Scribd
Upload
23 views8 pages

Proposal 1

Shorya Dixit is a B.Tech student in Electrical Engineering at IIT Jammu, with a focus on developing a Smart Facial Attendance System for teachers in government schools. The project involves researching open-source facial recognition models, creating a user registration pipeline, and addressing challenges such as accuracy, data security, and user acceptance. Shorya has prior experience with facial recognition systems and machine learning, and is motivated to apply these skills in a practical setting.

Uploaded by

21bch040
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views8 pages

Proposal 1

Shorya Dixit is a B.Tech student in Electrical Engineering at IIT Jammu, with a focus on developing a Smart Facial Attendance System for teachers in government schools. The project involves researching open-source facial recognition models, creating a user registration pipeline, and addressing challenges such as accuracy, data security, and user acceptance. Shorya has prior experience with facial recognition systems and machine learning, and is motivated to apply these skills in a practical setting.

Uploaded by

21bch040
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Name Shorya Dixit

Email ID [email protected]

Phone Number 7838076433

GitHub ID Shorya-Dixit (github.com)

Discord ID 1098949519015673856

Current occupation Student

IIT Jammu
Education Details B.Tech
Electrical Engineering

Python, C++, HTML, Javascript, CSS, Tensorflow, Keras,


Flask, PyTorch, OpenCV, Scikit-learn, NumPy, Pandas,
Google Firebase: - Intermediate
Technical skills with level
NLTK, ReactJS, Docker, MatPlotlib, MySQL, PL/SQL,
VHDL:- Novice
Title:- Smart Facial Attendance System

Summary:-
● Research on Open-Source Models: I will leverage my experience with dlib
for facial recognition and explore other open-source libraries like DeepFace and
InsightFace, that work effectively in both online and offline settings.

● Benchmarking and Dataset Preparation: To ensure an objective


evaluation, I will prepare a benchmark dataset of images under various
conditions (lighting, poses) to ensure model robustness. This dataset will be used
to assess the performance of shortlisted models based on predefined criteria,
including recognition accuracy, processing speed, and memory requirements.

● User Registration Pipeline: A user registration pipeline will be developed to


capture and store facial data (images) and facial landmarks (key points on the
face) of teachers. This pipeline will ensure secure and efficient data collection.
Facial images and extracted landmarks will be securely stored in a database with
access controls. Encryption can be employed for additional security.

● Face Lookup Algorithm: An efficient face lookup algorithm will be


implemented to compare live captured faces with the stored data. This algorithm
will output confidence scores indicating the likelihood of a match.

● Demo Application: A user-friendly demo application will be built to showcase


the system's functionalities. This application will facilitate both user registration
and user lookup, with geo-tagging.

Project Details:

1. Project Overview:

Understanding of the project:


This project aims to develop a facial recognition attendance system for teachers in
government schools. It will offer the following functionalities:

● User Registration: Teachers will be able to register themselves by capturing


their facial image for recognition.
● Facial Recognition Attendance Marking: The system will utilize Face
recognition models from libraries like DeepFace or dlib to recognize teachers'
faces and mark their attendance.
● Geo-Tagging: As a potential security measure, the system will capture the user's
location data during attendance marking to verify their physical presence at the
school (requires additional user consent and privacy considerations).
● Offline Functionality: The system must be able to operate in both online and
offline environments, increasing flexibility for various school settings.

Possible Problems:

1. Facial Recognition Accuracy:

● Pre-trained models might not achieve perfect accuracy, especially under varying
lighting conditions, poses, or facial occlusions (e.g., glasses).
● Teachers with similar facial features could lead to misidentification.

2. Data Security and Privacy:

● Storing sensitive facial data requires robust security measures to prevent


unauthorized access or breaches.
● User consent and data privacy regulations need to be addressed regarding data
collection, storage, and usage.

3. System Performance:

● Facial recognition algorithms can be computationally expensive, potentially


impacting performance on resource-constrained devices.
● Network connectivity issues might hinder system functionality in offline
environments (if implemented).

