Mouth Tracking on Aerosol Suction Robot to Assist Dentist Surgery Using Deep Learning

COVID-19 disease is an unprecedented health condition and has never been diagnosed in humans before. According to scientific evidence, COVID-19 can spread from person to person through respiratory droplets, which are released when an infected individual coughs or sneezes. Those at the highest risk of contracting the disease include healthcare professionals, teachers, general practitioners, and dental hygienists. In dental practice, the transmission of COVID-19 often occurs through aerosols and splatters from the patient’s oral cavity. Aerosol may be generated during upper respiratory tract instrumentation of the oral and nasal cavities. Extra-oral suction (EOS) is a device for sucking patient aerosols during surgery or dental treatment. However, must move the tool manually to position the EOS chimney above the patient’s mouth. So that when the patient changes the position of his face, there is a possibility that some particles from the aerosol are not sucked in by Extra-oral Suction (EOS). Therefore, this research will build a system for tracking faces using Deep Learning on aerosol suction robots to assist dentist operations. To perform mouth detection, the MobileNET SSD method is used, and the centroid tracking algorithm is used to find the midpoint of the detected object, namely the human mouth, which will then enter the calculation process to look for ErrorX to obtain the tracking position of the human mouth. Mouth tracking on aerosol suction robots can be implemented automatically to assist dental health professionals in preventing the spread of COVID-19. To detect the mouth using the SSD MobileNet method with an accuracy rate of 94% from 815 testing images and the final results of the mouth tracking test, humans on the Extra-oral suction (EOS) robot get an accuracy of 93%.