Do we have to replace our cameras to run forklift and PPE detection?

Usually no. Camera-agnostic analytics run on existing fleets, including five-to-seven-year-old cameras, as long as the stream clears a basic resolution and framerate threshold, roughly 1080p at 15 frames per second over ONVIF or RTSP. Inference runs on a separate server or in the cloud, so the cameras are never touched. The consultation walks the floor and flags only the handful of cameras that need a better sightline before the model will work.

What makes this different from the analytics built into our VMS?

A standard VMS records and lets you search after an incident. It answers what happened. A safety-vision build is tuned to prevent the incident: it watches the spatial relationship between forklifts and pedestrians, enforces PPE rules zone by zone, and flags an unsafe condition in real time so a supervisor can act before someone is hurt. When a packaged analytics pack does not fit your specific floor geometry or rules, Tec-Tel builds the detection logic to your site rather than forcing a stock product to approximate it.

How is the model tuned to our specific floor?

It starts with a walk-through. Tec-Tel maps travel lanes, blind corners, dock approaches, exclusion zones, and the PPE rules that change from bay to bay, then defines detection rules per zone and per shift. A pilot zone goes live first and the confidence thresholds are tuned against your real false-positive baseline until the alerts are trustworthy, because a false alarm trains the floor to ignore the real one. Expansion happens zone by zone once supervisors are acting on what they already get.

Where do the alerts actually go?

Wherever your team already works. A forklift-pedestrian conflict or a missing hard hat can route to the floor supervisor on shift, drop a daily digest to the safety lead, write to a coaching log, or trigger a reaction in the access-control system. A detection that lands somewhere nobody reads is worthless, so routing is part of the design, not an afterthought.

Is this employee surveillance?

Detection runs on classification, a person without a hard hat, a pedestrian in a forklift lane, rather than identity, a named employee. Most operator-side EHS programs treat that as fundamentally different from facial recognition, and Tec-Tel defaults away from biometrics unless the customer has a documented, consent-based program. The data flow is documented during deployment so HR, legal, and any union can review it before go-live. Illinois sites get extra scrutiny under BIPA even when face geometry is never touched.

Will our insurance carrier give us a premium credit for this?

Many carriers offer premium credits when an operator can document camera-based safety controls, though the credit is never automatic and the amount varies by carrier and program. The broker submits the written control summary, the retention policy, and the alert-routing matrix to the underwriter as evidence. Tec-Tel produces those deliverables as part of the install. The broker handles the carrier conversation.

How long does a build take to stand up?

A pilot zone goes from kickoff to first tuned alerts quickly. A floor-wide rollout takes longer, gated mostly by the time it takes to walk every zone with the safety lead and define the rules. Tec-Tel handles the deployment and the first round of tuning, then hands off a written runbook so the safety team owns the alert definitions going forward.

Do we have to replace our cameras to run forklift and PPE detection?

Usually no. Camera-agnostic analytics run on existing fleets, including five-to-seven-year-old cameras, as long as the stream clears a basic resolution and framerate threshold, roughly 1080p at 15 frames per second over ONVIF or RTSP. Inference runs on a separate server or in the cloud, so the cameras are never touched. The consultation walks the floor and flags only the handful of cameras that need a better sightline before the model will work.

What makes this different from the analytics built into our VMS?

A standard VMS records and lets you search after an incident. It answers what happened. A safety-vision build is tuned to prevent the incident: it watches the spatial relationship between forklifts and pedestrians, enforces PPE rules zone by zone, and flags an unsafe condition in real time so a supervisor can act before someone is hurt. When a packaged analytics pack does not fit your specific floor geometry or rules, Tec-Tel builds the detection logic to your site rather than forcing a stock product to approximate it.

How is the model tuned to our specific floor?

It starts with a walk-through. Tec-Tel maps travel lanes, blind corners, dock approaches, exclusion zones, and the PPE rules that change from bay to bay, then defines detection rules per zone and per shift. A pilot zone goes live first and the confidence thresholds are tuned against your real false-positive baseline until the alerts are trustworthy, because a false alarm trains the floor to ignore the real one. Expansion happens zone by zone once supervisors are acting on what they already get.

