Color Constancy is a research website focused on color constancy and illuminant estimation: methods for determining the color of the light illuminating a scene. It explains the topic, outlines approaches for estimating illumination, and provides evaluation resources for researchers working in computer vision and imaging. The site publishes results from multiple methods across publicly available datasets, including recovery and reproduction error measurements, and links to source code and dataset downloads.

Its content also highlights recent research, benchmarks, and challenges. Featured topics include learning-based methods designed to generalize across cameras, approaches for estimating one or more light sources, and temporal color constancy using sequences of images. The site covers datasets such as Cube++, which contains 4,890 images with known illumination colors and additional semantic information, as well as a benchmark comprising 600 real-world image sequences.

Rather than offering commercial products or services, Color Constancy serves as an information and research resource.