Detection of Stress in IT Employees using Machine Learning Technique | Python Final Year Project

Summary

This video introduces a Python project for stress detection in IT employees using machine learning. The project, based on an IEEE 2022 conference paper, deviates from the original KNN classification approach by implementing a CNN model architecture. The model boasts a training accuracy of 87.34% and a validation accuracy of 98.45%. It includes a MySQL database, user registration and approval processes, live camera image capture, and an admin panel for stress assessment. The video demonstrates the system's functionality and highlights the performance analysis with metrics like accuracy, precision, recall, and F-measure, along with a dynamic confusion matrix.

Takeaways

• 📘 The video introduces a Python project for stress detection in IT employees using machine learning techniques, based on an IEEE 2022 conference paper.
• 🔍 The original paper proposes using KNN classification for stress detection, but the project in the video implements a CNN model architecture instead.
• 🎯 The CNN model achieved a training accuracy of 87.34% and a validation accuracy of 98.45%, indicating high performance on the validation set.
• 🛠️ Before executing the project, viewers are advised to install necessary libraries and ensure the correct version of Python is used, as well as attach a database using MySQL administrator.
• 📸 The project involves capturing images from a webcam for testing facial expressions, and it's important to adjust image sizes for optimal recognition.
• 🌐 The project includes a web interface accessible via a URL, featuring a homepage, registration, login, and admin functionalities.
• 🔑 There is a registration process for new users, which includes validation criteria for the password to ensure security.
• 👤 The admin has the ability to approve new user registrations, which is a crucial step before users can log in and use the system.
• 🖼️ Users can upload images to the system, and the model predicts the emotional state of the individual in the image, such as happy, angry, or neutral.
• 📊 The project provides performance analysis with metrics like accuracy, precision, recall, F-measure, and a confusion matrix to evaluate the model's performance.
• 📈 A dynamic chart is included to visually represent the classification results, updating in real-time as new images are uploaded and classified.

Q & A

What is the main topic of the video?

The main topic of the video is a Python project for stress detection in IT employees using machine learning techniques.

What machine learning technique was originally proposed in the paper discussed in the video?

The original paper proposed using the KNN (K-Nearest Neighbors) classification technique for stress detection.

What model architecture does the video suggest implementing instead of KNN?

The video suggests implementing a CNN (Convolutional Neural Network) model architecture instead of KNN.

What are the reported training and validation accuracies of the CNN model mentioned in the video?

The reported training accuracy of the CNN model is 87.34%, and the validation accuracy is 98.45%.

What is the first technical requirement mentioned in the video for executing the project?

The first technical requirement mentioned is to install the necessary libraries on the required version of Python as specified in the requirements file.

How does the project handle the database attachment?

The project handles database attachment by using MySQL administrator, going to the restore option, and restoring from a specified database file in the project's database folder.

What is the process for an employee to register in the system as described in the video?

The employee must fill out a registration form with details such as username, password, email, mobile number, login ID, address, company, and state. The registration must also satisfy password validation criteria.

What validation is in place for the password during registration?

The password must contain at least one number, one lowercase letter, one uppercase letter, one special symbol, and be between 6 to 10 characters long.

Why might a newly registered employee be unable to log in immediately after registration?

A newly registered employee might be unable to log in immediately because their account must first be approved by an admin.

What are the two main parts or entities of the project mentioned in the video?

The two main parts or entities of the project are the employee part, where employees upload their images, and the admin part, where the stress of the employees is identified.

How does the project categorize the different expressions for stress detection?

The project categorizes expressions into classes such as angry, discourse, fearful, happy, neutral, sad, and surprised.

What performance analysis parameters are mentioned in the video?

The performance analysis parameters mentioned are accuracy, precision, recall, F-measure, and a confusion matrix.

How does the video demonstrate the dynamic nature of the project's performance analysis?

The video demonstrates the dynamic nature by showing how uploading a new image and getting a new classification result updates the chart in real-time.

Outlines

00:00

📚 Introduction to the Stress Detection Project

This paragraph introduces a Python project focused on detecting stress in IT employees using machine learning techniques. The project is based on an IEEE 2022 conference paper and proposes a Convolutional Neural Network (CNN) model instead of the K-Nearest Neighbors (KNN) classification mentioned in the original paper. The model boasts a training accuracy of 87.34% and a validation accuracy of 98.45%. The video script guides viewers on setting up the project environment, including installing necessary libraries and attaching a database using MySQL administrator. It also instructs on how to take sample images using a web camera and compressing them for better performance.