Where do the alerts actually go?

Wherever your team already works. A forklift-pedestrian conflict or a missing hard hat can route to the floor supervisor on shift, drop a daily digest to the safety lead, write to a coaching log, or trigger a reaction in the access-control system. A detection that lands somewhere nobody reads is worthless, so routing is part of the design, not an afterthought.

Is this employee surveillance?

Detection runs on classification, a person without a hard hat, a pedestrian in a forklift lane, rather than identity, a named employee. Most operator-side EHS programs treat that as fundamentally different from facial recognition, and Tec-Tel defaults away from biometrics unless the customer has a documented, consent-based program. The data flow is documented during deployment so HR, legal, and any union can review it before go-live. Illinois sites get extra scrutiny under BIPA even when face geometry is never touched.

Will our insurance carrier give us a premium credit for this?

Many carriers offer premium credits when an operator can document camera-based safety controls, though the credit is never automatic and the amount varies by carrier and program. The broker submits the written control summary, the retention policy, and the alert-routing matrix to the underwriter as evidence. Tec-Tel produces those deliverables as part of the install. The broker handles the carrier conversation.

How long does a build take to stand up?

A pilot zone goes from kickoff to first tuned alerts quickly. A floor-wide rollout takes longer, gated mostly by the time it takes to walk every zone with the safety lead and define the rules. Tec-Tel handles the deployment and the first round of tuning, then hands off a written runbook so the safety team owns the alert definitions going forward.

Solution · Custom computer vision

Catch the forklift near-miss before it becomes a recordable.

Custom computer vision for forklift-pedestrian conflict and PPE compliance, tuned to your floor and running on the cameras you already own. When the off-the-shelf VMS only records the incident, Tec-Tel builds the layer that prevents it.

Start the conversation

or reach us directly

Call us855-577-0400

NDAA-compliant
Platform-agnostic
1,000+ deployments over 15 years

Forklift and PPE safety vision is a custom computer-vision layer that flags forklift-pedestrian conflicts, speeding, and missing PPE in real time so a supervisor can act before someone is hurt. The National Safety Council counts 84 forklift-related deaths in 2024 and OSHA estimates roughly 70 percent are preventable. Tec-Tel builds the model to your floor's zones and rules when an off-the-shelf VMS does not fit, and runs it on the cameras you already own. Free consultation.

forklift-related worker deaths in 2024 (National Safety Council, Injury Facts)

~25,110

nonfatal forklift injuries involving days away from work (NSC, Injury Facts)

~70%

of forklift incidents OSHA estimates are preventable with stronger controls and training

No rip-and-replace

the model runs on the cameras already mounted over the floor

§01 What the model detects

Six detection domains tuned to the floor.

Not every floor needs every domain. The consultation maps your incident log and travel lanes, then picks the few that move the recordable rate. All of them run on the cameras already mounted over the floor, no rip-and-replace.

→

Forklift and pedestrian conflict A pedestrian inside a forklift travel lane, a near-miss at a blind corner, a person on foot in an active charging or staging aisle. The model watches the spatial relationship between trucks and people in real time and flags the conflict before it becomes a struck-by or a pinned-against recordable.

→

PPE compliance by zone Hard hats, hi-vis vests, safety glasses, hearing protection, gloves. Rules run zone by zone, so an office wing with no hard-hat requirement never fires, and a hot-work or chemical bay enforces every item. A missed item routes to the supervisor on shift, not into a void.

→

Speed and unsafe operation Forks raised while traveling, speeding inside the building, no-honk at a marked intersection, a truck entering a pedestrian-only zone. The model turns what a floor supervisor can only catch by standing there into a logged, coachable event.

→

Pedestrian exclusion zones A drawn boundary around a press, a robot work envelope, a moving conveyor, or a lockout-tagout area. Anyone crossing the line is flagged immediately. The same boundary logic covers a forklift entering a zone it should not.

→

Dock and staging-lane safety Trailer-creep while a truck is loading, a person between a forklift and a fixed object at the dock, an unsafe approach to an open dock door. The areas where the most expensive incidents happen get watched the same way the floor does.

