A war of words seems to have erupted over AI involving not just tech luminaries but also notable personalities from the world of academia. At the bone of contention is the argument that artificial intelligence also has a potential for misuse. Google CEO Sundar Pichai in a June 2018 blog on Google has taken the lead and put forward 7 key objectives of AI applications and has also gone as far as mentioning the areas where AI applications will not be developed by the company.
Coral is one such newly launched platform by Google which will abide by these AI principles as set by the CEO himself. Google’s Coral was launched to help developers build intelligent devices with on-device AI. It comes with a complete toolkit which helps speed up the time it takes to move from prototype to production. The tool kit includes the necessary software and hardware components.
One of the key hardware components is the Edge TPU model compiler which was updated recently to remove architecture-specific restrictions. Edge TPU performs high-level machine learning extrapolations for low power devices with ease. You can now submit any model architecture that you want through the improved Edge TPU model compiler.
Another interesting hardware components is the Coral Dev Board. The development board helps in quick prototyping on-device machine learning products. A camera can then easily connect with the Dev Board over a Mobile Industry Processor Interface (MIPI).
The software components include TensorFlow and TensorFlow Lite. The TensorFlow ops supported on the Edge TPU are designed to take full advantage of the Edge TPU at runtime.
Connected devices can be seamlessly built using Coral as it can be used with Google Cloud IoT as well which is considered as an intelligent IoT platform.