Augmented Reality-based Sudoku Solver with Training Module to Improve Cognitive Skills

This paper presents an interactive Augmented Reality (AR) Sudoku Scanner application that leverages Computer Vision and a dancing link algorithm to solve Sudoku puzzles in real-time. The application employs various techniques, including thresholding, greyscaling, and a Convolutional Neural Network (CNN) model, to accurately recognize Sudoku grids and individual digits. Beyond the scanner, the proposed app integrates a playable Sudoku game with multiple difficulty levels, hint functionality, and sign-up/login-based user authentication for a personalized experience. A user leaderboard further incentivizes engagement. Furthermore, a feedback mechanism is implemented to gather user data and perform basic data analysis. This analysis, visualized through charts and diagrams, enables us to understand user interactions and optimize the application’s user interface and experience. It was observed that the proposed app helped to improve cognitive skills in younger users compared to older ones.