Implementation:Kornia Kornia Hsv Conversion

Knowledge Sources	Kornia
Domains	Vision, Color_Processing
Last Updated	2026-02-09 15:00 GMT

Overview

This module provides differentiable conversion between RGB and HSV (Hue, Saturation, Value) color spaces.

Description

hsv.py is a module in the Kornia library's color subpackage implementing bidirectional RGB-to-HSV color space conversion. The rgb_to_hsv function computes Hue (in radians, range [0, 2*pi]), Saturation (range [0, 1]), and Value (range [0, 1]) from an RGB image. The conversion handles edge cases with an epsilon parameter for numerical stability. The hsv_to_rgb function converts HSV back to RGB using a sectored lookup approach. Both functions support arbitrary batch dimensions and produce differentiable outputs. Each function has a corresponding nn.Module wrapper (RgbToHsv and HsvToRgb).

Usage

Import this module when you need to separate or manipulate hue, saturation, and brightness independently, for example in color augmentation, color filtering, or artistic style transfer.

Code Reference

Source Location

• Repository: Kornia
• File: kornia/color/hsv.py
• Lines: 1-174

Signature

def rgb_to_hsv(image: torch.Tensor, eps: float = 1e-8) -> torch.Tensor: ...
def hsv_to_rgb(image: torch.Tensor) -> torch.Tensor: ...

class RgbToHsv(nn.Module):
    def __init__(self, eps: float = 1e-6) -> None: ...
    def forward(self, image: torch.Tensor) -> torch.Tensor: ...

class HsvToRgb(nn.Module):
    def forward(self, image: torch.Tensor) -> torch.Tensor: ...

Import

from kornia.color import rgb_to_hsv, hsv_to_rgb
from kornia.color import RgbToHsv, HsvToRgb

I/O Contract

Inputs

Name	Type	Required	Description
image (rgb_to_hsv)	torch.Tensor	Yes	RGB image with shape (*, 3, H, W). Values in range (0, 1).
eps (rgb_to_hsv)	float	No	Epsilon for numerical stability. Default: 1e-8.
image (hsv_to_rgb)	torch.Tensor	Yes	HSV image with shape (*, 3, H, W). H in [0, 2*pi], S in [0, 1], V in [0, 1].

Outputs

Name	Type	Description
rgb_to_hsv return	torch.Tensor	HSV image with shape (*, 3, H, W). Channel order: H (hue in radians [0, 2*pi]), S (saturation [0, 1]), V (value [0, 1]).
hsv_to_rgb return	torch.Tensor	RGB image with shape (*, 3, H, W).

Usage Examples

Basic Usage

import torch
from kornia.color import rgb_to_hsv, hsv_to_rgb

# Convert RGB to HSV
rgb = torch.rand(1, 3, 128, 128)
hsv = rgb_to_hsv(rgb)
print(hsv.shape)  # torch.Size([1, 3, 128, 128])

# Modify saturation (channel index 1)
hsv[:, 1, :, :] = hsv[:, 1, :, :] * 0.5  # reduce saturation by 50%

# Convert back to RGB
rgb_desaturated = hsv_to_rgb(hsv)
print(rgb_desaturated.shape)  # torch.Size([1, 3, 128, 128])

# Using nn.Module wrappers in a pipeline
from kornia.color import RgbToHsv, HsvToRgb
to_hsv = RgbToHsv(eps=1e-6)
to_rgb = HsvToRgb()
hsv_out = to_hsv(rgb)
rgb_out = to_rgb(hsv_out)

Related Pages

• Principle:Kornia_Kornia_HSV_Color_Space