Chicken Scratch

Inspiration

When professors go over their syllabuses, there's a consistent note, that writing out your notes is better than typing them out when it comes to learning and remembering the content. Source However, if you follow this idea, you may end up in the situation where at the end of the semester you can't clearly read what you wrote down, or can't find the note you want to find. Our goal was to utilize the Google Cloud Vision API to solve this issue.

What it does

Our website reads in an image or a list of images, and calls on the Cloud Vision API to access their pre-trained Vision AI. From there, we process these images into a clear and concise pdf edition of the given notes, ready to use for review.

How we built it

We utilized Node.js to build up our server example, and our website example. The Google Cloud Vision's AI processes and identifies the words in these images, and we take those and format them into a pdf.

Challenges we ran into

Many! We started out wanting to do a purely JavaScript program, and that quickly came upon the issue of the Vision AI API not actually being able to be set up in straight JavaScript. We then had to pivot to a Node.js system to get it working, and even then it was only working with hardcoded values, and not on the website. We realized that the Vision AI wasn't actually meant to be run client side, and this was why it was having the previous issue. While we first thought the solution would be to use Webpack to compile the javascript before loading in the site, but that wasn't the correct direction either. We were able to solve the issue by utilizing Express, which allowed us to create a local webserver with which we could run the program we had written originally, and have the website trigger it. With this, we were finally able to get the front end we had initially started with to talk to the back end we ended up needing.

Accomplishments that we're proud of

We're very proud that we were able to get our website working at all! With all of the issues we came up against, it was awesome that we were able to get our site working in its most basic form, even if we weren't able to get to all of the fancy stylizations that we had planned and designed for.

What we learned

Neither of us had ever touched a Google API that was specifically for AI before and hadn't ever used the Google Cloud hosting system. Because of this, we learned in the same way you would learn to swim in the deep end, to do it or die! We had to learn quickly how to manipulate and manage these systems and even though we struggled through it, we managed to get it situated in a way we're proud of. It's certainly nothing to scoff at after starting with nothing.

What's next for Chicken Scratch

We want to implement all of the user interface design that we had planned at the beginning of the hackathon, to add in automatic sending of the pdfs (like sending it to your email, or to a specific google drive folder), and to figure out camera access, so that we won't need to rely on people taking the photos first and uploading them after, and can instead do everything in one simple place.

Built With

• cloud-vision
• css
• google-cloud
• html5
• javascript
• node.js