Use Tensorflow on Raspberry Pi

This article describes how to install tensorflow on the Raspberry Pi and how to run the tensorflow model on the Raspberry Pi.
This article applies to Raspberry Pi 3B, 4B

Tensorflow environment installation

1. pip replaces domestic sources. If the
download speed of foreign sources is too slow, it is easy to time out and make mistakes.
Tsinghuayuan

pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

Doubanyuan

pip config set global.index-url http://pypi.douban.com/simple/

Aliyuan

pip config set global.index-url http://mirrors.aliyun.com/pypi/simple/

University of Science and Technology of China

pip config set global.index-url https://pypi.mirrors.ustc.edu.cn/simple/

Baidu source

pip config set global.index-url https://mirror.baidu.com/pypi/simple/

Note: The default download timeout time can be changed

pip --default-timeout=600 install **

2. Install tensorflow

Download link: https://github.com/lhelontra/tensorflow-on-arm/releases

Choose a suitable version to download.
This article uses tensorflow-2.0.0-cp37-none-linux_armv7l.whl to
download and upload it to the Raspberry Pi .

pip install tensorflow-2.0.0-cp37-none-linux_armv7l.whl

There may be many timed out during the installation process. You can download the corresponding package first, and then drag it to the Raspberry Pi to install.

Note:
Encountered problems:

solutions:

pip3 install wrapt --ignore-installed

Raspberry Pi runs Tensorflow model

After repeated attempts, I found a way to run the model on the Raspberry Pi
1. Save the model

model.save_weights('model/path_to_saved_model/test.ckpt')

2. Model reading

model = creat_model()
model_to_be_restored.load_weights("model/path_to_saved_model/test.ckpt")