AdaptAdept

Logo
Login
Signup
Dashboard
Learn Biology
Learn Chemistry
Quiz with Neural Network Analysis
Upload Files

Inspiration

One of our platform's key features, using hand-tracking via webcam to pinch and move objects on-screen, was inspired by a smart glasses demonstration.

The idea of using API calls to generate mock questions from self-uploaded resources was a result of taking modules where recent past year papers were not made available to students. Coming up with practice questions that could potentially show up on the exam became an area of interest.

What it does

AdaptAdept offers plug-and-play components for diverse learner groups, analyses students' content mastery using neural networks, and supports real-time revision content generation from uploaded files.

How we built it

We used a Streamlit-based browser client for the frontend, Python for the backend, and sqlite3 for the database. We developed the neural network using Tensorflow, with pandas used for data preparation and preprocessing.

Challenges we ran into

One of the biggest challenges was getting our example lessons to work within a Python-based framework. While Streamlit provided basic HTML support via its components module, it was not feasible to design a complex webpage with many interactive objects in the current setup. We eventually settled for smaller examples that still effectively conveyed our ideas.

Accomplishments that we're proud of

Implementing a functional neural network and webcam-based hand tracking from scratch was a surprising and rewarding achievement for us.

What we learned

Balancing accessibility and security was a recurring dilemma for us. While avoiding accounts or personal data minimises security concerns, educators may prefer to limit access to their materials or restrict exposure to unrelated content. A possible solution could involve temporary credentials or private keys that grant access without storing sensitive personal information.

What's next for AdaptAdept

We aim to further improve support for auditory learners and students who benefit more from slower-paced lessons, not just by easing question difficulty, but by tailoring delivery style.

We could also look further into prompt engineering, to generate not just revision questions, but also related content and in-depth summaries from student-uploaded notes.