NE301 - STM32N6570 AI Vision Camera

High-performance AI vision camera system based on STM32N6570 Discovery Kit, featuring real-time video processing, neural network acceleration, and modern web interface.

✨ Key Features

• 🎥 Real-time Video Processing - Multiple camera sensor support (OS04C10, IMX335)
• 🧠 AI Accelerated Inference - NPU hardware acceleration, YOLOv8/YOLOX
• 🌐 Modern Web Interface - Preact + TypeScript, real-time preview and configuration
• 📡 Multi-network Support - Ethernet, WiFi, Cat1, BLE connectivity
• 🔒 Secure Boot - TrustZone secure partitioning, firmware signature verification
• 🔄 OTA Updates - Secure over-the-air firmware and model upgrade

🎬 Preview

Take a Closer Look at CamThink NeoEyes NE301

Deploy NeoEyes NE301 Anywhere Outdoors

Always Awake. Always Ready for Action

Detecting Every Critical Moment

Learn More

🏗️ Project Structure

Low Power, Performance, and Edge AI NE301 is an AI vision camera system based on STM32N6570, featuring a multi-core architecture:

• STM32N6 (Main MCU): Cortex-M55, handles video processing, AI inference, and network communication
• STM32U0 (WakeCore): Power control unit, manages low-power operation and wake-up
• Web Frontend: Preact + TypeScript, provides real-time preview and configuration interface
• AI Models: YOLOv8/YOLOX, supports object detection and pose estimation

┌─────────────────────────────────────┐
│   Web Service (HTTP/WebSocket)      │  ← User Interface Layer
├─────────────────────────────────────┤
│   Service Layer                     │  ← Service Layer
│   ├─ AI Service                     │
│   ├─ System Service                 │
│   ├─ Communication Service          │
│   ├─ Web Service                    │
│   ├─ MQTT Service                   │
│   └─ OTA Service                    │
├─────────────────────────────────────┤
│   Core Layer                        │  ← Core Layer
│   ├─ Framework                      │
│   ├─ Video Pipeline                 │
│   ├─ Configuration Manager          │
│   └─ Event Bus                      │
├─────────────────────────────────────┤
│   HAL Layer                         │  ← Hardware Abstraction Layer
│   ├─ Camera Driver                  │
│   ├─ Network Driver                 │
│   ├─ NN Driver                      │
│   └─ Storage Driver                 │
└─────────────────────────────────────┘

ne301/
├── Appli/                  # Stm32n6 Main application
├── FSBL/                   # Stm32n6 First stage bootloader
├── WakeCore/               # Power Ctrl Unit
├── Web/                    # Web frontend 
├── Model/                  # AI models
├── Script/                 # Build and packaging scripts
└── Makefile                # Root build orchestration

🚀 Quick Start

Go to WIKI

🛠️ Development Guide

Development Environment Setup

🐳 Method 1: Docker (Recommended)

Prerequisites: Docker 20.10+ Disk Space > 10GB+

# 1. Build (Or pull)Docker image
docker build -t ne301-dev:latest .
# or pull (faster)
docker pull camthink/ne301-dev:latest
# 2. Run container
docker run -it --rm --privileged \
  -v $(pwd):/workspace \
  -v /dev/bus/usb:/dev/bus/usb \
  camthink/ne301-dev:latest
# 3. Inside container
make                        # Build all

💻 Method 2: Source Installation

Prerequisites:

• ARM GCC 13.3+
• GNU Make 3.81+
• Python 3.8+
• Node.js 20+
• pnpm 9+
• STM32CubeProgrammer(v2.19.0)
• STM32_SigningTool_CLI(v2.19.0)
• stedgeai(v2.2,stedgeai0202.stneuralart)

# 1. Check environment
./check_env.sh
# 2. Install as prompted
./setup.sh                  # Linux/macOS
setup.bat                   # Windows
# 3. Verify
./check_env.sh

See SETUP.md for detailed installation instructions.

Hardware Connect

Prerequisites:

• NE301 Board * 1
• ST-Link V2 * 1
• 4P 1.25mm pitch male to 2.54mm dupont female adapter (Used for flash N6 chips)
• 3P 2.54mm pitch dual female header dupont wires (Used for flash U0 chips)
• Type C USB cable compatible with the computer's USB port (for e.g., type C to type A if the computer has a type A USB port)

The mainboard contains two MCUs： stm32n6 and stm32u0

Ready for Flashing apps, web, or models to stm32n6

1. Turn on the dip switch 2 on the board to enter the flash mode.(After the flash is completed, please turn it off and power it back on or reset it to enter the running mode)

2. Connect ST Link to the DEBUG port on the board using a 4P adapter cable and connect ST Link to the computer.

3. Connect the board to a computer or adapter using a type-c USB cable, and the onboard DEBUG indicator light will remain on, indicating that it has entered the flash mode.

Ready for Flashing wakecore to stm32u0

1. Connect ST-LINK to STM32U0 chip using 3P DuPont wire and connect ST-LINK to computer.

2. Connect the board to a computer or adapter using a type-c USB cable.

Build

# Build
make                        # Build all (FSBL + App + Web + Model)
make app                    # Build application only
make web                    # Build web frontend
make model                  # Build AI model
make pkg                    # package for flash or OTA
make info                   # help

Flash

1. Firmwares List

  ne301_FSBL_signed.bin       --> use for stm32n6 FSBL        --> flash addr 0x70000000
  ne301_App_signed_pkg.bin    --> use for stm32n6 App         --> flash addr 0x70100000
  ne301_Web_pkg.bin           --> use for web gui             --> flash addr 0x70400000
  ne301_Model_pkg.bin         --> use for AI model            --> flash addr 0x70900000
  ne301_WakeCore.bin          --> use for stm32u0 wakecore    --> flash addr 0x08000000 

2. Flash tools supported

• STM32CubeProgrammer
• Script/maker.sh

Script/maker.sh flash <bin-name> <flash-addr>

• make

  # Flash all components
  make flash
  # Flash specific component
  make flash-fsbl
  make flash-app
  make flash-web
  make flash-model
  make flash-wakecore

📄 License

This software is released under a Dual-License model.

• Community Edition License
• Commercial Edition License

Please see the full terms in LICENSE

---

Development Team: CamThink AI Camera Team
Contact: [email protected]
Last Updated: 2025-10-21