A Training Report

Submitted to partial fulfilment of the requirements

for the award of the degree of

Bachelor of Technology
In

INFORMATION TECHNOLOGY
by

VIVEK MAHENDRABHAI SALIYA (19SE02IT053)

Under the supervision of

Ms. Khushbu Chauhan, Assistant Professor

MAY, 2023

P P SAVANI SCHOOL OF ENGINEERING

P P SAVANI UNIVERSITY
NH NO.: 8, VILLAGE: DHAMDOD, TA. MANGROL, NEAR KOSAMBA, SURAT – 394
125. (GUJARAT).
CERTIFICATE
 ACKNOWLEDGEMENT

It is indeed with a great sense of pleasure and gratitude that I acknowledge the help of
these individuals. I am highly indebted to our dean, Dr. Niraj Shah, of the School of
Engineering at P. P. Savani University, for the facilities provided to accomplish this
industrial training.
I feel elated in expressing our sense of gratitude to our institute's training guide, Ms.
Khushbu Chauhan. She has been a constant source of inspiration for me, and I am very
deeply thankful to her for her support and valuable advice.
I am extremely grateful to our departmental staff members, lab technicians, and
nonteaching staff members for their extreme help throughout my training.
Finally, I would like to express my thanks to the company's CEO and all employees of the
WP Infotech company who helped me in the successful completion of this industrial
training.

Name of Student:
VIVEK MAHENDRABHAI SALIYA (19SE02IT053)
 ABSTRACT

I have gained a lot of knowledge throughout my internship by finishing projects with some
extra work. Real-time object identification, 8D-Audio, and an Alexa interview bot are the
current projects I'm working on. Computer vision is connected to object detection. The
ability to detect occurrences of objects in photos and movies is known as object detection.
Instead of using conventional object detection techniques, it recognizes the features of
images and creates an intelligent knowledge of them that mimics how human vision
functions. In 8D-Audio, we investigate how audio affects users' experiences in virtual
environments. Sound appears to emanate from all directions when using 8D audio
processing, which creates an immersive audio experience. To create a three-dimensional
audio environment, it requires the use of many audio effects and techniques. It is feasible
to replicate the influence of sounds coming from different directions and distances relative
to the listener's head by applying HRTF filters to audio signals, and Interviewing is
typically used to determine which candidates are most qualified for a position. This
procedure differs from business to business. However, in general, the two basic
methods—the technical interview and the human resource interview—remain the same.
The conventional interviewing process is labour- and resource-intensive. By using the
Alexa interview bot in place of the human interviewer, we can automate this procedure.
The process will be somewhat facilitated by the bot. Our plan is to offer a chatbot that will
interview candidates, compile a report, and shortlist applicants based on the findings.

I've gained a lot of knowledge from this internship overall. not just in terms of academics
but also in terms of the workplace. I learnt more in-depth information about machine
learning and python, as well as about a number of python libraries. In addition to this, I
learnt about managing tasks in my working life, cooperating with others in a group,
finishing projects on time, and many other professional ethics.
 CONTENTS

I. Certificate i
II. Acknowledgement ii
III. Abstract iii
IV. Content iv
V. List of Figures v
VI. List of Tables vi
1.0 Overview of Company 1
1.1 Introduction of Company
1.2 Company Details
1.3 Organization Chart
1.4 Company Overview
1.5 Projects Details
1.6 Skill requirements for employment in the company

2.0 Introduction of Project 10

2.1 Introduction of Training
2.2 Scope of Project
2.3 Study of Existing System
2.4 Methodology
2.5 Software Requirements
2.6 Improvement proposed in the current system
2.7 Testing of New System
2.8 Projects Screenshot

3.0 Learning Outcomes & Discussion 25

3.1 Learning Outcomes
3.2 Details of reporting schedule
3.3 Glimpses of Training
3.4 Limitations & Future Improvement Suggestions

4.0 Conclusions 32

5.0 References 38
 LIST OF FIGURES

FIGURE NO. DESCRIPTION PAGE NO.

