Inspiration
Scrambling images is an interesting topic that brings novel challenges with it.
What it does
Takes a scrambled image and unscrambles it.
How we built it
Implemented using the Python language and Numpy numerical library. I have used statistical methods to decide on the correct image.
Challenges we ran into
The image dataset has a vast variety of images, and there are multiple objects in some of them. It caused the object detection-based approaches to fail.
Accomplishments that we're proud of
The algorithm obtains high accuracy rates (>90%) on the training dataset.
What we learned
Overfitting problems of the deep learning classifiers.
What's next for Puzzle Solver
The dimensionality of the puzzle can be increased from 2x2 to higher dimensions.
Log in or sign up for Devpost to join the conversation.