Overview

Advantech Container Catalog provides pre-integrated, hardware-accelerated containers that simplify edge AI development and deployment.

This plug-and-play YOLOv8 object detection container is optimized for Qualcomm® QCS6490, abstracting SDKs, runtimes, and toolchains—so developers can focus on building real-world AI applications.

Built with full DSP/GPU acceleration, the container integrates QNN SDK, SNPE, and LiteRT in a fully preconfigured environment, delivering real-time inference out of the box on Advantech AOM-2721.

Key Capabilities & Benefits

Hardware-Accelerated Edge AI
INT8 inference on Hexagon™ DSP 770 with optional Adreno™ 643 GPU acceleration via QNN, SNPE, and LiteRT.
YOLOv8 Ready Out-of-the-Box
Supports Ultralytics and Qualcomm® AI Hub workflows for fast testing and optimized deployment.
Dual Workflow Flexibility
Easily switch between rapid prototyping and production-grade optimization using script-based pipelines.
Multi-Model Format Support
Compatible with TFLite, SNPE DLC, and QNN .so model formats.
End-to-End Tooling Included
Preloaded export, quantization, and benchmarking scripts for streamlined development.
Real-Time Vision Pipeline
GStreamer + OpenCV preconfigured for responsive video inference.
ROS-Ready Robotics Integration
Compatible with Qualcomm® Robotics Reference Distro and ROS 2.0 (ROS 1.3-ver.1.1).

What’s Included

YOLOv8 Export & Optimization

Ultralytics Export – Rapid testing with TFLite
Qualcomm® AI Hub Conversion – INT8-optimized deployments

Integrated Runtime Stack

QNN, SNPE, LiteRT for DSP/GPU acceleration
GStreamer + OpenCV for vision pipeline development

Preloaded Scripts & Tools

advantech-coe-model-export.sh – Model export & conversion
advantech-aihub-model-export.sh – AI Hub optimization
wise-bench.sh – Runtime verification & benchmarking

Container Demo

Edge-Ready Use Cases

Industrial Automation – Defect detection, safety zone monitoring, predictive maintenance
Smart Retail – Customer analytics, shelf monitoring, automated checkout
Intelligent Transportation – Vehicle/pedestrian detection, traffic and in-cabin monitoring
Robotics & Drones – Autonomous navigation, obstacle detection, infrastructure inspection
Smart City & Surveillance – Crowd analysis, parking management, perimeter security
Healthcare & Assistive Systems – PPE compliance, patient activity monitoring
Agriculture – Crop/livestock monitoring, pest detection, yield estimation
Edge AI R&D – Model benchmarking, INT8 vs FP32 evaluation, custom YOLOv8 training

Host Device Prerequisites

Component	Specification
Target Hardware	Advantech AOM-2721
SoC	Qualcomm® QCS6490
GPU	Adreno™ 643
DSP	Hexagon™ 770
Memory	8GB LPDDR5
Host OS	Yocto 4.0 (LE1.3)

Container Environment Overview

Software Components on Container Image

Component	Version	Description
LiteRT	1.3.0	Provides QNN TFLite Delegate support for GPU and DSP acceleration
SNPE	2.29.0	Qualcomm’s Snapdragon Neural Processing Engine; optimized runtime for Snapdragon DSP/HTP
QNN	2.29.0	Qualcomm® Neural Network (QNN) runtime for executing quantized neural networks
GStreamer	1.20.7	Multimedia framework for building flexible audio/video pipelines
Python	3.10.12	Python runtime for building applications
OpenCV	4.11.0	Computer vision library for image and video processing

Container Quick Start Guide

For container quick start, including the docker-compose file and more, please refer to README.

Supported AI Capabilities

Vision Models

Model	Format	Note
YOLOv8 Detection	TFLite INT8	Downloaded from Ultralytics` official source and exported to TFLite using Ultralytics Python packages
YOLOv8 Segmentation	TFLite INT8	Downloaded from Ultralytics` official source and exported to TFLite using Ultralytics Python packages
YOLOv8 Pose Estimation	TFLite INT8	Downloaded from Ultralytics` official source and exported to TFLite using Ultralytics Python packages
Lightweight Face Detector	TFLite INT8	Converted using Qualcomm® AI Hub
FaceMap 3D Morphable Model	TFLite INT8	Converted using Qualcomm® AI Hub
DeepLabV3+ (MobileNet)	TFLite INT8	Converted using Qualcomm® AI Hub
DeepLabV3 (ResNet50)	SNPE DLC TFLite	Converted using Qualcomm® AI Hub
HRNet Pose Estimation (INT8)	TFLite INT8	Converted using Qualcomm® AI Hub
PoseNet (MobileNet V1)	TFLite	Converted using Qualcomm® AI Hub
MiDaS Depth Estimation	TFLite INT8	Converted using Qualcomm® AI Hub
MobileNet V2 (Quantized)	TFLite INT8	Converted using Qualcomm® AI Hub
Inception V3 (SNPE DLC)	SNPE DLC TFLite	Converted using Qualcomm® AI Hub
YAMNet (Audio Classification)	TFLite	Converted using Qualcomm® AI Hub
YOLO (Quantized)	TFLite INT8	Converted using Qualcomm® AI Hub

Supported AI Model Formats

Runtime	Format	Compatible Versions
QNN	.so	2.29.0
SNPE	.dlc	2.29.0
LiteRT	.tflite	1.3.0

Hardware Acceleration Support

Accelerator	Support Level	Compatible Libraries
GPU	FP32	QNN, SNPE, LiteRT
DSP	INT8	QNN, SNPE, LiteRT

Best Practices

Prefer INT8 quantized models for DSP acceleration
Ensure fixed batch sizes when converting models
Use lower GST_DEBUG levels for stable multimedia handling
Always validate exported models on-device after deployment