Welcome to PeekingDuck’s Documentation!

Contents

• Introduction
  • What is PeekingDuck?
  • Features
    • Build realtime CV pipelines
    • Leverage on SOTA models
    • Create custom nodes
  • How PeekingDuck Works
  • Acknowledgements
  • License
  • Communities
• Getting Started
  • Documentation Convention
  • Standard Install
    • Install PeekingDuck
    • Verify PeekingDuck Installation
  • Custom Install
    • Arm64
    • Apple Silicon Mac
• Tutorials
  • “Hello Computer Vision”
    • Pose Estimation
    • Object Detection
    • Using a WebCam
    • Pipelines, Nodes and Configs
    • Bounding Box vs Image Coordinates
  • Duck Confit
    • More Object Detection
    • Record Video File with FPS
    • Configuration - Behind the Scenes
    • Augmenting Images
  • Custom Nodes
    • Recipe 1: Object Detection Score
    • Recipe 2: Keypoints, Count Hand Waves
    • Recipe 3: Debugging
    • Other Recipes to Create Custom Nodes
      • CLI Recipe
      • Pipeline Recipe
  • Peaking Duck
    • Interfacing with SQL
    • Counting Cars
    • Object Tracking
      • Tracking People within a Zone
  • Calling PeekingDuck in Python
    • Using PeekingDuck’s Pipeline
      • Setting Up
      • Creating a Custom Node for Debugging
      • Creating the Python Script
      • Running the Python Script
    • Integrating with Your Workflow
      • Import the Modules
      • Initialize PeekingDuck Nodes
      • Create a Dataset Loader
      • Create a License Plate Parser Class
      • The Inference Loop
  • Using Your Own Models
    • Model Training
      • Setting Up
      • Update Training Script
      • Training the Model
    • Using Your Trained Model with PeekingDuck
      • Converting to a Custom Model Node
      • Using the Classifier in a PeekingDuck Pipeline
    • Custom Object Detection Models
• PeekingDuck Ecosystem
  • PeekingDuck Viewer
    • Running the Viewer
    • Navigating the Output Video
    • Using the Pipeline Playlist
    • Exiting the Viewer
• Model Resources & Information
  • Object Detection Models
    • List of Object Detection Models
    • Benchmarks
      • Inference Speed
        • Hardware
        • Test Conditions
      • Model Accuracy
        • Dataset
        • Test Conditions
    • Object Detection IDs
      • General Object Detection
      • Face Detection
  • Pose Estimation Models
    • List of Pose Estimation Models
    • Benchmarks
      • Inference Speed
        • Hardware
        • Test Conditions
      • Model Accuracy
        • Dataset
        • Test Conditions
    • Keypoint IDs
      • Whole Body
  • Object Tracking Models
    • List of Object Tracking Models
    • Benchmarks
      • Inference Speed
        • Hardware
        • Test Conditions
      • Model Accuracy
        • Dataset
  • Crowd Counting Models
    • List of Crowd Counting Models
    • Benchmarks
      • Model Accuracy
        • Dataset
  • Instance Segmentation Models
    • List of Instance Segmentation Models
    • Benchmarks
      • Inference Speed
        • Hardware
        • Test Conditions
      • Model Accuracy
        • Evaluation on masks
        • Evaluation on bounding boxes
        • Dataset
        • Test Conditions
    • Instance Segmentation IDs
      • General Instance Segmentation
  • Bibliography
    • Legend
    • Object Detection
    • Pose Estimation
    • Crowd Counting
    • Object Tracking
    • Instance Segmentation
• Edge AI
  • Installing TensorRT
  • Using TensorRT Models
  • Performance Speedup
    • NVIDIA Jetson Xavier NX with 8GB RAM
    • NVIDIA Jetson Xavier AGX with 16GB RAM
      • Test Conditions
  • References
• Use Cases
  • Privacy Protection
    • Privacy Protection (Faces)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
    • Privacy Protection (License Plates)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
    • Privacy Protection (People & Computer Screens)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
  • Smart Monitoring
    • Crowd Counting
      • Overview
      • Demo
      • How It Works
      • Nodes Used
    • Object Counting (Present)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
    • Object Counting (Over Time)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
      • Counting Objects Within Zones
    • People Counting (Over Time)
      • Overview
      • Demo
      • How It Works
      • Nodes Used
      • Counting People Within Zones
    • Zone Counting
      • Overview
      • Demo
      • How It Works
      • Nodes Used
  • COVID-19 Prevention and Control
    • Face Mask Detection
      • Overview
      • Demo
      • How It Works
      • Nodes Used
    • Group Size Checking
      • Overview
      • Demo
      • How It Works
      • Nodes Used
      • Using with Social Distancing
    • Social Distancing
      • Overview
      • Demo
      • How It Works
      • Nodes Used
      • Using with Group Size Checker
• FAQ and Troubleshooting
  • How can I post-process and visualize model outputs?
  • How can I dynamically use all prior outputs as the input at run time?
  • How do I debug custom nodes?
  • Why does input.visual progress stop before 100%?
  • Why does the output screen flash briefly and disappear on my second run?
• Glossary

API Documentation

input	Reads data from a given input.
augment	Performs image processing.
model	Deep learning model nodes for computer vision.
dabble	Algorithms that perform calculations/heuristics on the outputs of model.
draw	Draws results/outputs to an image.
output	Writes/displays the outputs of the pipeline.