AI Project Logbook

PROJECT NAME: IMAGE CLASSIFICATION

SCHOOL NAME: Dev Memorial Public SchooL

YEAR/CLASS: 2024-25/Xll

TEACHER NAME: Vikas Kaushik

TEACHER EMAIL:-kaushikvikas.2008@[Link]

TEAM MEMBER NAMES:

[Link] .
[Link] .
[Link] Saini.
[Link] Garg.
[Link] .

1. Introduction
Image classification is a fundamental task in
computer vision and machine [Link]
involves categorizing and labeling images based
on specific rules applied to their pixels or
vectors. Simply put, it enables computer to
recognize and classify the content within the
images.

2
 2. Team Roles
2.1 Who is in your team and what are their roles?

Role Role description Team Member Name

● Distribute tasks amongst Chhavi Chaudhary

team members.
PROJECT LEADER
● Ensure all deadlines are
met.

● Organize and manage

logbook
PROTOTYPE ● Work to build the model. Yatika Garg
BUILDER,
CODER,TESTER ● Train the model and
testing the efficiency
and accuracy of the
model.

INFORMATION ● Asses data and refine it Varsha Garg

RESEARCHER
● Work with end user and
understand their problems

DATA EXPERT ● Decides upon data and its Garima Saini

type of training model to be
used

● Collects data and draw the

insights required

● Perform analysis to assess

quality and meaning of data

● Analyze trends in data

DESIGNER ● Create designs and decide Vanshika

flow of how to make
solution for problem
statement

● Makes sure visuals are

impactful

● Plot roadmap for what is

desired and manage
making of website

3
4
 2.2 Project plan

The following table is a guide for your project plan. You may use this or create your own
version using a spreadsheet which you can paste into this section. You can expand the
‘Notes’ section to add reminders, things that you need to follow up on, problems that need to
be fixed urgently, etc.

Phase Task Planned Plann Planned Actual Actual Actual Who is Notes/Remarks
start date ed duration start end date duration responsible
end (hours, date (hours,
date minutes) minutes)

Preparing for Coursework, 15/6 15/6 6hr 15/6 15/6 6hr All team
the project readings member

Set up a 15/6 15/6 1 hr 15/6 15/6 1 hr Chhavi

team folder
on a shared
drive

Defining the Background 15/6 15/6 1hr 15/6 15/6 1 hr All team
problem reading evening eveni member
ng

Thinking 16/6 16/6 2 hr 16/6 16/6 2 hr All team

about the morning member
Problems

Team 16/6 16/6 1 hr 16/6 16/6 1 hr All team

meeting to afternoon member
discuss
issues and
select an
issue for the
project

Complete 16/6 16/6 1 hr 16/6 16/6 1 hr

section 3 of
the Project
Logbook

Rate 3/3
yourselves

Understanding Identify users 22/6 22/6 1 hr 22/6 22/6 1 hr All team

the users morning member

Meeting with 22/6 22/6 2 hrs 22/6 22/6 2 hrs All team
users to afternoon members
observe them

Interview with 23/6 23/6 1 hr 23/6 23/6 1hr Varsha

user (1)

Interview with 23/6 23/6 1 hr 23/6 23/6 1 hr GARIMA

user (2), and
etc… VANSHI
KA
Complete 23/6 23/6 30 23/6 23/6 30 All team
section 4 of evening minutes minutes member
the Project
Logbook

Rate 3/3
yourselves

Brainstorming Team 25/6 25/6 2-3 hr 25/6 25/6 3 hr All team

meeting to member
generate

5
 ideas for a
solution

Complete 25/6 25/6 1 hr 25/6 25/6 1 hr Sw apnil

section 5 of
the Project
Logbook

Rate 3/3
yourselves

Designing Team 1/7/2022 1/7/20 2 hr 1/7/2022 1/7/202 2 hr All team

your solution meeting to 22 member
design the
solution

Complete 1/7/2022 1/7/20 30 1/7/2022 1/7/2022 30 Varsha

section 6 of evening 22 minutes minutes
the logbook eveni
ng

Rate 3/3
yourselves

Collecting and Team 4 july 2022 4 july All team

preparing data meeting to 2022 members
discuss data
requirements

Collecting and Data 5 july 2022 6/7/22 24 hours 5/7/22 6/7/22 24 hours VANSHIKA
preparing data collection
Prototyping

