python3-tensorflow-lite-example.md

TensorFlow Lite Python image classification demo

Reference

• TensorFlow Lite Python image classification demo - tensorflow/tensorflow

How to

Build sample on Raspberry Pi 4 AArch64 (core-image-weston).

Clone repositories and oe-init-build-env.

git clone git://git.yoctoproject.org/poky.git
git clone git://git.yoctoproject.org/meta-raspberrypi
git clone git://git.openembedded.org/meta-openembedded
git clone https://github.com/NobuoTsukamoto/meta-tensorflow-lite.git
source poky/oe-init-build-env build

Add layer

bitbake-layers add-layer ../meta-openembedded/meta-oe/
bitbake-layers add-layer ../meta-openembedded/meta-python/
bitbake-layers add-layer ../meta-openembedded/meta-networking/
bitbake-layers add-layer ../meta-openembedded/meta-multimedia/
bitbake-layers add-layer ../meta-raspberrypi/
bitbake-layers add-layer ../meta-tensorflow-lite/

Create conf/auto.conf file and write config

Add python3-tensorflow-lite, python3-tensorflow-lite-example and python3-pillow recipes to conf/auto.conf file.

FORTRAN:forcevariable = ",fortran"
IMAGE_INSTALL:append = " python3-tensorflow-lite-example"

Bitbake

MACHINE=raspberrypi4-64 bitbake core-image-weston

Write image

Write image to micro-SD card.

Boot Raspberry Pi

Power on your Raspberry Pi.

Run example.

Launch a terminal and run the example.

cd /usr/share/tensorflow/lite/examples/python/
python3 label_image.py \
  --image ./grace_hopper.bmp \
  --model_file ./mobilenet_v1_1.0_224.tflite \
  --label_file ./labels.txt

The following results can be obtained.

0.919721: 653:military uniform
0.017762: 907:Windsor tie
0.007507: 668:mortarboard
0.005419: 466:bulletproof vest
0.003828: 458:bow tie, bow-tie, bowtie
time: 369.129ms

Code changes

The original code imports tensorflow. I am changing this to import tflite_runtime.interpreter. This has been modified to work with TensorFlow Lite.
See below for changes to the original code.

• 001-v2.15_label_image_py.patch