experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

TensorFlow Lite for Microcontrollers docs update

Browse Source 
PiperOrigin-RevId: 259163029
This commit is contained in:
Daniel Situnayake 2019-07-20 19:30:23 -07:00 committed by TensorFlower Gardener
parent 20562226f0
commit b0bc0ead62
3 changed files with 83 additions and 74 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
tensorflow/lite/experimental/micro/examples/hello_world/README.md
View File

@ -32,11 +32,17 @@ Microcontrollers.
### Build the code ### Build the code
To compile and test this example on a desktop Linux or MacOS machine, download To compile and test this example on a desktop Linux or macOS machine, first
[the TensorFlow source code](https://github.com/tensorflow/tensorflow), `cd` clone the TensorFlow repository from GitHub to a convenient place:
into the source directory from a terminal, and then run the following command:
```bash
git clone --depth 1 https://github.com/tensorflow/tensorflow.git
``` ```
Next, `cd` into the source directory from a terminal, and then run the following
command:
```bash
make -f tensorflow/lite/experimental/micro/tools/make/Makefile test_hello_world_test make -f tensorflow/lite/experimental/micro/tools/make/Makefile test_hello_world_test
``` ```

5
tensorflow/lite/g3doc/microcontrollers/build_convert.md
View File

@ -9,6 +9,11 @@ This document explains the process of converting a TensorFlow model to run on
microcontrollers. It also outlines the supported operations and gives some microcontrollers. It also outlines the supported operations and gives some
guidance on designing and training a model to fit in limited memory. guidance on designing and training a model to fit in limited memory.
For an end-to-end, runnable example of building and converting a model, see the
following Jupyter notebook:
<a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/hello_world/create_sine_model.ipynb">create_sine_model.ipynb</a>
## Model conversion ## Model conversion
To convert a trained TensorFlow model to run on microcontrollers, you should use To convert a trained TensorFlow model to run on microcontrollers, you should use

