Installation Guide - People Counting by bzh37299

VIEWS: 14 PAGES: 4

									   Installation Guide - People Counting
Survey the installation environment

It is essential to get familiar with the environment where the cameras should be
installed. Each element, such as door’s width, ceiling’s height, passageway,
should be taken into consideration as variable which may affect the object
counting procedure.


Important things to be noticed before installation:

         Door width
         Ceiling height
         Distance between object
         Camera Lens type and View Angle
                 307.00



                          175.00




                Fig A – Sample installation using single camera
                  Fig B – Sample installation using two cameras
257.00
         175.00




                  Fig C – Sample installation using three cameras
                         to cover larger area with EAS detector
Factor that might affect counting procedure:

          Sunlight
          Lightning
          Emergency exit sign
          Interior decoration
          Pedestrian


Advantages using GeoVision DVR system

Cost effectiveness:    Integrated people counting application
Ease of usage:         Unified management interface to use
Space saving:          Single PC to accomplish multiple tasks
Implementation:        Customer behavior analysis, Personnel Management,
                       Marketing Strategy, Pedestrian Flow Counting, etc


Test the installation environment

Based on actual scenario determine the required camera lens type and view
angle, perform the necessary cabling and system setup procedure. Afterward
conduct a real counting process for later verification and fine tuning process for
better accuracy.


    First verification

Due to insufficient sample acquired during setup process, it is recommended
to conduct the verification on a usual business activity day in the real world.
The counting process is done on both sides, manually by observer in a specific
time range whose result is compared later with the information acquired by the
DVR. After achieving the error percentage, make a system fine tuning.


                          People Counting at CPS
   Date            Time       Camera       Observer       System     Error   Note
          th
 June 10         Single Day   Cam1.2.3        223            246     -10%
          th
 June 11         Single Day   Cam1.2.3        215            236     -10%
 June 12th       Single Day   Cam1.2.3        184            204     -11%
                       People Counting at WBS
   Date           Time        Camera       Observer      System      Error     Note
           th
 June 10        Single Day    Cam1.2          479          415       13%
           th
 June 11        Single Day    Cam1.2          546          508        7%
June 12th       Single Day    Cam1.2          431          379       12%


                         People Counting at BZS
   Date           Time        Camera       Observer      System      Error     Note
          th
 June 8         Single Day     Cam1           428          368       14%
June 11th       Single Day     Cam1           326          259       21%


    Second verification

If the error percentage fells behind acceptance, please conduct the necessary
fine tuning and perform the same verification process as the first time.


                         People Counting at BZS
   Date           Time        Camera       Observer      System      Error     Note
           th
 June 17        Single Day     Cam1           174          182       -5%
June 18th       Single Day     Cam1           38            36        5%


                       People Counting at WBS
   Date           Time        Camera       Observer      System      Error     Note
          th
June 18         Single Day    Cam1,2          356          366       -3%
June 19th       Single Day    Cam1,2          218          215        1%


    Final verification

If the result is acceptable otherwise repeat the fine tuning process


People counting have always been difficult to prove its reliability in the past. It
varies from day to day, high season to low season; only performing random
verification like this can get a real accuracy in the real world.

								
To top