Convert RGB and HSV in a differentiable form with PyTorch

I couldn't find it unexpectedly, so it is written in HSV Color Space-Wikipedia. Implemented based on the expression.

If you just want to convert RGB and HSV, you can use other libraries (PIL, OpenCV, etc.), but you can't differentiate with that, so you can't incorporate it inside the neural network. So I wrote a function to convert RGB and HSV to each other using only PyTorch functions.

code

color_convert.py

import torch


def rgb2hsv(input, epsilon=1e-10):
    assert(input.shape[1] == 3)

    r, g, b = input[:, 0], input[:, 1], input[:, 2]
    max_rgb, argmax_rgb = input.max(1)
    min_rgb, argmin_rgb = input.min(1)

    max_min = max_rgb - min_rgb + epsilon

    h1 = 60.0 * (g - r) / max_min + 60.0
    h2 = 60.0 * (b - g) / max_min + 180.0
    h3 = 60.0 * (r - b) / max_min + 300.0

    h = torch.stack((h2, h3, h1), dim=0).gather(dim=0, index=argmin_rgb.unsqueeze(0)).squeeze(0)
    s = max_min / (max_rgb + epsilon)
    v = max_rgb

    return torch.stack((h, s, v), dim=1)


def hsv2rgb(input):
    assert(input.shape[1] == 3)

    h, s, v = input[:, 0], input[:, 1], input[:, 2]
    h_ = (h - torch.floor(h / 360) * 360) / 60
    c = s * v
    x = c * (1 - torch.abs(torch.fmod(h_, 2) - 1))

    zero = torch.zeros_like(c)
    y = torch.stack((
        torch.stack((c, x, zero), dim=1),
        torch.stack((x, c, zero), dim=1),
        torch.stack((zero, c, x), dim=1),
        torch.stack((zero, x, c), dim=1),
        torch.stack((x, zero, c), dim=1),
        torch.stack((c, zero, x), dim=1),
    ), dim=0)

    index = torch.repeat_interleave(torch.floor(h_).unsqueeze(1), 3, dim=1).unsqueeze(0).to(torch.long)
    rgb = (y.gather(dim=0, index=index) + (v - c)).squeeze(0)
    return rgb

How to use

Both rgb2hsv and hsv2rgb take images in a mini-batch of images (NCHW format) as input. The range of H (hue) is from 0 to 360 (the outside of the range loops). The range of RGB and SV is 0 to 1.