With the recent boom in Artificial Intelligence technologies, there’s a huge demand for hardware capable of performing AI on the edge. So, a lot of electronics manufacturers are coming up with their own products, flooding the market with numerous AI boards. To stand out from the crowd, Himax launched the WE-I Plus EVB board. This AI development board packs an ultra-low-power VGA mono camera along with an accelerometer and two microphones. At its heart is the HX6537-A ASIC microcontroller which enables the board to run in ultra-low power mode. With the support for all TensorFlow Lite Micro examples, the board comes up as an all-in-one solution for your TinyML applications.
“Himax WE-I Plus, coupled with Himax AoS image sensors, broadens TensorFlow Lite ecosystem offering and provides developers with possibilities of high performance and ultra low power,” said Pete Warden, Technical Lead of TensorFlow Lite for Microcontrollers at Google. The WE-I Plus board can run examples including “Hello World”, “Micro Speech”, “Person Detection”, and “Magic Wand”, all with inference speeds in the order of a few tens of milliseconds.
WE-I Plus EVB board for your AI-based Computer Vision
For running AI Vision applications, the board features an integrated Himax HM0360 ultra-low-power VGA mono camera. This vision sensor can provide an output of up to 60FPS with a resolution of 640×480 pixels while consuming a 7.8mA current and may even go as low as 240µA in monitor mode. This AoS (Always on Sensor) CMOS sensor makes use of the proprietary sensor interface available on the HX6537-A processor. In addition to that, the HX6537-A comes with a power management scheme that periodically wakes up the AoS sensor while running ultra-low power applications. Thus, it eliminates the trouble of interfacing external cameras while consuming an average power of less than 1mW.
Discussing further, the WE-I Plus board is built around Himax’s HX6537-A high-performance, ultra-low-power microcontroller. The ARC 32-bit EM9D DSP comes with 2MB flash, 2MB SRAM and runs at a frequency of 400MHz. The microcontroller also comes with “Floating` Point Unit and XY local data memory architecture to accelerate convolution operation of neural network algorithm”. Additionally, the HX6537-A also provides a wide range of peripherals including CMOS sensor interface, audio I2S, PDM interface, along with UART, I2C, SPI, GPIO, and ADC.
Accelerate AIoT solutions with WE-I Plus’s LoRa connectivity support
As discussed earlier in this article, AI is spreading its roots in every field with one of them being IoT, giving rise to AIoT. The WE-I Plus board comes up as a platform where developers can build their AIoT edge device concept quickly to support such applications. For that purpose, the board supports the LoRa REYAX RYLR896 module for communicating wirelessly. One potential application would be to deploy the board for remote surveillance where the vision sensor can detect people and then send the data with help of the LoRa wireless module. Himax also provides a similar example on how to do that on its GitHub repository.
Additionally, the WE-I Plus EVB board features an STM LSM9DS1 IMU which contains a 3-axis accelerometer. This sensor is useful for applications involving vibrations such as the “Magic wand” example from Tensorflow. There are also two onboard PDM MEMS microphones, one on the left and the other on the right side of the board. In addition to that, there is one reset button and two user LEDs for indicating classification results. The size of the board is 40mm x 27mm which is quite small considering the number of sensors and peripherals which it packs.
Finally, talking about the pricing, the Himax WE-I Plus EVB board is available for $65 on SparkFun’s store. The price is quite reasonable for the processing power and the camera module which it provides. You can learn more about WE-I Plus on Himax’s product page.
The product page is the source of all the technical information and images used.