Color Group @ CVC

The Color Group at the Computer Vision Center (CVC) conducts research at the intersection of computer vision, computational imaging, and visual perception. Our work explores the foundations of color science while addressing practical challenges in image enhancement, relighting, restoration, and scene understanding.

Over the past decade, the Color Group has published more than 60 papers in leading international journals and conferences. We regularly publish at top-tier computer vision venues such as CVPR, ICCV, and ECCV, as well as specialized conferences including computer graphics, CIC, and 3DV. We actively collaborate with researchers from academia and industry worldwide, developing projects sponsored by both public and private funding.

Faculty and students in the Color Group are affiliated with the Universitat Autònoma de Barcelona, within the Department of Computer Science.

Why Work in Color?

Color is a fundamental aspect of visual perception and image understanding, yet it remains one of the most challenging areas in computer vision. Our research addresses critical problems where color plays a central role:

Color Editing and Enhancement: We develop methods for interpretable and controllable color manipulation, from color harmonization to image retouching, enabling photographers and artists to achieve their creative vision while maintaining photorealistic results.

Image Relighting: Light and color are intrinsically entangled in images. Understanding and manipulating illumination is therefore key to realistic rendering and scene understanding. We work on techniques to decompose, edit, and synthesize lighting conditions in images, creating fine-grained datasets and methods that advance the field.

Computational Color: From camera image signal processing (ISP) pipelines to color correction and gamut mapping, we tackle the technical challenges that bridge the gap between physical light capture and perceptually accurate display.

Color in Large Models: As foundation models reshape computer vision, we investigate how these systems perceive and reason about color, addressing questions of color perception, semantic understanding, text-to-image generation, and biases in large-scale visual models.