Enabling Automated Pedestrian Data Collection using Computer Vision
Tarek Sayed, Mohamed H. Zaki, Houman Hediyeh · 20th ITS World CongressITS Japan · 2013
There is a significant need to explore methods of increasing the availability of pedestrian data. One of these methods is to develop reliable automated techniques for data collection such as the use of computer vision techniques. This paper demonstrates the use of a set of computer vision techniques for the automated collection of pedestrian data. The paper addresses three distinct issues in data collection. The first issue deals with the identification of gender attribute using information extracted on the walking characteristics of the pedestrians. The demonstrated case study reported correct classification rates (CCR) of 78%. The second issue is related to the conformance of pedestrian crossings. Automated spatial violation detection is demonstrated with accuracy greater than 90% is reported on the data set. The last data collection problem addressed in this paper is the automated classification of road-users. A classification approach relying on the movement characteristics of the road users is proposed. The approach is validated on real world data collected in Oakland California. The application of the classifier has shown a correct classification rate of around 90%. Overall, the three case studies demonstrate the considerable potential of using video-based computer vision techniques for automated pedestrian data collection.