The Age and Gender Mix of a Crowd, Never Who They Are
With the AI faces profile switched on, CrowdSense Desktop estimates an age band and a gender for each face in view and sends only the counts. It does not recognise, identify or match anyone, and nothing about a face is stored or sent.
Sponsors Ask Who the Audience Is. Surveys Answer a Week Later.
Sponsors, programmers and councils want to know the make-up of a crowd: is the family day reaching families, does the late-night stage draw the age group it was booked for. The usual answer is an intercept survey of a few hundred people, collected after the event.
An estimate of the age band and gender mix in front of a camera, sent as counts, gives a broad picture across the whole crowd during the event. It is an estimate about a crowd, never a fact about a person.
What It Estimates, and What It Never Does
Age band estimate
Each face in view is given an estimated age band. The portal reports how many fell in each band.
Gender estimate
Each face is given an estimated gender. Treat it as a trend across a crowd, since it is a guess from appearance and will be wrong for some people.
Counts on lines and zones
Estimates attach to line crossings and zone entries, so you see the mix arriving at a gate or standing at a stage.
No recognition or matching
The software does not recognise faces, compare them with any list or match a face across cameras or visits.
Nothing about a face kept
No face image, template or measurement is stored or sent. Only the counts leave the PC.
Off unless you choose it
The faces profile is a choice per camera. Every other camera stays on people or vehicles only.
Switch It On Deliberately, in Four Steps
- 01
Install on a Windows PC
Download CrowdSense Desktop and install it on a Windows 10, 11 or Server PC on the same network as your cameras, then sign in.
- 02
Add your cameras
Paste in each camera’s RTSP address, find it over ONVIF, or plug in a webcam. Choose the AI faces profile on the cameras that need it, and only those.
- 03
Draw a line or a zone
Draw a line across the entry or a zone in front of the stage or screen where faces are in view.
- 04
Report the mix
The portal shows counts by estimated age band and gender for each line and zone, live and over time.
Where to Put the Camera
Detection profile: AI faces. Any IP camera that streams over RTSP works, with ONVIF optional, and so does a USB webcam.
Faces need a near face-on view, 80° to 90°: the camera roughly at head height, looking along the direction people walk.
A face needs to fill at least 4% of the frame’s width. Run the stream at 720p or 1080p.
Place it where people walk towards it, such as an entry or a screen, rather than across a wide open space.
Check the camera against your privacy impact assessment and signage before you switch the faces profile on.
Pay per Camera, per Event or per Month
All prices in Australian dollars, excluding GST. One feed is one camera stream.
Event pass
$195
per feed, per event
One event is one primary location, one name and branding, up to 5 days. Feeds go live 14 days before the first day for set-up and testing; dashboards and the post-event report stay readable for 3 months after.
Always-on
From $39
a feed a month, yearly plan
For venues that count every day. Month to month it is $69 a feed for up to 9 feeds, and less at volume. Every band is on the pricing page.
Free trial
14 days
up to 4 feeds
A card is taken at sign-up. Nothing is charged until day 15, and you can cancel any time before then.
No cameras where you need them? For events, we hire Vision Nodes and the crew to install them.Get an events quote.
Who Uses Audience Estimates
Festival and event programmers
Whether a stage, a session or a day reached the audience it was planned for.
Sponsorship teams
A broad demographic picture for sponsor reports, alongside counts and dwell.
Media owners
The age and gender mix in front of a screen, as aggregate counts.
Councils and cultural venues
Who programs are reaching, for reporting and planning, without collecting anything personal.
Video Stays on Your PC. Only the Numbers Leave.
Face, gender and age detection is estimation, not identification. The software looks at a face in frame, estimates an age band and a gender, and sends only the count to the portal. It does not recognise or match anyone, it does not store or send any face image or measurement, and it is off unless you choose the faces profile for a camera. Even so, check it against your privacy impact assessment and your signage obligations under the Australian Privacy Principles before you switch it on.
The software detects people and vehicles so it can count them; it does not identify anyone. No facial recognition, no biometric identification of people. Ever. Your cameras, your recordings and your footage stay on your site, under your control.
Face, gender and age detection, Answered
Is this facial recognition?
No. Facial recognition identifies a person by matching their face. This estimates an age band and a gender from a face in frame and keeps only the count. Nothing is matched, compared or remembered.
Is any image of a face stored or sent?
No. The video is analysed on your PC and nothing about a face is stored or sent. The portal receives counts by estimated age band and gender.
Does it need a privacy assessment?
Check it against your own privacy impact assessment and signage obligations under the Australian Privacy Principles before you switch the faces profile on. Our compliance page sets out our position.
How reliable are the estimates?
They are estimates from appearance and will be wrong for some people. Use them as a broad mix across a crowd, never as a statement about an individual.
Is it on by default?
No. A camera only estimates age and gender if you choose the AI faces profile for it. All other cameras count people or vehicles only.
What camera set-up does it need?
A near face-on view (80° to 90°), faces at least 4% of the frame width and a 720p or 1080p stream.
More From the Same Cameras
Try It Free on Your Own Cameras
14 days free on up to 4 feeds. Install on a Windows PC, add a camera, draw a line.