Data 8/7/22 9/7/22 24 hours 8/7/22 9/7/22 24 hours GARIMA

preparation
and labelling

Complete 9/7/22 9/7/22 1 hr 9/7/22 9/7/22 1 hr Chhavi

Section 6 of evening eveni evening evening
the Project ng
Logbook

Team 11/7/22 11/7/2 2-3hr 11/7/221 1/7/22 3 hr All team

meeting to 2 members
plan
prototyping
phase

Prototyping Train your 13/7/22 13/7/2 2 hr 13/7/22 13/7/22 2hr . Yatika

Testing model with 2
input dataset

Test your 13/7/22 13/7/2 3 hr 13/7/22 13/7/22 3 hr Chhavi

model and 2
keep training
with more
data until you
think your
model is
accurate

Write a 14/7/22 14/7/2 2hr 14/7/22 14/7/22 2hr VANSHIK

program to 2 AAnd
initiate Yatika
actions
based on the
result of your
model

Complete 14/7/22 14/7/2 1 hr 14/7/22 14/7/22 1hr Chhavi

section 8 of evening 2 evening evening
the Project eveni
Logbook ng

6
 Rate 3/3
yourselves

Team 15/7/22 15/7/2 3hr 15/7/22 15/7/22 3hr All team

meeting to 2 members
discuss
testing plan

Testing Invite users 17/7/22 17/7/2 2hr 17/7/22 17/7/22 2hr VANSHIKA
Creating the to test your 2
video prototype

Conduct 18/7/22 18/7/2 5hr 18/7/22 18/7/22 5hr Yatika

testing with 2
users

Complete 19/7/22 19/7/2 2hr 19/7/221 19/7/22 2hr Varsha

section 9 of 2
the Project
Logbook

Rate 3/3
yourselves

Team 21/7/22 21/7/2 3-4hr 21/7/22 21/7/22 3.5hr All team

meeting to 2 members
discuss video
creation

Write your 22/7/22 22/7/2 3-4hr 22/7/22 22/7/22 3hr GARIMA

script 2

Film your 23/7/22 23/7/2 2-3hr 23/7/22 23/7/22 2hr All team
video 2 members

Edit your 25/7/22 25/7/2 2hr 25/7/22 25/7/22 2hr . Chhavi

video 2

Completing Reflect on 27/7/22 27/7/2 1 hr 27/7/22 27/722 1hr All team

the logbook the project 2 members
with your
team

Complete 27/7/222 27/7/2 1hr 27/7/22 27/7/22 1hr Varsha

sections 10 2
and 11 of the
Project
Logbook

Review your 28/7/22 28/7/2 2hr 28/7/22 28/7/22 2hr

Project 2
logbook and
video

Submission Submit your 30/7/22 30/7/2 30min 30/7/22 30/7/22 30min yatika
entries on the 2
IBM

7
 2.3 Communications plan

- Will you meet face-to-face, online or a mixture of each to communicate?

A mixture of both. Face-to-face whenever possible, and online for late hours.

- How often will you come together to share your progress?

As often as possible, ideally 2-3 times a week.

- Who will set up online documents and ensure that everyone is contributing?

Chhavi is responsible for setting up online documents,along with

ensuring appropriate contribution.

- What tools will you use for communication?

Google Meet, Whatsapp, Calling, Google Drive and shared MS Word file for
collaboration.

2.4 Team meeting minutes (create one for each meeting held)

Date of meeting:16 June 2022

Who attended:-All
Who wasn’t able to attend: None

Purpose of meeting: Research on topic, discussion of ideas and division of roles

Items discussed:
[Link] around us and how we can work on them.
[Link] we decrease the numbers of road accidents.
[Link] of team members-
—-->Editing-yatika
—-->Leader-Chhavi Chaudhary
—-->Research-Varsha garg

Things to do (what, by whom, by when)

[Link] people in locality and observe daily problems.(Vanshika and
Garima) [Link] online document for sharing data and start filling
logbook.(Chhavi)

8
 Date of meeting:25 June 2022
Who attended: All team members
Who wasn’t able to attend: None

Purpose of meeting: To discuss solutions for the problem

Items discussed: An image based model that would help in identification and
classification of objects.

Things to do: Learn and research on best AI powered framework that would help us
make our model

Date if meeting: 1 July 2022

Who attended:All team members
Who wasn’t able to attend:None

Purpose of meeting: To design the solution

Items discussed: Organizing and listing all the requirements required for construction and
designing of model and decided to opt for eye blink based recognition approach.

Things to do: Start with data preparation and collection

Date of meeting: 4 July 2022

Who attended: All the team members
Who wasn’t able to attend: None

Purpose of meeting: To discuss data requirements

Items discussed: Minimum amount of data required for good accuracy and precision as
well as the reliable sources we can collect it from.

