<![CDATA[The Covid-19 pandemic, which hinders various human activities, makes people keep themselves withdrawn from others. The massive social distancing also applies in hospitals, including between patients and nurses in the use of wheelchairs. The issued autonomous wheelchairs must do another task, that is detecting room plates and recognizing room nameplates. The detection in this research uses Faster Regional Convolutional Neural Network (Faster-RCNN) model made on Tensorflow. Meanwhile, room name recognition will actualize using PyTesseract. Testing was carried out on a smart wheelchair prototype using the Raspberry Pi 4B. The hardware integration result of the buzzer_1 function is 100%, the buzzer_2 function is 100%, the buzzer_3 function is 100%, the buzzer_4 function is 100%, the buzzer_5 function is 100%, and the motor function is 100%. While the integration of room plate detection software was 95% and room name recognition was 81%. Then performed image testing to measure accuracy, prediction ratio, and computation time. The results of detection accuracy using Faster-RCNN are 87%, the predictive ratio for recognition using PyTesseract is 77.73%, and the average computation time for detection is 6.825 seconds per image and for recognition of 2.54 seconds per image.]]>
Copyrights © 2021
