IV Sentinel AI — IntelliVision

BANKING SECURITY

NEW

IV Sentinel AI for ATM Security

Edge-based Smart Motion and Vision Language AI for detecting abnormal ATM behavior, generating in-context scene descriptions, and keeping AI video analysis on-premise.

Schedule a Demo

Download Datasheet

* Smart Motion + VLM Reasoning * On-premise edge appliance

VLM SCENE DESCRIPTION

“A hooded individual wearing a mask and dark clothing is forcefully interacting with the ATM. The behavior indicates tampering is in-progress.”

THE CHALLENGE

ATM attacks are more than motion events

Traditional motion detection can identify movement, and deep learning models can detect known objects. But these systems cannot explain whether behavior is normal customer activity or abnormal, nefarious conduct. IV Sentinel AI combines IntelliVision Smart Motion analytics with Vision Language Model reasoning — Smart Motion detects relevant activity, then VLM analysis evaluates selected video for ATM-specific context, abnormal behavior, and scene-level detail.

HOW IT WORKS

Multi-stage intelligence, all at the edge

ATM-area cameras

Existing IP cameras and RTSP streams

Edge AI appliance

Installation-ready,
fully integrated data pipeline

Smart Motion

Detects meaningful
activity, filters noise

VLM Reasoning

Behavior & context on selected events

Alert to VMS

Metadata, API
escalation, scene
summary

[SUSPICIOUS] [TAMPERING]

Individual in dark hoodie and face covering, actively prying open the lower service panel of an ATM; casing already breached. Behavior consistent with forced entry/tampering with ATM hardware.

VLM reasoning describes behavior in context: a person repeatedly manipulating the card reader, a hooded individual forcing the fascia, two people crowding a customer, or someone lying down inside a vestibule. It also captures visible attributes for review.

  Shirt style & color
  Clothing patterns
  Pants
  Vehicle type
  Vehicle color

BEHAVIORAL SCENE UNDERSTANDING

Descriptions of what is happening — and why it matters

Built for the ATM attack surface

ATM Vandalism

Striking, kicking, forced interaction, defacement.
Hook-and-Chain

Object-to-ATM interaction with chains, cables, ropes.
Skimming Activity

Suspicious card-reader interaction vs. normal use.
Jackpotting & Tampering

Unauthorized panel/cabinet interaction, tool-like behavior.
Cash Trapping

Suspicious activity around the cash dispenser area.
Vestibule Loitering

Prolonged occupancy, sleeping, unauthorized presence.
Threat Detection

Visible weapons and threatening behavior.
Tailgating & Collusion

Crowding, unauthorized entry, shoulder surfing.