What's color got to do with it? Face recognition in grayscale

Simple item page
dc.catalogadordfo
dc.contributor.authorBhatta, Aman
dc.contributor.authorMery Quirozm, Domingo Arturo
dc.contributor.authorWu, Haiyu
dc.contributor.authorAnnan, Joyce
dc.contributor.authorKing, Micheal C.
dc.contributor.authorBowyer, Kevin W.
dc.date.accessioned2025-03-06T12:18:58Z
dc.date.available2025-03-06T12:18:58Z
dc.date.issued2024
dc.description.abstractState-of-the-art deep CNN face matchers are typically created using extensive training sets of color face images. Our study reveals that such matchers attain virtually identical accuracy when trained on either grayscale or color versions of the training set, even when the evaluation is done using color test images. Furthermore, we demonstrate that shallower models, lacking the capacity to model complex representations, rely more heavily on low-level features such as those associated with color. As a result, they display diminished accuracy when trained with grayscale images. We then consider possible causes for deeper CNN face matchers "not seeing color". Popular web-scraped face datasets actually have 30 to 60% of their identities with one or more grayscale images. We analyze whether this grayscale element in the training set impacts the accuracy achieved, and conclude that it does not. We demonstrate that using only grayscale images for both training and testing achieves accuracy comparable to that achieved using only color images for deeper models. This holds true for both real and synthetic training datasets. HSV color space, which separates chroma and luma information, does not improve the network's learning about color any more than in the RGB color space. We then show that the skin region of an individual's images in a web-scraped training set exhibits significant variation in their mapping to color space. This suggests that color carries limited identity-specific information. We also show that when the first convolution layer is restricted to a single filter, models learn a grayscale conversion filter and pass a grayscale version of the input color image to the next layer. Finally, we demonstrate that leveraging the lower per-image storage for grayscale to increase the number of images in the training set can improve accuracy of the face recognition model.
dc.format.extent15 páginas
dc.fuente.origenWOS
dc.identifier.doiarXiv:2309.05180
dc.identifier.urihttps://doi.org/arXiv:2309.05180
dc.identifier.urihttps://repositorio.uc.cl/handle/11534/102385
dc.identifier.wosidPPRN:84951990
dc.information.autorucEscuela de Ingeniería; Mery Quiroz Domingo Arturo; 0000-0003-4748-3882; 102382
dc.language.isoen
dc.nota.accesoContenido parcial
dc.revistaArxiv
dc.rightsacceso restringido
dc.subjectFace recognition
dc.subjectColor space
dc.subjectSkin color
dc.subjectNeural network training
dc.subjectSynthetic dataset evaluation
dc.subject.ddc600
dc.subject.deweyTecnologíaes_ES
dc.subject.ods09 Industry, innovation and infrastructure
dc.subject.odspa09 Industria, innovación e infraestructura
dc.titleWhat's color got to do with it? Face recognition in grayscale
dc.typepreprint
sipa.codpersvinculados102382
sipa.trazabilidadWOS;2024-07-27
Files
Collections
Artículos de revistas