Project Plan

Product-based Capstone Project

Team ID : C242-PS037

Team Member :

1. (ML) M288B4KY1398 – Fardian Zahri Chaniago – Universitas Pamulang -

[Active/Inactive]
2. (ML) M254B4KY3032 – Muhammad Reza Al Fatah – Universitas Mercu Buana -
[Active/Inactive]
3. (ML) M254B4KY4405 – Vemas Adi Pratama – Universitas Mercu Buana -
[Active/Inactive]
4. (CC) C010B4KY0595 – Anugerah Fauzan Azzimah – Universitas Indonesia -
[Active/Inactive]
5. (CC) C294B4KX3739 – Rekha Inaya Putri – Universitas Pembangunan Jaya -
[Active/Inactive]
6. (MD) A254B4KX2130– Kaisyarani Zahra Adinomo – Universitas Mercu Buana -
[Active/Inactive]
7. (MD) A294B4KY2828 – Muhammad Faris Hafizh – Universitas Pembangunan Jaya -
[Active/Inactive]
 Project Plan
Product-based Capstone Project
Final Selected Themes:
Health Innovation: Empowering Vulnerable Communities for Health and Well-being

Title of the Project:

Pawers Cat Disease Detection

Excecutive Summary/Abstract:

According to data from Pet Fair SEAsia, 72% of Indonesian households own pets, with cats
being the most popular choice at 47%. However, novice cat owners often struggle to
recognize signs of illness in their pets. This application is specifically designed to help
beginner cat enthusiasts detect diseases in their cats through modern technology.

The project aims to detect cat diseases using artificial intelligence technology. This
application employs a Convolutional Neural Network (CNN) model trained on a dataset of
cat skin disease images, optimized with MobileNetV2 to ensure smooth operation on mobile
devices. For other diseases that cannot be detected through images, the application uses a
custom Vertex AI chatbot based on structured data provided to the model.

Key features of the application include skin disease classification through computer vision,
detection of other diseases via the chatbot, and additional information on care tips,
prevention, and nearby pet shops. The application achieved a validation accuracy of 80% in
detecting cat skin diseases, which is considered reliable for initial use.

This application aims to improve cat owners’ awareness in recognizing diseases early and
reducing the risk of delayed treatment. Future development plans include adding more
features and enhancing accuracy.

How did your team come up with this project?

For our capstone project, we developed a mobile application that utilizes computer vision to
classify various types cats of diseases . In this process, we trained a Convolutional Neural
Network (CNN) model using TensorFlow on a dataset comprised of labeled images of cat
skin diseases, allowing us to achieve high accuracy in classifying different skin conditions.

To enhance the performance and efficiency of the application on mobile devices, we

applied transfer learning by fine-tuning the MobileNetV2 model, which is specifically
designed to perform well on resource-constrained devices. In addition, We plan to integrate
a chatbot using Vertex AI to provide interactive support to users. This chatbot is designed to
 Project Plan
Product-based Capstone Project
answer user inquiries regarding symptoms, offer prevention tips, and guide them on the
necessary steps to take if a disease is detected. With the implementation of this chatbot,
we aim to improve the user experience by making information easily accessible and
understandable, ultimately helping cat owners provide timely and appropriate care for their
pets.

Project Scope & Deliverables :

The scope of the project is to develop a mobile application for detecting skin diseases in
cats using computer vision. The application will utilize a trained Convolutional Neural
Network (CNN) model to classify various skin conditions based on images uploaded by
users. Additionally, we plan to integrate a chatbot using Vertex AI to provide interactive
support, answering users' questions about symptoms, prevention tips, and guiding them on
next steps if a disease is detected. The application aims to enhance the user experience by
making information accessible and easy to understand.

Deliverables:
ML:
1. Prepare dataset & data gathering
2. Preprocessing & Cleaning data
3. Create Model Architecture
4. Model Optimization
5. Deployment & Integration
CC:
1. Google Cloud Service
2. Express.js
MD:
1. Create UI display design
2. Create an application
3. Insert the model into the application and ensure the model can run properly
 Project Plan
Product-based Capstone Project

Project Schedule:

Start
Task PIC Status End date Priority Link
date
Implement a basic workflow Not 11/15/20 11/20/20
MD Main Task
to access the main feature. started 24 24
Integrate AI/ML capabilities
as the app's main feature, Not 11/16/20 11/21/20
MD Main Task
either via on-cloud or directly started 24 24
on-device
Implement networking calls
Not 11/17/20 11/22/20
to interact with the project's MD Main Task
started 24 24
API.
Ensure the implementation
of the main features you aim
Not 11/18/20 11/23/20
to address in the project MD Main Task
started 24 24
without causing the
application to crash.
https://www.figm
a.com/design/0fq
In SqyUs7ZMIC98Ur
Add a custom app icon to 10/28/20 11/11/20
MD progres Main Task Z6BeJ/Untitled?n
your application. 24 24
s ode-id=0-1&t=yoZ
CsHdpBWq6VYit-
1
Provide a downloadable APK Not 12/1/202 12/7/202
MD Main Task
file of the app. started 4 4
https://www.figm
a.com/design/0fq
In SqyUs7ZMIC98Ur
10/28/20 11/11/20
App Mockup MD progres Side Task Z6BeJ/Untitled?n
24 24
s ode-id=0-1&t=yoZ
CsHdpBWq6VYit-
1
Add Login and Registration Not
MD Side Task
Feature started
Not
Implement OTP MD Side Task
started
 Project Plan
Product-based Capstone Project
Add Feature Dashboard,
Not
Settings, Homepage, MD Side Task
started
Articles, Histoy
Add Unique Component:
Not
Notification, Location, and MD Side Task
started
Widgets
Utilize TensorFlow
Not 11/9/202 11/15/20
architecture for building ML Main Task
started 4 24
machine learning models.
Vertex Al can only be used if
Not 11/10/20 11/16/20
you intend to build ML Main Task
started 24 24
Generative Al
In
10/28/20 11/8/202
Buat First Model ML progres Side Task
24 4
s
Not
Buat Second Model ML Side Task
started
Hosting API or other web Not 11/11/20 11/30/20
CC Main Task
service started 24 24
Not 12/1/202 12/8/202
Build cloud arcitecture CC Main Task
started 4 4
Not
Creat git teams repository CC Side Task
started
Create backend using Not 11/11/20 12/5/202
CC Main Task
express started 24 4

