Google has been striving hard to make local AI possible by applying AI to products and other domains as well as developing tools. In local AI, the speed of local inference allows considerable savings on bandwidth and cloud compute costs while keeping data local preserves user privacy. These are the few reasons as to why organisations prefer to use the local inferencing platform.
Last year, the tech giant launched Coral, an AI platform to build products with local AI. The platform has gained lots of traction since then and has witnessed frequent updates. Popular organisations like Asus, Florida-based care.ai, among others have also been using this platform for the development of AI. The multinational company, Asus has chosen the Coral SOM as the base of its Tinker Edge T product. The company will be using the Google Edge TPU to speed up its processing efficiency and build connected devices and intelligent applications.
Google Coral provides hardware components and software tools that make it easy to prototype and scale local AI products. In this platform, the solutions for various on-device intelligence include object detection, pose estimation, image segmentation, and key-phrase detection. Last year, the tech giant also released Coral Dev Board which is a hardware board to accelerate AI edge computing and a USB accelerator which is a USB accessory featuring the Edge TPU, and Mini PCIe Accelerator
Recently, the tech giant announced new additions to the Coral platform. The unveiled hardware products for Coral include Coral Accelerator Module, Coral Dev Board Mini and some new variations to the Coral System-on-Module.
About The Products
Coral Accelerator Module
This is an easy to integrate multi-chip package that encapsulates the Google Edge TPU ASIC. Coral Accelerator Module exposes both PCIe Gen 2 and USB 2.0 interfaces and can easily integrate into custom PCB designs. The mounting type is SMT, 120-pin LGA and the control interface is 12C (optional). It has a number of functionalities such as performs high-speed ML inferencing, supports Tensorflow Lite as well as supports AutoML vision edge. At CES 2020, the tech giant will be producing a demo of this module and it will be available publicly by the first half of this year.
Coral Dev Board Mini
Coral Dev Board Mini is a development board which provides a smaller form-factor, lower-power, and lower-cost alternative to the Coral Dev Board. This board has the capability to perform high-speed ML inferencing, support Tensorflow Lite, support AutoML vision edge as well as working as a single-board computer with SoC, Ml, and wireless connectivity. It combines the new Coral Accelerator Module with the MediaTek 8167s SoC to create a board that excels at 720P video encoding/decoding and computer vision use cases.
This year, the tech giant has offered new variations to the Coral System-on-Module. The System-on-Module is a fully integrated system for accelerated ML applications which includes CPU, GPU, Edge TPU, Wi-Fi, Bluetooth, and secure element in a 40mm x 48mm pluggable module. This module is now available with 2GB and 4GB LPDDR4 RAM in addition to the original 1GB LPDDR4 configuration.
Ahead of Consumer Electronics Show (CES) 2020 which is going to be held on 7th-10th Jan, the tech giant unveiled these new hardware products of Coral. Google will display these hardware products as well as showcase demo at the upcoming event. Further, the developers at Google will be showcasing how the System-on-Module (SoM) can be used in a smart city, manufacturing sector, and healthcare applications.