🛡️ WhizSmart AI Vision Safety

AI Vision Safety Monitoring for Indoor Water Parks

Name: WhizSmart AI Safety Edge Box
Brand: WhizSmart
SKU: WS-AI04
Availability: InStock

WhizSmart AI Safety Edge Box turns your existing cameras into a 24/7 safety assistant

Fall · Running · Fighting · Smoking · Smoke & Fire — alerts in under 500 ms

Get a Quote WhatsApp

Works with existing cameras <500 ms latency Open API

Indoor water park monitored by WhizSmart AI vision with detection boxes

The Challenge

Why manual CCTV is not enough

Manual monitoring is tiring

High visitor flow and many screens: staff cannot watch every area in real time, so incidents get missed.

Mist and glare confuse detection

Water mist, reflections and changing light make abnormal behaviour hard to identify accurately.

No early warning

Slips, running and smoking go unwarned; teams can only react after an accident.

Solution Architecture

Cameras → Edge Box → Safety Platform

Cameras capture video, the AI edge box analyses it in real time and pushes alerts to the management platform — fully automated. Edge boxes suit small-to-medium deployments with low-latency local processing; deploy one box per park zone (e.g. Zone A / Zone B).

Existing IP camerasMainstream brands, no replacement

Switch / NVRVideo streams per zone

AI Safety Edge BoxLocal real-time analysis <500 ms

Safety Management PlatformAlert display & data dashboard

Pool edgesSlidesLazy riverWet walkwaysLobby & corridorsMonitoring centre

5 Detection Scenarios

What the system detects

Millisecond-level response: real-time stream analysis latency under 500 ms, so risks are found first.

Fall Detection

Where: Wet walkways and slip-prone areas; outdoor activity areas, lobbies, corridors, walkways, entrance stairs, lift halls.
Capability: Full frame or up to 4 closed zones per camera; alarm when a person is detected in a fallen state.
Definition: Hands and knees on the ground, or half the body down. Squatting does not count.

Running

Where: Strictly controlled areas such as wet walkways and slip-prone zones.
Capability: Full frame or up to 4 closed zones; detects front, back and side views.
Definition: Moving at about 2 m/s — light jogging with bent, swinging arms.

Fighting

Where: Public activity areas.
Capability: Full-frame detection.
Definition: 2–4 people in close physical contact with violent interaction.

Cannot tell whether weapons are involved.

Smoking

Where: Flammable equipment and facility areas; indoor offices and meeting rooms.
Capability: Per camera, up to 4 no-smoking zones and 4 masked (smoking-allowed) zones.
Definition: A cigarette held to the mouth or held in the mouth.

Objects are not distinguished — a pen held in the mouth may also be flagged.

AI smoke and fire detection: heavy black smoke above a building

Smoke & Fire

Where: Indoor fires, open-air smoke venting, heavy black smoke; indoor and outdoor.
Capability: Real-time video analysis alarms as soon as fire appears.
Definition: Red flames; black, yellow and white smoke.

High accuracy

End-to-end false-alarm suppression, from algorithm production to on-site delivery.

Full coverage

Highly adaptable to indoor and outdoor environments.

Open API

Alert data can be reported to third-party business platforms via API.

Edge Box Capabilities

More than five scenarios

System functions

Face capture and recognition, video structuring, perimeter alerts and behaviour alerts; multiple algorithms run in parallel and can be selected as needed.

Perimeter alerts

Person/vehicle line crossing, wall climbing, intrusion, loitering, on/off-post checks, no-parking and vehicle departure.

Behaviour alerts

Smoking, running, falling, fighting, phone calls, phone use, under/over-staffing, sleeping on duty, crowd gathering.

Video structuring & custom algorithms

Capture and attribute extraction for faces, people, vehicles and non-motor vehicles, licence-plate recognition; industry algorithms customised on demand, new risk types can be added later.

Specifications

WhizSmart AI Safety Edge Box

Models

WS-AI04 / WS-AI08 / WS-AI16 (4 / 8 / 16 channels)

Analysis capacity

Up to 16 × 2MP video streams + 32 face-camera image streams; 16-ch face, structuring, perimeter & behaviour; 50 QPS face capture

Latency

< 500 ms per real-time video stream

Detection zones

Full frame or up to 4 custom zones per camera

Camera access

Mainstream IP camera brands via ONVIF / RTSP / GB28181, H.264 / H.265; GA/T 1400 for face-capture cameras

Processor

Single AI SoC, fanless

Enclosure

All-metal cast aluminium alloy

Interfaces

2 × Gigabit Ethernet, 4 × USB, 4-in / 4-out alarm I/O, 1 × HDMI, 1 × Line-in, 1 × Line-out

Deployment Modes

Fits your network

Intranet platform

Cameras connect to the box via RTSP / GB28181 (H.264/H.265); the box reports to the on-premises platform over WebAPI.

Extranet / cloud platform

The box actively registers with the cloud platform and reports alerts over WebAPI — no fixed public IP needed on site.

Third-party integration

Open API lets alerts flow into your existing safety or smart-park systems.

Value

From passive response to active prevention

Safety management upgrade

From passive response to active prevention. Lower operating risk, better guest experience and reputation, and a solid base for sustainable operation.

Future direction

Personnel positioning for precise visitor location and multi-layer protection; integration with smart-park management for linked data and one unified operation.