Detection of human fatigue level based on changes in skin color using RGB sensor

• Published in: IOP conference series. Earth and environmental science Vol. 542; no. 1; pp. 12037 - 12048
• Main Authors: Priambudi, F A, Herodian, S, Saulia, L
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.07.2020
• Subjects: Cameras; Color; Correlation; Data capture; Data processing; Data retrieval; Fatigue; Fatigue tests; Heart rate; Image processing; Intake systems; Physical activity; Regression analysis; Sensors; Skin; Skin pigmentation; Work capacity
• ISSN: 1755-1307; 1755-1315
• DOI: 10.1088/1755-1315/542/1/012037

Decreased work capacity could be caused by fatigue, which is a state of decreased efficiency and endurance. Fatigue can be measured using various methods, one of them is based on heart rate. Measurement of fatigue levels based on heart rate has been widely carried out, one of them is the use of contactless measurement. The objective of this research was to detect human fatigue levels using an RGB sensor based on changes in skin color with image processing technology. This research started from testing of the data intake system, followed by image data retrieval in the form of images before and after physical activity and fatigue levels data using the step test method. Image data capture is performed using an RGB sensor in the form of a Kinect 360 and mirrorless camera, while fatigue levels data used a pulse monitor. The next step was processing data using a simple linear regression method to see the level of correlation between the results of image processing in the form of changes in the RGB color of the skin with changes in fatigue levels based on changes in heart rate. The results of simple linear regression showed significant RGB color changes toward changes in fatigue levels with a correlation value reaching 0.8 when using mirrorless camera then Kinect 360. This result showed that the use of mirrorless camera is better. The average correlation value of 20 subjects at all points above 0.6 on the two sensors used. Each of the RGB color components had a significant change. This showed that there was a relationship between changes in RGB with fatigue level changes so that the method of detecting fatigue levels based on changes in skin color can be done.