Running machine learning on embedded devices with scarce resources is one of the most exciting developments of the past years, allowing users to detect complex motions, recognize sounds or classify images. To make it possible to build and run these models on the smallest of form factors, we’ve partnered with Himax to support the new Himax WE-I Plus EVB board.
The WE-I Plus EVB is a versatile board with a low-power monochrome camera, a microphone and an accelerometer. The built-in WE-I Plus ASIC (HX6537-A) combines a Synopsys ARC EM9D DSP running at 400MHz and 2MB internal SRAM and 2MB Flash. And this chip is fast! As an example, you can classify a single image at 96x96 pixels in just over a 100ms.
But what is a fast chip without software? Thanks to the integration with Edge Impulse you can now quickly collect real-world sensor data, train ML models on this data in the cloud, and then deploy the model back to your WE-I Plus board. From there you can integrate the model into your applications with a single function call. Your sensors are then a whole lot smarter, being able to make sense of complex events in the real world. The built-in examples allow you to collect data from the accelerometer, the microphone and the camera. And, the WE-I Plus is actually the first device to support image collection directly from the Studio Data Acquisition tab!
Connecting the Himax WE-I Plus to Edge Impulse and collecting your first data sample.
Naturally the Himax WE-I Plus also supports EON - our neural network compiler, which runs the same machine learning models in a lot less RAM and flash. Thanks to EON you can fit two times larger models on the Himax WE-I Plus compared to TensorFlow Lite for Microcontrollers. And, we have added hardware-acceleration for the Himax WE-I Plus to our SDK, so models run as fast as possible without having to do any configuration yourself - making models run eight times faster on the same hardware.
Read the How to get started sectionHow to get started
Excited? This is how you build your first machine learning model with the Himax WE-I Plus board:
- Get an Himax WE-I Plus Board on Sparkfun.
- Connect the board to Edge Impulse.
- Follow one of the tutorials on:
- Go to the Deployment tab in Edge Impulse, and build a new binary that includes your machine learning model for the AI Sensor; or export as a C++ library and integrate the model in your own firmware.
Voila! You now have an efficient machine learning model running with minimal power consumption.
The Himax WE-I Plus is a low-power AI board with a dedicated DSP that makes machine learning models run with minimal latency around audio, image and motion recognition. Head over to the Himax WE-I Plus guide and start building your embedded machine learning model!
Want to learn more? Want to see the board in action? Our CTO, Jan Jongboom, is using the Himax WE-I Plus to build a computer vision model that runs on Himax WE-I Plus during his session at the IoT Online Conference this Wednesday. Also, we’ll be hosting a webinar together with Himax in January, so keep an eye out for our developer newsletter for the link!
Aurelien Lequertier, Lead User Success Engineer at Edge Impulse