ScribbleCheck

ScribbleCheck mobile demo

Inspiration

“Seventh graders. I can’t read anything they write!” my teacher grumbled to me as he poured over a stack of assignments.

Reading messy handwriting is a daily problem is teacher’s lives. Our application addresses this, and also adds an additional plagiarism and AI-check feature. With the rise of AI technology, academic dishonesty has become an increasingly prevalent issue as students use it to generate their homework.

To combat this, ScribbleCheck aims to uphold academic honesty and fairness, assisting teachers with grading tests and homework.

What it does

The application uses powerful artificial intelligence models to first transcribe messy handwritten text to digital text. It then does checks on plagiarism and unethical AI usage to keep students accountable.

How we built it

We used Microsoft Azure's Document Intelligence read model, a powerful neural network that does optical character recognition (OCR) on dense handwritten text. After deciphering student notes, essays, and other handwritten assignments, we fed the digital text into a plagiarism checkers and AI usage checker called Winston AI.

Challenges we ran into

We had a lot of trouble working with different APIs, especially the OCR API. We also tried building our own model from scratch but it had subpar accuracy, ultimately causing us to pivot to an API.