Profile Analyzer

Operating System Language
Recommended PC Specifications CPU : Intel Core-i7-4770 or greater (at least 4 cores, HT technology compatible) Memory : 8GB or greater
Operating System Windows 7 Ultimate/Professional/Enterprise/Home Premium SP1 64-bit
Windows 8.1/Windows 8.1 Pro/Windows 8.1 Enterprise 64-bit
Windows 10 Pro/Enterprise/Home 64-bit
Windows Server 2008 R2 Standard SP1 64-bit
Windows Server 2012 Standard 64-bit
Windows Server 2012 R2 Standard 64-bit

Windows Server 2016 is not supported.
Software .NET Framework 4.5.2
Operating System Language Japanese/English
Compatible Cameras

*information as of June 2017
VB-H761LVE-H, VB-H751LE-H, VB-M741LE-H, VB-S30D MK II, VB-S31D MK II, VB-S800D MK II, VB-S900F MK II, VB-S805D MK II, VB-S905F MK II, VB-H730F MK II, VB-H45/B, VB-M44/B, VB-S30VE, VB-S800VE, VB-S910F, VB-R13VE, VB-R13, VB-R12VE, VB-M50B, VB-H652LVE, VB-H651VE, VB-H651V, VB-H761LVE, VB-H760VE, VB-H751LE, VB-R11VE, VB-R11, VB-R10VE, VB-M641VE, VB-M641V, VB-M640VE, VB-M640V, VB-M741LE, VB-M740E, VB-H43/B, VB-H630VE, VB-H630D, VB-H730F, VB-M42/B, VB-M620VE, VB-M620D, VB-M720F, VB-S30D, VB-S31D, VB-S800D, VB-S900F, VB-S805D, VB-S905F
Requirements for the input file Video Format : JPEG
Resolution : 1920x1080, 960x540, 480x270
(The closest available image size offered by the camera, will be used as a source.)
Frame rate : 10fps (Fixed)
Other Setting : The camera’s quality and angle settings, other than the resolution and frame rate, need to be set using the camera beforehand.
Continuous : Not limited. Continues as long as the camera’s “Maximum Connection Time” setting allows.
Communications Protocol : IPv4 (Does not support IPv6), HTTP. Does not support SSL
User Authentication : Basic Authentication /Digest Authentication
Connection via proxy : Not supported
Functions
Age group estimation Gender estimation Input Canon Network Camera Live Video
Output Output: CSV file, log
Age : 0-2/ 3-9/ 10-19/ 20-29/ 30-39/ 40-49/ 50-59/ 60 or older (Unit : years old)
Gender : Male/ Female

*If a profile cannot be estimated (for example, only a side profile is visible and a forward looking face is undetectable), it is set to no profile.
*Counts people for each profile (Age Group, Gender), and aggregates at each specified time.
Detection area specification Sets using a polygon with three to eight points.
Multiple areas cannot be specified.
Detection area edit: Changes the shape of the detection area
CSV file aggregation interval Interval at which to aggregate the results of feature extraction: 10 min. / 15min. / 30 min. / 1 h
File Output Interval Works with file aggregation interval
Schedule Setting Sets analysis start time/end time (30 min. units)
Setting export/ import Export file are output into the specified folder
Log output
Processing status check People detection status (displays rectangles on detected faces)
Preview Display (displays latest profile estimation results for 6 people)
Specifications
Maximum number of people detectable per minute 30 people/min
Minimum analyzable face size 80 pixels
Number of concurrent activations per license 1
Number of simultaneous sessions on 1 PC No limitations (SIer controls the configuration)
Cameras Installation Conditions
Detectable tilt of human face Within +/- 15° in all directions (up/down/right/left rotation). When over/under +/- 15°, it is possible that the face cannot be detected.
Recommended declination The angle from directly in front of the face to the left or right is ±5°. The vertical angle from the face to the camera is 13° or less.
Scenes and Subjects that are difficult to detect
Situations where detection accuracy decreases
  • Backlit face appears dark. Contour of the person is fuzzy, due to backlight.
  • Face is blurry. (Shutter speed is slow)
  • Face is not facing camera.
    • E.g.: Looking at the ground
    • E.g.: Turning the head or holding cellphone which blocks the face
    • E.g.: Looking around
  • Part of face is hidden. (mask, sunglasses and facial hair that can not recognize ones face)
  • Inadequate light
    • Face is not recognized due to darkness
    • Due to unequal distribution of light, part of face may be shaded
  • People or face blocked or hidden by object
The following situations can cause non-human objects to be incorrectly detected as a person:
  • Things that are often mistakenly detected as people
    • Objects like dolls, or composed of colors/shapes similar to face
    • Reflections in water, mirrors, glass, etc.
    • Photos and illustrations in posters, etc.
    • People displayed on the monitor, etc.