Figure 1.1 Company Logo

Figure 1.3 Organization Chart

Figure 2.8.1 Create Annotation

Figure 2.8.2 Create Json file

Figure 2.8.3 Object Detection

Figure 2.8.4 Object Counting

Figure 2.8.5 Registration From

Figure 2.8.6 Login Page

Figure 2.8.7 Database of Login Users

Figure 2.8.8 Home page

Figure 2.8.9 YouTube URL page

Figure 2.8.10 Pop-Up for Email Verification

Figure 2.8.11 Database of Email Verification

Figure 2.8.12 Send Email for Email Verification

Figure 2.8.13 Email content for verification

 LIST OF TABLES

TABLE NO. DESCRIPTION PAGE NO.

Table 1.2 Company Details

Table 3.2 Reporting Schedule

 Chapter 1
Overview of Company

1.1 Introduction of Company

Figure 1.1 Company Logo

→ WP INFOTECH founded in 2012 as a customize software development

company with the philosophy to provide ethical solutions to satisfy the needs of
Diamond industry for Identifying, Authenticating and Securing critical business
information.
→ WP INFOTECH gives very much importance to their valuable clients for
maximum Return on their Investment. WP INFOTECH offers variety of services
in the field of Information Technology likes, Web development, Graphics design,
Android apps development, IOS apps, etc. Our company also works on Python
development as well as Machine learning.

1.2 Company Details

→ As WP Infotech is an IT sector company, the main objective of our company is
to provide great software, digital products, and brands to their customers. Our
skill development programs are designed considering the present expectations
of the clients/users.
→ Some other details are as mentioned below,
Company Name WP Infotech
Founder and CEO Mr. Dhruval Godhani
Android Development, Web designing, IOS
Core work area
Development, Python.
Scale Small to Medium Enterprise (SME)
Contact [email protected]
3004, Silver Business Point, VIP circle, Uttran,
Address
Surat, Gujarat 394105
Website https://www.wpinfotech.com/
Table 1.2 Company Details
1.3 Organization Chart

Figure 1.3 Organization Chart

→ Currently, WP infotech does not have a lot of staff members. So, the company
is not following any organizational chart company is working under managers of
“Production” and “Manager” and other team leaders and team members work
under those managers. Supervision of the project work is done by the company’s
CEO.

1.4 Company Overview

→ We are an innovative IT company specializing in mobile application
development, with a focus on Web development and IOS. Our talented team of
developers, designers, and project managers work closely with clients to deliver
custom solutions for their unique business needs. We also offer Python
technologies, UI/UX design, and quality assurance testing. With a commitment to
excellence and staying up-to-date with the latest industry trends, we provide
high-quality, user-friendly applications that exceed expectations.

1.5 Projects Detail

→ Real-time object detection is an essential task in computer vision, enabling
machines to perceive and interact with the surrounding environment. Our
project aims to utilize the Region-based Convolutional Neural Network (RCNN)
model to perform real-time object detection. RCNN is a state-of-the-art deep
learning technique that has proven to be highly effective in detecting and
classifying objects in images. It works by first generating a set of region
proposals, or potential object locations, within an image. Each of these proposals
is then individually processed by a convolutional neural network to extract
relevant features, which are used to classify the object.
 → 8D audio is an exciting new technology that creates an immersive, three-
dimensional audio experience for listeners. Our project focuses on using Head-
Related Transfer Function (HRTF) processing to achieve high-quality 8D audio.
HRTF is a technique that simulates how sound waves interact with the human
head and ears, creating a unique audio profile for each individual listener. By
capturing this profile and processing audio signals using advanced algorithms,
we can create 8D audio that sounds as though it is coming from all around the
listener, rather than just from traditional stereo channels.
→ Currently, I’m working on interview bot using Alexa voice service which is
provided by Amazon’s. Our project aims to develop an innovative interview bot
using Amazon's Alexa voice service. The interview bot is designed to streamline
the hiring process by conducting initial candidate screenings, saving time and
resources for companies.

