Bananarama

Bananas made more accessible for the visually-impaired

Comment

Inspiration

An aspect of life that most individuals take for granted is the ability to see with high precision and accurate colors. Unfortunately, a significant portion of the population lacks these qualities. Today, one out of every twelve males is color blind and the global population of color-blind individuals is equivalent to the population of the United States. This means there are 300 million people who cannot distinguish between a succulently sweet banana or a starchy plantain-like fruit. Our project strives to ensure that all people can make informed decisions when choosing the fruit they want to eat.

What it does

Bananarama categorizes bananas into the categories of unripe, ripe, and overripe given an image of the banana in question.

How we built it

Using the Color Thief API, we determined the RGB values of the top 5 dominating colors. Then, we averaged the values of each parameter by squaring each parameter value, taking the mean of the squared values, and finally square rooting that average. The summative color would then be referenced against the determined color ranges for unripe, ripe, and overripe.

Challenges I ran into

Originally, we worked with Tensorflow, a neural network that we primed with 100 images of unripe, ripe, and overripe images each. Unfortunately, we ran into multiple formatting issues. First, the images needed to have a very particular length by width. We tackled this issue by cropping each image to the top left corner in a 96*96 pixel square. Next, our images were all transferred from RGB to BGR since BGR is the default for Matplotlib. Unfortunately, this conversion turned our yellow bananas into blue ones.

Accomplishments that I'm proud of

This is my (Anand's) first hackathon, and both my partner and I are high school students who recently turned 18. We have been excited about this event for the past several months, and we finally made it! We developed our idea and fleshed it out despite several setbacks in a single day. Also, while gathering bananas to test our program, we took the initiative to travel to a nearby Target and ask for the ripest bananas they had in storage. As shy young-adults, this request took a modicum of courage.

What I learned

I basically took a crash course in how to apply my Java skills from AP Computer Science to a full-blown project. We also extensively researched the structure of the internet through its domains. We also familiarized ourselves with the specifications and mechanisms of Tensorflow despite it not appearing in our final product. In the USAF workshop, we learned the basics of "traditional hacking" and garnered a new appreciation for cybersecurity specialists. Finally, we learned how to play Soylent Pong, which will soon be an essential skill as we embark on our college careers.

Built With

Share this project:

Updates