SOS Grader

Teachers often grade repetitive Multiple Choice Questions, losing valuable teaching time. What if teachers could take a picture of students' responses and have them graded?

Comment

Inspiration

Teachers spend so much time grading simple multiple-choice sheets from students. What if they can get everything they used to get done in a fraction of the time?

What it does

SOS Grader allows users to upload images of student work and uses a CNN Model to identify the corresponding letter.

How I built it

Backend: Built using Python and Flask to handle image uploads and predictions. Based on a trained TensorFlow/Keras model for handwritten character recognition processes the uploaded image.

Frontend: Designed with HTML, CSS, and JavaScript to create an intuitive and responsive interface. It uses fetch API to send data to the backend and handle the results dynamically.

Integration: Flask serves as the bridge between the front end and the machine learning model.

Challenges I ran into

Frontend Communication: Debugging CORS issues and ensuring smooth communication between the front end and back end.

User Experience: Creating a visually appealing design while ensuring usability after spending most of the time on the backend and integration.

Accomplishments that I am proud of

Trained a whole neural network :D

What I learned

This was the first hack I had done on my own. I learned a lot about integration and full stack. This is also one of the first projects I made using AI, so I learned a lot about CNNs and other neural networks during my research.

What's next for SOS Grader

  1. Make it so that the user does not need to train the model so that it is more user-friendly Allow users to upload multiple images at once.
  2. Make it so that teachers can upload many quizzes at once and be able to post the quizzes and privately store student responses.
  3. Create an account management system so that teachers can follow each other to see assignments they made for their students.
Share this project:

Updates