People tracking in the warehouse: safety and flow, without watching anyone.
Computer vision that detects people and forklifts, alerts when their paths cross and shows how the operation really moves.
- 01Real-time alert when a person enters a forklift zone.
- 02Builds walking-path heat maps to redesign layout and picking.
- 03Works with anonymous silhouettes: it does not identify people.
The problem
Pedestrians sharing space with forklifts is one of the main risks in a warehouse. Signs and floor markings help, but they do not warn anyone in the moment.
Few operations also know where their teams actually walk. Computer vision answers both questions with the cameras you already have, or a few new ones.
We deploy it in warehouses in Argentina and Spain, adapting the design to each country’s privacy and workplace safety rules.
Moving silhouettes, a heat map and an alert when it matters.
- 01 · DetectionCameras detect people and equipment as silhouettes, without recognizing faces.
- 02 · PathsEvery step adds to the warehouse heat map.
- 03 · AlertIf a person enters a forklift zone, an alarm sounds.
Where it applies
- Pedestrian and forklift safetyAlerts at crossings, aisles and docks.
- Heat mapsReal walking paths to redesign layout and slotting.
- Picking timesDistance walked per zone and shift, in aggregate.
- PPE and zone controlVests and helmets in mandatory areas, restricted access.
The model is the easy part. The hard part is the operation.
- Privacy
- Anonymization, no facial recognition, and prior notice to workers and their representatives. In Argentina, Law 25.326 (personal data protection) and AAIP guidance (the data protection authority); in Spain, GDPR, LOPDGDD (the national data protection act) and an impact assessment.
- Cameras
- Coverage of aisles and critical crossings, with on-site processing where possible.
- Integration
- Beacons, lights or sirens on the floor, and an event log for health and safety / occupational risk prevention.
- Operations
- Data used in aggregate to improve processes, not to evaluate individuals.
What we build it with
- Ultralytics YOLOObject detection
- OpenCVImage processing
- NVIDIA JetsonEdge computing
- n8nAutomation and alerts
- GrafanaDashboards and monitoring
Reference tools. On each project we choose based on the architecture and systems the company already runs.
Yes, if designed properly. In Argentina, Law 25.326 on personal data protection and AAIP Provision 10/2015 on video surveillance apply: clear purpose, signage and proportionate data. In Spain, GDPR and LOPDGDD apply: a legal basis, prior notice to workers and an impact assessment. In both cases we design it with your legal team from day one.
No. It detects silhouettes and movement. Nobody needs to be identified to trigger alerts or measure flows.
Often yes, if position and image quality are good enough. We assess it in a site survey.