05:02

🔑 Registration and Login Process for Employees and Admins

The second paragraph explains the registration and login process for both employees and admins within the project. It details the registration fields required for new employees and the validation rules for the password, emphasizing the need for a strong password. After registration, new employees must be approved by an admin before they can log in. The admin panel includes features to approve user registrations and view registered employee details. The paragraph also describes the employee's ability to upload images for stress detection after successful login and authentication.

10:04

🖼️ Image Upload and Stress Detection Results

This paragraph demonstrates the process of uploading images by employees and the subsequent stress detection analysis performed by the admin. It explains how employees upload images, which are then analyzed by the admin to determine the emotional state of the employee, such as happy, angry, or neutral. The script includes a walkthrough of uploading different images and observing the predicted results in the admin panel. It also mentions the potential inaccuracies of the model and the importance of not expecting 100% accuracy.

15:07

📊 Performance Analysis and Dynamic Chart Update

The final paragraph discusses the performance analysis of the stress detection model, including metrics such as accuracy, precision, recall, F-measure, and a confusion matrix. It also introduces a dynamic chart that updates in real-time according to the classes predicted by the model. The script shows how uploading different images affects the chart's representation of the predicted emotional states. The paragraph concludes with a demonstration of logging out and a summary of the project's purpose, which is to detect stress in IT employees using machine learning techniques.

Keywords

💡Machine Learning

Machine learning is a subset of artificial intelligence that enables systems to learn from and make decisions based on data. In the context of this video, machine learning is utilized to detect stress in IT employees by analyzing their expressions through a model. The script mentions the use of a CNN model architecture for this purpose, which is a type of deep learning algorithm known for its effectiveness in image recognition tasks.

💡CNN Model Architecture

CNN, or Convolutional Neural Network, is a deep learning model widely used in computer vision tasks. It is designed to process data with a grid-like topology, such as images. The script refers to a CNN model being implemented instead of the KNN classification mentioned in the base paper, indicating a shift towards a more sophisticated approach for stress detection in IT employees based on facial expressions captured by cameras.

💡Stress Detection

Stress detection in this video pertains to the identification of stress levels in individuals, specifically IT employees, through observable cues such as facial expressions. The script describes a project that aims to automate this process using machine learning techniques, which is significant for understanding employee well-being and potentially improving workplace conditions.

💡KNN Classification

KNN, or K-Nearest Neighbors, is a simple algorithm used for both classification and regression. It is mentioned in the script as the method used in the base paper for stress detection. However, the video's project opts for a CNN model instead, suggesting an attempt to enhance accuracy and effectiveness in detecting stress through facial expressions.

💡Training Accuracy

Training accuracy refers to the performance of a machine learning model when it is trained on a dataset. The script states that the model has a training accuracy of 87.34%, which indicates how well the model is learning from the training data to detect stress in IT employees.

💡Validation Accuracy

Validation accuracy is a measure of a model's performance on a separate dataset that was not used during the training phase. With a validation accuracy of 98.45% mentioned in the script, it demonstrates the model's ability to generalize and accurately detect stress when presented with new, unseen data.

💡MySQL Administrator

MySQL Administrator is a tool used for managing MySQL databases. In the context of the video, it is used to attach a database for the project, which involves restoring a database from a backup file. This step is crucial for setting up the project's database requirements before execution.

💡Web Camera

A web camera is a digital camera that captures images or videos and is connected to a computer or laptop, often used for video conferencing or in this case, for capturing facial expressions of IT employees. The script describes the process of taking sample images using a web camera for testing the model's stress detection capabilities.

💡Image Compression

Image compression is the process of reducing the size of an image file, often to improve loading times or to fit storage constraints. The script mentions compressing image sizes to ensure that they are not too large, which could affect the model's ability to accurately identify facial expressions.

💡Registration and Login

Registration and login are processes that allow users to create an account and access a system, respectively. The script describes a user registration process with validation for password strength and an approval step by an admin before the user can log in, which is part of the project's system for managing users.

💡Admin Approval

Admin approval is a step in the user registration process where an administrator reviews and approves new user accounts. In the script, it is mentioned that a new employee must be approved by an admin before they can log into the system, ensuring a controlled and secure user management process.

💡Confusion Matrix

A confusion matrix is a table used in machine learning to visualize the performance of a classification model. It shows the true positives, false positives, true negatives, and false negatives for each class. The script refers to a confusion matrix as part of the performance analysis, indicating the model's accuracy in classifying different emotional states.

Highlights

Introduction of a Python project for stress detection in IT employees using machine learning techniques.

The paper presented at an IEEE 2022 conference proposes a novel concept for stress detection.

The authors used KNN classification for stress detection, but the project will implement a CNN model architecture.

The model achieved a training accuracy of 87.34% and a validation accuracy of 98.45%.