Things to do:
1. Collect data
2. Check it for patterns and statistics

Date of meeting: 11 July 2022

Who attended: All the members
Who wasn’t able to attend: None

Purpose of meeting: Team meeting to discuss prototyping phase

Items discussed: Deciding on different types of model we will make

Things to do: Input the data obtain desirable prototype

9
 Date of meeting: 15 July 2022
Who attended: All the members
Who wasn’t able to attend: None

Purpose of meeting: Team meeting to discuss testing plan

Items discussed:
1. Decide on how to proceed with testing and calculate accuracy,F1 score and prediction.
2. Adding diversity to ideas with the help of all team members and end users.

Things to do:
1. Testing
2. Testing by team members followed by end users.

Date of meeting: 21 July 2022

Who attended: All the members
Who wasn’t able to attend: None

Purpose of meeting: Team meeting to discuss video creation

Items discussed:
1. What all materials should be used?
2. Overall theme and script of video
3. Any voice over required ?

Things to do:
1. Film and edit the video
[Link] changes

3. Problem Definition
3.1 List important local issues faced by your school or
The problem faced by counselor is that a large number of students
come up for counseling or with career related queries. They are
attended on a First come first serve basis. The counsellor sorts the
queries classwise. It is seen that most of the class 12th students have
almost the same career related [Link] the same questions
becomes the lengthy process for the counselor.
Main points:
1. Lengthy Process. [Link] consuming.
2. Delayed responses.
3. Human efforts.
community

10
 3.2 Which issues Matter to you and Why?

It is a question of our safety so the mechanism of the image classification has to receive each
and every detail about the face or image recognition for each person. It is a time consuming
task.
The issues in the online process should be addressed if we want to improve the efficiency and
effectiveness of the system. The issues should be specified So the new improved system can
remove these obstacles.

3.3 Which issue will you focus on?

The on which we want to focus is the online approach to make this

process automated and more safer, privacy concerns, The
necessity of quality training data,computational demands,and
potential algorithm biases.

3.4 Write your teaM’s probleM stateMent in the forMat

beloW.

The problems and challenges which our team want to address are dealing with noisy and
unrestructured data, imbalance data, and variation in lighting, angle, and background.

Rate yourself 5

Problem Definition

1 point - A local problem is described

2 points - A local problem which has not been fully solved before is described.
3 points - A local problem which has not been fully solved before is explained in detail with supporting
research.

11
 4. The Users
 Who are the users and how are they affected by the problem?

students
counselor
teachers
Takes time to answer individual queries.

 What have you actually observed about the users and how the
problem affects them?
From what I’ve observed, users often face the following issues due to these
problems:
● Frustration with Inaccurate Results: Users may become frustrated when the
model frequently misclassifies images, especially if they rely on it for critical
tasks.
● Increased Time and Effort: Users might need to spend additional time
correcting errors or providing more data to improve the model.
● Lack of Trust: Persistent inaccuracies can lead to a lack of trust in the system,
making users hesitant to rely on it for important decisions.
● Resource Constraints: Addressing these issues often requires significant
computational resources and expertise, which may not be readily available to
all users.

12
 Record your interview questions here as well as
responses from users.

Interview Questions
Responses:
1. "Almost daily, especially on long drives."
2. "I drink coffee and take breaks, but they
aren't always effective."
3. "Yes, I'd be interested if it improves
safety."

13
  Empathy Map
Map what the users say, think, do and feel about the problem in this table

What our users are saying What our users thinking

 “I need the system to
recognize objects
accurately.”  “Will this system save
 “It should work well my time?”
with different lighting  “Is the accuracy of this
conditions.” system better than
 “I want it to be fast manual classification?”
and reliable.”  “Can I trust the results
for critical tasks?

What our users are How our users feel

doing
 Frustration when the
 Uses the image system misclassifies
classification tool for images.
sorting and organizing  Satisfaction when the
images. system works
 Tests the system with accurately and
various image types and efficiently.
conditions.
 Compares the system’s
 Anxiety about the
output with manual reliability of the
classification results. system in different
scenarios.

14
 What are the usual steps that users currently take related to the
problem and where are the difficulties?

 Data Quality and Quantity: High-quality, labeled data is crucial but

often hard to obtain. Insufficient or poor-quality data can lead to
inaccurate models.
 Computational Resources: Image classification, especially with
deep learning, requires significant computational power, which
can be costly.
 Model Selection and Tuning: Choosing the right model and tuning
it for optimal performance is complex and requires expertise.
 Generalization: Ensuring the model performs well on new, unseen
data is challenging. Overfitting to the training data is a common
problem.