1.6 Skill requirements for employment in the company

→Depending on the exact work role and needs, our organization may have
different skill requirements for its employees. There are certain essential abilities
that are highly regarded, like as
➢ Technical proficiency: This refers to mastery of programming languages,
database administration, and other technical abilities pertinent to the
employment.
➢ Problem-solving skills: Locate the source of problems and create
workable solutions.
➢ Communication skills: Excellent communication skills are needed to
cooperate with team members, explain technical concepts to non-
technical stakeholders, and interact with clients.
➢ Creativity: We place a great priority on innovation and creativity in our
workplace, thus professionals must be able to solve challenges creatively
and creatively.
➢ Time management: In order to successfully prioritise jobs and fulfil
deadlines when working on numerous projects at once, we have great
time management skills.
➢ Continuous learning: Continuing education and professional
development, staying current with emerging technologies and market
tendencies.
 Chapter 2
Introduction of Projects

2.1 Introduction of Training

→ The company has been a pioneer in the field when it comes to embracing
technological development and incorporating it into their projects to come up
with creative ways to present answers to diverse issues. I had the good fortune to
work on a project module throughout my internship. I had the chance to work on
a fascinating machine learning project during my internship that involves
Python- based real-time object detection. I had to create a convolutional neural
network (CNN) model for the project that could recognize and categorize
objects from a video stream in real-time. In order to enhance the user
experience, I developed 8D-Audio, which produced a 360-degree sound
sphere surrounding the listener, allowing for a more immersive audio
experience. I successfully finished the creation of this project module during my
internship. By raising the caliber and efficacy of their products and services, this
has the potential to revolutionize a number of industries, including security and
entertainment.

2.2 Scope of Project

→The goal of the project is to build a model that can recognize objects in real time
and use that knowledge to activate various capabilities. For instance, the
fundamental idea is to use object detection to identify the exact person and
immediately send their photos to their email if they attend a ceremony where
numerous photographers take pictures. We were able to automate the process of
identifying people and delivering their images to their emails by implementing
this object detection algorithm into the functionality set, reducing the
requirement for manual work, and enhancing the user experience.
→The potential for 8D-Audio processing with Head-Related Transfer Function
(HRTF) is enormous. A 360-degree sound field can be created around the listener
using HRTF processing, creating an immersive audio experience that can
improve a variety of media types, including music, movies, and video games. The
main goal of 8D-Audio is to mimic how we perceive sound in the real
environment, giving users a more authentic and natural experience. For the user,
this technology can be applied in a variety of sectors, including entertainment and
virtual reality, to improve their experiences and make them more participatory.

2.3 Study of Existing System

→ Real-time object detection systems typically use deep learning algorithms and
computer vision techniques to analyze and identify objects in real-time video
 feeds. These systems can be broken down into three primary components: input,
processing, and output.
→The input component of the system involves capturing and preprocessing video
data from various sources, such as webcams or surveillance cameras. The video
data is then fed into the processing component of the system, which analyzes the
frames in real-time to identify objects.
→ The existing system works by recording and mixing the sound in a way that
simulates a 3D experience. This is achieved by using different frequencies and
reverberations, as well as panning the sound from left to right, to create an illusion
of depth and spatial positioning. However, the sound is still limited to the two-
channel stereo format, and the experience can sometimes feel flat or lacking in
depth.

2.4 Methodology
→For, Real-time object detection
➢ Data Collection: Collect a dataset of images with objects that you want to
detect. You can use popular object detection datasets like COCO or create
your own dataset.
➢ Preprocessing: Resize the images to a fixed size and normalize the pixel
values.
➢ Train the RCNN model: Use the preprocessed dataset to train an RCNN
(Region-based Convolutional Neural Network) model. An RCNN model is a
two-stage approach to object detection that involves generating region
proposals in the first stage and then classifying those regions in the second
stage.
➢ Real-time Object Detection: Use the trained model to detect objects in
real-time video or images. For this, the input frames are passed through the
RCNN model, which generates region proposals. Then, the regions are
classified into object categories, and the bounding boxes are drawn around
them.

