Cargando…
Deep neural models for color classification and color constancy
Color constancy is our ability to perceive constant colors across varying illuminations. Here, we trained deep neural networks to be color constant and evaluated their performance with varying cues. Inputs to the networks consisted of two-dimensional images of simulated cone excitations derived from...
Autores principales: | , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
The Association for Research in Vision and Ophthalmology
2022
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8976922/ https://www.ncbi.nlm.nih.gov/pubmed/35353153 http://dx.doi.org/10.1167/jov.22.4.17 |
_version_ | 1784680667634728960 |
---|---|
author | Flachot, Alban Akbarinia, Arash Schütt, Heiko H. Fleming, Roland W. Wichmann, Felix A. Gegenfurtner, Karl R. |
author_facet | Flachot, Alban Akbarinia, Arash Schütt, Heiko H. Fleming, Roland W. Wichmann, Felix A. Gegenfurtner, Karl R. |
author_sort | Flachot, Alban |
collection | PubMed |
description | Color constancy is our ability to perceive constant colors across varying illuminations. Here, we trained deep neural networks to be color constant and evaluated their performance with varying cues. Inputs to the networks consisted of two-dimensional images of simulated cone excitations derived from three-dimensional (3D) rendered scenes of 2,115 different 3D shapes, with spectral reflectances of 1,600 different Munsell chips, illuminated under 278 different natural illuminations. The models were trained to classify the reflectance of the objects. Testing was done with four new illuminations with equally spaced CIEL*a*b* chromaticities, two along the daylight locus and two orthogonal to it. High levels of color constancy were achieved with different deep neural networks, and constancy was higher along the daylight locus. When gradually removing cues from the scene, constancy decreased. Both ResNets and classical ConvNets of varying degrees of complexity performed well. However, DeepCC, our simplest sequential convolutional network, represented colors along the three color dimensions of human color vision, while ResNets showed a more complex representation. |
format | Online Article Text |
id | pubmed-8976922 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2022 |
publisher | The Association for Research in Vision and Ophthalmology |
record_format | MEDLINE/PubMed |
spelling | pubmed-89769222022-04-04 Deep neural models for color classification and color constancy Flachot, Alban Akbarinia, Arash Schütt, Heiko H. Fleming, Roland W. Wichmann, Felix A. Gegenfurtner, Karl R. J Vis Article Color constancy is our ability to perceive constant colors across varying illuminations. Here, we trained deep neural networks to be color constant and evaluated their performance with varying cues. Inputs to the networks consisted of two-dimensional images of simulated cone excitations derived from three-dimensional (3D) rendered scenes of 2,115 different 3D shapes, with spectral reflectances of 1,600 different Munsell chips, illuminated under 278 different natural illuminations. The models were trained to classify the reflectance of the objects. Testing was done with four new illuminations with equally spaced CIEL*a*b* chromaticities, two along the daylight locus and two orthogonal to it. High levels of color constancy were achieved with different deep neural networks, and constancy was higher along the daylight locus. When gradually removing cues from the scene, constancy decreased. Both ResNets and classical ConvNets of varying degrees of complexity performed well. However, DeepCC, our simplest sequential convolutional network, represented colors along the three color dimensions of human color vision, while ResNets showed a more complex representation. The Association for Research in Vision and Ophthalmology 2022-03-30 /pmc/articles/PMC8976922/ /pubmed/35353153 http://dx.doi.org/10.1167/jov.22.4.17 Text en Copyright 2022 The Authors https://creativecommons.org/licenses/by/4.0/This work is licensed under a Creative Commons Attribution 4.0 International License. |
spellingShingle | Article Flachot, Alban Akbarinia, Arash Schütt, Heiko H. Fleming, Roland W. Wichmann, Felix A. Gegenfurtner, Karl R. Deep neural models for color classification and color constancy |
title | Deep neural models for color classification and color constancy |
title_full | Deep neural models for color classification and color constancy |
title_fullStr | Deep neural models for color classification and color constancy |
title_full_unstemmed | Deep neural models for color classification and color constancy |
title_short | Deep neural models for color classification and color constancy |
title_sort | deep neural models for color classification and color constancy |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8976922/ https://www.ncbi.nlm.nih.gov/pubmed/35353153 http://dx.doi.org/10.1167/jov.22.4.17 |
work_keys_str_mv | AT flachotalban deepneuralmodelsforcolorclassificationandcolorconstancy AT akbariniaarash deepneuralmodelsforcolorclassificationandcolorconstancy AT schuttheikoh deepneuralmodelsforcolorclassificationandcolorconstancy AT flemingrolandw deepneuralmodelsforcolorclassificationandcolorconstancy AT wichmannfelixa deepneuralmodelsforcolorclassificationandcolorconstancy AT gegenfurtnerkarlr deepneuralmodelsforcolorclassificationandcolorconstancy |