United States Department of Transportation
[PDF-11.78 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:9/2018 – 11/2019
-
Corporate Publisher:
-
Abstract:This study evaluates the automatic counting feature of two sensors to count pedestrians and bicyclists. The first sensor (sensor A) is a hardware/software system that is used to detect, classify, and count different objects. It is composed of two components: the sensor (hardware) and the software (the data platform). All the components of sensor A are designed to be weatherproof and to easily conform to the specifications of the city planners and the traffic engineers. The second sensor (sensor B) is a professional sensor (or camera) that is also capable of automatically counting objects and capturing videos. Six different locations at New Orleans and Baton Rouge with different conditions (weather, time of the day, traffic volume, and density of pedestrians and bicyclists), were selected for the evaluation process. Sensor B was tested only in one location in New Orleans; that location has a high-traffic volume of pedestrians. The evaluation of sensor A showed that the overall total observations median and mean Absolute Percentage of Error (APE) of the pedestrians during the day-time are 119.72% and 119.15%, and during the night-time are 69.10% and 111.90%. The overall observations median and mean APE of the bicyclists during the day-time are 69.62% and 80.03% and during the night-time are 89.47% and 80.15%. The evaluation of sensor B showed that the overall total observations median and mean APE of the pedestrians and bicyclists are 89.9% and 86.1%, respectively.
-
Format:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
