-
-
when testing zone
-
when testing zone
-
training Dataset listout in pdf
Upasthiti AI – Smart Attendance System
Inspiration
We were inspired by the everyday challenges of managing attendance in schools and workplaces.
Manual systems are:
- Time-consuming
- Error-prone
- Not real-time
We realized that many classrooms already have CCTV or smart cameras. Why not use them smartly?
So we built an AI solution that automates attendance through face recognition, making existing camera infrastructure smarter and more useful.
What it does
Upasthiti AI is a facial recognition-based smart attendance system.
Features:
- Passive attendance tracking using classroom CCTV cameras
- Live face detection and recognition
- Admin dashboard for viewing attendance
- Real-time insights and analytics
- Privacy-conscious data processing
- Beta: Windows software for native classroom camera integration
We have not deployed it yet because it is a big system and currently under testing.
How we built it
- Frontend: React + Next.js
- Backend: Node.js + Express.js
- Face Recognition: InsightFace
- Database: MongoDB
- Windows Integration: Native module to use existing CCTV
- Deployment: Currently published on GitHub and YouTube for demonstration, but not deployed live due to ongoing testing
- AI Model Training: Trained using real student image data
Challenges we ran into
- Lighting and camera angle variation across classrooms
- Real-time CCTV feed integration on Windows
- Managing false positives in large datasets
- Making it work without GPU
- Ensuring privacy and secure data handling
Accomplishments that we're proud of
- Developed and deployed a real-time facial recognition attendance system
- Achieved over 95% accuracy
- Built a responsive web dashboard for institutions
- Created a beta version of our Windows camera software
- Designed a clean UI that runs on both desktop and mobile
- Ensured data is stored securely and exportable
What we learned
- How to implement facial recognition in real-world web apps
- Building cross-platform tools (web + Windows native)
- Importance of quality datasets and preprocessing
- Designing for speed, privacy, and usability
- Real-time logging and feedback are critical for stability
- Rapid development under time constraints
What's next for Upasthiti AI
- Official release of Windows-compatible camera module
- Large-scale rollout to institutions with existing CCTV
- Add voice and gesture-based recognition
- Build mobile and tablet support
- ERP system integration
- Create offline-first version for rural deployment
Link
Visit Upasthiti AI Introduction video
Visit Upasthiti AI website Github Visit Upasthiti AI website video
