The Recombobulator

• 

  "Inspiration"

Inspiration

Like all the greats in coding history, a combination of grit, cat pictures, and inadvisably high doses of caffeine led to this magnum opus.

What it does

Organizes your thoughts, conveniently scattered in a 4-panel jumble, into a beautiful square, probably.

How we built it

We used an algorithm to determine the “distance” (quantified difference) between the RGB values of potentially adjacent pixels. Developed with Keras in TensorFlow using our predetermined activation function and learning function, then trained the model on the data set using a predetermined loss function and developed a working model for the sorting and output of the puzzles.

Challenges we ran into

Even with only 4 segments, a given side of a starting piece had 3 unique combinations it could be placed with. We needed to determine a methodology for finding relationship between a relationship between potentially matching sides. The inclusion of RGB values makes the code more rigorous, but has potential for more accuracy.

Accomplishments that we're proud of

We built something that does something!

What we learned

How to develop models using the Keras library in TensorFlow… but also that maybe the real treasure was the friends we programmed along the way.

What's next for The Recombobulator

Negotiating a deal for a major motion picture. Look for it in Summer of 2024.

Built With

• python
• tensorflow