Skip to content
This repository has been archived by the owner on Dec 8, 2021. It is now read-only.

Latest commit

 

History

History
86 lines (68 loc) · 2.99 KB

setup_tflite.md

File metadata and controls

86 lines (68 loc) · 2.99 KB
Raw

How to use the Tflite-Interpreter on a fresh Raspberry OS install.

The process is the same for fresh Raspberry OS 32 bit and 64 bit versions.

The non Raspberry OS specific parts can also be used for other Linux distributions, such as Ubuntu.

BASICS

Install Raspberry OS on your device and boot to desktop. The site Raspberry Pi OS offers tools to install the OS quickly.

Connect to a wifi or ethernet of your choice using the GUI or the wpa supplicant. The wifi setup process via GUI is described here.

To check your IP adress enter

ifconfig

It is always recommended to run

sudo apt update
sudo apt upgrade -y
sudo apt autoremove

on a fresh install of any Linux distribution to bring your system up to date. This may take a while, depending on the age your image and the speed of your device.

(OPTIONAL) ssh on the RPi

if you want to work remotely, enable ssh by entering

sudo raspbi-config

then under

  1. Interface Options
  2. P2 SSH
  3. <Yes> <No>
  4. choose the option Yes.

SSH will be available after a restart which can be invoked either with

sudo reboot

or through the GUI.

Tflite interpreter setup

There are a few methods to get the Tflite-Interpreter to run on Linux. It can be built from source via cross compilation, built directly on the device, or a pre-compiled Python wheel can easily be installed via pip, if it is available for your platform. It is recommended to use a Python virtual environment (venv) for better packet control. Here, we will use Python 3 as base. To check which Python 3 version you are running and setup a venv, go to a directory of your choosing where the virtual environment will be setup and enter

python3 --version
python3 -m venv venv_tflite

Now a directory venv_tflite has been created in the current folder. The virtual environment can be activated by sourcing the activate file via

source venv_tflite/bin/activate

Now every package that will be installed via pip, will be contained in this directory. To check which packages are installed enter

pip3 list

Pip in your venv will most likely be outdatet. To update pip enter

pip3 install --upgrade pip setuptools

Now everything is ready to install the tflite interpreter For example for Python3.7 and ARM 32 enter

pip3 install https://dl.google.com/coral/python/tflite_runtime-2.1.0.post1-cp37-cp37m-linux_armv7l.whl

and after a few seconds it should be completed. Links to other Tflite packages and how to run inference can be found in https://www.tensorflow.org/lite/guide/python

To build Tflite from source, consider following this guide https://qengineering.eu/install-tensorflow-2-lite-on-raspberry-pi-4.html

Issues

Should any issues arise during the completion of the guide or any errors noted, please let us know by filing an issue and help us keep up the quality.