Tensorflow/ML/IoT Hardware

Embedded Development Boards, Kits, Programmers
,
1

Hello,

I just wanted to have your advice about a product design using components from Digikey.

I am trying to put together a small system/board using a microcontroller that could support high calculations such as deep learning implementation taking into consideration the memory too. Could you please guide me to a product with such capabilities in addition to the ability to use a wifi connection for IoT features?

Is there a processor that can support Tensorflow/Tensorflow Lite as a language?

I thank you very much for your help.

2

Hi @Mounir one of our best option at the moment are these Google based “Coral” accelerators:

https://www.digikey.com/en/products/filter/evaluation-boards-embedded-mcu-dsp/786?s=N4IgTCBcDaIMYHsBOBDANiAugXyA

They also come in stand-alone m.2 slot based devices…

Regards,

3

Hello @RobertCNelson,

Thank you for your reply. I went through its datasheet, I could not find the information on which programming language it does support.
Is it compatible with Tensorflow Lite? Also, Is there any “reduced” version of it because I only need the processor and Wi-Fi for connectivity only.
Thank you in advance for your help.

Regard,

4

There are multiple modules with different SBC’s the libraries can be found here:

Here’s all the different devices with this Module:

Regards,

5

Hi Mounir,

According to the TensorFlow website, the following platforms support TensorFlow Lite and have WiFi capability:

ESP-EYE - Espressif ESP-EYE
1597-102991299-ND - Seeed Wio Terminal: ATSAMD51

6

Hello @David_1528 @RobertCNelson,

Thank you for your replies. They are very helpful. The proposed options are interesting and fall into my project requirements.
I still have an inquiry about one board in this forum topic; Do you have in your stock the AZ 3166 or any other helpful information regarding it.

I really appreciate your time.

Regards,

7

Hi @Mounir

17 in stock today:

Seeed has good directions here:

and Microsoft here:

https://docs.microsoft.com/en-us/samples/azure-samples/mxchip-iot-devkit-get-started/sample/

Regards,

8

Hi @RobertCNelson,

Thank you for the quick reply.
Do you know how the AZ 3166 compares with the previously proposed Google-based “Coral” accelerators by you and the Espressif ESP-EYE and the ATSAMD51 proposed by @David_1528 ?

Thank you in advance.

Regards,

9

Of those 3, Espressif’s WiFi is more supported in third part applications then RTL8720DN (SAMD51) and finally the AZ3166’s WiFi…

The “Coral”, is a pure AI accelerator…

Regards,

10

@RobertCNelson and @David_1528,

Thank you very much for your help and time.

Best Regards,