→For, 8D-Audio
➢ Data Collection: The CIPIC dataset is a popular dataset for HRTF
processing. It contains HRTF measurements for 125 different locations
around the head of a human listener. Download and extract the dataset.
➢ Preprocessing: The CIPIC dataset includes measurements for different
elevations and azimuths, so you can select a subset of measurements that
correspond to the elevation and azimuth angles you want to use.
➢ Implement HRTF processing: Once you have selected the subset of HRTF
measurements, you can implement HRTF processing. This involves
convolving an audio signal with the HRTF filters to simulate how sounds
would reach the ears of a human listener in real life.
 ➢ Spatial Audio Mixing: Once you have processed the audio signals with
HRTF filters, you can mix the signals to create an 8D audio experience. The
mixing process involves using different amplitudes for each HRTF-filtered
signal, based on the sound's location in 3D space.
➢ Audio Playback: Finally, you can play the mixed audio signals through an
8D audio playback system, which can be headphones or a speaker array.
The playback system will spatially distribute the mixed signals to create an
immersive audio experience.

2.5 Software Requirements

→ Deep learning framework such as TensorFlow or PyTorch.
→ Computer vision library such as OpenCV.
→ Python libraries such as Librosa, Sox, Pysndfx, and NumPy can be used for
audio signal processing.
→ Head-Related Transfer Function (HRTF) database or library such as the MIT
KEMAR or CIPIC HRTF databases.
→ Overall, implementing real-time object detection using the RCNN model and
8D audio using HRTF processing with Python requires a combination of
computer vision, deep learning, and audio signal processing libraries, as
well as access to an HRTF database or library for spatial audio rendering.

2.6 Improvement proposed in the current system

→ Real-time object detection involves building a model that takes an input name
and uses it to detect objects in all images. Once the objects are detected, the images
are moved to another folder. The folder containing the detected images is then
zipped and automatically sent via email.
→ To use 8D-Audio, you can take input in the form of a song or a YouTube URL. If
you choose to use a YouTube URL, you can select the format in which you want to
download the song. This can include options like MP3 only, MP4 only, 8D MP3, or
8D MP4.
→Once you have the song in the desired format, you can use HRTF processing to
create an 8D-Audio experience. This involves convolving the audio signal with the
HRTF filters to simulate how sounds would reach the ears of a human listener in
real life. The mixing process involves using different amplitudes for each HRTF-
filtered signal, based on the sound's location in 3D space.

2.7 Testing of New System

→To test the real-time object detection system, we can follow the steps given
below:
➢ Input testing images: We need to input various images with different
objects in them to test the system's accuracy. These images can be sourced
from various online image databases or captured using a camera.
 ➢ Run the detection model: Once the input images are available, we can run
the real-time object detection model on them.
➢ Verify the detected objects: After the detection model has run on the
input images, we need to verify that all the detected objects are correctly
identified. We can do this by comparing the detected objects with the actual
objects present in the images.
➢ Check the movement of images to another folder: Once the objects are
detected, the images should be moved to another folder. We can verify this
by checking the folder and verifying that the detected images are present
in it.
➢ Check email with zipped folder: Finally, we can check the email to which
the zipped folder of detected images was sent. The email should contain the
zipped folder with all the detected images.

→To test the 8D-Audio system, we can follow the steps given below:
➢ Input testing songs: We need to input various songs in MP3 and WAV
formats to test the 8D-Audio system.
➢ Download the songs: After inputting the songs, we need to download
them in the desired format. We can choose from the available formats,
including MP3 only, MP4 only, 8D MP3, or 8D MP4.
➢ Process the songs with HRTF filters: Once the songs are downloaded, we
need to process them with HRTF filters to create an 8D-Audio experience.
We need to ensure that the processing is done correctly and that the audio
signals are convolved with the HRTF filters accurately.
➢ Verify the 8D-Audio experience: After the processing is complete, we can
listen to the songs to verify the 8D-Audio experience. The audio should
sound immersive, and the listener should be able to hear the different
sounds coming from different directions.
➢ Check the sound location in 3D space: The mixing process involves using
different amplitudes for each HRTF-filtered signal, based on the sound's
location in 3D space. We can verify that the mixing process is done
correctly by checking the location of the sound in 3D space. We should be
able to hear the sound coming from the correct location.
2.8 Projects Screenshot
→For, Real-time object detection

