TY - GEN
T1 - Implementation of Single Shot Detector (SSD) MobileNet V2 on Disabled Patient's Hand Gesture Recognition as a Notification System
AU - Nurfirdausi, Annisaa F.
AU - Soekirno, Santoso
AU - Aminah, Siti
N1 - Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - The development of smart healthcare system has a potential to improve the quality of health services. Disabled patients need to express their needs non-verbally to their families or medical workers to fulfill their needs immediately. In this study, we developed hand gesture recognition using web camera as a notification system. Image acquisition was done on 12 subjects with various gender and ages. Based on human's basic daily needs, there are 5 gestures as the object of study: need to eat, need to drink, need to go to the toilet, need help and need medicines. The collected images were trained using Single Shot Detector (SSD) algorithm on MobileNet V2 architecture. Among the various deep learning techniques, SSD MobileNet V2 was chosen because it has good ability in object detection and needs low computation. Therefore, it is suitable to be applied on realtime detection. This study results on mean Average Precision (mAP) 44.7% and detection rate 85% which means 85 out of 100 images were well-detected. The mAP showed better result than previous studies. Frame rate per second (FPS) provided in this study was ±2 FPS. The gestures detected also triggered the notification on telegram to notify family or nurse who take care of disabled patients.
AB - The development of smart healthcare system has a potential to improve the quality of health services. Disabled patients need to express their needs non-verbally to their families or medical workers to fulfill their needs immediately. In this study, we developed hand gesture recognition using web camera as a notification system. Image acquisition was done on 12 subjects with various gender and ages. Based on human's basic daily needs, there are 5 gestures as the object of study: need to eat, need to drink, need to go to the toilet, need help and need medicines. The collected images were trained using Single Shot Detector (SSD) algorithm on MobileNet V2 architecture. Among the various deep learning techniques, SSD MobileNet V2 was chosen because it has good ability in object detection and needs low computation. Therefore, it is suitable to be applied on realtime detection. This study results on mean Average Precision (mAP) 44.7% and detection rate 85% which means 85 out of 100 images were well-detected. The mAP showed better result than previous studies. Frame rate per second (FPS) provided in this study was ±2 FPS. The gestures detected also triggered the notification on telegram to notify family or nurse who take care of disabled patients.
KW - deep learning
KW - disabled patient communication
KW - hand gesture recognition
KW - single shot detector mobilenet v2
KW - smart healthcare system
UR - http://www.scopus.com/inward/record.url?scp=85123858814&partnerID=8YFLogxK
U2 - 10.1109/ICACSIS53237.2021.9631333
DO - 10.1109/ICACSIS53237.2021.9631333
M3 - Conference contribution
AN - SCOPUS:85123858814
T3 - 2021 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2021
BT - 2021 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2021
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 13th International Conference on Advanced Computer Science and Information Systems, ICACSIS 2021
Y2 - 23 October 2021 through 26 October 2021
ER -