Resources:
Seeed Studio Grove Vision AI Module V2 with Dual Cortex-M55, Ethos-U55 microNPU and SenseCraft AI Support
The Seeed Studio Grove Vision AI Module V2 is a compact MCU-based edge-AI vision module built around the Himax WiseEye2 HX6538 processor. It combines 2 Arm Cortex-M55 processing cores with an integrated Arm Ethos-U55 microNPU, allowing image processing and compatible machine-learning inference to run locally without relying on a cloud service or the connected host controller. The high-performance Cortex-M55 core operates at up to 400 MHz, while the efficiency-oriented Cortex-M55 core reaches 150 MHz, providing a flexible architecture for always-on vision and embedded-AI applications.
The module features a standard CSI camera interface, onboard PDM digital microphone, SD card slot, USB Type-C, Grove connector, I2C, UART, and SPI connectivity. Through SenseCraft AI, developers can deploy supported ready-made or custom models, including MobileNet V1/V2, EfficientNet-Lite, YOLOv5, and YOLOv8 workflows, with TensorFlow and PyTorch support. Compatibility with Seeed Studio XIAO, Arduino, Raspberry Pi, ESP32-based boards, and other controllers makes the module suitable for object detection, pose estimation, visual monitoring, smart sensors, industrial automation, and TinyML development.
Key Features
- WiseEye2 HX6538 AI Processor: Uses the Himax WiseEye2 HX6538 architecture developed for always-on smart sensing and embedded-AI applications.
- Dual Arm Cortex-M55 Cores: Combines a high-performance Cortex-M55 core with a power-efficient Cortex-M55 core for flexible edge processing.
- 400 MHz High-Performance Core: Main Cortex-M55 processor operates at up to 400 MHz for image processing and application workloads.
- 150 MHz Efficiency Core: Secondary Cortex-M55 processor operates at up to 150 MHz for lower-power processing tasks.
- Arm Ethos-U55 microNPU: Integrated neural-processing accelerator executes compatible machine-learning workloads efficiently on-device.
- 400 MHz microNPU: Ethos-U55 neural-processing unit operates at up to 400 MHz.
- 64 MACs/Cycle: Ethos-U55 implementation provides 64 multiply-accumulate operations per cycle for accelerated neural-network inference.
- Arm Helium Technology: Cortex-M55 cores include vector-processing extensions designed to accelerate DSP and machine-learning workloads.
- Up to 2432 KB System Memory: Configurable internal memory supports model execution, image processing, and embedded applications.
- 64 KB Boot ROM: Integrated boot memory supports device startup and system operation.
- SenseCraft AI Support: Enables no-code or low-code deployment of compatible trained AI models to the module.
- Multiple AI Models: Supports compatible MobileNet V1/V2, EfficientNet-Lite, YOLOv5, and YOLOv8 models.
- TensorFlow Support: Supports deployment workflows based on compatible TensorFlow models.
- PyTorch Support: Compatible PyTorch models can be prepared and deployed through supported AI toolchains.
- On-Device AI Processing: Image processing and inference are performed locally on the HX6538 instead of relying on the host controller.
- Standard CSI Camera Interface: Supports compatible CSI cameras, including selected Raspberry Pi camera modules.
- Onboard PDM Microphone: Enables audio acquisition for multimodal sensing, voice detection, and acoustic AI applications.
- SD Card Slot: Provides removable storage for data, model resources, images, and application files.
- Grove Interface: Simplifies integration with Arduino-compatible boards and other Grove-based systems.
- I2C Communication: Allows processed inference results to be transferred efficiently to compatible host controllers.
- UART Interface: Provides serial communication for integration, debugging, and data transfer.
- SPI Interface: Supports high-speed synchronous peripheral communication.
- USB Type-C: Used for power, firmware downloading, configuration, and development.
- 5 V Input: Designed for straightforward integration with common development-board power systems.
- XIAO Compatible: Can operate as an AI expansion module for compatible Seeed Studio XIAO boards.
- Broad Host Compatibility: Supports integration with Arduino, Raspberry Pi, ESP32-based boards, and other embedded controllers.
- Open-Source Hardware Platform: Code, schematics, design files, and related development resources are available for customisation.
- Camera Purchased Separately: The standard module does not include a CSI camera unless purchased as part of a camera or XIAO kit.
Applications
- Object detection
- Person detection
- Facial key-point detection
- Human pose estimation
- Smart cameras
- TinyML applications
- Edge-AI sensing
- Industrial quality inspection
- Predictive-maintenance systems
- Machine-vision prototypes
- Smart agriculture
- Equipment monitoring
- Smart-city sensors
- Occupancy monitoring
- Embedded vision systems
- Robotics
- Mobile IoT devices
- Audio and visual sensing
- Automation systems
- AI-enabled sensor nodes
Hardware Overview
Package Includes:
- Grove Vision AI V2 (camera not included)x1
- Cablex1
