DRP-AI

This section provides an overview of the DRP-AI (Dynamically Reconfigurable Processor - AI Matrix Arithmetic Circuit) Driver available on the RZ/V2H platform, along with instructions on how to utilize its features effectively.

Note

This section focuses on the DRP-AI Driver and its capabilities for compiling AI models with the DRP-AI extension package.

• Overview and Concepts of DRP-AI
  • What is Dynamically Reconfigurable Processor (DRP)?
  • What is DRP-AI?
  • DRP-AI Driver Architecture
  • DRP-AI Driver Execution Flow
• BYOM AI model support
  • Getting Started
  • Install the DRP-AI TVM extension package
  • BYOM Development Flow
• How to Compile Sample Model
  • Getting Started
  • Work flow
  • Setup Environment
  • Compile AI models
    • Download an ONNX model
    • Compile an ONNX model
• How to compile Your Own Model
  • Compile your own model for RZ/V2H
  • Tutorial
  • Workflow
  • Set up the environment
  • Confirm the model information
  • Modify the sample script
    • Calibration
    • Pre-runtime
  • Compile the AI model
• How to Deploy Model Files
  • Collect the model files
  • Calculate the memory size used by the model
  • Final model configuration for deployment
• DRP-AI model compilation tutorials
  • YOLOX Object Detection Tutorial
    • Compile with DRP-AI TVM Extension Package
  • YOLOv8 Object Detection Tutorial
    • Installation Requirements
    • Train the Model Using Ultralytics YOLOv8 and Transfer Learning
    • Export and Cut the Model
    • Compile with DRP-AI TVM Extension Package
  • MediaPipe Hand Landmark Tutorial
    • Download Models
    • Set Up Conversion Environment
    • Conversion Process
    • Compile with DRP-AI TVM Extension Package
  • RTMPose Hand Landmark Tutorial
    • Compile with DRP-AI TVM Extension Package
  • HRNetV2 Hand Landmark Tutorial
    • Compile with DRP-AI TVM Extension Package