Building with TensorFlow Lite for microcontrollers | Workshop

preview_player
Показать описание
Today, people use TensorFlow to develop large scale machine learning models. But did you know that TensorFlow can now run on microcontrollers? In this Workshop, the speaker discusses the potential of building with TensorFlow Lite for microcontrollers. He debuts demos, shows you how to train a model, and explains where TensorFlow fits in the TinyML ecosystem. Participants should have an Arduino Sense 33 BLE and install the Arduino IDE, but it’s not required.

Resources:

Speaker: Pete Warden

Watch more:

#GoogleIO #AI/ML #OpenSource #IoT

product: TensorFlow - TensorFlow Lite; event: Google I/O 2021; fullname: Pete Warden; re_ty: Livestream;
  1. TensorFlow
  2. purpose: Educate
  3. type: Conference Talk (Full production)
  4. pr_pr: Google I/O
  5. TensorFlow
  6. TensorFlow and microcontrollers
Рекомендации по теме
Building with TensorFlow 0:01:49

Building with TensorFlow Lite for microcontrollers - Air Pencil

Building a Home 0:01:09

Building a Home HIIT Workout Trainer (Demo) | TensorFlow Lite | Edge TPU | Raspberry Pi | James Wong

C++ : How 0:01:12

C++ : How to build TensorFlow Lite as a static library and link to it from a separate (CMake) projec

How to build 0:22:06

How to build TensorFlow Lite GPU with a single script for Android

SensiML + TensorFlow 0:05:49

SensiML + TensorFlow Lite Tutorial: Using TensorFlow Lite for Microcontrollers Classification

building tensorflow lite 0:36:34

building tensorflow lite for android (failed)

TensorFlow Lite Demo 0:04:18

TensorFlow Lite Demo apps - 14: On-device Recommendation

AMB23 TensorFlow Lite 0:00:31

AMB23 TensorFlow Lite for Ameba RTL8722

Project Data Mining 0:10:49

Project Data Mining - Aplikasi Digit Classifier Tulisan Tangan dengan TensorFlow Lite (CNN)

TensorFlow Lite C 0:45:29

TensorFlow Lite C API on Windows

TensorFlow Lite in 1:16:49

TensorFlow Lite in Action with Google Colab

TensorFlow Lite Demo 0:04:56

TensorFlow Lite Demo apps - 3: Object detection - keyboard, mouse, bike in traffic. (no coding)

Extract tensorflow lite 0:05:34

Extract tensorflow lite for use in platform io with the ESP32

TensorFlow Lite with 0:12:19

TensorFlow Lite with XNNPACK on windows

SensiML + TensorFlow 0:01:19

SensiML + TensorFlow Lite Tutorial: Building a Boxing Sport Wearable Device Application

Tensorflow lite on 0:00:19

Tensorflow lite on the Raspberry Pi 4 for image classification

#TFShowcase: Building a 1:47:27

#TFShowcase: Building a TensorFlow Lite App

Flutter & Google 0:03:25

Flutter & Google Gemini - Build Chatbots for Android and IOS - Use Gemini API in Flutter

Build iOS & 0:01:13

Build iOS & Android Object Detection Application using Flutter & TensorFlow Lite Tutorial 20...

TensorFlow lite course 1:46:33

TensorFlow lite course for Android App Development | train tensorflow lite models for android

TensorFlow Lite Demo 0:03:36

TensorFlow Lite Demo apps - 10: Style Transfer

Code walkthrough - 0:03:34

Code walkthrough - Device-based Models with TensorFlow Lite

TensorFlow Lite Demo 0:06:13

TensorFlow Lite Demo apps - 13: Natural Language Question and Answering

Object Detection using 1:07:08

Object Detection using Tensorflow Lite C API on Windows

Комментарии
Автор

Very interesting presentation!! It was very usefull for me! thanks!!

bdoriandasilva
Автор

Thank you so much this was really interesting

xecoq
Автор

Can we modify the lib to work with esp32? And is there anyway to download the model directly to the microcontroler without flashing a whole new program with new model? if no way to do it I just only thinking about upload the model to firebase and use esp32 download the model so that we can use it right way.

johnnietien
Автор

Would love to see an overview of the tools/steps involved.. for example, how do you generate the training data? Edge Impulse Studio?

Choosing the right framework for your application? Pytorch, Tensorflow/TinyML, Keras?

How can you use cloud-services to speed-up the process? Google Colab, kaggle ? And how do you go from this output to something that can be deployed on a microcontroller (ESP32), FPGA, SBC (Raspberry Pi/BeagleBoneBlack), etc? Fixed vs floating point considerations?

sentdex has a video on using cloud-based GPU (paperspace, linode)

bennguyen
Автор

can we have a tinyML game hackaton? that would be suit to see how it will work. and also interested in viewing the limiations!

tomasbarrios
Автор

29:44 Arduino_TensorFlowLite is not available via the Library Manager, what should be used instead?

DAYSGG
Автор

Hi can I use this for Camera Capture ?

checkerface-
Автор

"Compilation error: ble_file_transfer.h: No such file or directory" - anyone else getting this error, any fix?

benthompson
Автор

is it a requirement to have Arduino NANO 33 BLE Sense ??

afiqkazuhiko
Автор

I am just a hobbyist, interested learning of what is possible. Taught by a genius. Thank you for

marcush.