Instructions on installing required libraries and setting up the database for the project.

Guide on attaching the database using MySQL administrator and restoring from a backup file.

The necessity of capturing sample images with a webcam for testing the model.

Demonstration of resizing and compressing images for optimal model performance.

Execution of the project using the 'stress.py' script and accessing it through a web browser.

Overview of the project's user interface, including registration, login, and admin functionalities.

Registration process validation, including password strength requirements.

Explanation of the approval process for new employee registrations by the admin.

Demonstration of the live camera feature for real-time image capture.

Process for uploading images and receiving stress detection results.

Admin panel functionality to review and approve employee registrations and view stress detection results.

Discussion of the model's accuracy and the potential for incorrect predictions.

Performance analysis section covering accuracy, precision, recall, F-measure, and confusion matrix.

Interactive chart that dynamically updates based on the predicted stress classes.

Final summary and acknowledgment of the project's demonstration and its contributions to stress detection in IT employees.

Transcripts

welcome viewers if still you're not

subscribed to this channel kindly

subscribe and also click the Bell icon

to get technology updates regularly on

this JP info Channel

hi in this video we are going to see

about a python project which is n title

as

detection of stress in it employees

using machine learning technique which

is an IEEE 2022 conference paper

so in this paper the authors have

proposed a concept of detecting the

stress of an I.T employee so

here the authors have used the KNN

classification for finding the stress

detection of an I.T employee instead

we are going to

Implement our model as CNN model

architecture so our proposed algorithm

or model is CNN model architecture

instead of the k n n which is mentioned

in the base paper so our training

accuracy of this model is

87.34 percentage and the validation

accuracy is of

98.45 percentage

so in order to see the execution of the

project before executing the project

make sure that your satisfy all the

requirements that is mentioned in the

requirements file by installing the

libraries on the required version of the

Python first thing is like you wanted to

attach the database so just go to mySQL

administrator for attaching the database

as this project consists of database

so just go to mySQL administrator and

then go to the restore option just click

the open backup file option now go to

the project source code location

and go to the database folder and select

the database and then click Start

restore

now the ray the restore operation was

finished successful message is being

displayed means the database has been

attached successfully in the MySQL now

you can close this MySQL part

the first thing is like copy the source

code location

go to the command prompt

now go to the source code location

so now we have go to the source code

location now before executing the

project make sure that you have taken

some sample of your images from your

camera so let me go to my camera so

first things like you need to connect

your web camera in case if you are

executing this project in your in your

PC in case if it does laptop make sure

that you have the web camera and you can

open this app so this is this is not the

project this is generally that is

available in your camera or PC that is

your laptop or PC the first thing is

like take some sample images for testing

the expression part

so now I have taken three pictures from

my web camera

so here you can see those pictures that

I have taken right now

if the image is very bigger size kindly

make sure that

to reduce the size of the image

so if it is very big

there may be some problem in

identifying the expression so just let

me compress the picture size

so let me make it somewhat smaller

and save it

same let me do it for other two images

also

so now let me execute the project so now

you can type python

stress.py after going to the source code

location and click enter

and after some time you can see this URL

just copy this URL and go to the browser

any of the browser I'm going to Firefox

and just paste this URL in the browser

part and then click enter and now you

can see the home screen and it comes to

the project the project title direction

of stress the 90 employees using machine

learning technique so this is the

home page of the project so there you

can see register login and admin part so

here we have two entities that is the

employee part as admin part so employee

part is the part where the employee will

be uploading their images and admin part

is the part who identifies the stress of

the employee so first thing is like

there should be a valid employee for

that purpose so whenever a new employee

has been registered it is it will be

approved only by that mean after that

only they can able to login into their

system so let me show you that execution

part the first thing is let's go to the

registration path so first new user

should register their details so here

you can see username password email

mobile number login id address company

and state so the important thing of this

registration page is like here we have

validation for the password part in case

if you are giving a smaller password it

will be not registering let me show you

that so first thing is like let me enter

the username as JP let me give the

password as JP

let me give my email ID

mobile number login ID

address

company

state

so now let us see what happens for this

details and click submit

so now you can see it has been the same

register paid and if you scroll down the

page you can see password should contain

at least one number one lawyer case

character one uppercase character one

special symbol and must be between six

two six six to ten characters long so

this is the validation that we have done

so it should satisfy this condition The

password should satisfy this condition

so now let me enter

password satisfying this condition

so now I have entered the details and

click submit

now you can see you have registered

successfully you proceed for the login

so only if you get this message the the

registration is success and click ok so

now it will be navigated to the login

part so now the registration is success

but even the registration is Success if

we are entering the correct username and

password also

you can see a message like this