4. User Acceptance:

● Teachers might be hesitant to use a facial recognition system due to privacy


concerns or lack of technical familiarity.
Solutions:

1. Model Selection and Training:

● Evaluate different pre-trained models to find the one offering the best balance
between accuracy and resource efficiency.
● Consider fine-tuning a pre-trained model on a dataset of teachers' faces from
your specific school environment to improve recognition accuracy.
● Implement techniques like facial landmark detection to enhance recognition
robustness.

2. Data Security and Privacy:

● Utilize client-side encryption to encrypt facial data before storing it on the server.
● Implement access controls to restrict unauthorized access to data.
● Obtain informed consent from teachers regarding data collection and usage.
● Adhere to relevant data privacy regulations.

3. System Optimization:

● Choose libraries and frameworks optimized for performance on the target


devices.
● Consider implementing lightweight versions of facial recognition models if
necessary.
● Explore offline processing techniques for attendance marking if internet
connectivity is unreliable.

4. User Training and Communication:

● Clearly communicate the benefits and security measures of the system to


teachers.
● Offer training sessions to familiarize teachers with the registration and
attendance marking process.

2. Implementation Details with timelines:

Milestone 1: Investigation (Required - 2 weeks)

● Research on open-source facial recognition models.


● Define evaluation criteria (accuracy, speed, memory footprint).
Deliverables:

● Project report outlining model research and evaluation plan.

Milestone 2: Benchmarking and Dataset Preparation (Required - 3 weeks)

● Prepare a benchmark dataset (if chosen) or explore publicly available options.


● Implement data preprocessing techniques (cleaning, alignment).

Deliverables:

● Prepared benchmark dataset or justification for utilizing a public dataset.


● Code for data preprocessing.

Milestone 3: User Registration Pipeline (Required - 2 weeks)

● Design and implement the user registration process.


● Establish secure storage for facial data and landmarks.

Deliverables:

● Functional user registration pipeline with secure data storage.


● Code for facial landmark detection and data storage.

Milestone 4: Face Lookup Algorithm (Required - 2 weeks)

● Choose the optimal facial recognition model based on benchmarking results.


● Implement the face lookup algorithm with confidence score output.

Deliverables:

● Integrated face lookup algorithm using the chosen model.


● Code for face recognition and confidence score generation.

Milestone 5: Demo Application (Required - 2 weeks)

● Develop a user-friendly application for registration and user lookup


functionalities.
● Implement optional geo-tagging feature (requires additional planning).

Deliverables:

● Functional demo application showcasing registration and attendance marking.


● Code for the demo application (including optional geo-tagging if implemented).
Milestone 6: Documentation (Required - 1 week)

● Prepare comprehensive project documentation.


● Include user manuals, code comments, and deployment instructions.

Deliverables:

● Project documentation for future reference and potential deployment.

Availability:-

Number of hours available to dedicate to this project per 25 hours


week

Do you have any other engagements during this period ?


Yes
(projects/internships)

I am starting a remote internship at DRDO from 21st May,2024. (For 6 weeks)

Personal Information:-

About Me: 3rd year student at IIT Jammu, currently enthralled by the potential of
Machine Learning. I'm a perpetual learner, always seeking to expand my knowledge
and tackle novel problems.

What is your motivation to apply for this project?


Developing a facial recognition attendance system previously ignited my passion for
applying Machine Learning's power to real-world challenges, particularly in facial
recognition. This project presents a captivating opportunity to delve deeper into the
intricacies of these algorithms and tailor them for optimal performance in a practical
setting.
Previous experience/open source projects:-

Project Name Project Description Links (if any)

Tracks student attendance using facial Shorya-Dixit/Face_


recognition. Integrates Firebase for Attendance_Syste
Face Attendance System real-time updates and data storage. Offers m (github.com)
instant verification and detailed student
profiles.

Developed sentiment analysis model. Shorya-Dixit/Senti


Deployed model as web app for real-time ment-Analysis
Sentiment Analysis
sentiment analysis. (github.com)

Contribution in C4GT’s open community:-

Have you contributed to tickets in C4GT’s open No


community? (Mandatory to answer)

Have you successfully completed C4GT’s GitHub No


Classroom Assignment? (Mandatory to answer)

Enter your DPG points (Mandatory to answer, 0


Enter 0 if not applicable)

Screenshot of leaderboard with your GitHub ID 0


(Mandatory to answer, enter 0 if not applicable)

You might also like