# Towards Spectral Estimation from a Single RGB Image in the Wild

## TL;DR

This paper introduces deep learning methods for estimating spectral information from a single RGB image taken in unconstrained, real-world conditions, achieving state-of-the-art results and demonstrating feasibility in wild settings.

## Contribution

The authors develop novel deep learning techniques for spectral estimation from single RGB images in the wild, addressing unknown camera and scene parameters.

## Key findings

- Achieved state-of-the-art results on standard spectral reconstruction benchmarks.
- Demonstrated accurate spectral estimation from a single RGB image in real-world conditions.
- Expanded spectral reconstruction to work with RGB images taken in the wild.

## Abstract

In contrast to the current literature, we address the problem of estimating the spectrum from a single common trichromatic RGB image obtained under unconstrained settings (e.g. unknown camera parameters, unknown scene radiance, unknown scene contents). For this we use a reference spectrum as provided by a hyperspectral image camera, and propose efficient deep learning solutions for sensitivity function estimation and spectral reconstruction from a single RGB image. We further expand the concept of spectral reconstruction such that to work for RGB images taken in the wild and propose a solution based on a convolutional network conditioned on the estimated sensitivity function. Besides the proposed solutions, we study also generic and sensitivity specialized models and discuss their limitations. We achieve state-of-the-art competitive results on the standard example-based spectral reconstruction benchmarks: ICVL, CAVE, NUS and NTIRE. Moreover, our experiments show that, for the first time, accurate spectral estimation from a single RGB image in the wild is within our reach.