Abstract
|
|
A previous study shows that there appears to be a correlation between the time of day and and the number of people in a waiting room. [1] Is there also a correlation between different waiting rooms? Are they all busy at the same times of day? In order to facilitate this goal the project created a computer program to monitor the number of people in a room with no or minimal human interaction.
|
Introduction
|
|
There are many uses for knowing how busy a room is at different times of the day. Business owners might use this information in order to schedule their employees ensuring peak hours are covered, or customers may use it time their visits for when the office is least busy, hoping to have a shorter wait. Large differences from the usual number of people in the room may indicate that a special event is going on or that something is wrong.
|
Methods
|
|
Images were collected every five minutes from each of 6 cameras for one business day. The number of people in each image were counted by hand and by 2 different naive computer programs. One that compared each image to an empty image of the room and one that compared the room to an image that was the average of all the images in the room.
|
Results
|
|
Some pairs of offices matched very closely in number of people while between others there was almost no similarity. More study is necessary to determine why this is and whether similarities hold over longer periods of time. The computer programs cannot detect the precise number of people that are in a waiting room but they can detect changes in the number of people present.
|
Privacy
|
|
For pairs of locations that are correlated in the number of people present at a given time of day it is possible to make inferences about one by observing the other. A person who had access to a camera at one location could gain some information about what was going on at another location even though they weren't able to see it. Even though a person can't see the room they may be able to tell how crowded it is.
|
References
|
|
Camera: https://www.state.ak.us/dmv/DMVwebcams.htm
Project Paper (PDF)
[1] Bedford, Virginia. Use of Publicly Available Webcams in Naturalistic Observation Studies.
dataprivacylab.org/courses/dp1/refs/surveillance/samples/Bedford.pdf
|
