<![CDATA[Humans have a sense of sight that is able to obtain visual information in capturing objects far and near, but it is different from blind people. Blind people find it difficult to carry out daily activities such as walking in unfamiliar places and finding a room that is done independently. Several studies have made object detection devices to help visually impaired people through sound using a camera, lighting conditions affect the accuracy of detecting an object. The solution is to create a tool for blind people to recognize room signage in various lighting conditions using the Nvidia Jetson Nano-based YOLOv3 algorithm. This system adapts to 3 lighting conditions, namely dark (0-50 lux), dim (51-100 lux) and bright (101-500 lux) using a distance of 1 to 2.5 meters. The result of recognizing this room's nameplate is in the form of voice output. The test results are based on the LDR sensor response to lighting conditions of 100%, room nameplate recognition voice output of 100%, computational time obtained on LED light response and CLAHE mode of 1.891 seconds and room nameplate recognition of 0.40008 seconds. The average value of the whole system in recognizing the nameplate of the room under lighting conditions is 95.67%.]]>
Copyrights © 2022
