This is a simple try and learn demonstration project for some topics such as;

• Raspberry Pi - Single-board small computer
• Raspberry Pi Sense Hat - Add-on board for Raspberry PI to have some sensors and LED
• Machine Learning
  • Lobe - Machine Learning Tool to train and generate models - https://lobe.ai/
  • Tensorflow
• Python
• Lobe Python API - https://github.com/lobe/lobe-python

Basically the project is about very well-known game, Rock-Scissors-Paper game. Small Python script make some random choices between rock, scissors and paper. A set of images are trained with Lobe and exported as Tensorflow model. With a Python script, user's hand gestures are predicted within that trained model for rock,scissors and paper. So user tries to beat Raspberry PI. 😀

Very simple project but it opens lots of doors to learn new things. If you believe some fancy staff can be added, please free to add...

• Lobe Python API required Tensorflow to be installed, so;
  • To install Tensorflow 2.0.0 for ARM based boards such as Rasberry PI, I have used following build

wget https://github.com/lhelontra/tensorflow-on-arm/releases/download/v2.0.0/tensorflow-2.0.0-cp37-none-linux_armv7l.whl
python3 -m pip install tensorflow-2.0.0-cp37-none-linux_armv7l.whl

• If you do not want to use Lobe Python API, just Tensorflow API's are also fine. Just check example *.py script when you exported your Lobe model as Tensorflow model
• A simple set of images that are trained with Lobe are also in model folder. Feel free to use it, but for best results, just download the Lobe(https://lobe.ai/) and create your own model