 Write your team’s problem statement in the format below.

Our team is focused on developing an advanced image classification

system to address the growing need for accurate and efficient image
recognition in various real-world applications, such as healthcare,
security, and retail.

Rate yourself 5

The Users

1 point - The user group is described but it is unclear how they are affected by the
problem.
2 points - Understanding of the user group is evidenced by completion of most of the
steps in this section.
3 points - Understanding of the user group is evidenced by completion of most of the
steps in this section and thorough investigation

15
 5. Brainstorming

5.1 Ideas
Table
Category Brainstorming Ideas

Data Collection - Automated data labeling using AI tools

- Synthetic data generation with GANs

- Crowdsourced labeling platform

- Transfer learning with pre-trained models (ResNet, Inception,

Model Selection VGG)

- Custom CNN architectures tailored to the dataset

- Hyperparameter tuning using Bayesian optimization

Handling Challenges - Addressing class imbalance with SMOTE

- Noise reduction using AI-based filters

- Data augmentation with AI-driven techniques

- Advanced metrics (ROC-AUC, confusion matrix, precision-

Evaluation recall curves)

- k-fold cross-validation for robust evaluation

Deployment - Edge AI deployment for real-time classification

- Continuous learning pipeline for model updates

- User feedback loop for model refinement

Applications - Medical imaging for diagnostics (e.g., tumor detection)

- Retail inventory management and product identification

- Security surveillance for real-time activity detection

Ethical
Considerations - Bias mitigation strategies

- Ensuring privacy and data security

16
17
18
6. Design

3.1What are the steps that users will now do using your AI solution
to address the problem?

19
 7. Data

7.1 What data will you need to train your AI solution?

7.2 Where or how will you source your data?

Rate yourself 3/3

Data

1 point – Relevant data to train the AI model have been identified as well as how the data will be
sourced or collected.
2 points - Relevant data to train the AI model have been identified as well as how the data will be
sourced or collected. There is evidence that the dataset is balanced.
3 points - Relevant data to train the AI model have been identified as well as how the data will be
sourced or collected. There is evidence that the dataset is balanced, and that safety and privacy have
been considered.

20
 8. Prototype

8.1 Which AI tool(s) will you use to build your prototype?

Eye blink sensor

8.2 Which AI tool(s) will you use to build your solution?

A type of computer vision

8.3 What decisions or outputs will your tool generate and what further
action needs to be taken after a decision is made?

Rate yourself 3/3

Prototype

1 point – A concept for a prototype shows how the AI model will work.
2 points - A prototype for the solution has been created and trained.
3 points - A prototype for the solution has been created and successfully trained to meet users’
requirements.

21
9. Testing
9.1 Who are the users who tested the prototype?

9.2 List your observations of your users as they tested your solution.

22
 9.3 Complete the user feedback grid

What works What needs to change

More data needs to be added to increase the
Able to classify most of chances of dowrsy accuracy.
position

Questions? Ideas
You can add more functions to the model
Would it affect our lives in any case??
Is it actually of any use??
What are the daily uses?

9.4 Refining the prototype: Based on user testing, what needs to be

acted on now so that the prototype can be used?

23
 9.5 What improvements can be made later?

Rate yourself 3/3

Testing

1 point – A concept for a prototype shows how it will be tested.

2 points - A prototype has been tested with users and improvements have been identified to meet
user requirements.
3 points - A prototype has been tested with a fair representation of users and all tasks in this section
have been completed.

24
 10. Team collaboration
10.1 How did you actively work with others in your team and with stakeholders?

- Held regular team meetings to discuss progress and

address challenges.
- Collaborated with design and tech teams to align on
functionality and comfort.
- Engaged with potential users for feedback through
surveys and focus groups.
- Consulted with safety experts to ensure compliance
with industry standards.
- Maintained open communication channels via email
and project management tools.

Rate yourself 3/3

Team collaboration

1 point – There is some evidence of team interactions among peers and stakeholders.
2 points - Team collaboration among peers and stakeholders is clearly documented in this section.
3 points - Effective team collaboration and communication among peers and stakeholders is clearly
documented in this section.

11. Individual learning reflection

11.1. Team Reflections

A good way to identify what you have learned is to ask yourself what surprised you during
the project. List the things that surprised you and any other thoughts you might have on
issues in your local community.