Figure 2.8.1 Create Annotation Figure 2.8.2 Create Json file

Figure 2.8.3 Object Detection

Figure 2.8.4 Object Counting

→For, 8D-Audio

Figure 2.8.5 Registration Form

Figure 2.8.6 Login Page

Figure 2.8.7 Database of Login Users

 Figure 2.8.8 Home page

Figure 2.8.9 YouTube URL page

Figure 2.8.10 Pop-Up for Email Verification

 Figure 2.8.11 Database of Email Verification

Figure 2.8.12 Send Email for Email Verification

Figure 2.8.13 Email content for verification

 Chapter 3
Learning Outcomes & Discussion

3.1 Learning Outcomes

→ The real-time object detection system involves building a machine learning
model to detect objects in images. By working on this project, one can learn
about various machine learning concepts, including data preprocessing, model
building, and evaluation.
→ By working on this project, one can gain practical experience with computer
vision concepts such as feature extraction, object detection, and image
classification.
→ There are several object detection libraries available, such as Tensorflow
Object Detection API, YOLO (You Only Look Once), and Faster R-CNN.
→ Building an accurate object detection model requires fine-tuning various
hyperparameters such as learning rate, batch size, and optimizer. By working
on this project, one can gain experience with model tuning techniques to
optimize the model's performance.
→ Training deep learning models on large datasets can be resource-intensive,
requiring high-end GPUs and specialized hardware.
→The 8D-Audio system involves processing audio signals to create an
immersive listening experience. By working on this project, one can learn about
various audio processing concepts, including Fourier transforms, filters, and
spatial audio.
→By working on this project, one can gain an understanding of human
perception, including how humans localize sounds and the effects of binaural
hearing.
→ There are various audio formats available, including MP3, WAV, and FLAC.
By working on this project, one can gain familiarity with these formats and learn
how to convert audio between them.
→By working on this project, one can learn about HRTFs, which are filters that
simulate how sound waves interact with the listener's head and ears.
→ The 8D-Audio System involves mixing multiple audio signals to create a
spatial audio experience. By working on this project, one can gain experience
with audio mixing techniques such as panning, equalization, and reverb.
→The 8D-Audio System requires visualizing the audio signals in 3D space. By
working on this project, one can gain familiarity with audio visualization tools
such as Audacity, Sonic Visualizer, or MATLAB for analyzing and visualizing
audio signals.
3.2 Reporting Schedule
Sr. No. Reporting Dates Points Discussed
1. 31 Dec, 2022 →Introduction about company
→Give me task of learning basic Python,
Machine Learning, and Database
2. 28 Jan, 2023 →On the first reporting, as the project
was given to me which was ‘Real-time
object detection’ I discussed it with our
mentor that what the project is, scope of
our project and applications of it in real-
life, flow of the making this project and
the technologies which we are going to
use in this.
→After that, I discussed about data
gathering and labelling of objects to
make our own dataset.
→How I trained the dataset with the
help of some software and environment
in RCNN model.
→Also, discussed about weights files,
how it works and why we choose the
weight file.
→Afterwards, I talked about my work
which was performed detection of the
objects and how it took automatically as
per the requirement.
3. 11 Feb, 2023 →On the next reporting, I discussed
about how we can detect object outside
of the COCO dataset.
→After that, I create Annotation file and
make weight file for all the objects.
→Afterwards, I build a model for male
and female for gender identification.
→Also, I discussed about my testing
period and how much it was accurate.
4. 25 Feb, 2023 →Third reporting taken by industrial
expert.
→ I show all the work done on my
project while at the company.
→Also, I discussed of the 8D-Audio for
immersive sound experience.
 5. 11 Mar, 2023 →On the fourth reporting, I discussed
with my mentor the new project
assigned to me at the company. I
presented all the functionalities of the
project and demonstrated a small demo.
6. 25 Mar, 2023 →At the next reporting session, my
mentor from another institution took
charge. I presented both projects and
discussed their functionalities, as well as
the workflow.
→Additionally, I implemented the
suggestions provided by my mentor.
7. 15 Apr, 2023 →In the final reporting, I presented all
the work completed during my time at
the company. My mentor provided me
with some suggestions regarding my
project.
Table 3.2 Reporting Schedule

