Beyond Color Correction : Skin Color Estimation In The Wild Through Deep Learning

2020 ◽  
Vol 2020 (5) ◽  
pp. 82-1-82-8
Author(s):  
Robin KIPS ◽  
Loïc TRAN ◽  
Emmanuel MALHERBE ◽  
Matthieu PERROT
Keyword(s):  
Skin Color ◽  
Estimation Method ◽  
Color Correction ◽  
Facial Image ◽  
Calibration Target ◽  
Region Segmentation ◽  
Color Calibration ◽  
Skin Region ◽  
In The Wild ◽  
Image Pixels

Estimating skin color from an uncontrolled facial image is a challenging task. Many factors such as illumination, camera and shading variations directly affect the appearance of skin color in the image. Furthermore, using a color calibration target in order to correct the image pixels leads to a complex user experience. We propose a skin color estimation method from images in the wild, taken with unknown camera, under an unknown lighting, and without a calibration target. While prior methods relied on explicit intermediate steps of color correction of image pixels and skin region segmentation, we propose an end-to-end color regression model named LabNet, in which color correction and skin region segmentation are implicitly learnt by the model. Our method is based on a convolutional neural network trained on a dataset of smartphone images, labeled with L*a*b* measures of skin colors. We compare our method with standard skin color estimation approaches and found that our method over-perform these models while removing the need of color calibration target.

A MODEL-BASED APPROACH TO HAIR REGION SEGMENTATION

2003 ◽  
Vol 03 (03) ◽  
pp. 481-501
Author(s):  
ZHI-QIANG LIU ◽  
JESSICA Y. GUO
Keyword(s):  
Facial Image ◽  
Region Segmentation ◽  
Manual Intervention ◽  
Geometrical Features ◽  
External Feature ◽  
Image Pixels ◽  
Human Faces ◽  
Segmentation Problem ◽  
Manual Extraction

Hair regions represent an important external feature in many tasks involving the processing of human faces. Currently, the task of locating the hair region in a facial image requires manual intervention. In this paper we examine the different aspects of the hair region segmentation problem and develop an automatic system for such a problem. The system incorporates the human knowledge on where the hair is usually located in a facial image. Segmentation is performed via the classification of image pixels and is based on both textural and geometrical features. Experiments have shown that the segmentation results are generally satisfactory and are at least comparable to the performance of manual extraction.

scholarly journals PKT: fast color-based spatial model for human skin detection

2021 ◽  
Author(s):  
Grace L. Samson ◽  
Joan Lu
Keyword(s):  
Spatial Structure ◽  
Human Skin ◽  
Skin Color ◽  
Skin Detection ◽  
Quad Tree ◽  
Learning Tasks ◽  
Color Detection ◽  
Speed Up ◽  
Image Pixels ◽  
Speed Accuracy

AbstractWe present a new detection method for color-based object detection, which can improve the performance of learning procedures in terms of speed, accuracy, and efficiency, using spatial inference, and algorithm. We applied the model to human skin detection from an image; however, the method can also work for other machine learning tasks involving image pixels. We propose (1) an improved RGB/HSL human skin color threshold to tackle darker human skin color detection problem. (2), we also present a new rule-based fast algorithm (packed k-dimensional tree --- PKT) that depends on an improved spatial structure for human skin/face detection from colored 2D images. We also implemented a novel packed quad-tree (PQT) to speed up the quad-tree performance in terms of indexing. We compared the proposed system to traditional pixel-by-pixel (PBP)/pixel-wise (PW) operation, and quadtree based procedures. The results show that our proposed spatial structure performs better (with a very low false hit rate, very high precision, and accuracy rate) than most state-of-the-art models.

A Face Detection Algorithm in Color Image Based on Skin Color Segmentation

2013 ◽  
Vol 811 ◽  
pp. 417-421
Author(s):  
Shi Lei
Keyword(s):  
Face Detection ◽  
Skin Color ◽  
Color Image ◽  
Detection Algorithm ◽  
Color Segmentation ◽  
Skin Region ◽  
Otsu Method ◽  
The Face ◽  
Segmentation Image ◽  
Ellipse Area

Aiming at color images under complex background, this paper put forward a face detection algorithm based on skin color segmentation, combining the geometric characteristics. The skin region can be obtained by using skin color model and OTSU method to automatically optimize threshold segmentation image. By analyzing the characteristics of skin color region, the face position is determined by criterion of ellipse area.

