Do I Have a Banana?

• 

  Unfortunately, I don't own any bananas at home

• 

  Fortunately, I can Google an image of a banana

• 

  Do you like those accuracies? Because I certainly do

Inspiration

There were a lot of bananas here, I also ate a lot of those bananas.

What it does

Classifies an image from you webcam to determine whether it contains a banana or not.

How I built it

Trained an image classification model for bananas using Nvidia Digits and AWS. For dataset collection, I used images from ImageNet and some of my own from the bananas available. Then I ported the model to a more javascript and web friendly format, and built a website to interact with. I used github-pages to host the site.

Challenges I ran into

Getting the model to run on a web environment was very difficult. Also, planning around training time is key, as it takes more than 4 or 5 hours to get a good and accurate model.

Accomplishments that I'm proud of

Used Deep Learning on a Static Webpage

What I learned

How to detect bananas

What's next for Do I Have a Banana?

Getting a model to run on the web is extremely useful. Due to the ever growing presence of deep learning, having it run on something as accessible as the web allows much more application

Built With

• amazon-web-services
• digits
• domains.com
• github
• javascript
• nvidia