Overview

YOLO Vision Applications on AMD Ryzen™ AI NPU

Short summary: Real-time YOLO-based computer vision applications (object detection, classification, and segmentation) optimized for AMD Ryzen AI NPU hardware acceleration.

About Advantech Container Catalog (ACC)

Advantech Container Catalog is a comprehensive collection of ready-to-use, containerized software packages designed to accelerate the development and deployment of Edge AI applications. By offering pre-integrated solutions optimized for embedded hardware, it simplifies the challenges often faced with software and hardware compatibility, especially in GPU/NPU-accelerated environments.

Feature / Benefit	Description
Accelerated Edge AI Development	Ready-to-use containerized solutions for faster prototyping and deployment
Hardware Compatible	Reduces hardware and package incompatibility issues
GPU/NPU Access Ready	Supports passthrough for efficient hardware acceleration
Model Conversion & Optimization	Built-in model conversion and quantization recommendations
Optimized for CV & LLM Applications	Optimized stacks for vision and language workloads

Container Overview

This container provides a complete YOLO vision application environment optimized for AMD Ryzen AI NPU acceleration. It includes pre-configured tools for running YOLO11 models in multiple tasks (object detection, classification, and segmentation) with real-time performance on embedded AMD Ryzen systems. The container simplifies deployment by bundling the Ryzen AI runtime, necessary dependencies, and example applications.

Demo

Segmentation:

Detection:

Use Case

Real-time object detection on embedded systems
Video analysis and monitoring applications
Instance segmentation for autonomous systems
Image classification at the edge
Performance comparison between NPU and CPU acceleration
Model conversion and optimization for edge deployment

Key Features

NPU acceleration for real-time YOLO inference
Support for YOLO11 detection, classification, and segmentation models
Easy model format conversion (PyTorch to ONNX)
Dual execution modes: NPU-accelerated and CPU-only for comparison
Live webcam and video file processing
Built-in performance diagnostics and benchmarking
Headless mode for server deployments
Integrated virtual environment with pre-installed dependencies

Host Device Prerequisites

Item	Specification
Compatible Hardware	AMD Ryzen AI NPU-enabled systems (e.g., AIMB-RN)
Platform Version	Ryzen AI NPU Runtime 1.6+
Host OS	Linux (Ubuntu 22.04 LTS recommended)
Required Packages	Docker, XRT (Xilinx Runtime), XDNA driver

Required Software Packages on Host Device

Component	Version	Description
XDNA Driver	Latest	AMD Ryzen AI NPU kernel driver and firmware
XRT (Xilinx Runtime)	Latest	Runtime for NPU execution and device management
Docker	20.10+	Container runtime platform
Docker Compose	2.0+	Multi-container orchestration
Boost Libraries	1.71+	Required dependency for NPU runtime

Container Environment Overview

Software Components in the Image

Component	Version	Description
Ryzen AI Runtime	1.6+	AMD Ryzen AI NPU execution engine
Python	3.10+	Programming language runtime
ONNX Runtime	Latest	Cross-platform inference engine
OpenCV	Latest	Computer vision library for image processing
YOLO11	Latest	Object detection/segmentation models
PyTorch	Latest	Deep learning framework
GStreamer	Latest	Multimedia framework for video processing

Container Quick Start Guide

For installation, setup, build scripts, and detailed usage instructions, please refer to the Vision-Applications-on-AMD-Ryzen README in the repository.

Supported AI Capabilities

Vision Models

Model Family	Supported Versions	Notes
YOLO	YOLO11 (nano, small, medium, large)	Object detection and segmentation
YOLO Classification	YOLO11-cls	Image classification tasks
YOLO Segmentation	YOLO11-seg	Instance segmentation

Supported Model Formats

Format	Support Level	Notes
ONNX	Full	Primary format for NPU inference
PyTorch	Full	Native YOLO format, export to ONNX recommended
TensorFlow SavedModel	Supported	Via conversion utilities

Hardware Acceleration Support

Accelerator	Support Level	Compatible Libraries	Notes
AMD Ryzen AI NPU	Full	Ryzen AI Runtime, ONNX Runtime	Primary acceleration target
CPU Fallback	Full	OpenCV, PyTorch, ONNX Runtime	For comparison and compatibility

Troubleshooting & Notes

NPU test falls back to CPU execution:
- Check that required Boost runtime libraries are installed
- Ensure /opt/xilinx/xrt exists and is mounted into the container
General considerations:
- The Docker container reuses host-installed XRT and firmware
- Reboot the host after any NPU driver or firmware update
- For detailed setup and execution instructions, refer to the README in the repository