The problem
Manual counts and end-of-shift tallies are slow and error-prone. Discrepancies between produced, packed and shipped quantities are often found days later, when it is hard to see where they came from.
How it works
- 1A counting line or zone is defined on the camera view.
- 2Items are detected and tracked so that each one is counted once.
- 3Counts are aggregated by time, station, shift or shipment.
- 4Discrepancies against targets or orders can be flagged with clips for review.
Existing IP cameras / NVR
RTSP / ONVIF streams
Edge device on-site
AI models + rules; video stays local
Events
Time, camera, zone, evidence frame
Dashboard, alerts, API
Notifications, relays, REST/webhook
What can be detected
- Item count per line, station or shift
- Throughput per hour
- Count discrepancy against a target or order
- Loading progress at docks
Typical scenarios
- Conveyor output
- Packing and palletising stations
- Dock loading and unloading
- Workstation output in labour-intensive processes
Deployment options
On-premise / edge
Analysis runs on hardware inside your network. Video does not leave the site; only events and metrics are stored or shared.
Cloud
Suitable for distributed, low-camera-count sites where central management matters more than keeping video on-site.
Hybrid
Video is processed on-site, while events, dashboards and multi-site reporting are managed centrally.
Working with existing cameras
- Items should be separable in the image; heavy overlap on conveyors reduces accuracy.
- A view perpendicular to the flow usually gives the most reliable counts.
- Most IP cameras and NVRs that provide RTSP or ONVIF streams can be connected, regardless of brand.
- Before a pilot, sample footage from each candidate camera is reviewed for resolution, angle, distance and lighting.
- Where a view is not suitable, the recommendation is usually to reposition the camera or add one — not to replace the whole system.
Integration
- REST API / webhook to MES, ERP or WMS
- Shift report exports
- Dashboard
Privacy and security
- No face recognition is used in safety and operations analytics; events describe what happened, not who.
- On-premise processing keeps video inside your network.
- Evidence clips, retention periods and user access are configurable and should follow your GDPR/KVKK policies.
- Workplace monitoring should be introduced with clear employee information and, where applicable, consultation with employee representatives.
Limitations
- Stacked or fully overlapping items may need a different camera angle or process change.
- Count accuracy is validated against manual counts during the pilot.
How a pilot works
- 01
Discovery
We agree on the problem, the owner, the sites and what a successful outcome would look like.
- 02
Camera assessment
Sample footage from candidate cameras is reviewed for angle, resolution, distance and lighting.
- 03
Use-case definition
Events, zones, rules, alert recipients and success metrics are written down before deployment.
- 04
Pilot deployment
Edge hardware or cloud processing is set up and connected to the selected cameras.
- 05
Validation
Detections are checked against reviewed footage; true detections, false alarms and misses are counted per camera.
- 06
Report
Results, limitations and recommended changes are documented and shared with all stakeholders.
- 07
Rollout decision
Based on measured results, you decide whether and how to expand to more cameras, sites or use cases.
Frequently asked questions
Can it count different product types separately?
If the product types are visually distinguishable, a model can be trained to classify them. Feasibility is assessed on sample footage.
How is counting accuracy checked?
Camera counts are compared to manual or system counts over defined periods in the pilot, and the difference is reported per camera.
Related solutions
Case studies
Start with the cameras you already have
Send us a few frames from your cameras and the problem you want to solve. We will tell you honestly what is feasible.