scholarly journals The Muscadine Grape (Vitis rotundifolia Michx)

EDIS ◽  
2020 ◽  
Vol 2020 (6) ◽  
Author(s):  
Peter C. Andersen ◽  
Ali Sarkhosh ◽  
Dustin Huff ◽  
Jacque Breman
Keyword(s):  
Mississippi River ◽  
Skin Color ◽  
Gulf Coast ◽  
Calcareous Soils ◽  
Vitis Vinifera L ◽  
Vitis Rotundifolia ◽  
Grape Berries ◽  
In The Wild ◽  
Muscadine Grape ◽  
Muscadine Grapes

The muscadine grape is native to the southeastern United States and was the first native grape species to be cultivated in North America. The natural range of muscadine grapes extends from Delaware to central Florida and occurs in all states along the Gulf Coast to east Texas. It also extends northward along the Mississippi River to Missouri. Muscadine grapes will perform well throughout Florida, although performance is poor in calcareous soils or in soils with very poor drainage. Most scientists divide the Vitis genus into two subgenera: Euvitis (the European, Vitis vinifera L. grapes and the American bunch grapes, Vitis labrusca L.) and the Muscadania grapes (muscadine grapes). There are three species within the Muscadania subgenera (Vitis munsoniana, Vitis popenoei and Vitis rotundifolia). Euvitis and Muscadania have somatic chromosome numbers of 38 and 40, respectively. Vines do best in deep, fertile soils, and they can often be found in beside river beds.  Wild muscadine grapes are functionally dioecious due to incomplete stamen formation in female vines and incomplete pistil formation in male vines. Male vines account for the majority of the wild muscadine grape population. Muscadine grapes are late in breaking bud in the spring and require 100-120 days to mature fruit. Typically, muscadine grapes in the wild bear dark fruit with usually 4 to 10 fruit per cluster. Bronze-fruited muscadine grapes are also found in the wild, and they are often referred to as scuppernongs. There are hundreds of named muscadine grape cultivars from improved selections, and in fact, one that has been found in the Scuppernong river of North Carolina has been named Scuppernong. There are over 100 improved cultivars of muscadine grapes that vary in size from 1/4 to 1 ½ inches in diameter and 4 to 15 grams in weight. Skin color ranges from light bronze to pink to purple to black. Flesh is clear and translucent for all muscadine grape berries. Originally published 1994 by Peter C. Anderson and Timothy E. Crocker. Published on EDIS June 2003. Revised November 2010, October 2013, January 2017. This revision with Sarkhosh and Huff.

Monitoring System for Persons With Alzheimer's Disease via Video-Object Tracking

2018 ◽  
Vol 9 (2) ◽  
pp. 18-36
Author(s):  
Haitham Asaad Al-Anssari ◽  
Ikhlas Abdel-Qader ◽  
Maureen Mickus
Keyword(s):  
Alzheimer’S Disease ◽  
Alzheimer's Disease ◽  
Optical Flow ◽  
Monitoring System ◽  
Skin Color ◽  
Moving Objects ◽  
Color Space ◽  
Color Information ◽  
Region Detection ◽  
Skin Region

This article presents a framework for a food intake monitoring system intended for use with persons with Alzheimer's disease and other dementias. Alzheimer's disease has a significant impact on the individual's ability to perform their daily activities including eating. Providing assistance with feeding is a major challenge for caregivers, including a significant time commitment. We present a vision-based system that tracks moving objects, such as the hand, using a combined optical flow and skin region detection algorithms. Skin detection is implemented using two different methods. Hue, saturation, and value (HSV) color space, which is on separation of the illuminance component from chrominance one as the first method and skin color information is extracted from subject's face detected using Viola-Johns algorithm for the second method. Once face and other moving skin regions are detected, bounding boxes are created and used to track all moving regions over the video frames, recognizing eating behavior or the lack of it. Based on experimental results the proposed method using optical flow and skin regions segmentation using HSV color detects the hand to mouth eating motion with 92.12% accuracy. The optical flow and skin region segmentation based on face color information achieves a higher accuracy of 94.29%.

Skin region segmentation using an image-adapted colour model

2015 ◽  
Vol 52 (2/3) ◽  
pp. 142 ◽  
Author(s):  
Seok Woo Jang ◽  
Kee Hong Park ◽  
Gye Young Kim
Keyword(s):  
Region Segmentation ◽  
Skin Region
Sign in / Sign up

Export Citation Format

Share Document