Description Grove Vision AI Module V2 is an MCU-based smart vision module powered by the Himax WiseEye2 HX6538 processor. It combines dual Arm Cortex-M55 cores with an Arm Ethos-U55 neural network accelerator for compact edge AI vision projects. The module supports TensorFlow and PyTorch model workflows, Arduino IDE development, and no-code or low-code deployment through SenseCraft AI. It is designed to pair with Seeed Studio XIAO boards while also offering CSI, IIC, UART, SPI, Type-C, Grove, microphone, and SD card interfaces for embedded development. Note The package does not include a camera. To use the full vision features, prepare a compatible CSI camera separately; Seeed recommends an OV5647-62 FOV camera module or the matching kit option. Features Edge AI processor: Himax WiseEye2 HX6538 with dual-core Arm Cortex-M55 and integrated Arm Ethos-U55 microNPU. Model support: Works with SenseCraft AI, Mobilenet V1/V2, EfficientNet-lite, YOLO v5/v8, TensorFlow, and PyTorch workflows. Rich onboard hardware: Includes a PDM microphone, SD card slot, Type-C, Grove interface, and standard CSI camera interface. Development compatibility: Works with XIAO series boards, Arduino workflows, Raspberry Pi, ESP32 development boards, and other embedded platforms. Open development: Code, design files, SDK examples, and circuit documentation are available for deeper customization. Performance and AI Models The official performance comparison highlights a 33 ms inference time, 30.30 FPS frame rate, and about 0.35 W power consumption, positioning the module for low-power vision AI tasks. Suitable model types include object detection, image classification, facial keypoint detection, gesture recognition, and human pose estimation. SenseCraft AI can be used to deploy off-the-shelf or custom models and view recognition results in the browser. Hardware Overview The board provides a standard CSI camera connector, onboard digital microphone, SD card slot, Type-C port, Grove connector, IIC, UART, and SPI interfaces. With a connected XIAO board, it can be used with Arduino, MicroPython, CircuitPython, PlatformIO, Home Assistant, cloud services, or private servers depending on the project. Specifications Processor Himax WiseEye2 HX6538 with dual Arm Cortex-M55 cores and Arm Ethos-U55 microNPU Onboard peripherals PDM microphone and SD card slot Interfaces CSI, IIC, UART, SPI, Type-C, Grove, and dual 7-pin connector Input voltage 5V Communication rate 115200 Big Cortex-M55 frequency Up to 400 MHz Little Cortex-M55 frequency Up to 150 MHz Ethos-U55 microNPU frequency Up to 400 MHz Memory card interface SD/SDIO host, DS mode, up to 25 MHz Internal system memory Configurable system memory up to 2432 KB plus 64 KB boot ROM Applications Object detection and classification at the edge. Facial keypoint detection and gesture recognition. Human pose estimation for posture, activity, and motion analysis. Compact camera sensing projects connected to XIAO, Grove, Raspberry Pi, or ESP32-based systems. Shipping List Grove Vision AI V2, camera not included 1 Cable 1 Other Content When connecting a CSI camera, check the cable direction before powering the device. If bootloader recovery is needed, keep the BOOT button pressed while connecting the data cable, then release it after connection. The Wiki notes that this may require several tries. The Wiki FAQ notes that live frame output and XIAO result output are not supported at the same time. The Wiki camera guide lists Raspberry Pi OV5647-series cameras as supported; other CSI cameras may need driver or color-processing work and can affect recognition accuracy. Resources Grove Vision AI Module V2 Wiki Guide Grove Vision AI V2 Software Support Guide Grove Vision AI V2 Supported Camera Guide Deploy Models to Grove Vision AI V2 with SenseCraft AI Grove Vision AI Module V2 HX6538 Datasheet Grove Vision AI Module V2 Circuit Diagram Grove Vision AI Module V2 SDK Examples Grove Vision AI V2 Kit Case 3D Model FAQ Is a camera included?No. The package includes the Grove Vision AI V2 module and a cable. A compatible CSI camera is required for full vision use. Which camera should I use?Seeed recommends an OV5647-62 FOV camera module. The Wiki also lists Raspberry Pi OV5647-series cameras as supported. Can it run custom AI models?Yes. It supports TensorFlow and PyTorch workflows, and SenseCraft AI can deploy off-the-shelf or custom models without coding. Can it output a live screen and XIAO recognition results at the same time?No. The Wiki FAQ notes that these two tasks cannot be used simultaneously.
Related
Related
Related
Related