DoodlePhonics

Inspiration

We were inspired by how kids naturally learn through interaction and play. We wanted to build a fun, intuitive way for children to connect with the alphabet through drawing and hearing, helping reinforce early literacy skills in an engaging way. With AI becoming more accessible, we thought why not bring it into early education in a meaningful way!

What it does

Our app is built for children starting to learn the English alphabet. When the child writes a letter on the screen, our AI model analyzes the handwriting and pronounces the letter out loud, helping the child associate sounds with shapes and improve recognition through feedback.

How we built it

We used Python for the backend and used custom dataset of hand-drawn letters to train our model. In just two days, we created a working prototype that connects the user’s input to a trained neural network and outputs the recognized letter and its pronunciation.

Challenges we ran into

One challenge was collecting and cleaning enough hand-drawn letter samples to train our model effectively children’s writing can be unpredictable! Another was getting consistent predictions across a variety of drawing styles. Integrating real-time drawing recognition with a smooth, responsive UI and audio playback was also tricky under a short timeframe.

Accomplishments that we're proud of

We’re proud that our model can accurately recognize most handwritten letters, even from messy or uniquely drawn inputs. We also managed to build a user-friendly interface that kids could start using immediately without any instructions. It felt great to watch it actually teach something.

What we learned

We learned a lot about how handwriting recognition works, especially the preprocessing steps needed to make sense of varied input data. We also gained experience working with machine learning models, integrating them into mobile apps, and thinking about usability from a child’s perspective.

What's next for DoodlePhonics

We want to expand the app so it can also recognize letters from other languages like Japanese, Russian, and Hindi — helping kids all over the world learn in their native scripts. Beyond letters, we plan to teach kids basic object recognition things like cars, houses, animals, etc.

PLEASE NOTE I am sorry. I couldn't get hold of my teammate in the last coupe hours. I don't know what happened. She was responsible for the code. I am sure she was done, with them, but for unknown reasons, she was not able to share them to me. As my teammate wasn't able to share her code on time, we are unable to provide you with a "try it out" link

Built With

• python