Loitering Detection

GitHub project Readme Installation guide

Loitering Detection leverages advanced AI algorithms to monitor and analyze real-time video feeds, identifying individuals lingering in designated areas. By proactively detecting suspicious behavior, this system helps to address potential security threats effectively.

By utilizing cutting-edge technologies and pre-trained deep learning models, this application enables real-time processing and analysis of video streams, making it an ideal solution. Its modular architecture and integration capabilities ensure that users can easily customize and extend its functionalities to meet their specific needs.

Get Started

• Overview
  • Key features
  • How it works
    • Components
  • Learn More
• System Requirements
  • Supported Platforms
  • Minimum Requirements
  • Software Requirements
  • Compatibility notes
  • Validation
• Get Started
  • Prerequisites
  • Set up and first use
  • Run the application
  • Access the Application and Components
    • Nginx Dashboard
    • Grafana UI
    • NodeRED UI
    • DL Streamer Pipeline Server
  • Stop the Application:
  • Other Deployment Option
  • Next Steps
  • Troubleshooting
  • Supporting Resources
• Release Notes
  • v1.3.0 (Upcoming release)
    • Updates
  • Version 1.2 - March 28th, 2025
    • High-Level Features

How to

• Customize Application
  • Overall System Architecture
  • DL Streamer Pipeline Server
    • Overview of Deep Learning Streamer Pipeline Server
    • Key DL Streamer Pipeline Server Components
      • Logging and general configuration
      • Video processing pipelines
        • Object detection pipelines (YOLOv10 Series)
        • Object tracking pipelines
      • Configurable parameters
      • Messaging Interface (MQTT)
    • DL Streamer Pipeline Server workflow
  • Node-RED Flow for Data Processing
    • Overview of the Node-RED flow
    • Key Node-RED components
      • MQTT Input nodes
      • Data extraction nodes
      • Euclidean and IoR Nodes (Loitering Detection)
      • Data Table Output Nodes
      • MQTT Output Nodes
    • Node-RED workflow
  • Grafana visualization
    • Overview of Grafana tool
    • Key Grafana Components
      • Data Sources
      • Dashboards and Panels
    • Setting up Grafana
      • Install and launch Grafana
      • Add Your data source
      • Create your dashboard
    • Grafana use cases for object tracking and Loitering Detection
  • End-to-End integration
  • Conclusion
• How to Deploy with Helm
  • Get Started
  • Prerequisites
  • Step 1: Download the Helm chart
  • Step 2: Configure and update the environment variables
  • Step 3: Deploy the application and Run multiple AI pipelines
  • Step 4: End the demonstration
  • Summary
  • Troubleshooting
    • Deploying with Intel GPU K8S Extension on Intel® Tiber™ Edge Platform
    • Deploying without Intel GPU K8S Extension
    • Error Logs
• How to Deploy with the Edge Orchestrator
  • Deploy with Edge Orchestrator
    • Prerequisites
    • Make the Deployment Package Available
    • Deploy the Application onto the Edge Nodes
    • Access the Loitering Detection Sample Application
• How to Benchmark Performance
  • Prerequisites
  • Step 1: Understand the Benchmarking Script
    • Stream Density Logic
    • Average FPS Calculation
    • Recommended Pipeline Parameters
  • Step 2: Prepare for Benchmarking
  • Step 3: Run the Benchmark
    • Example: Running Stream Density Benchmark for Loitering Detection
  • Step 4: Stop the Benchmark
• View Open Telemetry Data
  • Visualizing metrics in Prometheus
    • For Docker
    • For Helm
    • Try Out the Following Queries
• How to use GPU for inference
  • Pre-requisites
    • Provide GPU access to the container
    • Hardware specific encoder/decoders
    • GPU specific element properties
    • Selecting the GPU render device of your choice if there is more than one GPU device on the system
  • Tutorial on how to use GPU specific pipelines

Source Code

• GitHub

Community and Support

• Get Help
  • Troubleshooting Steps
  • Support