Pedestrian-Vehicle Interaction
Pedestrians entering active vehicle lanes, vehicles reversing toward people, parking that blocks sightlines.
Near-Miss Intelligence and Proactive Safety Training for HSE teams
See the near-miss risk in your facility, and act on it before it becomes an incident.
OTDMT connects to the CCTV your facility already runs and identifies near-miss precursors as they happen. It groups those events into site-specific risk intelligence, shows the HSE team the anonymized evidence behind each pattern, drafts targeted training built from the facility's own situations, and measures whether the same risk declines after the training is delivered.
- 0readable faces in 5,906 measured frames
- 14 dayscontinuous run, 8 cameras, 0 dropped frames
- On-premvideo is processed and stored on site
The visibility gap
Most dangerous situations never reach a report.
Safety programs record two kinds of events well. The first is the incident that causes injury, damage, or a stoppage. The second is the near miss that a worker takes the time to write up. Between those two sits a large set of events that happen and then disappear, with no structured record and no learning attached to them.
These unrecorded situations carry leading-indicator information that HSE teams rarely get to use. A vehicle enters a pedestrian path at the wrong moment. A forklift passes close behind a pedestrian at a blind corner. A delivery vehicle moves while loading is still underway. Each of these is a signal, and most of them are gone by the end of the shift. A facility cannot improve a risk it cannot observe with any regularity.
Why existing CCTV matters
Your cameras already watch the areas where risk happens.
Industrial sites, warehouses, and fuel stations already operate extensive camera coverage over the exact areas where vehicles, people, equipment, and hazardous materials interact. In most facilities this footage is reviewed only after something goes wrong. OTDMT reads relevant safety situations from that existing coverage and writes them as structured events, without new hardware on the forecourt or the warehouse floor.
Existing cameras
The coverage you already own
RTSP streams from the site NVR.
OTDMT
One server, inside the facility
Detection runs in memory. Faces and plates are removed before anything is stored or shown.
Relevant safety events
Class, zone, time, evidence
Only in the areas and hazard categories the HSE team defines.
The OTDMT closed-loop safety engine
One loop, from what the cameras observe to whether the risk went down.
OTDMT runs as a continuous loop with five stages. Each stage produces something the next stage uses, and the HSE professional stays in control of every decision that leaves the platform.
MEASUREMENT RETURNS TO DETECTION
- 01
Observe and Detect
OTDMT connects to existing CCTV and detects near-miss precursors using computer vision that runs at the edge, inside the facility. Faces and identifying markers are anonymized. Each relevant situation becomes a structured safety event with a class, a location, and a time.
- 02
Understand and Prioritize
Events are grouped by location, frequency, severity, and recency. OTDMT maintains a Site Risk Index that ranks where attention is needed, using a five-level severity scale from L1 to L5, and surfaces hotspots and recurring patterns.
- 03
Learn and Draft
OTDMT can use anonymized site footage to draft short training material tied to the situations it has recorded: 60 to 90 second micro-training modules, toolbox talks, and knowledge checks that reflect the facility's real layout.
- 04
Teach and Review
The HSE professional reviews every draft, edits it, adds local rules and context, approves it, and decides how it is delivered. OTDMT prepares the material. The specialist owns the content and the delivery.
- 05
Verify and Measure
After training is delivered, OTDMT tracks whether the same class of near-miss precursor occurs less often at that location. Measurements can be normalized against operating hours so busy and quiet periods compare fairly.
Hazard and event taxonomy
A structured taxonomy for forecourt, warehouse, and industrial risk.
OTDMT organizes detected situations into a consistent set of classes. Consistent classes make events countable, comparable across locations, and usable as training inputs. The examples under each class are representative.
Vehicle and Equipment Incidents
Vehicles grazing bollards or dispenser islands, stretched hoses across paths, departures with a connected line.
Fuel and Hazardous Material Release
Improper nozzle handling during active flow, coupling leaks, pooling on the surface, overfill trips.
ATEX and Ignition Sources
Smoking, e-cigarette use, or open flame in a hazardous zone, engines left running during refueling.
Delivery and Loading Bay Operations
Unregistered vehicles in delivery zones, foot traffic near manifolds, missing barriers or cones during offloading.
PPE and Lone Worker Safety
Missing high-visibility or eye protection in drop zones, entry to restricted areas, lone-worker fall or immobility.
PHOTOS: REPRESENTATIVE ENVIRONMENTS FROM A LIVE FORECOURT, FACES AND PLATES REDACTED.
Privacy by design and worker trust
The system observes situations, and it is built to protect the people in them.
OTDMT processes video at the edge, inside the facility. Faces and identifying visual markers are anonymized, so the records the HSE team works with show the situation and the environmental context rather than an identifiable person. OTDMT is designed with privacy considerations aligned with GDPR and EU AI Act expectations for workplace systems.
Customers, attendants and drivers are redacted before any frame is stored or shown. Vehicle plates too.
Human review and HSE control
OTDMT prepares the work. The HSE professional decides.
- Named review. Every event is reviewed, labelled and approved by a named person before it becomes an action.
- Dry-run by default. Every automated response ships switched off until an operator turns it on.
- Full audit trail. Every review, approval and configuration change is written to the audit log.
Measuring whether the intervention worked
Detection has value when you can see what changed after you acted.
OTDMT is built to connect detection, intervention, and the measurement that follows. It supports a small set of indicators that an HSE team can track over time.
- Actionable Near-Miss Precision. How consistently detected events lead to a useful safety action rather than noise.
- Repeat-Precursor Reduction. Whether the same class of situation occurs less often after an intervention.
- Verified Risk Drop per Operating Hours. How a risk pattern trends when normalized against operating activity.
- Training-to-Risk Relationship. Whether targeted training coincides with a measurable change in the pattern it addressed.
Repeat-precursor rate · one pattern · one locationIllustrative
Before interventionAfter, same location, same class
ILLUSTRATIVE CURVE. YOUR SITE'S TREND IS BUILT FROM ITS OWN EVENTS.
FAQ
Questions HSE teams ask about OTDMT.
Does OTDMT require new cameras?
OTDMT is designed to work with the CCTV a facility already operates in the relevant areas. The camera coverage a site has becomes the input for detection.
How does CCTV integration work?
OTDMT connects to existing camera sources for the monitored areas. The specific integration is set up with the site during deployment, based on the cameras and areas the HSE team wants to cover.
Where is video processed?
Video is processed at the edge, inside the facility.
How are faces handled?
Faces and identifying visual markers are anonymized during processing, so the records the HSE team works with show the situation and its context.
Does OTDMT monitor individual employees?
OTDMT is designed to observe safety-relevant situations and environmental conditions in defined areas. With anonymization applied at the edge, the platform is oriented toward hazards and interactions rather than tracking individual workers.
How does the HSE team review events?
The team works from a prioritized view of events, with anonymized evidence, class, location, and severity for each pattern, and reviews and approves any action before it reaches a worker.
See the recurring risk, understand it, train against it, and measure what changes.
OTDMT turns the situations your cameras already observe into structured near-miss intelligence, targeted site-specific training, and a measurement of whether the risk declined. Your HSE team stays in control of every decision. The next step is a conversation about your site and the areas you want to cover.
Prefer email? hello@otdmt.tech