Dashboard Overview — Real-time overview of residents, entries, exits, and recent activity.
Add New Guest — Form for registering a new resident with their details and face data.
Guest List — Centralized list of all registered residents and guests.
Live Door Monitoring — Live entry and exit camera monitoring through the dashboard.
Manual Entry — Option to manually record or correct a resident's entry or exit.
Outpass Management — Interface for creating and managing resident outpasses.
Resident Profile — Detailed profile and activity history of a registered resident.
Unknown Faces — Displays faces that could not be identified by the system.
Parent Approval Email — Email interface allowing parents to approve or decline an outpass.
Generated Outpass — Digitally generated outpass containing resident and approval details.

About the Project

PG Register is a smart entry and exit management system designed to make PG management easier and more secure. It uses face recognition and computer vision to automatically identify residents and maintain their entry and exit records.

Inspiration

We were inspired by the limitations of traditional PG registers, where residents have to manually enter their details every time they enter or leave. This process can be time-consuming and may also lead to incorrect or missing records.

We wanted to create a system that could automate this process using AI while also providing useful features for the PG warden and residents.

What it does

Our system automatically recognizes registered residents using cameras and records their entry and exit.

It also provides features such as:

Resident and guest management
Automatic entry and exit tracking
Unknown face detection
Live camera monitoring
Manual entry and correction
Outpass creation and management
Parent email notifications
Parent approval for outpasses
History of resident movements

This makes the overall process faster and reduces the need for manual record keeping.

How we built it

We divided the project into three main parts: AI/Computer Vision, Backend, and Frontend.

For the AI part, we used Python, OpenCV, InsightFace, NumPy, and ONNX Runtime for face detection, tracking, and recognition.

The backend was developed using Node.js and Express.js. It handles resident records, entry/exit logs, outpasses, and communication between the AI service and dashboard.

For the frontend, we used React, Vite, and Tailwind CSS to create a simple dashboard where the warden can manage residents, view logs, monitor cameras, and handle outpasses.

Challenges we ran into

One of our main challenges was getting reliable face recognition in real-world conditions. Different lighting, camera angles, movement, and image quality could affect recognition.

We also had to make sure that an unknown person was not incorrectly identified as a registered resident. To handle this, we used similarity thresholds and face tracking.

Another challenge was determining whether a resident was entering or leaving. We solved this by supporting separate cameras for the entry and exit sides.

Accomplishments that we're proud of

We are proud that we were able to build a complete working system instead of only a face-recognition model. We successfully connected computer vision with a backend and an interactive web dashboard.

We are also proud of the outpass system, where residents can request an outpass and parents can approve or decline it through email.

What we learned

Through this project, we learned that developing a real-world AI application involves much more than training a model. We gained practical experience in computer vision, face recognition, API integration, React, backend development, testing, and system design.

We also learned the importance of handling real-world problems such as incorrect predictions, unknown users, camera failures, and privacy of biometric data.

What's next for Untitled

In the future, we want to improve the face recognition accuracy and make the system more reliable under different lighting and camera conditions.

We also plan to add a proper database, stronger authentication and security, resident activity analytics, and better notifications. Our long-term goal is to test the system in a real PG environment and make it scalable for multiple PGs.