3.3 Glimpses of Training

→ I had the chance to work on a Real-Time Object Detection System and an 8D-
Audio System project during my internship. Both initiatives gave me great
learning opportunities and gave me the chance to put my academic
understanding into practice
→ I oversaw developing a deep learning model that could recognize things in
real-time as part of the Real-Time Object Detection System project. I gained
knowledge of data preprocessing, neural network construction and training,
and model optimization. My abilities in computer vision, deep learning, and
model tweaking have all improved as a result of this research.
→ As part of the 8D-Audio System project, I was responsible for creating an
immersive audio experience using head-related transfer functions (HRTFs) and
audio mixing techniques. I learned how to process audio signals, convolve them
with HRTF filters to simulate 3D audio, and use mixing techniques such as
panning and equalization to create a spatial audio experience. This project
helped me improve my skills in audio signal processing, psychoacoustics, and
audio mixing.
→ Overall, my internship gave me invaluable practical experience in signal
processing and machine learning. It assisted me in honing my technical abilities
and provided a sneak peek into what it's like to work on actual projects
involving artificial intelligence. I am appreciative of the chance to work on these
projects as well as the advice and assistance my mentors gave me while I was
an intern.
3.4 Limitations & Future Improvement Suggestions
→ Limitations:
➢ The Real-Time Object Detection System may struggle with detecting
objects in low-light conditions or if the object is partially obscured.
➢ The 8D-Audio System may not work optimally for all types of music or
audio content, as the effectiveness of the HRTF processing can vary
depending on the characteristics of the audio.

→ Future Improvement Suggestions:

➢ To improve the Real-Time Object Detection System, additional data
augmentation techniques could be employed to help the model
generalize better to different lighting conditions and partial occlusions.
➢ For the 8D-Audio System, future work could focus on developing more
accurate HRTF filters to improve the spatial audio experience.
Additionally, machine learning techniques could be used to train the
system on different types of audio content and optimize the mixing
process for different genres of music.

→ In both projects, future work could also focus on improving the user interface
and making the system more user-friendly for non-technical users. This could
involve developing a web or mobile application for the systems to simplify the
input and output processes. Additionally, the systems could be integrated with
other software and hardware to enable more advanced use cases, such as real-
time streaming or integration with smart home devices.
 Chapter 4
Conclusions

→ The chance to work on such intriguing projects as the Real-Time Object Detection
System and the 8D-Audio System makes me proud. We have acquired excellent
knowledge and abilities through the Glimpses of Training internship that will certainly
be helpful in our future jobs.
→Working on these projects has allowed me to develop my team's technical skills
while also teaching me the value of teamwork, communication, and flexibility in the IT
sector. We were able to collaborate directly with knowledgeable mentors and peers
during the internship, which allowed us to share knowledge, opinions, and
suggestions.
→ Despite the limitations of these initiatives, I'm enthused about the prospect of
further advancements and innovations in these areas. I'm motivated and driven to
keep learning about and experimenting with new technologies and to make a positive
impact on the development of the tech sector.
→I'm glad for the chance to take part in the Glimpses of Training internship, and my
team and I are eager to put the skills and information we've acquired to good use in
the IT sector and beyond.
 Chapter 5
References

1. https://github.com/matterport/Mask_RCNN
2. https://www.youtube.com/watch?v=bUoWTPaKUi4
3. https://sox.sourceforge.net/
4. https://pypi.org/project/pysndfx/
5. https://openai.com/blog/chatgpt
6. https://phpmailer.en.softonic.com/?ex=DINS-635.3
7. https://www.w3schools.com/whatis/
8. https://www.codingnepalweb.com/email-subscription-form-php/