→

Leading-indicator trend reporting Every flagged event feeds a weekly report: which aisle, which shift, which intersection produces the most near-misses. The reduction in recordables comes from acting on that trend before an event happens, not from the alert alone.

§02 The operational problem

The off-the-shelf VMS sees the incident. It does not prevent it.

A standard video management system records everything and answers questions after the fact. When a forklift clips a pedestrian at a blind corner, the footage exists, so the investigation has a clip. That is useful for the claim. It does nothing for the worker who was hit.

The National Safety Council records 84 forklift-related worker deaths in 2024 and roughly 25,110 nonfatal injuries serious enough to cost days away from work. OSHA estimates close to 70 percent of forklift incidents are preventable with stronger controls and training. The gap between what a plain VMS does and what the floor needs is exactly that prevention layer.

Most off-the-shelf analytics packs are built for a different question: a retail people-counter, a license-plate reader, a perimeter line-cross. They were not trained on the geometry of a forklift sharing a 12-foot aisle with a picker, or on the PPE rules that change from one bay to the next inside the same building. When the stock product does not fit the operational problem, the usual answer is to buy more cameras and hope. That is the wrong answer.

§03 How the build works

Tec-Tel builds the model to your floor, on the cameras you already own.

Tec-Tel is a one-stop integrator, and the software-and-monitoring layer includes building custom computer vision when the off-the-shelf stack does not solve the problem. For forklift and PPE safety, that usually means starting from a strong EHS-focused analytics platform and tuning it to your specific zones, sightlines, and rules, or building the detection logic directly where a packaged product has no answer for your layout.

The model reads the existing camera feeds. Cameras built in the last several years that output a usable stream over ONVIF or RTSP at roughly 1080p and 15 frames per second carry the analytics with no hardware change. Inference runs on a separate server or in the cloud, so the cameras themselves are never touched. Where a sightline is genuinely missing, the consultation flags the handful of cameras to add, not the fleet to replace.

A detection is only useful if it lands somewhere a person acts on it. Tec-Tel wires the output into where your team already works: an alert to the floor supervisor on shift, a daily digest to the safety lead, an entry in a coaching log, or a reaction in the access-control system. The event becomes part of the safety program, not a notification nobody reads.

→ Walk the floor with the EHS lead. Map travel lanes, blind corners, dock approaches, PPE zones, and exclusion areas before a single rule is written.
→ Define the rules per zone and per shift, because a model tuned site-wide produces noise and a model tuned zone by zone produces action.
→ Pilot one zone first. Tune confidence thresholds against your real false-positive baseline until the alerts are trustworthy, because a false alarm trains the floor to ignore the real one.
→ Expand zone by zone once supervisors are acting on what they already get, then hand off a written runbook the safety team owns.

§04 The ROI angle

The return is the recordable that never happens.

A single medically-consulted workplace injury is expensive before the claim, the lost production, and the replacement training are counted. A forklift struck-by event sits at the severe end of that range. The math on a custom safety-vision build is straightforward: the system pays for itself the first time it prevents one serious incident, and leading-indicator trend reporting drives that prevention across a year, not a one-time alert.

A second return shows up on the insurance line. Carriers increasingly offer premium credits when an operator can document camera-based safety controls, and evidence-grade video at the moment of an incident changes the negotiating posture on a workers-compensation claim. Tec-Tel produces the written control summary, the retention policy, and the alert-routing matrix a broker submits to the underwriter. The credit is never automatic and varies by carrier, but the documentation makes the conversation possible.

§05 Why custom, not a SKU

When the packaged product fits, we say so. When it does not, we build.

Tec-Tel has no house analytics product to push. For many floors, a tuned EHS platform from an established vendor is the right answer, and the consultation will say so plainly. The custom build matters when your operational problem falls in the gap a packaged product leaves: an unusual aisle geometry, a mixed-traffic zone the stock model misreads, a PPE rule specific to your process, or a detection nobody sells off the shelf.

Because Tec-Tel owns the cameras, the network, and the software layer end to end, the build is not a science project bolted onto someone else's install. The same team that runs the cabling and mounts the cameras tunes the model and stays on for service. One partner, one accountable line, instead of a camera vendor pointing at an analytics vendor pointing at the integrator.

