|
|
NRC-CNRC |
IIT-ITI | Computational
Video Group | Video
Recognition Systems Demonstrations:
|
| Premises | |
| There are
many premises that are important to monitor. Most of
all, this includes entrances/exits to/from the
building where no human personnel present and where an
intruder or thief can pass-by unnoticed.
Two of such entrance spaces in one of the NRC
buildings are shown here.
|
In order to tighten the security in such places, NRC
has purchased (from ADT security) and installed
several surveillance cameras (in December 2006)
|
|
Normal Monitoring (without ACE): Every surveillance camera comes with a monitor, at which a dedicated human personnel has to look at all times in order to aware of everything happening in the observed premises. ![]() |
But what happens, when the officer has to leave the desk or is not looking at the monitor ? Will intruders pass unnoticed? - NO, if you have ACE installed on your computer!
|
|
Monitoring with ACE: In real-time mode - Watch closely when alarm sounds. |
In archival mode - Zoom on a day, on hour, then point and click |
![]() |
|
Examples:
See ACE-Surveillance in action: real-life CES summarizations obtained by ACE users on a variety of monitoring assignments. Easily browse over the detected events and objects by day and hour, see their annotations and selected best-shot captures (data courteously provided by our clients)
Days viewed in this demo correspond to different real monitored assignments: difference is seen as a new station, camera and/or setup (e.g different angles of view, lighting conditions, camera image quality, complexity of observed events, number of of tracked objects, their velocities, shapes and sizes)
Output of the ACE: The entire activity captured by the surveillance system over several hours (from the office window) is summarized into 2 minutes (600Kb) of annotated video comprised of Critical Evidence Snapshots (CES).
These are the results from several real-life monitoring tasks as obtained by the recognition module of the CES client. Note the quality of the video image for different cameras and lighting conditions.
(Timestamp is shown at top right of the video image. Detected objects/actions are highlighted with coloured boxes. Blue box - the center of computed foreground. Red box - motion center. Green box - enclosing around the object)
1. Monitoring the premises in front of an office:
|
Hours of operation: 19 hours (16:45 - 10:00). Cam 1: USB web-cam. |
# of extracted CES: 148 Summarization: wmv video (600Kb), CES |
2. Monitoring the premises in front of a house:
|
Hours of operation: 20 hours (22:10 - 18:09) and
20 hours next day (22:06 - 17:34) Cam 1: wireless CCTV cam + Video2USB converter. |
# of extracted CES:
177 and 222 Summarization: video part1 (900Kb) and video part2 (1.9Mb), annotated CES (6Kb each), original CES. |
3. Three-cam monitoring around a house:
|
Hours of operation: 15 hours (21:00 - 14:30) Cam 1 - overviews carport in the front of the house: USB web-cam (summary avi - 74Kb) Cam 2 - overviews carport in the front of the house: wireless CCTV cam+ Video2USB converter (summary avi - 800Kb) Cam 3 - overviews the road in the back of the house: USB web-cam (summary avi - 4Mb) |
# of extracted CES: X, Y, Z
|
4. Example of automated surveillance annotation:
|
Hours of operation: 10 hours (22:00 - 8:00). Cam 1: USB web-cam. |
# of extracted CES: 23
Enter here. Password: "c.e.s." (lowercase). |
www.perceptual-vision.com
(http://ace.vrs.iit.nrc.ca)
Last updated: 2006-XII-05
Copyright (R), IIT-ITI, NRC-CNRC
Project Leader:
Dmitry Gorodnichy