Crowd Parameter Extraction from Video at the Main Gates of Masjid al-Haram

摘要

Gathering information on the large crowd of pilgrims attending Hajj is beneficial for management and safety of the event. Capturing information from videos can provide an effective method to gather different kind of data. However extracting information from such a large crowd is a challenging process. Computer vision allows for capturing of such information without the need for specialized devices or manual marking (labeling). In this paper we will focus on extracting pedestrian information such as speed from a video of Gate I (King Abdul Aziz) of Masjid al-Haram. Getting such information is non-trivial for many reasons, one of which is high occlusions. A literature review of the methods used for detection and tracking of human motion is presented. A methodology for the pedestrian parameter extraction is also proposed and the preliminary results of the tests on the algorithms are presented.