Conditional GANs for Multi-Illuminant Color Constancy: Revolution or yet Another Approach? (2018-11-15T00:00:00.000000Z)

TL;DR

A novel view on this classical problem via generative end-to-end algorithm based on image conditioned Generative Adversarial Network is proposed and the largest existing shadow removal dataset is rendered and made publicly available.

Abstract

Non-uniform and multi-illuminant color constancy are important tasks, the solution of which will allow to discard information about lighting conditions in the image. Non-uniform illumination and shadows distort colors of real-world objects and mostly do not contain valuable information. Thus, many computer vision and image processing techniques would benefit from automatic discarding of this information at the pre-processing step. In this work we propose novel view on this classical problem via generative end-to-end algorithm based on image conditioned Generative Adversarial Network. We also demonstrate the potential of the given approach for joint shadow detection and removal. Forced by the lack of training data, we render the largest existing shadow removal dataset and make it publicly available. It consists of approximately 6,000 pairs of wide field of view synthetic images with and without shadows.

Authors

Oleksii Sidorov

2 papers

TL;DR

Abstract

Authors

References94 items

Artificial Color Constancy via GoogLeNet with Angular Loss Function

Video-to-Video Synthesis

Multimodal Unsupervised Image-to-Image Translation

Leave-One-Out Kernel Optimization for Shadow Detection and Removal

Color Constancy Algorithm for Mixed-Illuminant Scene Images

Direction-Aware Spatial Context Features for Shadow Detection

DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks

Progressive Growing of GANs for Improved Quality, Stability, and Variation

Shadow Detection with Conditional Generative Adversarial Networks

Recurrent Color Constancy

Fast Shadow Detection from a Single Image Using a Patched Convolutional Neural Network

Unsupervised Visual Attribute Transfer with Reconfigurable Generative Adversarial Networks

Improving CNN-Based Texture Classification by Color Balancing

FC^4: Fully Convolutional Color Constancy with Confidence-Weighted Pooling

DeshadowNet: A Multi-context Embedding Deep Network for Shadow Removal

Multiple illuminant estimation from the covariance of colors

Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks

Learning to Discover Cross-Domain Relations with Generative Adversarial Networks

Learning Residual Images for Face Attribute Manipulation

Stacked Generative Adversarial Networks

StackGAN: Text to Photo-Realistic Image Synthesis with Stacked Generative Adversarial Networks

Semantic Segmentation using Adversarial Networks

Image-to-Image Translation with Conditional Adversarial Networks

Multiple Illuminant Color Estimation via Statistical Inference on Factor Graphs

Conditional Image Synthesis with Auxiliary Classifier GANs

Deep Identity-aware Transfer of Facial Attributes

Large-Scale Training of Shadow Detectors with Noisily-Annotated Shadow Examples

Deep Specialized Network for Illuminant Estimation

Energy-based Generative Adversarial Network

Mixed pooling neural networks for color constancy

Semi-Supervised Learning with Generative Adversarial Networks

Generative Adversarial Text to Image Synthesis

New spectrum ratio properties and features for shadow detection

Automatic Shadow Detection and Removal from a Single Image

Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Learning to Remove Soft Shadows

Shadow Remover: Image Shadow Removal Based on Illumination Recovering Optimization

Effective learning-based illuminant estimation using simple features

U-Net: Convolutional Networks for Biomedical Image Segmentation

Shadow optimization from structured deep edge detection

Color constancy using CNNs

Conditional Generative Adversarial Nets

Pixel-wise illuminant estimation for mixed illuminant scenes based on near-infrared camera information

Going deeper with convolutions

Interactive Shadow Removal and Ground Truth for Variable Scene Categories

Automatic Feature Learning for Robust Shadow Detection

Segmentation and Estimation of Spatially Varying Illumination

Fast Shadow Removal Using Adaptive Multi‐Scale Illumination Transfer

ImageNet classification with deep convolutional neural networks

Shadow Removal Using Bilateral Filtering

Color constancy using faces

Color Constancy for Multiple Light Sources

What characterizes a shadow boundary under the sun and sky?

Color constancy and non-uniform illumination: Can existing algorithms work?

Illuminant color estimation for real-world mixed-illuminant scenes

Illumination estimation and cast shadow detection through a higher-order graphical model

Single-image shadow detection and removal using paired regions

Shadow Removal Using Intensity Surfaces and Texture Anchor Points

Detecting Ground Shadows in Outdoor Consumer Photographs

Learning to recognize shadows in monochromatic natural images

User-assisted intrinsic images

Color constancy based on local space average color

Texture-Consistent Shadow Removal

Light mixture estimation for spatially varying white balance

Bayesian color constancy revisited

Detecting Illumination in Images

Edge-Based Color Constancy

Color Constancy using Natural Image Statistics

Natural shadow matting

Editing Soft Shadows in a Digital Photograph

Color Constancy for Multiple-Illuminant Scenes using Retinex and SVR

Cast shadow segmentation using invariant color features

Color constancy through inverse-intensity chromaticity space.

Gamut Constrained Illuminant Estimation

A Large Image Database for Color Constancy Research

Estimating the scene illumination chromaticity by using a neural network.