incorrect username password please login

with current details because

any employee which is who are

registering the for the first time

should be approved by the Admin so first

let me go to the admin part so let me

open this in a new tab so this is the

admin page so this consists of default

username and password as admin and

so just enter the same user detail for

the admin and admin as username and

password as admin and I mean and click

login

you can see the login as access and

click ok

and once the login is successful if you

be navigated to the admin home page

where you can see the admin home page

where you can see the

e-register details so in the register

details you can see the details of the

users who have been registered that is

the employees of the register so here

you can see the user name JP which I

have done right now with the details of

the username email ID Mobile company

name State and everything and status you

can see it has been waiting so only if

the admin is been approved this this

user then only the user employee can

able to login into their system so let

me click approve

so now you can see the status has been

updated to the uproot part now let me go

to the user login page

now let me enter the correct username

and password and then click login

and now you can see it has been

successful it will be navigated to the

upload image part so if once after the

user

login is successful that is

authentication is successful you can see

the in the user part upload live camera

logout so this live camera is just to

check out the

their images in the live camera for the

user part this is not the main Consular

project so first let me click this live

camera and show you make sure that the

camera is connected in your PC or laptop

so here you can see the user details

that is the user camera

so this is the user part this is not the

main concept this is an additional

concept so let me go to the upload part

so here you the user that is employee

should upload their image so I have

already taken three sample images right

so let me upload one of the image and

then click upload

so once after the image is uploaded

again it will be navigated to the upload

image part now let me go to the admin

part and check so let me log into the

admin in the new tab and I'm checking

out it

so now after the admin login just go to

the user details part so in the user

detail part you can see the details of

the thing that is the user ID is two the

username is JP and the email ID the

company name what is the image that is

and what is the printed result of it so

the printed result is happy

so now let me upload some other image

and check what happens and let me upload

this image and then click upload

so now the image is uploaded again it

will navigate it to the upload image

part now let me go to the next tab that

is in the admin part and Let me refresh

the user details part

so now here you can see the again the

user the username JP with the details of

it which I have uploaded now it has

shown the predicted result is angry

so let me show with some other image

then click upload in the user part

and now it is uploaded and now let me go

to the admin part of it and let me

refresh it the user details

so now you can see the third image that

is a neutral one so in this way you can

check with the your own images or you

can check with the data set images also

so let me now we will see about the

other classification results other than

them that we have taken from the web

camera so here in the base paper you can

see the

other classes as angry discourse fearful

happy

neutral

sad and surprised

so we have already uploaded our live

images using happy angry and you tell us

being detected now let us see about all

other data set image so let me log in to

the user part and let me upload some

other images and check with it

we have already provided you the

the test images that we have to be

uploaded with these classes also so just

go to the test folder image of the

project

so in the test data you can find these

angry discus fearful happy neutral sad

and surprise folders where you can just

upload any of the image and then upload

it

so now if I refresh this page in admin

side

so here you can see the image and the

predicted result is angry so

let me check with the discuss part

and upload it

now let me refresh it

so here you can see the printed result

is discussed

and let me check with the other class

with fear

and upload it

in the admin part the printed result

part you can see it is printed as

neutral

so the fear part is not predicting

correctly

kindly note that any model or this

particular model is not 100 accurate so

there may be some

wrong calculations of it so here fear

has not been printed correctly so now

let me check with uploading some other

image

so let me check with this happy image

so what does character correctly

predicting that the classification of

this employee is happy

now let me go for the other part the

neutral part

load

and in the

admin side you can find it is predicting

as neutral

okay in the sad part

and click upload

and in the user details you can find

that it has been rated as sad

and the final classification just

surprise

and click upload

now let me refresh in admin part

so now you can find the printed result

is surprise so I have shown you with the

all the classification of the

things that is discard so you can just

uh make your own images from your webcam

and check whether this this and never

let me move to the next part that is

performance analysis part so in the

performance analysis part we will be

having performance analysis parameters

of accuracy precision value recall value

and F measure value and

comes the confusion Matrix with the

true table and treated variable where

you can find with the classes of angry

discuss few happy neutral sad and

surprise of it

and final comes the chat part and this

chart is a dynamic chart it will be

showing according to the

the classes that has been predicted so

here you can see there is too angry one

discussed two happy three neutral one

sad and one surprise so for example let

me upload other surprise

so let me upload this surprise image

so now if I

go to the admin part

and now let me check the user details

so here you can see that I have uploaded

this image now it is surprise and in the

graph part

the chart part you can find that now you

can see it has been updated so according

to the printed result this chart will be

varied

and now let me log out

and this is all about the project

detection of stress in it employees

using machine learning technique and

thank you all for watching