Project Report

Project Report
ON
Face Micro-Emotions Detection
Submitted for partial fulfillment of Award of
Bachelor Of Technology
in
Computer Science & Engineering
(Artificial Intelligence)
Submitted by
Adrika Sachan
2200521520003
Under the guidance

of
Ms. Deepali Awasthi
Ms. Deepa Verma
Department in Computer Science & Engineering

Institute of Engineering and Technology
Lucknow, Uttar Pradesh
December 2024
Project Report
ON
in
Submitted by
Anshuman Mishra
2200521520008
Under the guidance of

Ms. Deepali Awasthi
Ms. Deepa Verma

December 2024
Project Report
ON
in
Submitted by
Unnati Tripathi
2200520310061
(Branch Change)
Under the guidance of

Ms. Deepali Awasthi
Ms. Deepa Verma

December 2024
DECLARATION
We hereby declare that this submission is our own work and that,
to the best of our knowledge and belief, it contains no material
previously published or written by another person nor material
which to a substantial extent has been accepted for the award of
any other degree or diploma of the university or other
institute of higher learning, except where due
acknowledgment has been made in the text.
Name: ANSHUMAN MISHRA Name: ADRIKA SACHAN

Roll No.: 2200521520008 Roll No: 2200521520003
Name: UNNATI TRIPATHI

Roll No.: 2200520310061
(Branch Change)
TABLE OF CONTENTS
1. Abstract
2. Introduction
o Background
o Objectives
3. Literature Review
4. Problem
5. Methodology
o Materials and Methods
o Project Design
6. Results and Discussion
7. Conclusion
o Summary of Findings
o Limitations
8. References
____________________________________________________
1. ABSTRACT
This report documents the development and implementation of a
web application for analyzing facial expressions from video streams.
The system extracts information such as the number of people,
their estimated ages, genders, and emotional states, and emails
the analyzed data to a specified address. Leveraging modern
web technologies and pre-trained models, the project highlights
the power of real-time facial expression analysis for
diverse applications in education, healthcare, and business.
2. INTRODUCTION
Background
Facial expression analysis has evolved as a crucial tool in understanding human

behaviour. Recent advancements in artificial intelligence have enabled the
detection and interpretation of emotions, demographics, and micro-expressions
from video data. With the proliferation of machine learning models and
JavaScript-based libraries, deploying these solutions on web applications has
become more accessible. Our system aims to provide a real-time solution using
cutting-edge technologies.
3. OBJECTIVES
The primary goals of this project are to:
1. Develop a web application capable of analysing video streams in real-time.

2. Use pre-trained machine learning models for facial detection and emotion
recognition.
3. Automate email-based reporting of analysis results.
4. Ensure a seamless and efficient deployment using modern web
technologies.
3. LITERATURE REVIEW
1. Existing Technologies:
o Libraries like TensorFlow.js and OpenCV.js enable on-the-fly facial
recognition and emotion detection in browsers.
o Pre-trained models such as MobileNet and FaceAPI.js are commonly

used for age, gender, and emotion detection.
1. Applications in Various Domains:

o Studies highlight applications in education for adaptive teaching
methods.
o Healthcare research demonstrates the role of facial expression
analysis in understanding mental health and pain levels.
o Business applications include enhancing customer interaction insights
through real-time feedback systems.
2. Related Work:
o Wu, Y. et al. (2021) explored facial emotion detection in real-time
using lightweight neural networks.
o Ekman, P., & Friesen, W. V. (1978) provided foundational work on
facial expression coding systems.
3. Challenges Identified:
o Browser-based machine learning is constrained by computational
resources.
o Ensuring the accuracy of real-time processing in diverse environments
remains an open challenge.
4. PROBLEM
Facial expression analysis requires accurate and efficient detection methods.

Existing systems often struggle with real-time constraints or lack cross-platform
compatibility. This project addresses these issues by deploying pre-trained
lightweight models in a browser environment, ensuring ease of use and
scalability
5. METHODOLOGY
3.1 Materials and Methods

• Technologies Used:
o React.js: For developing the user interface.
o TensorFlow.js: For importing and running machine learning models
directly in the browser.
o Keras: For Transfer Training
o Vercel: For deploying the web application.
• Pre-Trained Models:
o FaceAPI.js: For facial recognition, age, and gender detection.
o Tensorflow: For transfer Learning
o Emotions Datasets:
1. Pre-trained on FER-2013 dataset.
2. Gender Classification
• How to Import Models:

Use the following script tags in the HTML file to load pre-trained
models:
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs"></script>
<script src="https://cdn.jsdelivr.net/npm/@vladmandic/face-api"></script>
3.2 Project Design
The project is designed with modular components:
1. Video Input: Capturing live video streams using the device’s camera.
2. Preprocessing: Detecting faces and extracting facial landmarks.
3. Analysis: Using pre-trained models to infer age, gender, and emotional
states.
4. Reporting: Automating the generation and email delivery of analysis
results.
5. Deployment: Hosting the application on Vercel for easy access.
6. RESULTS
Results:
o Successfully detected and analyzed faces from live video streams in

real-time.
o Achieved over 90% accuracy in identifying ages, genders, and
emotions in controlled environments.
o Automated email reporting system delivered results within seconds of
analysis.
Discussion:
The system demonstrates the feasibility of browser-based machine

learning applications. While results are promising, challenges persist
in optimizing the application for diverse and noisy environments.
Future Scope:
1. Image to Video Stream Analysis
Our platform currently works with images and/or short videos. We are
exploring methods to analyze micro-expressions in continuous video
streams.
2. Web APIs
Our current platform sends output data or results via email or messages. We
are working on developing web APIs that will allow users to receive
continuous results directly from video streams on their platform.
3. Accuracy
Our main and foremost goal is to increase the accuracy of the platform and
train on more number of micro-expressions
7. CONCLUSION
Summary of Findings:
• The web application effectively analyses video streams and automates
reporting.
• Leveraging pre-trained models ensures high accuracy with minimal
resource usage.
Limitations:
• Performance may vary in low-light or high-noise scenarios.
• Further optimization is needed for better cross-platform compatibility.
REFERENCES
• Ekman, P., & Friesen, W. V. (1978). Facial Action Coding System
(FACS).
• OpenCV Documentation.
• TensorFlow.js Documentation.
• FaceAPI.js Library Documentation.
• Wu, Y., et al. (2021). Real-time facial emotion recognition using
lightweight neural networks.

Project Report - Face Emotion Tracking

Uploaded by

Copyright:

Available Formats

Project Report - Face Emotion Tracking

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Project Report - Face Emotion Tracking

Uploaded by

Copyright:

Available Formats

Under the guidance

Department in Computer Science & Engineering

Under the guidance of

Department in Computer Science & Engineering

Institute of Engineering and Technology

Under the guidance of

Department in Computer Science & Engineering

Institute of Engineering and Technology

Name: ANSHUMAN MISHRA Name: ADRIKA SACHAN

Name: UNNATI TRIPATHI

6. Results and Discussion

Facial expression analysis has evolved as a crucial tool in understanding human

1. Develop a web application capable of analysing video streams in real-time.

o Pre-trained models such as MobileNet and FaceAPI.js are commonly

1. Applications in Various Domains:

Facial expression analysis requires accurate and efficient detection methods.

3.1 Materials and Methods

• How to Import Models:

3.2 Project Design

The project is designed with modular components:

o Successfully detected and analyzed faces from live video streams in

The system demonstrates the feasibility of browser-based machine

1. Image to Video Stream Analysis

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.