Gantt Chart Models:

Oct Nov Des

Task Week Week Week Week Week Week
4 1 2 Week3 4 1 2
Creat git teams repository
Add a custom app icon to your
application.
App Mockup
Create First Model
Implement a basic workflow to access the
 Project Plan
Product-based Capstone Project
main feature.
Integrate AI/ML capabilities as the app's
main feature, either via on-cloud or directly
on-device
Implement networking calls to interact
with the project's API.
Ensure the implementation of the main
features you aim to address in the project
without causing the application to crash.
Vertex Al can only be used if you intend
to build Generative Al
Hosting API or other web service
Build cloud arcitecture
Create backend using express

Based on your team’s knowledge, what tools/IDE/Library and resources that your
team will use to solve the problem?
- Android Studio
- Firebase
- Retrofit
- Glide
- Figma
- Tensorflow
- Tensorflow Lite
- Jupyter Notebook
- Google Collab
- Google Maps API
- Vertex AI

Based on your knowledge and explorations, what will your team need support for?
We need veterinarians to validate the predictions our model generates. This includes
reviewing the model’s output to confirm that the identified disease or condition aligns with
clinical symptoms and known diagnostic criteria.

Once the model identifies a disease, we will need expert veterinary guidance on the
appropriate treatment methods. Veterinarians can help us compile specific and effective
 Project Plan
Product-based Capstone Project
treatment protocols for each disease our model can predict, covering medication types,
dosages, and alternative treatments.

Based on your knowledge and explorations, tell us the Machine Learning Part of your
Capstone!
For our capstone project, we developed a mobile application that uses computer vision to
classify types of skin diseases in cats. We trained a Convolutional Neural Network (CNN)
model using TensorFlow on a dataset of labeled cat skin disease images, achieving accurate
classification of various conditions. To enhance performance and efficiency on mobile
devices, we applied transfer learning by fine-tuning the MobileNetV2 model. Additionally,
We planned to integrate a chatbot using Vertex AI to provide users with interactive support,
answering questions about symptoms, prevention tips, and guiding users on next steps if a
disease is detected. This chatbot improves user experience by making information
accessible and easy to understand

Based on your knowledge and explorations, tell us the Mobile Development Part of
your capstone?
We will create prototype using Figma for mockup design and user flow, use Kotlin
programming language and Android Studio to build app and we use Firebase for the
authentication. For the machine learning model, we use TensorFlow Lite, and we also utilize
the Google Maps API.

Based on your knowledge and explorations, tell us the Cloud/Web/Frontend/Backend

Part of your capstone?
The backend will be developed with a serverless deployment approach, using Cloud
Functions and Cloud Run for flexible and efficient service handling, we may specifically use
Node.js and/or Flask as the frameworks for building our services. We will create private
RESTful APIs with authentication and authorization handled via Firebase Authentication. The
real-time database will be managed with Firestore, and Vertex AI will be used for deploying
the trained CNN model for disease detection. Cloud CDN may be used for faster content
delivery, depending on the need for optimized media distribution.

Based on your team’s planning, is there any identifiable potential Risk or Issue related
to your project?
1. Model Accuracy
Due to limited datasets, we are afraid that something bad could happen in the
future. However, the model we have created so far can detect cat disease very well.
2. User Adoption
 Project Plan
Product-based Capstone Project
The other challenge is user adoption, where we have to be able to create a display
that is user friendly and attractive. We have to ensure that the UI we create is
suitable and can be used well. Users may be skeptical about the accuracy and
reliability of an AI-driven diagnosis, fearing misdiagnoses that could harm their pets.
So pawers must be able to give users confidence by explaining the details of the
disease, how to prevent it and how to treat it
3. Technical Integration
Ensuring the smooth integration of MobileNetV2, Vertex AI, and other technologies
on mobile devices presents several challenges, primarily due to the limited
computational resources and varying hardware specifications of these devices.
MobileNetV2, a lightweight Convolutional Neural Network (CNN) designed for
mobile applications, must be carefully optimized to run efficiently.

Any other notes/remarks we should consider on your team’s application

Our project has great potential in terms of animal disease detection. We apply a simple UI
display, so that new users who want to know their cat's disease can use our application
properly without complications. We also provide an explanation of the disease and how to
treat it directly from a Veterinarian. So all cat owners don't need to worry about their cat's
condition.