Inspiration
With cyber threats on the rise, we wanted to create a fast, secure, and user-friendly way to verify identity. Traditional methods are often slow or insecure, so we imagined a solution that leverages everyday devices, like webcams, for secure and human authentication.
What it does
DigiSign captures a unique digital signature by tracking the user's finger through their webcam. It allows users and organizations to:
- Verify identity instantly
- Reduce reliance on passwords
- Ensure tamper-proof signing
- Prevents brute force methods
The user selects either DigiShield or OTP as their verification method. If they are new, they’re prompted to register, and their data is securely stored in the backend. When they return to make a purchase, they submit their signature or OTP code, which the backend verifies against stored data and then labels the purchase as approved or rejected, sending the result back to the user.
How we built it
- Frontend: React for website, webcam capture, and responsive UI
- Backend: Node.js and Google Gemini APIs for signature processing and encrypted storage
Challenges we ran into
- Implementing correct hand tracking and drawing
- Meshing detection program with React website
- Ensuring proper detection and comparison despite natural variation
Accomplishments that we're proud of
- Capturing accurate finger position and tracking
- High verification accuracy while tolerating natural variation
- Secure, scalable backend with package installation
- Implementation with Stripe API for successful wallet transactions
What we learned
- How to implement the API in our program to help create our program
- User experience is as critical as technical security
- How to connect Stripe to our program to make it work in the real world
- How to use Google Gemini for object detection and accuracy checks
What's next for DigiSign
- Mobile platform support
- Integration with enterprise identity systems
- Exploring additional biometrics like gestures or keystrokes
- Conducting user studies to improve usability and accuracy
Log in or sign up for Devpost to join the conversation.