Qualcomm AI Hub Models - Particle Blueprint

Introduction

Welcome to the Qualcomm AI Hub Models Blueprint App! This project enables AI model execution from Qualcomm's AI Hub Models on Linux-based edge devices using Particle's IoT platform.

With this blueprint, you can:

Deploy and run AI models dynamically from the AI Hub.
Use Docker to containerize execution, making it easy to manage and update.
Leverage Particle's IoT platform for remote deployment and model selection.

This serves as an entry point for running AI workloads at the edge using Tachyon and other supported Linux devices.

Project Structure

The repository is structured as follows:

/ai-hub
│── /ai-hub-models/        # The AI Hub Models repository (submodule)
│    ├── qai_hub_models/   # Core AI Hub framework
│    ├── configs/          # AI model configuration files
│    ├── datasets/         # Dataset handling for AI training/inference
│    ├── models/           # Model definitions and implementations
│    ├── scripts/          # Utility scripts for setup and execution
│    ├── Dockerfile        # Docker container setup
│    ├── README.md         # AI Hub-specific documentation
│── docker-compose.yaml    # Docker Compose setup
│── output/                # Stores inference results
│── images/                # Example images for model testing
│── blueprint.yaml         # Particle Blueprint definition
│── README.md              # This document

AI Hub Models as a Submodule

The AI Hub Models are pulled in as a Git submodule within this repository. This allows easy updates and synchronization with the latest developments from Qualcomm's AI Hub.

To ensure you have the latest submodule contents, run:

git submodule update --init --recursive

Getting Started: Running AI Models

This blueprint allows you to execute AI models using Docker containers on Particle-supported edge devices.

1. Set Up Your Particle Linux Device

Ensure your Tachyon or Raspberry Pi device is:

Running a compatible Linux distribution.
Connected to the Particle IoT platform.

2. Clone & Initialize the Submodules

Since this repo includes a submodule, use:

git clone --recurse-submodules https://github.com/particle-iot/blueprint-ai-hub-models.git
cd blueprint-ai-hub-models

If you've already cloned the repository but forgot the submodule, run:

git submodule update --init --recursive

3. Build & Run the Docker Container

Use Docker Compose to build and run the container:

docker-compose up --build

To run a specific model:

particle app run

For details on available models, refer to the original AI Hub Models.

Contributions

We welcome contributions! If you have improvements or fixes, please open a pull request in the AI Hub Models GitHub Repository.

Supported Devices

This blueprint supports:

Tachyon (Particle Linux Edge Device)
Raspberry Pi
Other Linux-based edge computing platforms