RetaliQ
STORE PERFORMANCE · PILLAR 03

AI Video that reveals real in-store behaviour.

RetaliQ uses AI Video to transform camera images into actionable insights — from customer flows and dwell time to staff movements and real conversion moments.

Live retail intelligence layer
2 fps inference
on every IP camera you already own
$26
CPU baseline
per camera · per month
$5.50
GPU optimized
per camera · per month
0
New cameras required
connects to Dahua, Hikvision, Axis — any IP brand
What it delivers

Turns passive cameras into a live retail intelligence layer.

People Counting & RPV

Counts entries and exits in real time. Calculates Revenue Per Visitor by store, hour, zone and cashier. When RPV drops unexpectedly the system flags it immediately.

Heat Maps & Shopper Behavior

Tracks movement through the store. Identifies zones with high traffic but low conversion. Updated continuously — not twice a year by a consultant.

Hot Zone Management

Detects unusual crowding or conversion drops by zone. Sends a push notification to the floor supervisor's phone. Supervisor acts in the moment, not at end of day.

Crowd & Density Alerts

Continuous occupancy monitoring. Triggers proactive alerts when crowding patterns exceed thresholds — for safety, conversion, or staffing decisions.

Queue Management

Detects checkout queue length and dwell. Triggers staffing alerts before customers walk away. Measurable impact on conversion.

Digital Twin

Reconstructs a real-time floor plan with approximate person positions. Cross-referenced with planogram: identifies ignored zones and layout bottlenecks.

Custom Exception Detection

Define the exception your operation needs in plain language: missing PPE, unauthorized access, raised arms, abandoned objects. The model handles the rest.

Shelf AI (ShelfBreak)

Out-of-stock detection from any camera or smartphone. AI reads the surrounding packaging and identifies the specific SKUs and brands likely missing.

Architecture

Connects to existing IP cameras. No new hardware required.

Step 1
IP Cameras
RTSP / RTMP
CCTV existing infrastructure
Step 2
MediaMTX Server
Stream broker
RTMP in · RTSP / HLS out
Step 3
AI Worker
YOLOv8n · 2 fps
People counting · Zone detection
Step 4
TimescaleDB
PostgreSQL + ext.
People counts · Queue events
Step 5
Dashboard + PWA
Role-based alerts
No app installation needed

Deployed on GCP, AWS, Azure or on-premises. GPU optional. CPU baseline is enough for production.

AI Vision · brand-agnostic

AI Vision for any camera. Any brand. Any VMS.

The AI Video pillar is built as an intelligence layer that sits on top of your existing CCTV — not next to it. Plug into any IP camera or RTSP/ONVIF stream, define the exception your client cares about in plain language, and route alerts to the channels they already live in.

100% brand-agnostic

Hikvision, Dahua, Axis, Bosch, or any RTSP/ONVIF source. No proprietary lock-in. Perfect for the mixed environments that are the reality of most installations.

RTSP · ONVIF · HLS · RTMP

Exception intelligence

Not a generic detection box. We configure the exact exception each client needs — raised arms, missing PPE, unauthorized access, abandoned objects — whatever matters to them.

Precision · Low false-positive rate

Acts where teams already work

Alerts arrive in Telegram, incident management systems, webhooks, or email — wherever the operations team already lives. Zero new software for end users.

Telegram · Webhooks · SIEM · Email
Retail
Queue overflow · Abandoned carts · POS suspicious behavior · Stockroom access
Manufacturing
PPE compliance · Restricted zones · Unsafe postures · Near-miss events
Logistics
Cargo security · Forklift zone violations · Loading dock validation
Banking
ATM loitering · Tailgating · Branch unusual behavior
AI Video Lab

Extensions of the AI Video pillar — built on real retail problems.

The AI Video architecture is extensible by design. As partners and customers surface new problems, RetaliQ codifies them into new AI capabilities that share the same backend, the same PWA, and the same role-based alerting.

NEW AI Video Lab

ShelfBreak — Out-of-Stock Detection

On-shelf availability without RFID, without robots.

An employee walks the aisle with a phone. On-device computer vision flags candidate gaps; cloud AI confirms, classifies the category, identifies the visible brands and the SKUs likely missing. Auto-routed restock task to the floor team — loop closed in under five minutes.

  • No new cameras. No IT integration.
  • Runs on the Android phones stores already have.
  • Tested in supermarkets across Chile and Morocco.
  • Identifies real brands: Soprole, Colgate, Coca-Cola, Whiskas, Parents Choice…
PILOT AI Video Lab

Human Intelligence Layer

Every person in the store can document an incident.

Floor staff use the PWA to capture a photo or short video. The system auto-pulls nearby camera clips for the reported timestamp and routes the case to the right role. For staff without phones — IoT buttons (WiFi or LoRa) trigger the same workflow.

  • Photo or short video, captured on the phone.
  • Auto video enrichment from adjacent cameras.
  • IoT buttons with 3 press protocols (assistance / security / emergency).
  • Same incident queue as auto-detected events.
Featured Lab solution · ShelfBreak

From walkthrough to restock — four steps, one phone, no extra hardware.

  1. 1

    Walk

    Staff walks the aisles with the mobile app in Employee Mode.

  2. 2

    Detect

    Real-time on-device analysis flags empty shelves automatically.

  3. 3

    Capture

    Three photos (zoom, normal, wide) plus metadata are saved as an incident.

  4. 4

    Resolve

    Manager reviews in Management Mode, assigns and tracks the restock.

No new cameras. No IT integration. Runs on the Android phones stores already have.

Live preview

The actual production app, running below.

Real incidents, real AI classifications, captured in supermarkets last week.

app.retaliq.net / manage