Color Constancy Using CNNs
arXiv:1504.04548
Abstract
In this work we describe a Convolutional Neural Network (CNN) to accurately predict the scene illumination. Taking image patches as input, the CNN works in the spatial domain without using hand-crafted features that are employed by most previous methods. The network consists of one convolutional layer with max pooling, one fully connected layer and three output nodes. Within the network structure, feature learning and regression are integrated into one optimization process, which leads to a more effective model for estimating scene illumination. This approach achieves state-of-the-art performance on a standard dataset of RAW images. Preliminary experiments on images with spatially varying illumination demonstrate the stability of the local illuminant estimation ability of our CNN.
Accepted at DeepVision: Deep Learning in Computer Vision 2015 (CVPR 2015 workshop)
Cited by in corpus (9)
- Improving Color Constancy by Discounting the Variation of Camera Spectral Sensitivity
- Color Constancy by Learning to Predict Chromaticity from Luminance
- Deep Structured-Output Regression Learning for Computational Color Constancy
- Fully Point-wise Convolutional Neural Network for Modeling Statistical Regularities in Natural Images
- Color Constancy by GANs: An Experimental Survey
- CroP: Color Constancy Benchmark Dataset Generator
- Illumination Estimation Challenge: experience of past two years
- Fast Color Constancy with Patch-wise Bright Pixels
- Single and Multiple Illuminant Estimation Using Convolutional Neural Networks