অক্ষি OKKHI

On-premise vision AI

000Loading

Loading OKKHI

SOLUTIONS / OKKHI SENTINEL

Okkhi Sentinel

Surveillance that understands.

Safety and security analytics that turn the CCTV you already own into a system that acts. Temporal voting suppresses false alarms; an on-premise vision-language model verifies high-stakes alerts before they reach a human — and fails closed, never open.

BRANCH / FACTORY / CAMPUS · EDGE GPU · HUNDREDS OF CAMERAS PER SERVER

CAM_31 · ASSEMBLY FLOOR · PPELIVE

Capabilities

Threats and hazards, caught while they matter.

Each module runs on the CCTV you already own, verifies what it sees before raising an alert, and sends the snapshot to whoever needs to act.

Fire & Smoke Detection detection view SEN-01

Fire & Smoke Detection

Early visual detection of flame and smoke signatures — your cameras see the fire before a ceiling sensor smells it.

PPE Compliance detection view SEN-02

PPE Compliance

Helmets, vests, harnesses — continuous compliance monitoring with per-zone rules and shift reports.

Weapon & Threat Detection detection view SEN-03

Weapon & Threat Detection

Detects weapons and threat behavior, with on-premise VLM verification of every alert before escalation.

Intrusion & Zone Watch detection view SEN-04

Intrusion & Zone Watch

Person-in-restricted-area detection with calibrated zones, schedules, and instant notification.

Camera Tampering Watch detection view SEN-05

Camera Tampering Watch

Detects covered, defocused, or redirected cameras — the blind spot every attacker creates first.

Banking & ATM Security detection view SEN-06

Banking & ATM Security

Branch counter and ATM booth analytics: vandalism, suspicious entry, after-hours activity — built so footage and metadata never leave the branch.

Vehicle Overspeed Detection detection view SEN-09

Vehicle Overspeed Detection

Measures vehicle speed from ordinary cameras across calibrated zones and flags overspeed instantly — no radar required.

Collision & Proximity Warning detection view SEN-10

Collision & Proximity Warning

Predicts vehicle–pedestrian and vehicle–vehicle near-misses from time-to-collision and warns before contact, not after.

Fall & Lying-Person Detection detection view SEN-11

Fall & Lying-Person Detection

Detects a person down — collapsed, fallen, or lying where no one should be — and raises help in seconds.

Face Recognition Access detection view SEN-12

Face Recognition Access

Recognizes enrolled faces for door access, watchlists, and VIP service — entirely on-premise, no cloud matching.

Unsafe Behavior Detection detection view SEN-13

Unsafe Behavior Detection

Spots risky acts as they happen — skipped handrails, phone use in risk zones, objects thrown from height.

Visitor & Service Analytics detection view SEN-14

Visitor & Service Analytics

Counts visitors at every entrance and desk, anonymously, from the cameras already on the ceiling.

Theft & Asset Protection detection view SEN-15

Theft & Asset Protection

Catches theft as it happens — a hand reaching across a counter, an asset walking away, a pickpocket pattern forming.

Suspicious Entry & Unusual Activity detection view SEN-16

Suspicious Entry & Unusual Activity

Flags the entry that doesn't look right — covered faces, helmets indoors, loitering, out-of-hours presence — before anything happens.

Distress & Aggression Detection detection view SEN-17

Distress & Aggression Detection

Reads visible distress and aggression at service points and raises help while the situation is still unfolding.

Crowd & Mob Detection detection view SEN-18

Crowd & Mob Detection

Tracks crowd density in real time and flags the moment a gathering turns into a mob formation.

Staircase & Handrail Safety detection view SEN-19

Staircase & Handrail Safety

Watches staircases for handrail use and unsafe movement — the fall-prevention layer most safety programs skip.

Don't see your exact use case here?

Most of these started as a customer asking for something that wasn't on the list. Describe the problem and we'll tell you honestly whether vision AI is the right tool for it.

Let's Talk

Why teams choose it

Alerts your team will actually trust.

Verified before it pages

Temporal voting requires a detection to hold across frames, and an on-premise vision-language model checks high-stakes events before a human is disturbed.

Fails closed, never open

If a camera drops or a model is unsure, the system flags it rather than silently passing. Silence never means 'all clear' by accident.

Hundreds of feeds per server

One edge GPU server covers a large camera fleet, so adding a second or third module costs a fraction of the first.

Nothing leaves the building

Decode, inference, and storage all happen on-premise. Data-residency requirements are met by architecture, not by promise.

How a deployment runs

From first call to live system.

01

Camera audit

We review your existing fleet and tell you honestly which views work today and which need repositioning.

02

Zone and rule setup

Restricted areas, PPE requirements, and operating hours are configured per camera with your safety and security teams.

03

Tuning against your footage

Thresholds and voting windows are tuned on your real conditions until the alert volume is one your team will act on.

04

Handover

Alerts route to your channels, and the health dashboard shows every camera and module at a glance.

Free 2-week pilot · you keep the report Let's Talk

Proof beats promises.
See it on your own cameras.

We run a two-week pilot on your footage, with your incidents, measured against your numbers. If the results don't convince you, you keep the full report and owe nothing.

Book a live demo

For teams deploying vision AI

Field notes from real deployments.

Once a month: what we learned shipping AI monitoring on factory floors and branch networks in Bangladesh — what worked, what broke, what we'd do differently. No product pitches.

No spam. Unsubscribe any time.