Cross-Camera Convolutional Color Constancy

TL;DR

This paper introduces C5, a learning-based method that accurately estimates scene illuminant color across different cameras by dynamically adapting to new camera spectral properties using unlabeled test images, enabling practical calibration-free white balance.

Contribution

C5 extends convolutional color constancy with a hypernetwork approach, enabling cross-camera adaptation through transductive inference with minimal computational resources.

Findings

Achieves state-of-the-art accuracy on multiple datasets.

Fast inference times (~7 ms on GPU, ~90 ms on CPU).

Requires minimal memory (~2 MB).

Abstract

We present "Cross-Camera Convolutional Color Constancy" (C5), a learning-based method, trained on images from multiple cameras, that accurately estimates a scene's illuminant color from raw images captured by a new camera previously unseen during training. C5 is a hypernetwork-like extension of the convolutional color constancy (CCC) approach: C5 learns to generate the weights of a CCC model that is then evaluated on the input image, with the CCC weights dynamically adapted to different input content. Unlike prior cross-camera color constancy models, which are usually designed to be agnostic to the spectral properties of test-set images from unobserved cameras, C5 approaches this problem through the lens of transductive inference: additional unlabeled images are provided as input to the model at test time, which allows the model to calibrate itself to the spectral properties of the test-set camera during inference. C5 achieves state-of-the-art accuracy for cross-camera color constancy on several datasets, is fast to evaluate (~7 and ~90 ms per image on a GPU or CPU, respectively), and requires little memory (~2 MB), and thus is a practical solution to the problem of calibration-free automatic white balance for mobile photography.