Poké-Pi-Dex - A Real Pokédex
Demo video of the project made by Karina Chichifoi and Michele Righi, for Digital Systems M (Alma Mater Studiorum - Bologna University).
GitHub Repository: https://github.com/TryKatChup/Poke-Pi...
Poké-Pi-Dex implements a physical Pokédex, capable of recognizing the 151 Pokémon of the first generation given an image, showing the main details (name, id, type, evolution, description and basic statistics).
The application uses a convolutional neural network (CNN) and runs on a Raspberry Pi4 with several components attached, including a 3.5" LCD display, Pi Camera 1.3, speaker, push buttons and an analog joystick. We made the case, inside which it is mounted, in recycled cardboard, inspired by the third generation Pokédex model (Hoenn). To implement the application we used Python3 with the following packages:
tkinter for the graphical interface;
opencv to get images from the camera and for calibration;
tensorflow (+ keras) for the realization of the model.
Other packages used: pygame, numpy, pillow, gpiozero, tflite_runtime.
Video made and edited by Lorenzo Castriota "Brian"