Questions buyers ask us

FAQ

Do we have to replace our cameras to run forklift and PPE detection?: Usually no. Camera-agnostic analytics run on existing fleets, including five-to-seven-year-old cameras, as long as the stream clears a basic resolution and framerate threshold, roughly 1080p at 15 frames per second over ONVIF or RTSP. Inference runs on a separate server or in the cloud, so the cameras are never touched. The consultation walks the floor and flags only the handful of cameras that need a better sightline before the model will work.
What makes this different from the analytics built into our VMS?: A standard VMS records and lets you search after an incident. It answers what happened. A safety-vision build is tuned to prevent the incident: it watches the spatial relationship between forklifts and pedestrians, enforces PPE rules zone by zone, and flags an unsafe condition in real time so a supervisor can act before someone is hurt. When a packaged analytics pack does not fit your specific floor geometry or rules, Tec-Tel builds the detection logic to your site rather than forcing a stock product to approximate it.
How is the model tuned to our specific floor?: It starts with a walk-through. Tec-Tel maps travel lanes, blind corners, dock approaches, exclusion zones, and the PPE rules that change from bay to bay, then defines detection rules per zone and per shift. A pilot zone goes live first and the confidence thresholds are tuned against your real false-positive baseline until the alerts are trustworthy, because a false alarm trains the floor to ignore the real one. Expansion happens zone by zone once supervisors are acting on what they already get.
Where do the alerts actually go?: Wherever your team already works. A forklift-pedestrian conflict or a missing hard hat can route to the floor supervisor on shift, drop a daily digest to the safety lead, write to a coaching log, or trigger a reaction in the access-control system. A detection that lands somewhere nobody reads is worthless, so routing is part of the design, not an afterthought.
Is this employee surveillance?: Detection runs on classification, a person without a hard hat, a pedestrian in a forklift lane, rather than identity, a named employee. Most operator-side EHS programs treat that as fundamentally different from facial recognition, and Tec-Tel defaults away from biometrics unless the customer has a documented, consent-based program. The data flow is documented during deployment so HR, legal, and any union can review it before go-live. Illinois sites get extra scrutiny under BIPA even when face geometry is never touched.
Will our insurance carrier give us a premium credit for this?: Many carriers offer premium credits when an operator can document camera-based safety controls, though the credit is never automatic and the amount varies by carrier and program. The broker submits the written control summary, the retention policy, and the alert-routing matrix to the underwriter as evidence. Tec-Tel produces those deliverables as part of the install. The broker handles the carrier conversation.
How long does a build take to stand up?: A pilot zone goes from kickoff to first tuned alerts quickly. A floor-wide rollout takes longer, gated mostly by the time it takes to walk every zone with the safety lead and define the rules. Tec-Tel handles the deployment and the first round of tuning, then hands off a written runbook so the safety team owns the alert definitions going forward.

Book a walkthrough

Have a forklift-pedestrian problem an off-the-shelf system won't solve?

The free consultation walks your floor, maps the blind corners and PPE zones against the cameras already on site, and tells you whether a tuned platform fits or a custom build is the right call. Either way, one partner owns the cameras, the network, and the model.

Tell us how many sites you run and what's already in place. We'll show you what a build or upgrade looks like.
Straight answers from the team that does the work. We're platform-agnostic, so you get the system that fits your sites, not one brand's catalog.

Since 2010 · 1,000+ deployments nationwide · ISN-accredited

See it on your cameras 855-577-0400

Or send the details

How can we help?

What you're looking for, plus any details. We review it and follow up, usually the same day.

Related from Tec-Tel

Hub

Catch the forklift near-miss before it becomes a recordable.

Six detection domains tuned to the floor.

The off-the-shelf VMS sees the incident. It does not prevent it.

Tec-Tel builds the model to your floor, on the cameras you already own.

The return is the recordable that never happens.

When the packaged product fits, we say so. When it does not, we build.

FAQ

Have a forklift-pedestrian problem an off-the-shelf system won't solve?

How can we help?

AI & analytics hub

Manufacturing security

Logistics & warehousing

Vision Intelligence

AI software & analytics

Workplace safety