How a dual-MCU architecture combining STM32N6 and STM32U0 creates an open-source vision solution that balances AI performance with power efficiency.


The CamThink AI Camera NeoEyes NE301 represents a significant advancement in low-power edge AI vision solutions. Built on an innovative hybrid architecture featuring the STM32N6 MCU for AI processing and STM32U0 MCU for power management, this open-source camera delivers 0.6 TOPS of compute power while maintaining exceptional power efficiency

.

With its modular design and web-based interface, the NE301 enables developers to rapidly prototype and deploy intelligent vision applications across diverse scenarios—from smart cities to industrial automation—without sacrificing battery life or performance.

Core Innovation: Dual-MCU Architecture for Balanced Performance

Edge AI Computing (Powered by STM32N6)

At the heart of the NE301's AI capabilities lies the STM32N6 MCU, which integrates a Neural-ART™ accelerator delivering 0.6 TOPS of computational power

. This dedicated neural processing unit enables real-time inference for lightweight models like person detection and gesture recognition directly on the device.

The processor achieves an impressive 3 TOPS/W power efficiency, allowing full-speed operation without additional cooling systems. The Cortex-M55 core running at 800 MHz with Arm Helium vector processing technology handles general computing tasks while the NPU accelerates neural network inference for complex AI workloads.

Precision Power Management (Powered by STM32U0)

The STM32U0 MCU specializes in ultra-low-power operation, implementing sophisticated power gating and dynamic frequency scaling techniques. This dedicated power management controller enables the system to achieve deep sleep power consumption of just 7-8μA with millisecond-fast wake-up responses.

This intelligent power architecture allows the camera to maintain extended operation on battery power, making it suitable for deployment scenarios where continuous AC power is unavailable or impractical.

Key Features and Specifications

The NE301 combines hardware flexibility with developer-friendly software interfaces:

  • Modular Design: Interchangeable communication modules, camera modules, and power solutions allow customization for specific application requirements
  • Web-Based Interface: Built-in Wi-Fi AP enables browser-based configuration for AI model management, parameter adjustment, and real-time preview
  • Multi-Sensor Triggering: Supports PIR, radar, and audio triggers for event-based capture, significantly reducing power consumption during idle periods
  • Comprehensive Connectivity: Wi-Fi 6, BLE 5.4, LTE Cat-1, and PoE options provide flexibility for various deployment scenarios
  • Open Hardware: Available design files and open-source firmware accelerate customization and product development

Technical Specifications

The camera supports 4-megapixel image capture with customizable field-of-view options (60°, 90°, and 137°) and includes H.264 hardware encoding capable of 1080p@30fps video processing. With 4.2 MB of SRAM, 128 MB HyperFlash, and 64 MB PSRAM, the device has sufficient resources for complex AI models and temporary data storage.

The operating temperature range of -20°C to +50°C and IP67 protection (with appropriate housing) make the NE301 suitable for challenging environmental conditions.



Application Scenarios

The NE301's combination of AI capabilities and power efficiency enables diverse implementation scenarios:

Smart City Applications: Public space occupancy detection, unauthorized accumulation alerts, and customer flow analysis for retail environments.

Industrial Automation: Safety equipment verification (helmet detection), production line monitoring, and equipment compliance checks.

Agricultural Monitoring: Crop growth analysis through periodic capture and livestock activity monitoring.

Home Security: People detection for smart doorbells and visitor identification systems.

Embedded Vision Integration: As an AI vision module for existing equipment like industrial instruments and vehicle systems.

Modular Expansion Capabilities

The camera's flexibility extends through its modular expansion system:

  • Communication Options: Default Wi-Fi 6/BLE with optional LTE Cat-1 or PoE modules
  • Camera Modules: Standard CPI camera (OS04C10) or optional USB camera (SC200AI) with various field-of-view options
  • Power Solutions: 4xAA battery compartment, Type-C power delivery, solar power (via expansion interface), or Power over Ethernet

Development Resources

CamThink provides comprehensive resources for developers, including open-source firmware on GitHub (https://github.com/camthink-ai) and complete commercial-grade SDKs supporting secondary development.

The company's commitment to "making edge AI simpler" reflects in the extensive documentation and hardware customization options available, helping developers overcome technical barriers and accelerate AI model deployment.

Conclusion

The CamThink NeoEyes NE301 demonstrates how thoughtful architecture combining specialized processing units can deliver practical edge AI solutions. By balancing the AI performance of the STM32N6 with the power management expertise of the STM32U0, this open-source camera platform addresses critical challenges in deploying intelligent vision systems where power constraints and computational requirements must be carefully balanced.

For developers seeking to implement low-power AI vision applications, the NE301 offers a flexible foundation that can adapt to diverse requirements while maintaining the power efficiency necessary for extended battery operation.