If you’ve read Harry Potter, you’ll definitely know what the spell “I solemnly swear that I am up to no good” evokes. For the uneducated uninitiated, that spell activates a magical map that shows you where people are and where they are heading inside Hogwarts—like everyone in the building has a live GPS tag on them. I’m pretty sure fans of the series must have wished for such a Marauder's Map at one point or another, but our boring Muggle world just never allowed for such cool luxuries.
Not to be left behind, researchers at the Carnegie Mellon University have managed to create a real-life “Marauder’s Map” using CCTV footage, an algorithm that uses facial recognition, colour recognition and location tracking tools. The algorithm uses CCTV footage to identify individuals, track their movements and plot their positions on the map, giving you what can be called the Muggle equivalent of the Marauder’s Map. Take a look at this video demonstrating the Map.
The system maps out people's movements based on what they're wearing, how they look and other factors.
The researchers have presented their work in a paper “Harry Potter’s Marauder’s Map: Localizing and Tracking Multiple Persons-of-Interest by Nonnegative Discretization”, which will be presented at the Computer Vision and Pattern Recognition conference scheduled to be held late this month.
The researchers, Shoou-I Yu, Yi Yang and Alexander Hauptmann, found that a lot of factors can also be used to track individuals in complex environments where doors, walls or furniture may block a camera’s view. The Marauder’s Map records the facial features, colours of the clothing and prior location, and uses the data to determine if a person who left the room is the same one entering the hallway.
“To perform reliable localization and tracking, important cues such as color, person detection, face recognition and non-background detection should all be utilized. Also, the tracking algorithm has to deal with typical yet complex indoor scenes consisting of different rooms, many walls and corridors. Therefore, an ideal Marauder’s Map algorithm should integrate different sources of information in a seamless way to perform reliable localization and tracking of persons-of-interest in complex indoor environments,” the research paper reads.
Yu and his team tested the system in two scenarios: one had 19 people in an outdoor scene with seven cameras being used, and the other was inside a nursing home with 13 people and 15 cameras were used for surveillance. In the especially complex second scenario, the system managed to track people within one metre of their position.
However, this Marauder’s Map can’t handle crowds very well at the moment. If the system is further developed, it could prove to be a powerful tool for surveillance in places such as railway stations, airports or hospitals.
Published Date: Jun 06, 2013 05:32 pm | Updated Date: Jun 06, 2013 05:32 pm