📸 Biometric Attendance Engine: Real-Time Identity Verification

🏆 Award-Winning Innovation: Selected as a Finalist among 100+ competing teams (Team Mavericks)

📖 The "Problem-to-Solution" Narrative

Manual attendance tracking is a high-friction administrative task prone to data latency and "proxy" errors. As a core developer for Team Mavericks, I collaborated on building an AI-powered ecosystem designed to make identity verification seamless and totally automated.

Our mission was to move facial recognition out of a static script and into a functional, secure web application capable of handling diverse input streams in real-time.

---

🏗️ System Architecture: Three-Mode Versatility

To ensure the system was practical for real-world institutional environments, we engineered 3 specialized processing modes:

Live Feed: Real-time identification via active webcam streams for immediate classroom logging.

Static Image: Batch-processing of photographs for post-event verification and archival.

Video Playback: Asynchronous analysis of recorded footage to register attendance from pre-captured video files.

---

🛠️ The Technical Stack

Core Developer & Backend Integration: Shreya Malogi

I contributed to the full-cycle development of this project, with primary ownership of the Backend Architecture and Data Integration Layer:

AI & Computer Vision

• Methodology: Utilized HOG (Histogram of Oriented Gradients) for robust facial feature encoding, ensuring the model identifies structural shapes rather than simple pixel intensity.
• Libraries: Implemented OpenCV and Dlib for the 68-point landmark detection required for biometric precision.
• Identity Logic: Integrated face_recognition libraries to map detected landmarks against a trained encoding database.

Backend & Systems Engineering

• Framework: Built the core application logic using Flask to manage multi-part real-time video streaming.
• Database (ORM): Engineered a relational schema using SQLAlchemy to manage Users, Students, and Semesters.
• Persistence: Leveraged SQLite for structured data storage of all attendance logs and user credentials.
• Security: Integrated Werkzeug for secure password hashing and Flask-Login for protected administrative sessions.

Data & UI Layer

• Reporting: Utilized Pandas to automate the transformation of raw biometric signals into structured CSV attendance reports.
• Frontend: Designed a responsive interface using HTML, CSS, Bootstrap, and Jinja2 templating for dynamic data rendering.

---

🤯 Engineering Challenges & Optimization

• Latency Mitigation: We encountered significant processing lag when integrating external IP cameras. We optimized the frame-buffer cycle to maintain real-time performance.

State Management: Orchestrating the logic to switch between live, static, and video inputs without memory leaks required a modular system architecture.

---

🚀 Impact & Recognition

By focusing on a user-centric design—creating an interface for both Admins and Teachers—our project was selected as a Finalist among 100+ teams. This project proves my ability to collaborate in a high-pressure environment to deliver a secure, scalable AI product.

---

📥 Setup & Technical Deployment

1. Clone & Install: pip install -r requirements.txt
2. Database: The system initializes DataBase.db on launch using the SQLAlchemy models.
3. Run: Execute python main.py and access the dashboard at localhost:5000.