Google keeps on increasing the range of AI products and services with a trio of new hardware devices aimed at AI hardware development. The gadgets have not been officially announced yet and it was first spotted by Hackster.
It is presented under the under a newGoogle Coral brand (in Beta), includes a development board that sells for $149.99, a USB accelerator that goes for $74.99, and a 5-megapixel camera that’s available for $24.99. Both Dev board and accelerator are powered by Google’s Edge TPU chips, which is ASIC processors no bigger than your fingernail that is designed to run AI models without breaking a sweat whereas the camera is an add on for the dev board.
None of these products are useful for the training of ML algorithms, it is a heavy duty task which required GPUs or cloud processing power. Alternatively, they facilitate the inference” part of AI which comes after you have trained your model and using it in the wild. Both of the accelerator and dev board support TensorFlow Lite, which is the lightweight version of the organization’s AI framework system designed specifically for mobile and IoT gadgets.
You can also run AI models on any old CPU but they would be slow and cranky. Moreover, specially designed chips like Google’s TPU processors do the job much more efficiently without the need of internet connection or without any power source. The USB accelerator can boost inference on any Linux machine, whereas the pin ports arrangement of Dev’s board makes it perfect for prototyping the hardware and other experimental applications.
Although, Google is not the first software to introduce these kinds of products of AI hardware development. Intel released their own USB accelerator a few years ago, the Movidius-powered Neural Compute Stick. If you want more information on these devices, you can find full spec sheets and resources, including tutorials and test models on Google coral homepage.