Controlled bandwidth
Analyzing locally avoids streaming continuous video over the internet.
How it works
From camera to dashboard, step by step
A clear architecture: intelligence sits as close to the cameras as possible, and only structured data is sent to the platform.
- 01
Camera infrastructure
Existing or new cameras, positioned for the target metric.
- Assessment of angle, mounting height and lighting
- IP cameras, or analog cameras via encoder
- Respect for private areas and privacy masking
- 02
Video streams
Streams retrieved using industry-standard protocols.
- RTSP / ONVIF
- Low-resolution substream when sufficient
- Isolated local network
- 03
Edge AI
An on-site computer analyzes streams in real time.
- Multiple streams per device, depending on capacity
- Keeps running even if the internet goes down
- Images not transmitted by default
- 04
Detection & tracking
Vision models detect and track the objects that matter.
- People, vehicles, configured objects
- Multi-object tracking to avoid double counting
- Virtual lines and zones
- 05
Business events
Detections become meaningful events.
- Entry, exit, line crossing
- Start and end of occupancy, duration
- Threshold breach
- 06
Cloud sync
Aggregated events are transmitted with encryption.
- Lightweight structured data
- Queued during outages
- Controlled hosting
- 07
Dashboards
Metrics are consolidated by zone, site and period.
- History and comparisons
- Correlation with POS/ERP, subject to study
- Role-based access
- 08
Reports & alerts
The right information reaches the right people.
- Real-time threshold alerts
- Scheduled reports
- Exports for your tools
Architecture
Smart Vision architecture diagram
Local processing, transmission of structured events, secure delivery.
Why Edge AI?
Stronger privacy
Images stay on site; only aggregated data is sent upstream.
Responsiveness
Alerts are generated locally, within moments.
Resilience
Analysis keeps running even if the connection drops.
Frequently asked questions
Do I need a permanent internet connection?
No. Analysis happens on site. The connection is used to sync events and view dashboards; if it drops, events are queued and sent once it is restored.
How many cameras can one Edge AI unit analyze?
It depends on the hardware chosen, the resolution and the analytics required. Sizing is carried out during the technical study.
Does VIDEO.MA record video?
Not by default. Video recording is handled by your video surveillance system and governed by your legal obligations. Our analysis produces events and metrics.
Is the client platform already available?
The self-service client platform is in preparation. Today, dashboards and reports are delivered as part of each project.
Ready to turn your cameras into business intelligence?
Tell us about your business: we’ll show you what your cameras can measure, backed by an honest feasibility study.