Autograder

Login/Register with Auth0
Login/Register with Auth0
Upload answer and students' submissions
Upload answer and students' submissions
Wait for automated scoring
Automated question-wise analysis of student's mistakes

Inspiration

Grading handwritten student responses is time-consuming and prone to inconsistencies. We wanted to create a solution that automates this process while maintaining accuracy, efficiency, and fairness.

What We Learned

Throughout this project, we gained valuable insights into AI-powered handwriting recognition, model evaluation, and workflow automation. Additionally, we explored the challenges of integrating AI into real-world educational applications.

How We Built It

We developed Texas A&M AutoGrader using:

Google's Gemini AI for analyzing and comparing handwritten responses
Python/Streamlit to build a interactive UI and backend for processing submissions
Auth0 for secure user authentication and management

The AI model extracts text from handwritten responses, compares it with a digital answer key, and assigns scores based on predefined grading criteria.

Challenges Faced

Handwriting Variability: Different handwriting styles affected accuracy, requiring fine-tuning and preprocessing.
Performance Optimization: Processing multiple student submissions simultaneously demanded efficient backend optimization.
Integration Issues: Ensuring smooth interaction between the AI model, backend, and front-end required extensive debugging and testing.

Despite these challenges, we successfully built an AI-powered grading tool that simplifies the grading process for educators.