sipeed/MaixPy

Easily create AI projects with Python on edge device

What it solves

MaixPy is designed to simplify the development of AI projects on edge devices. It removes the complexity of embedded systems programming by providing a high-level Python API, allowing developers to implement AI vision, audio, and LLM/VLM capabilities on hardware like the MaixCAM series without needing deep expertise in hardware or complicated software stacks.

How it works

MaixPy acts as a Python SDK that integrates hardware peripheral operations, video streaming, and AI algorithms. It leverages the NPU (Neural Processing Unit) of the MaixCAM hardware to accelerate AI models (such as YOLO and MobileNet) and provides a C/C++ SDK with identical APIs for those who need to move from prototype to production deployment.

Who it’s for

  • AI Algorithm Engineers: For deploying AI models to embedded devices.
  • STEM Teachers and Students: For learning AI and embedded development through easy-to-use APIs and tutorials.
  • Makers and Engineers: For rapid prototyping of intelligent projects.
  • Enterprises: For developing AI vision products (e.g., production line QA systems) without needing a large team of embedded engineers.

Highlights

  • Rapid Development: Create AI vision projects in as few as 10 lines of Python code.
  • Comprehensive Ecosystem: Includes the MaixVision IDE, MaixHub online training platform for one-click AI model training, and supporting hardware (MaixCAM, MaixCAM2, MaixCAM-Pro).
  • High Performance: Supports high-fps AI detection (e.g., YOLO11n) and classification (e.g., MobileNetv2) on edge hardware.
  • Edge LLM Support: Capable of running small language models like Qwen/DeepSeek 0.5B and 1.5B, as well as VLMs (InterVL 1B) on compatible hardware.
  • Hardware Integration: Built-in support for peripherals like UART serial ports, camera, and display.

Related

  • Project
    livekit/python-sdks

    A Python SDK for integrating real-time video, audio, and data features into applications, enabling the creation of multi-modal AI and live streaming tools.

  • Project
    PaddlePaddle/PaddleX

    A low-code AI development tool based on PaddlePaddle that provides pre-trained model pipelines for rapid training, inference, and deployment across diverse hardware.

  • Project
    Seeed-Studio/ModelAssistant

    A toolchain for deploying optimized AI algorithms on low-cost embedded devices and microcontrollers, supporting computer vision, anomaly detection, and scenario-specific tasks.

  • Project
    runpod/runpod-python

    The official Python library for RunPod, providing an SDK to build serverless AI workers and a GraphQL wrapper to manage GPU cloud pods and endpoints.

  • Project
    huangjunsen0406/py-xiaozhi

    py‑xiaozhi is a Python‑based, async‑first framework for building multimodal AI assistants that can listen, speak, see, and control hardware. It offers real‑time Opus voice streaming, offline wake‑word detection, vision‑language integration, a JSON‑RPC tool server, and secure WebSocket/MQTT communication. Runs on Windows/macOS/Linux desktops and on edge boards (Raspberry Pi, Jetson Nano). Provides GUI (PySide6/QML), CLI, and GPIO modes, and a plugin system for easy extension.