Automatic Passenger Counting (APC) for Online Event Data Recorder (EDR)

The passengers counting is performed to provide data related to passenger number and characteristic. However, the traditional passenger counting needs effort and leads to inaccurate data. Automatic Passenger Counting (APC) is the system to help resolving those problem. It can be developed using computer vision with Haar Cascade method. Those method has efficient computation, so it is proper to be applied in Raspberry Pi as the main device of the proposed system. The proposed system also involves OpenCV as main library to develop passenger detection application. The OpenCV is utilized to accomplish computer vision and image processing tasks. According to the system testing, the proposed system can detect passengers with mask and eyeglasses. The scenarios for testing are using the model of pretrained Haar cascade for detecting faces only and using the model of pretrained Haar cascade for detecting faces and eyes. However, this system only has accuracy about 60% with setting up the variables such as scaleFactor (1.10, 1.30, 1.50), minNeighbors (1, 5, 10), and minSize (30,30). The best result is produced from minimum value of scaleFactor and minNeighbors variables. The result of APC can be integrated to complete the Event Data Recorder (EDR) features so it can provide real-time information related to vehicle condition.