Inspiration
It all started when I was experimenting with computer vision and realized how fascinating it is to make computers “see.” One day, while I was stuck trying to format math properly in my code, I thought:
“What if I could just write equations in the air instead of typing them?”
That idea led to AirMath. The concept felt like something out of Harry Potter — waving your fingers and watching equations appear and solve themselves.
How I Built It
I used Python with mediapipe and opencv to track the fingertip position in real time.
The fingertip coordinates were used to draw numbers and operators on a virtual canvas.
I connected this backend to a Flask server, allowing it to send the recognized data to a frontend built with HTML, CSS, and JavaScript. Then it uses Gemini's API to recognize the text and solve it .
What I Learned
How computer vision can detect and process hand gestures.
How to connect a Python backend with a web-based frontend for real-time updates.
How to use MathJax to render LaTeX math beautifully in HTML.
The importance of optimizing UI updates to remove flickering and ensure smooth transitions.
Challenges I Faced
The finger tracking initially flickered a lot, causing unstable drawings. I tried Computer Vision for the first time so I had to use AI's to learn new concepts as I implemented them
Keeping the frontend display smooth and responsive while fetching new results every second was difficult, but adjusting CSS transitions and update timing solved it.
Reflection
Building AirMath showed me how creative ideas can merge with technical skills to make something truly interactive. It’s more than a math solver — it’s a step toward blending imagination, design, and computer vision into real-world experiences.
Log in or sign up for Devpost to join the conversation.