Fence climbing and crossing
Evaluated in real time within the configured camera region.
COLT AI · PERIMETER & FACILITY SECURITY
Perimeter & Facility Security
Real-time events for fences, virtual boundaries and restricted areas.
OPERATIONAL PROBLEM
Continuous human monitoring across large perimeters makes timely detection of critical movement difficult.
COLT AI APPROACH
Colt AI relates configured fence and zone rules to human movement, creating time-stamped security events with visual evidence.
What Does Colt AI Detect?
Restricted-area entry
Evaluated in real time within the configured camera region.
Virtual boundary crossing
Evaluated in real time within the configured camera region.
Multiple-person intrusion
Evaluated in real time within the configured camera region.
Perimeter movement in low visibility
Evaluated in real time within the configured camera region.
COLT AI · F01—F04
Four Cameras. Progressive Event Analysis.
Real COLT AI product views from human tracking and safe thresholds to fence contact and confirmed boundary breaches.
Human & pose tracking
Bounding box, tracking identity and keypoints from a frontal camera.
01 / 06
TEMPORAL RULE EVALUATION
The process that separates approach from breach
Rather than deciding from a single frame, Colt AI evaluates a person's approach, fence contact, climbing behavior and boundary crossing as a sequence. A critical event is created only when the configured conditions are confirmed.
- 01Safe tracking
- 02Approach
- 03Fence contact
- 04Climbing
- 05Confirmed breach
RULE CALIBRATION
Not every approach is a breach.
Zone geometry, pose signals and temporal behavior are evaluated together to distinguish an early alert from a confirmed boundary breach. Thresholds and event rules are configured for the site scenario.
From Rule to Event Record
01Camera zone
02Detection & tracking
03Rule evaluation
04Time-stamped event
05Team notification
ACTION & MANAGEMENT
Event, evidence and notification in one flow
Camera and region rules are managed in the web dashboard. Events are recorded with camera, region, time, type and visual evidence. Notification behavior is configured by team and role.
DEPLOYMENT ARCHITECTURE
Configured for your site.
Camera count, network topology, latency needs and organizational data policies are assessed together. Colt AI is not a cloud-only solution.
On-Premise
- Processing location
- On the organization’s physical server infrastructure
- Suitable scenario
- Centralized camera topologies and strict data policies
- Data approach
- Video and event data can remain within organizational boundaries
Edge
- Processing location
- On processing units close to the site or camera groups
- Suitable scenario
- Distributed sites and local decision requirements
- Data approach
- Enables raw video to be processed on site
Hybrid
- Processing location
- Across edge processing and a central management layer
- Suitable scenario
- Multi-site and mixed network architectures
- Data approach
- Data flows are separated according to organizational policy
The required processing infrastructure is designed around active cameras, enabled AI modules and site requirements.
FAQ
Frequently Asked Questions
Essential questions about camera compatibility, configuration and project architecture.
01Can existing cameras be used?+
IP camera and NVR infrastructure can be included following a compatibility assessment.
02Can fence climbing be distinguished from simply approaching a fence?+
Separate event rules can be configured by evaluating region geometry, fence contact, movement and temporal behavior together.
03Can virtual boundaries be configured per camera?+
Yes. ROIs, virtual lines and restricted zones can be configured separately for each camera view and field scenario.
04What information is included in an event record?+
Depending on configuration, camera, region, time, event type and related visual evidence can be retained in event history.
05How is a low-visibility project assessed?+
Camera position, lighting, image quality and the target scenario are evaluated together during compatibility assessment.