scaling - bicubic and Lanczos resampling when scaling dataset and input style images accordingly provides sharper shrinking, whereas original uses nearest neighbour
Video Processing
The repo includes a bash script to transform your videos. It depends on ffmpeg. Compilation instructions
./genvid.sh input_video output_video model start_time duration
The first three arguments are mandatory and should contain path to files.
The last two are optional and indicate starting position and duration in seconds.
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 0
Generate
python generate.py <input_image_path> -m <model_path> -o <output_image_path>
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
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.