—observations
Current detections across camera views
OFFLINE RETAIL INTELLIGENCE
Retail, in full view.
Every camera, connected. Every space, understood.
Portfolio preview · Recorded retail benchmark. Run the local backend for live cameras.
0/ 0 configured
— FPS combined inference
0/ 0 zones
At or above your configured capacity
—seconds / visit
Time observed within configured zones
CONNECTED SPACES
Your camera network
Connect a camera to start collecting retail insights.
Camera view
AREASTANDBY
Connect your next insight.
Choose a camera, then start inference.
CAMERAYOLO11n / person detection
● YOLO11n— FPS0 detections
Entered—
Exited—
Area analytics camera
Area heatmap
Selected cameraSESSION ACTIVITY
Activity builds as people move through the view.
Less activityMore activity
Most active spotWaiting for observations
Heat is weighted by observed time, not frame rate.
Camera coordinates · relative intensity · person-seconds
Zone performance
0 ZONES| Zone / camera | People now | Visits | Avg. dwell | Peak people | Activity share | Status |
|---|
Draw zones on a camera view to compare activity.
People over time
Selected cameraPeak —
Start inference to collect activity data.
0 samplesPeople visible · selected camera
Edge inference
—FPS selected camera
Inference latency— ms
P95 latency— ms
Frames processed0
Compute deviceAwaiting runtime
Data freshness—
Identity buffer—
Actual completions / rolling 5 seconds
Counts and visits are local to each camera. Overlapping views can include the same person. Dwell measures observed time, and is not purchase conversion.
Demo footage: EC Funded CAVIAR project / IST 2001 37540 · CC BY-SA · converted replay, detection and heatmap overlays.