Vision-Based Sudoku Solver Works Through AI
Sudoku isn’t much of a challenge for today’s computers. However, getting the machine to recognize an image of a sudoku puzzle and translate it into an editable form that can then be solved is a bit more challenging. The GUI Smart Sudoku Solver is designed to collect the image, process it, and find a solution in a matter of seconds, regardless of where it gets the source image from. Anyone with a smartphone camera can now use AI to solve their sudoku puzzles for them.
How it Works
The sudoku solver is a python-based repository that’s freely available on GitHub. The brainchild of Neeramitra Reddy, a graduate of the National Institute of Technology Karnataka located in India, it offers both an AI neural network and an offline processing method to solve the puzzles. The offline process is currently the more accurate of both of the solution methodologies, but the more people use the AI, the better it will become.
Not the First Automatic Solver for Sudoku
Attempts have been made in the past to create an automatic solution for sudoku puzzles. A Swedish engineer developed a system using a LEGO robot to manually crawl the page that the problem was on and pick up the details. It would then go on to solve the puzzle on its own. The image recognition system is a lot more portable, even though its processing seems to be in a similar vein to the LEGO robot.
What This Means for the Future
AI recognition is a significant part of removing barriers between computers and the world around them. With better image processing and logical solution methods in place, we may see AI being used to perform more tasks that require sensory feedback and even learn from their mistakes. And advancement like this may be the first step in getting machine learning robots to operate more efficiently out in the real world.
