Delhi Technological University
(Formerly Delhi College of Engineering)
Shahbad Daulatpur, Bawana Road, Delhi-110042
Department of Software Engineering
Academic Year 2025-26
[Link] Project-I Synopsis, AY 2025-2026
Project Progress Report : SecureAuth Face
Title of Work : Design and Implementation of a Real-Time Biometric Access Control
System using Facial Recognition
Type of Work
Research and Development Project
This project focuses on engineering a secure, real-time access control system by leveraging
computer vision and machine learning. The primary goal is to replace traditional
authentication methods (keys, cards) with a robust facial recognition framework, thereby
enhancing security and convenience.
The project involves designing a complete hardware and software solution, from capturing
facial data to processing it through a trained deep learning model and actuating a physical
lock mechanism. The final output will be a functional prototype capable of identifying
authorized individuals and granting access automatically.
Approach
The project follows an end-to-end implementation pipeline, beginning with system design and
data collection, followed by model development, hardware integration, and real-world testing.
The core emphasis is on achieving high accuracy and low latency for a seamless user experience
while ensuring system reliability.
Stepwise Description of Work
1. Literature Review and System Design (Completed) Researched various facial detection
and recognition algorithms (e.g., Haar Cascades, HOG, MTCNN, FaceNet). The system
architecture has been finalized, specifying a Raspberry Pi as the core controller, a Pi Camera
for video input, and a solenoid lock for the physical access mechanism. The software stack
will primarily use Python with OpenCV and Dlib libraries.
2. Dataset Creation and Preprocessing (Ongoing) This phase involves building a custom
dataset by capturing facial images of authorized users. The data collection script is designed
to capture images under varied lighting, angles, and expressions to ensure model robustness.
Preprocessing will include face detection, cropping, conversion to grayscale, and histogram
equalization to normalize the images for training.
3. Facial Recognition Model Development (Planned) An initial baseline model will be
developed using the Local Binary Patterns Histograms (LBPH) algorithm available in
OpenCV. Subsequently, a more advanced model will be implemented by fine-tuning a pre-
trained Convolutional Neural Network (CNN), such as FaceNet or VGGFace, on our
custom dataset to learn unique facial embeddings for high-accuracy recognition.
4. Hardware Integration and Lock Control (Planned) The Raspberry Pi's GPIO pins will
be interfaced with a relay module to control the 12V solenoid door lock. A script will be
developed to trigger the relay, unlocking the door for a set duration upon successful
recognition of an authorized face by the software module.
5. System Integration and Real-Time Testing (Planned) All software and hardware
components will be integrated into a single, cohesive system. The final prototype will be
tested rigorously for performance metrics, including recognition accuracy, processing
latency (speed), and reliability under different operational scenarios.
Progress So Far (Around 30% Complete) - The project is currently in the data
acquisition and preliminary preprocessing stage. Approximately one-third of the planned
work has been completed, with data successfully collected and stored for multiple
cryptocurrencies over several weeks.
Work Completed So Far: (Around 35% Complete)
The project is currently in the dataset creation and preprocessing stage. The foundational work,
including system design and component selection, has been completed, and the initial data for
training the recognition model is being actively collected.
1. System Design and Component Sourcing: The overall system architecture has been
designed. All necessary hardware components, including a Raspberry Pi 4, Pi Camera
Module, 12V solenoid lock, and relay module, have been procured and tested
individually.
2. Environment Setup: The development environment on the Raspberry Pi has been fully
configured. Python, OpenCV, Dlib, and other required libraries have been installed and
verified. Version control is being managed using GitHub.
3. Data Acquisition Module: A functional Python script using OpenCV has been created to
detect faces from the camera feed, capture multiple images per user, and store them in a
structured directory for dataset creation.
The next stage involves completing the dataset preprocessing, training the baseline LBPH
recognition model, and starting the hardware integration by connecting the solenoid lock to the
Raspberry Pi. The project is progressing as per the defined [Link] successfully executed.
References
• Viola, P., & Jones, M. (2001). Rapid Object Detection using a Boosted Cascade of Simple
Features. CVPR.
• Dalal, N., & Triggs, B. (2005). Histograms of Oriented Gradients for Human Detection.
CVPR.
• Schroff, F., Kalenichenko, D., & Philbin, J. (2015). FaceNet: A Unified Embedding for
Face Recognition and Clustering. CVPR.
• Bradski, G. (2000). The OpenCV Library. Dr. Dobb's Journal of Software Tools.
SNo Group member name Roll No Signature
1 Piyush Yadav 2K22/SE/125
2 Satyapal 2K22/SE/158
Supervisor: Mr. Sanjay Patidar Supervisor Signature: