Inspiration - The inspiration for this project came from the need to easily share notes without sacrificing the benefits of writing by hand. With Scribble, students will be able to quickly share their notes with their classmates, and be able to use text searching to find certain topics. This will make using notes far easier for many.

What it does - The MVP of Scribble takes in an image file containing handwritten characters and outputs the digital equivalent after processing.

How we built it - It was built using react, which utilizes HTML and Javascript, along with Amazon's Rekognition software.

Challenges we ran into - Originally we planned to develop our own algorithm and training process to recognize digits. However, it was extremely difficult to accomplish this without experience in neural networks or linear algebra. As such, we moved towards implemented an existing, efficient algorithm in order to develop a better understanding of these programs.

Accomplishments that we're proud of - We are proud of the programs ability to take an image and process it accurately, while also accounting for line breaks.

What we learned - We learned a lot about machine learning and how these character recognition programs function. We also learned the valuable lesson of not being overly ambitious, as much time was spent on our own training program when we did not really know what we were doing.

What's next for Scribble - Scribble has a lot of potential for the future. First, we want to adapt it to automatically create a Google Doc containing their notes. We also want to add the ability to automatically classify the document based on the headings at the top. Finally, there are many features we can add to increase utility for the user, such as support for highlighting, colors, and underlining.

Share this project:
×

Updates