Skip to main content
Artificial Intelligence

AI Analytics Engine

10 production-ready AI models with hardware-adaptive inference. Enable face recognition, person tracking, ANPR, crowd analytics, and more on any stream with per-stream configuration granularity.

10
AI Models
2
Inference Runtimes
Per-Stream
Configuration
4+
Coming Soon
Models

10 Production-Ready AI Models

Each model is optimized for video analytics use cases with configurable parameters, threshold tuning, and per-stream activation.

Identification

Face Recognition

High-accuracy face detection, embedding extraction, and 1:N matching against enrolled galleries. Supports liveness detection and multi-angle recognition in challenging lighting conditions.

  • 1:N gallery matching
  • Liveness detection
  • Multi-angle support
  • Real-time embedding extraction
Tracking

Person Re-Identification

Track individuals across multiple cameras and timeframes using appearance-based feature vectors. Maintains identity consistency without requiring face visibility.

  • Cross-camera tracking
  • Appearance matching
  • Temporal search
  • Gallery-free operation
Identification

ANPR / LPR

Automatic Number Plate Recognition supporting 100+ country formats. High-speed capture at up to 200 km/h with infrared and visible-light camera support.

  • 100+ country formats
  • High-speed capture (200 km/h)
  • IR + visible light
  • Partial plate matching
Detection

Object Detection

General-purpose object detection identifying 80+ object classes including vehicles, animals, luggage, and personal items. Real-time bounding box output with confidence scoring.

  • 80+ object classes
  • Real-time bounding boxes
  • Confidence scoring
  • Custom class training
Analytics

Crowd Detection

Density-based crowd estimation and occupancy counting without individual tracking. Configurable thresholds trigger alerts when defined capacity limits are exceeded.

  • Density estimation
  • Occupancy counting
  • Threshold alerts
  • Zone-based counting
Detection

Person Detection

Specialized human body detection optimized for surveillance scenarios. Higher accuracy than general object detection for person-specific use cases including partially occluded individuals.

  • Occlusion handling
  • Pose awareness
  • Size filtering
  • Direction estimation
Security

Intrusion Detection

Define restricted zones on any camera view. Alerts trigger when persons or objects enter, exit, or dwell within configured polygonal boundaries.

  • Polygonal zone definition
  • Enter/exit/dwell rules
  • Multi-zone support
  • Schedule-based activation
Analytics

Line Crossing

Directional line crossing detection for counting, flow analysis, and perimeter monitoring. Supports multiple lines per camera with per-direction triggering rules.

  • Directional counting
  • Multi-line support
  • Flow analysis
  • Perimeter monitoring
Security

Loitering Detection

Identifies individuals who remain in a defined area beyond a configurable time threshold. Combines person detection with temporal analysis for accurate dwell-time measurement.

  • Configurable dwell time
  • Zone-based monitoring
  • Individual tracking
  • Alert escalation
Safety

PPE Detection

Detects presence or absence of personal protective equipment including hard hats, safety vests, gloves, and goggles. Designed for industrial and construction site compliance monitoring.

  • Hard hat detection
  • Safety vest detection
  • Gloves/goggles check
  • Compliance reporting
Inference

Hardware-Adaptive AI

Visylix automatically selects the optimal inference runtime based on available hardware. Deploy with or without GPUs.

TensorRT
NVIDIA GPU

Optimized inference on NVIDIA GPUs with FP16 and INT8 quantization. Batch processing across multiple streams maximizes GPU utilization and throughput.

  • FP16 / INT8 quantization
  • Multi-stream batching
  • CUDA-accelerated pre/post processing
  • Dynamic batch sizing
OpenVINO
Intel CPU

Efficient CPU inference using Intel OpenVINO toolkit. Enables AI analytics on servers without dedicated GPU hardware, reducing deployment cost for smaller installations.

  • No GPU required
  • AVX-512 optimization
  • Lower hardware cost
  • Edge deployment friendly
Configuration

Per-Stream AI Control

Enable, disable, and tune individual AI models on each video stream independently.

Granular Activation

Enable face recognition on entrance cameras, ANPR on parking streams, and crowd detection in lobbies. Each stream runs only the models it needs.

Threshold Tuning

Adjust confidence thresholds, minimum object sizes, and detection intervals per model per stream. Fine-tune sensitivity for each camera's field of view.

Zone Configuration

Define polygonal regions of interest within each camera view. AI models only process objects within configured zones, reducing false positives and compute load.

Event Routing

Configure where AI events are sent: WebSocket subscribers, webhook endpoints, recording triggers, or third-party integrations. Each model's events route independently.

Performance Budgeting

Set per-stream FPS targets for AI processing. The engine balances inference load across available hardware to meet quality-of-service requirements.

Roadmap

Coming Soon

New AI models under active development to expand Visylix's analytical capabilities.

Fire & Smoke Detection

6-12 months

Early-warning fire and smoke detection using thermal and visual analysis for critical infrastructure and industrial environments.

Vehicle Make & Model

6-12 months

Classify vehicle make, model, year, and color for parking management, toll collection, and law enforcement applications.

Anomaly Detection

12-24 months

Unsupervised anomaly detection that learns normal scene patterns and alerts on deviations without predefined rules.

Abandoned Object

12-24 months

Detect objects left unattended in public spaces. Distinguishes between placed objects and temporarily stationary items.

Power Your Video with AI

See how Visylix AI integrates with the VMS Core Engine and multi-protocol streaming to deliver actionable intelligence.