跳至主要内容

Make AI easier with Google Coral

Google Coral is a complete toolkit to build products with local AI. 
Among them, Coral Dev Board, Coral USB Accelerator and Coral 5-megapixel camera accessory are suitable for AI beginners and makers.
Google now also offers three Accelerators that include Mini PCIe, M.2 A+E key, and M.2 B+M key, which enable easy integration of the Edge TPU into existing systems helping create tailored solutions for your industry and make it easy to take your idea from sketch into proof-of-concept.

Google Coral Products
All three Edge TPU cards are now available to order for $34.99 on Gravitylink Store and you can also get the original Coral Edge SoM for $114.99.

1.Coral System on Module (SoM)

The Coral SoM is a fully-integrated Linux system that includes NXP's iMX8M system-on-chip (SoC), eMMC memory, LPDDR4 RAM, Wi-Fi, and Bluetooth, and the Edge TPU coprocessor for ML acceleration. It runs a derivative of Debian Linux(Mendel).
In other words, as long as power is supplied through the backplane, this Coral SoM itself is a hardware system that can run Linux completely.

Its mechanical dimensions（40x48mm）are shown in the figure below. 

2.Coral Mini PCIe card specifications:

The Mini PCIe Accelerator is a half-size Mini PCIe card designed to fit in any standard Mini PCIe interface. This form-factor enables easy integration into ARM and x86 platforms so you can add local ML acceleration to products such as embedded platforms, mini-PCs, and industrial gateways.

3.Coral M.2 Accelerator A+E key:

Considering that the PCIe interface has a variety of common standard specifications, the M.2 Accelerator also provides a version of the M.2 A + E Keyed slot (two notches in the slot and one screw in the back). It measures only 22 mm x 30 mm.

4.Coral  M.2 Accelerator, B+M Key:

If you have an M Keyed PCIe interface on your board, you can use this Coral M.2 Accelerator B+M key.
If your M.2 slot is not M Keyed but B Keyed,and the B Keyed supports PCIe signal, not SATA, then you can plug it in.
The mechanical dimensions of this module are consistent with normal M Keyed M.2-2280-B-M-S3 NVME SSDs.
Of course, the TPU acceleration module and NVME SSDs need to be traded off, considering that most boards only have one M Key interface. 
Its dimensions are 22 mm x 80 mm.
Coral is a complete local AI toolkit helping you bring on-device AI applications ideas from prototype to production easily.
It offers a platform of hardware components, software tools, and pre-compiled models allowing you to create, train, and run neural networks (NN) on your local device. It also provides fast neural network performance and enhances privacy protection. To help you bring your ideas to market, Google designed Coral components specifically to enable rapid prototyping and easy extension to production lines. When you are ready to expand to the production line, you can use the SoM and PCIe version of the development board for mass deployment. To further support your integration, Google will publish substrate circuit diagrams for users who want to build custom carrier boards.
标签：

评论

发表评论

此博客中的热门博文

10 questions you should know about Google Edge TPU

1. What is the Edge TPU? The Edge TPU is a small ASIC designed by Google that provides high performance ML inferencing for low-power devices. For example, it can execute state-of-the-art mobile vision models such as MobileNet V2 at almost 400 FPS, in a power efficient manner. Google offers multiple products that include the Edge TPU built-in. Two Edge TPU chips on the head of a US penny 2. What machine learning frameworks does the Edge TPU support? TensorFlow Lite only. 3. What type of neural networks does the Edge TPU support? The first-generation Edge TPU is capable of executing deep feed-forward neural networks (DFF) such as convolutional neural networks (CNN), making it ideal for a variety of vision-based ML applications. 4. How do I create a TensorFlow Lite model for the Edge TPU? You need to convert your model to TensorFlow Lite and it must be quantized using either quantization-aware training (recommended) or full integer post-training quantization. (To create a compatible model...
发表评论
阅读全文

How to Retrain an object detection model

This tutorial shows you how to retrain an object detection model to recognize a new set of classes. You'll use a technique called transfer learning to retrain an existing model and then compile it to run on an Edge TPU device—you can use the retrained model with either the Coral Dev Board or the Coral USB Accelerator. Specifically, this tutorial shows you how to retrain a MobileNet V1 SSD model (originally trained to detect 90 objects from the COCO dataset) so that it detects two pets: Abyssinian cats and American Bulldogs (from the Oxford-IIIT Pets Dataset). But you can reuse these procedures with your own image dataset, and with a different pre-trained model. The steps below show you how to perform transfer-learning using either last-layers-only or full-model retraining. Most of the steps are the same; just keep an eye out for the different commands depending on the technique you desire. Note: These instructions do not require deep experience with TensorFlow o...
发表评论
阅读全文

Introducing Google Coral Dev Board

Introducing Google Coral Dev Board (Part 1) Introducing Google Coral Dev Board (Part 2) Introducing Google Coral Dev Board (Part 3) Introducing Google Coral Dev Board (Part 4)
发表评论
阅读全文