Hashemifard, Kooshan, Flórez-Revuelta, Francisco
From Garment to Skin: The visuAAL Skin Segmentation Dataset
Hashemifard, K., Florez-Revuelta, F. (2022). From Garment to Skin: The visuAAL Skin Segmentation Dataset. In: Mazzeo, P.L., Frontoni, E., Sclaroff, S., Distante, C. (eds) Image Analysis and Processing. ICIAP 2022 Workshops. ICIAP 2022. Lecture Notes in Computer Science, vol 13373. Springer, Cham. https://doi.org/10.1007/978-3-031-13321-3_6
URI: http://hdl.handle.net/10045/131009
DOI: 10.1007/978-3-031-13321-3_6
ISSN: 
ISBN: 978-3-031-13321-3
Abstract: 
Human skin detection has been remarkably incorporated in different computer vision and biometric systems. It has been receiving increasing attention in face analysis, human tracking and recognition, and medical image analysis. For many human-related recognition tasks, using skin detection cue could be a proper choice. Despite the vast area of usage and applications for skin detection, not many large or reliable skin detection datasets are available, and many of the existing ones, are originally created for other tasks such as hand tracking or face analysis. In this paper, we propose a methodology for extracting skin pixels from garment segmentation and recognition datasets. This is achieved by using deep learning methods to generate automatic skin label masks from them by exploiting human body and hair segmentation and provided garment masks. Following this approach, a large human skin segmentation dataset is introduced. A validation set is also manually segmented in order to evaluate the accuracy of the output skin masks. Finally, usual methods for skin detection and segmentation are evaluated on this new dataset.
Keywords:Skin segmentation, Dataset
Springer, Cham
info:eu-repo/semantics/conferenceObject