25
 Team member name: Vanshika

This project showed me my weaknesses and strengths and made me work on them and
enhance my skills.I learnt how to work in a team and how to be time [Link] helped bring
change in my work ethics and taught me new skills.
I spent so much time on this project and all through it, I learnt that if I set my goals, I can
achieve anything. Hence this project gave me a new perspective on things.

Team member name:-Yatika and Chhavi

The project affected me as a person whole. It has shaped my personality in a way that now I
know what to expect when people say they are looking for people with skill and passion to work
with, not just some degree holding person. Working with Yatika and GARIMA who are the
most passionate people to work with, have taught me how to have fun without worrying too
much,
contributing not only to my team building skills but also my personality. The sense of
responsibility that I experienced to make sure that everything about the project is perfect was
fun not a burden. This project was practical in my school life that I didn’t take as a burden or a
compulsion but as an opportunity to learn and experience something new. Indeed it was a
project that really made me use my creative side of brain and for the first time I realized what it
is to put use of your theoretical knowledge practically and how different both the domains
[Link] Course it wasn’t easy but it was a refreshing and wholesome experience that I would
cherish.
Team member name:-Varsha and Garima

of working in a group,and how to tackle different ideas and ideologies on a particular topic.
This project helped me improve my time and stress management,the positives and negatives

26
 Rate yourself 3/3

Individual Learning Reflection

1 point – Some team members present an account of their learning during the project.
2 points - Each team presents an account of their learning during the project.
3 points - Each team member presents a reflective and insightful account of their learning during the
project.

12. Video link

Enter the URL of your team video :-

A real life accident which is caused by the sleepy

position of driver while drove a passenger bus here the
link of that video :-[Link]
AA4CvEjgT2VwbmFb

27
 Appendix
Recommended Assessment Rubric (for Teachers)

LOGBOOK AND VIDEO CONTENT

Steps 3 points 2 points 1 point Points
Given

Problem A local problem which has not A local problem which has not A local problem is
definition been fully solved before is been fully solved before is described
explained in detail with described.
supporting research.

The Users Understanding of the user group Understanding of the user The user group is
is evidenced by completion of all group is evidenced by described but it is unclear
of the steps in Section 4 The completion of most of the how they are affected by
Users and thorough steps in Section 4 The Users. the problem.
investigation.

Brainstorming A brainstorming session was A brainstorming session was A brainstorming session

conducted using creative and conducted using creative and was conducted. A solution
critical thinking. A compelling critical thinking. A solution was was selected.
solution was selected with selected with supporting
supporting arguments from arguments in Section 5
Section 5 Brainstorming. Brainstorming.

Design The use of AI is a good fit for the The use of AI is a good fit for The use of AI is a good fit
solution. The new user the solution and there is some for the solution.
experience is clearly documentation about how it
documented showing how users meets the needs of users.
will be better served than they
are today.

Data Relevant data to train the AI Relevant data to train the AI Relevant data to train the
model have been identified as model have been identified as AI model have been
well as how the data will be well as how the data will be identified as well as how
sourced or collected. There is sourced or collected. There is the data will be sourced or
evidence that the dataset is evidence that the dataset is collected.
balanced, and that safety and balanced.
privacy have been considered.

Prototype A prototype for the solution has A prototype for the solution A concept for a prototype
been created and successfully has been created and trained. shows how the AI model
trained to meet users’ will work
requirements.

Testing A prototype has been tested with A prototype has been tested A concept for a prototype
a fair representation of users with users and improvements shows how it will be
and all tasks in Section 9 have been identified to meet tested.
Testing have been completed. user requirements.

Team Effective team collaboration and Team collaboration among There is some evidence of
collaboration communication among peers peers and stakeholders is team interactions among
and stakeholders is clearly clearly documented in Section peers and stakeholders.
documented in Section 10 Team 10 Team collaboration.
collaboration.

Individual Each team member presents a Each team presents an Some team members
learning reflective and insightful account account of their learning present an account of their
of their learning during the during the project. learning during the project.
project.

28
 Total points

VIDEO PRESENTATION

Points Given
3 – excellent
Criteria 2 – very good
1 – satisfactory

Communication The video is well-paced and communicated, following a clear

and logical sequence.

Demonstrations and/or visuals are used to illustrate

Illustrative
examples, where appropriate.

Accurate The video presents accurate science and technology and

language uses appropriate language.

The video demonstrates passion from team members about

Passion
their chosen topic/idea.

Sound and
image quality
The video demonstrates good sound and image quality.

The content is presented in the video within a 3-minute

Length
timeframe.

Total points

29