Product Information
Product Overview
NeoEyes NE302 is a Mini AI Vision Camera for device integration. It combines 4 MP image capture, STM32N6 edge AI, wireless connectivity and local storage on a 38 × 38 mm Main Board. A complete NE302 shipment includes the Main Board and Interface Board: the two-board device supports continuous USB Type-C power or a USB Type-C external battery pack; the Main Board can also be used for custom integration with DC power.
Core Capabilities
- Edge AI inference: STM32N6 Cortex-M55 and Neural-ART perform INT8 inference on the device.
- Image capture and encoding: The product supports 4 MP image capture; H.264 video encoding supports up to 1920 × 1080 at 30 fps, with hardware JPEG encoding.
- Two-board expansion and custom integration: The Interface Board provides the complete device with USB Type-C power, MicroSD, control and development connections for installation, maintenance and further development; the Main Board can also be used for custom integration with DC power.
- Wireless connectivity: 2.4 GHz Wi-Fi 6 and Bluetooth Low Energy are supported; radio revision, antenna and regional configuration follow the delivered SKU.
- Local storage and data delivery: MicroSD storage and the Web console support preview, capture, model validation and device management.
- Lighting and extensions: Some hardware revisions may expose supplemental lighting or extension modules. Confirm availability against the delivered SKU and firmware.
- Development resources: The open engineering project contains FSBL, application, Web, model and WakeCore components for build, packaging, flashing and OTA development.
- Compact installation: The enclosure baseline is 42 × 42 × 20 mm and supports indoor magnetic or 3M-adhesive mounting according to the product design.
Specifications
The following specifications use the NE302 datasheet as their public baseline. For delivered interface and engineering-revision differences, see the Hardware Guide.
Core platform
| Item | Specification |
|---|---|
| Primary MCU | STM32N6, Cortex-M55, 800 MHz, Arm Helium |
| AI accelerator | Neural-ART, 1 GHz, up to 0.6 TOPS INT8 |
| Memory | 4.2 MB on-chip SRAM / NPU RAM; 32 MB PSRAM |
| Flash | 64 MB SPI Flash |
Imaging and sensing
| Item | Specification |
|---|---|
| Image sensor | 4 MP high-sensitivity CMOS |
| Maximum H.264 video mode | 1920 × 1080 at up to 30 fps |
| JPEG encoding | Hardware JPEG encoding |
| Lens | Standard M12; 88° / 137° HFOV options |
| Onboard lighting | White LED |
Wireless, storage and controls
| Item | Specification |
|---|---|
| Wireless | Wi-Fi 6, 802.11ax 2.4 GHz; BLE 5.3 |
| Antenna | Standard external SMA, 3–4 dBi |
| Storage and controls | MicroSD slot; Trigger and Reset buttons; dual-color indicators |
Power and mechanical
| Item | Specification |
|---|---|
| Power | Complete device (Main Board + Interface Board): continuous USB Type-C (5 V) power or USB Type-C external battery-pack power; Main Board custom integration: DC power |
| Low power | STM32U0 always-on domain; supports low-power sleep and multi-source wake-up |
| PCBA | 38 × 38 mm |
| Enclosure | 42 × 42 × 20 mm |
| Environment | Indoor, −20 °C to +50 °C |
| Mounting | Rear magnetic or 3M adhesive mounting |
Product Contents
NE302 is built from the following major parts:
| Part | Role | Notes |
|---|---|---|
| Main Board | Carries the STM32N6, WakeCore, camera and wireless-related circuits | 38 × 38 mm board baseline |
| Interface Board | Provides power, storage, control and development connections | Interface count and layout follow the hardware revision |
| Camera assembly | Provides the image input | Lens model and focal length follow the SKU |
| Antenna | Provides wireless connectivity | The antenna form and connector follow the SKU |
| Enclosure | Provides mechanical protection and mounting | Product dimensions follow the delivered mechanical record |
| USB Type-C cable | Provides device power | Use the compatible cable and power source supplied for the delivery |
Performance and Edge AI
Vision-AI Computing
NE302 uses the STM32N6 Cortex-M55 and Neural-ART accelerator for image processing and local AI inference. The engineering project contains object-detection, pose-estimation and model-switching support; usable models, input sizes and performance depend on the model package, firmware and hardware revision.
| Capability | NE302 boundary |
|---|---|
| Inference location | Local device inference |
| Acceleration | Neural-ART, up to 0.6 TOPS INT8 in product material |
| Model source | Model files, JSON configuration and packaging flow under Model/ |
| Runtime validation | Model Validation in the Web console |
| Performance claims | A single-device test or one displayed model is not a system-wide performance commitment |
Image and Data Flow
The Web console can show a preview, execute a capture, validate a model and inspect stored records. Result delivery and media-stream configuration depend on the MQTT, Webhook, RTSP or RTMP pages exposed by the current firmware. Protocol choices, fields and compatibility must follow the delivered release.
Hardware Introduction
Main Board
The Main Board integrates the STM32N6, STM32U0, camera, PSRAM, SPI Flash, and wireless-related circuits. See Components Overview to identify the board, camera, antenna, and hardware-version boundaries.
Interface Board
The Interface Board exposes the connections needed for use and development, including USB Type-C, MicroSD, debug/flash channels, serial access and other revision-dependent interfaces. It is the main reference for first assembly, flashing and field connection.
Version Notes
Delivered versions may differ in radio revision, lens, antenna, debug interfaces and power input. Do not infer a specific SKU from a product render or generic annotated board image. Check the delivered hardware record before installation, flashing or model deployment.
The current public wireless baseline is Wi-Fi 6 and Bluetooth Low Energy. A Wi-Fi HaLow-compatible version is planned, but it requires the corresponding wireless hardware and is not part of the current delivery baseline. Confirm the delivered version and hardware configuration when Wi-Fi HaLow is required.
Product Applications
NE302 is positioned around compact camera integration, local AI and device-level result delivery. The following are product-positioning examples, not verified customer cases or fixed SKUs. Validate the lens, lighting, model, trigger path and result protocol before deployment.
NE302 can act as a compact vision node inside an existing device, workstation or small terminal. Confirm the lens, lighting, model and network path before defining a deployment.
Device state and event detection
Compact vision integration
Product Resources
Wiki Guides
- Quick Guide: complete the first device-management and model-validation path.
- Capture and Storage: configure capture, storage and records.
- Data Transmission: configure result and media delivery.
- AI Model Validation: upload a compatible model and validate fixed inputs.
- System Maintenance: manage communication, firmware, storage resources and device information.
- Hardware Guide: identify boards, interfaces and installation boundaries.
- Software Guide: set up the source environment and build or flash the project.
Development Resources
- NE302 GitHub Repository: source, README, setup instructions and build scripts.
- CamThink Developer Center Models: downloadable model resources for device validation.
- NE302 GitHub Releases: released firmware and OTA packages.
Technical Support
When reporting a product or firmware difference, provide the hardware version, component versions, reproduction steps and a redacted result. Do not include passwords, Secret Keys, network keys, MAC addresses, serial numbers or complete device-configuration files.
