Skip to content

Fifteen puzzle variant with MNIST image recoginition

Notifications You must be signed in to change notification settings

shem-anton/8_puzzle

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

8 puzzle

Practicing the basics of tensorflow framework for recognizing MNIST digits.

The goal of "8 puzzle" is to shuffle the tiles on the board to order them from 1 to 8 with the empty tile in the bottom right corner. The player is only allowed to move tiles in an empty space. Seems like an easy puzzle, but it is not trivial! Fortunately an assistant helps you to find the next move. However an assistant first needs to recognize the boards, because digits are presented as images. Here is where a neural network comes into play.

To run the game you need tensorflow and tkinter. When the dependencies are installed simply run

python3 game/game.py

Unfortunately, the neural network very rarely learns to the point where it can recognize every digit. It gets pretty close, but a few tiles are typically guessed wrong. Maybe it is because of all incorrect labels in MNIST dataset ¯\(ツ)

About

Fifteen puzzle variant with MNIST image recoginition

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published