AI Maze Solver Project (UniPG)

This university project, developed and tested in Google Colab, focuses on combining computer vision with pathfinding algorithms.

1. Image Processing: From a hand-drawn grid representing a maze, extract the maze’s structural features using basic computer vision techniques (maze dimensions, contained letters, edges, etc.). Lettura_analisi_immagine.ipynb
2. Character Recognition: Apply a neural network trained on the MNIST dataset to recognize the letters detected in the maze. Classificazione_immagini.ipynb
3. Pathfinding: Reconstruct the maze and determine the optimal route using a set of classical search algorithms. Ricerca_percorso.ipynb

Technologies: Python, OpenCV, MNIST-trained neural networks, search algorithms.