Inspiration
As students who believe in reflection and more traditional forms of learning through handwriting, we believed that there is a gap in student tools which unite digital and physical forms, as well as personal and academic development. We wanted to make a simple, but effective hub that allows us to consistently reflect on our development as people and as students.
What it does
We have multiple features, including flashcards, converting handwriting into electronic formats for notes and journal entries, tracking mood and academic performance, task tracking, habit tracking, and a chatbot.
How we built it
We built the backend using Python, leveraging APIs like Google Cloud Vision OCR (Optical Character Recognition), Groq, and supabase. The frontend was also built with Python, specifically using the streamlit library.
Challenges we ran into
We had to learn to use the Google Cloud API, supabase API, and streamlit library from scratch and integrate them seamlessly together which was challenging. Additionally, some features we tried to implement had to be removed because we weren't able to get them to work in time. We also had difficulty with consistency with the OCR analysis of handwriting, especially for more unique styles of writing.
Accomplishments that we're proud of
Having a working website and a functional handwriting-to-text OCR model. The base Google Cloud OCR model for handwriting was slightly inconsistent in its output and we are proud of our solution of using LLaMA 3.2 to cross-reference and double check its results for a more consistent analysis of handwriting.
What we learned
We learned about all the libraries that we used, and about how to manage a big project among multiple people. We also learned specifically about how OCR models actually break text documents down and analyze them.
What's next for GAAT
More fleshed out features like better OCR detection (custom training based on user writing style, etc.) and support for uploaded user documents in the chatbot and flashcard widgets (have chatbot analyze the document and help answer questions, generate flashcards from handwritten notes).
Log in or sign up for Devpost to join the conversation.