Google introduced Coral, the platform that supports building AI IoT hardware integration

In addition to a series of highly appreciated machine learning projects in recent times, Google today has officially announced a new platform, supporting the construction of smart hardware devices with AI. Integrated named Coral. Coral is now entering the public beta, and emerged as a useful terminal solution for developers in creating IoT hardware, from prototype building to commercial production.

Coral offers a complete set of local AI tools, making it easy to develop your ideas from prototype to production. It includes hardware components, software tools and content to help you create, train and run neural networks (NNs) locally on your device.

1. Google launched TensorFlow Lite 1.0 for mobile devices and embedded devices

The core nature of this local AI toolkit comes from the idea for offline situations, where connectivity is limited, while these low-power devices have to work with the application. dip. In addition, the other practical benefit of this toolkit is that it can store user data more securely and privately on the device. Overall, Coral is designed to run TensorFlow Lite, a very popular open source AI framework version, which has also been rebuilt for low-power devices.

Coral is used with the Edge TPU, as well as specifically designed to run in so-called edge computing - a market that many experts believe may even be larger than cloud computing.

Designed by Google, this small ASIC provides high-performance ML inference - like mobile vision models at 100+ FPS - for low-power devices.

1. This is the interface of Lite OS, Microsoft's new operating system, competitor with Chrome OS

When officially launched, Coral will give developers 5 utility products, supporting from strengthening and improving the existing designs to the end of expanding production scale, including:

1. Coral Dev Board: Single board computer with detachable system module (SOM), along with Edge TPU, as well as Wi-Fi, Bluetooth, RAM and eMMC memory.
2. USB Accelerator (USB Accelerator): An Edge TPU-equipped USB accessory, which provides the ML reasoning capability for existing systems. In addition, it can also be easily integrated into any Linux system (including Raspberry Pi) via USB 2.0 and 3.0 connectivity.
3. Camera : Camera system in the form of compatible modules, 5 megapixel resolution.
4. PCI-E Accelerator (PCI-E Accelerator): The PCI-E device makes integrating the TPU Edge into existing systems more easily.
5. System-on-Module (SOM): A fully integrated system in one, designed as a pluggable module, with dimensions of 40mm x 40mm.

Coral will have good compatibility with other Google software tools like TensorFlow and TensorFlow Lite, as well as Google Cloud IoT in managing connected devices. Besides, this tool can also work in many projects with a wide range of scale, from start-up to large enterprises.

1. Google AI application in detecting eye diseases