140
tensorflow/lite/g3doc/microcontrollers/get_started.md
View File

@ -3,12 +3,54 @@
This document will help you start working with TensorFlow Lite for This document will help you start working with TensorFlow Lite for
Microcontrollers. Microcontrollers.
## Sample code Start by reading through and running our [Examples](#examples).
To get started, you can explore the following example: Note: If you need a device to get started, we recommend the
[SparkFun Edge Powered by TensorFlow](https://www.sparkfun.com/products/15170).
It was designed in conjunction with the TensorFlow Lite team to offer a flexible
platform for experimenting with deep learning on microcontrollers.
<a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/micro_speech">Micro For a walkthrough of the code required to run inference, see the *Run inference*
speech example</a> section below.
## Examples
There are several examples that demonstrate how to build embedded machine
learning applications with TensorFlow Lite:
### Hello World example
This example is designed to demonstrate the absolute basics of using TensorFlow
Lite for Microcontrollers. It includes the full end-to-end workflow of training
a model, converting it for use with TensorFlow Lite, and running inference on a
microcontroller.
In the example, a model is trained to replicate a sine function. When deployed
to a microcontroller, its predictions are used to either blink LEDs or control
an animation.
<a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/hello_world">Hello
World example</a>
The example code includes a Jupyter notebook that demonstrates how the model is
trained and converted:
<a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/hello_world/create_sine_model.ipynb">create_sine_model.ipynb</a>
The process of building and converting a model is also covered in the guide
[Build and convert models](build_convert.md).
To see how inference is performed, take a look at
[hello_world_test.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/experimental/micro/examples/hello_world/hello_world_test.cc).
The example is tested on the following platforms:
- [SparkFun Edge Powered by TensorFlow (Apollo3 Blue)](https://www.sparkfun.com/products/15170)
- [Arduino MKRZERO](https://store.arduino.cc/usa/arduino-mkrzero)
- [STM32F746G Discovery Board](https://www.st.com/en/evaluation-tools/32f746gdiscovery.html)
- Mac OS X
### Micro Speech example
This example uses a simple This example uses a simple
[audio recognition model](https://www.tensorflow.org/tutorials/sequences/audio_recognition) [audio recognition model](https://www.tensorflow.org/tutorials/sequences/audio_recognition)
@ -16,48 +58,43 @@ to identify keywords in speech. The sample code captures audio from a device's
microphones. The model classifies this audio in real time, determining whether microphones. The model classifies this audio in real time, determining whether
the word "yes" or "no" has been spoken. the word "yes" or "no" has been spoken.
The sample works end-to-end (including audio capture and inference) on the <a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/micro_speech">Micro
following platforms: Speech example</a>
The [Run inference](#run_inference) section walks through the code of the Micro
Speech sample and explains how it works.
The example is tested on the following platforms:
- [SparkFun Edge Powered by TensorFlow (Apollo3 Blue)](https://www.sparkfun.com/products/15170) - [SparkFun Edge Powered by TensorFlow (Apollo3 Blue)](https://www.sparkfun.com/products/15170)
- [STM32F746G Discovery Board](https://www.st.com/en/evaluation-tools/32f746gdiscovery.html) - [STM32F746G Discovery Board](https://www.st.com/en/evaluation-tools/32f746gdiscovery.html)
- Mac OS X - Mac OS X
### SparkFun Edge Note: To get started using the SparkFun Edge board, we recommend following
If you need a device to get started, we recommend the
[SparkFun Edge Powered by TensorFlow](https://www.sparkfun.com/products/15170).
It was designed in conjunction with the TensorFlow Lite team to offer a flexible
platform for experimenting with deep learning on microcontrollers.
To get started using the Edge board, we recommend following
[Machine learning on a microcontroller with SparkFun TensorFlow](https://codelabs.developers.google.com/codelabs/sparkfun-tensorflow), [Machine learning on a microcontroller with SparkFun TensorFlow](https://codelabs.developers.google.com/codelabs/sparkfun-tensorflow),
a codelab that introduces you to the development workflow. a codelab that introduces you to the development workflow using the Micro Speech
example.
## Workflow ### Micro Vision example
Using TensorFlow Lite for Microcontrollers involves four major steps: This example shows how you can use TensorFlow Lite to run a 250 kilobyte neural
network to recognize people in images captured by a camera. It is designed to
run on systems with small amounts of memory such as microcontrollers and DSPs.
1. Create or find a model architecture. <a class="button button-primary" href="https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/examples/micro_vision">Micro
2. Train a model. Vision example</a>
3. Convert the model.
4. Write code to run inference.
The first three steps are covered in the guide The example is tested on the following platforms:
[Build and convert models](build_convert.md). The sample code comes with a
pretrained model, and includes scripts to train a model that recognizes
different spoken words. Instructions on training are in
[README.md](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/experimental/micro/examples/micro_speech/README.md#creating-your-own-model).
In this document, we will focus on the code that will feed processed audio data - [SparkFun Edge Powered by TensorFlow (Apollo3 Blue)](https://www.sparkfun.com/products/15170)
into the model and execute it, resulting in a prediction of which word was - [STM32F746G Discovery Board](https://www.st.com/en/evaluation-tools/32f746gdiscovery.html)
spoken. This process is called *inference*. - Mac OS X
## Run inference ## Run inference
The sample's The following section walks through the [Micro Speech](#micro_speech) sample's
[main.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/experimental/micro/examples/micro_speech/main.cc) [main.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/experimental/micro/examples/micro_speech/main.cc)
contains the code that runs inference. We'll now walk through the key parts. and explains how it used TensorFlow Lite for Microcontrollers to run inference.
### Includes ### Includes
@ -277,48 +314,9 @@ recognition results across a number of frames. This is defined in
The same technique can be used to improve reliability when processing any The same technique can be used to improve reliability when processing any
continuous stream of data. continuous stream of data.
## Build the sample
The sample contains build scripts that will download all required dependencies
and compile a binary that can be run on a device.
Note: The build process has been tested on MacOS and Linux, but not on Windows.
To build the sample, take the following steps:
1. Clone the TensorFlow repository from GitHub to a convenient place.
```bash
git clone --depth 1 https://github.com/tensorflow/tensorflow.git
```
1. Enter the directory that was created in the previous step.
```bash
cd tensorflow
```
1. If you are using MacOS, run the following command. If you are using Linux,
you do not need to do this.
```bash
PATH=tensorflow/lite/experimental/micro/tools/make/downloads/gcc_embedded/bin/:$PATH
```
1. To download all of the required dependencies and initiate the build process,
issue the following command. You can set `TARGET` depending on which
platform you want to build for. Explore
[`targets/`](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/micro/tools/make/targets)
for the current options.
```bash
make -f tensorflow/lite/experimental/micro/tools/make/Makefile
TARGET=sparkfun_edge micro_speech_bin
```
## Next steps ## Next steps
Once you have built and run the sample, read the following documents: Once you have built and run the samples, read the following documents:
* Learn how to work with models in * Learn how to work with models in
[Build and convert models](build_convert.md). [Build and convert models](build_convert.md).