README.rst
1.. zephyr:code-sample:: tflite-hello-world
2 :name: Hello World
3
4 Replicate a sine wave using TensorFlow Lite for Microcontrollers.
5
6Overview
7********
8
9This sample TensorFlow application replicates a sine wave and
10demonstrates the absolute basics of using TensorFlow Lite Micro.
11
12The model included with the sample is trained to replicate a
13sine function and generates x values to print alongside the
14y values predicted by the model. The x values iterate from 0 to
15an approximation of 2π.
16
17The sample also includes a full end-to-end workflow of training
18a model and converting it for use with TensorFlow Lite Micro for
19running inference on a microcontroller.
20
21The sample comes in two flavors. One with TensorFlow Lite Micro
22reference kernels and one with CMSIS-NN optimized kernels.
23
24.. Note::
25 This README and sample have been modified from
26 `the TensorFlow Hello World sample for Zephyr`_.
27
28.. _the TensorFlow Hello World sample for Zephyr:
29 https://github.com/tensorflow/tflite-micro/tree/main/tensorflow/lite/micro/examples/hello_world
30
31Building and Running
32********************
33
34The sample should work on most boards since it does not rely
35on any sensors.
36
37Add the tflite-micro module to your West manifest and pull it:
38
39.. code-block:: console
40
41 west config manifest.project-filter -- +tflite-micro
42 west update
43
44The reference kernel application can be built and executed on QEMU as follows:
45
46.. zephyr-app-commands::
47 :zephyr-app: samples/modules/tflite-micro/hello_world
48 :host-os: unix
49 :board: qemu_x86
50 :goals: run
51 :compact:
52
53Exit QEMU by pressing :kbd:`CTRL+A` :kbd:`x`.
54
55The CMSIS-NN kernel application can be built and executed on any Arm(R) Cortex(R)-M core based platform,
56for example based on Arm Corstone(TM)-300 software. A reference implementation of Corstone-300
57can be downloaded either as a FPGA bitfile for the
58[MPS3 FPGA prototyping board](https://developer.arm.com/tools-and-software/development-boards/fpga-prototyping-boards/mps3),
59or as a
60[Fixed Virtual Platform](https://developer.arm.com/tools-and-software/open-source-software/arm-platforms-software/arm-ecosystem-fvps)
61that can be emulated on a host machine.
62
63Assuming that the Corstone-300 FVP has been downloaded, installed and added to
64the :envvar:`PATH` variable, then building and testing can be done with following
65commands.
66
67```
68$ west build -p auto -b mps3/corstone300/an547 samples/modules/tflite-micro/hello_world/ -T sample.tensorflow.helloworld.cmsis_nn
69$ FVP_Corstone_SSE-300_Ethos-U55 build/zephyr/zephyr.elf
70```
71
72Sample Output
73=============
74
75.. code-block:: console
76
77 ...
78
79 x_value: 1.0995567*2^1, y_value: 1.6951603*2^-1
80
81 x_value: 1.2566366*2^1, y_value: 1.1527088*2^-1
82
83 x_value: 1.4137159*2^1, y_value: 1.1527088*2^-2
84
85 x_value: 1.5707957*2^1, y_value: -1.0849024*2^-6
86
87 x_value: 1.7278753*2^1, y_value: -1.0509993*2^-2
88
89 ...
90
91The modified sample prints 10 generated-x-and-predicted-y pairs. To see
92the full period of the sine curve, increase the number of loops in :file:`main.c`.
93
94Modifying Sample for Your Own Project
95*************************************
96
97It is recommended that you copy and modify one of the two TensorFlow
98samples when creating your own TensorFlow project. To build with
99TensorFlow, you must enable the below Kconfig options in your :file:`prj.conf`:
100
101.. code-block:: cfg
102
103 CONFIG_CPP=y
104 CONFIG_REQUIRES_FULL_LIBC=y
105 CONFIG_TENSORFLOW_LITE_MICRO=y
106
107Note that the CMSIS-NN kernel sample demonstrates how to use CMSIS-NN optimized kernels with
108TensorFlow Lite Micro, in that is sets below Kconfig option. Note also that this
109Kconfig option is only set for Arm Cortex-M cores, i.e. option CPU_CORTEX_M is set.
110
111.. code-block:: cfg
112
113 CONFIG_TENSORFLOW_LITE_MICRO_CMSIS_NN_KERNELS=y
114
115Training
116********
117Follow the instructions in the :file:`train/` directory to train your
118own model for use in the sample.
119
