Color transfer between images

• Paper: Color Transfer between Images by Erik Reinhard, Michael Ashikhmin, Bruce Gooch and Peter Shirley.
• This is the re-implementation based on: https://github.com/chia56028/Color-Transfer-between-Images.
• The transfer process performed in numpy array instead of looping over pixel to improve performance.
• Support on pytorch and numpy.

Requirements

• numpy
• opencv-python>=4.7.0.72
• torch>=1.13.0

Usage

pip install color_transfer_py

Numpy usage

import cv2
from color_transfer import color_transfer

src = cv2.imread("examples/1/src.jpeg")
target = cv2.imread("examples/1/target.png")

out = color_transfer(src, target)

cv2.imwrite("examples/1/output.jpeg", out)

Pytorch usage

import torch
from color_transfer import color_transfer_pytorch

# In pytorch, you can run multiple images at once
# but images must have the same size

src = torch.rand(4, 3, 224, 224)
target = torch.rand(4, 3, 224, 224)

outs = color_transfer_pytorch(src, target)
# Outs shape: (4, 3, 224, 224), output range is [0, 1]

Transfer results

Source	Target	Output

Benchmark results

• The benchmark results can be reproduce by running python3 benchmark.py

This implementation	chia56028 implementation
0.02 second	1.337 second