The recently released Seeed Grove Vision AI v2, building on the foundation set by the original Grove Vision AI module, is a small microcontroller board designed for MCU-based computer vision and edge AI projects. And also like its predecessor, it’s now officially supported by Edge Impulse!!
This newer v2 version dramatically improves on the capabilities and speed versus the original, with its modern Himax WiseEye2 HX6538 processor containing a dual-core Arm Cortex-M55 and integrated Arm Ethos-U55 NPU for dedicated AI acceleration. There is an onboard microphone, SD Card slot, and CSI camera slot that supports the OV5647 found in the Raspberry Pi Cameras, making it a great option for computer vision applications. There is a USB-C port, and IIC, UART, and SPI connectivity. To easily accommodate sensors, there is also a Grove connector.
The Himax WiseEye2 HX6538 runs at up to 400mhz, and the addition of the Arm Ethos NPU not only boosts frames per second, but according to Seeed's own testing:
Reduces inference time over 10x versus the original Grove Vision AI module.
As an officially supported device, this means that you can build image classification or object detection projects in Edge Impulse, then deploy them to the Seeed Grove Vision AI v2 easily by building native firmware, ready to run on the device. You can also export your model as an Arduino library or C++, and then integrate it into your application.
To get started, you can follow along in the documentation, located here.