Conceptual Page

1|Page
Plagiarism Report

2|Page
Acknowledgement
I would like to express my deepest gratitude to my supervisor Mr. Manoj Shrestha, for their
guidance, support, and encouragement throughout this project. Their invaluable insights and
knowledge have been instrumental in the successful completion of this project. I would also
like to extend my sincere thanks to Mr. Shrawan Thakur for providing me with the necessary
resources to complete this research. Their support and assistance have been greatly
appreciated. I would like to express my gratitude to my friends and colleagues Mr. Milan
Yadav, Mr. Tulasi Kattel, Rohit Sah and Mr. Ajay Purbe, for their valuable suggestions and
feedback during the development of this project. Their support and encouragement have been
a great source of motivation for me.

Finally, I am deeply grateful to my family for their unwavering support and encouragement.
Their love and support have been a constant source of motivation throughout this journey. I
would like to express my appreciation to all those who have directly or indirectly contributed
to the successful completion of this project.

3|Page
Abstract
This project aims to develop a reliable and accurate solution for detecting various types of
rice plant diseases using the latest advancements in machine learning. The study begins with
data collection and pre-processing, where many images of different rice plant diseases were
collected and labelled manually. The data was then pre-processed to prepare it for model
training. The study then utilizes convolutional neural networks (CNNs) and the VGG19
architecture for feature extraction and model training. The model was trained for several
epochs and evaluated using metrics such as accuracy, precision, and recall. Hyperparameter
tuning was also performed to optimize the model's performance. The model was then
deployed in an application built using Streamlit, which made it easy for end-users to interact
with the solution. The application provides accurate results and helps farmers to identify the
diseases in the rice plants. The study also discussed the ethical challenges and
recommendations for future work. Overall, this solution provides a reliable and accurate
method for detecting rice plant diseases and is expected to have a significant impact in the
agricultural industry.

4|Page
Keywords

5|Page
Table of Contents
Conceptual Page....................................................................................................................................1
Acknowledgement.................................................................................................................................2
Keywords...............................................................................................................................................4
Table of figures......................................................................................................................................6
Introduction...........................................................................................................................................8
Aim......................................................................................................................................................11
Objectives............................................................................................................................................12
Justification..........................................................................................................................................12
Problem Statement.........................................................................................................................12
Solution...........................................................................................................................................14
Research questions..............................................................................................................................16
Scope...................................................................................................................................................17
Ethical Consideration...........................................................................................................................18
Research Methodology........................................................................................................................19
Literature Review................................................................................................................................21
Case study 1: - Google Lens.............................................................................................................21
Case study 2: - Tesla automated vehicle..........................................................................................23
Case study 3: - FASAL.......................................................................................................................25
Development Methodology:............................................................................................................27
Tools:...............................................................................................................................................29
Visual Studio Code:......................................................................................................................30
Jupyter Notebook:.......................................................................................................................31
Figma:..........................................................................................................................................32
Technology:.....................................................................................................................................33
Python:........................................................................................................................................34
TensorFlow:.................................................................................................................................35
Streamlit:.....................................................................................................................................35
Keras:...........................................................................................................................................35
Techniques......................................................................................................................................35
User Experience (UX)...................................................................................................................38
Availability, Backup, and recovery strategy.................................................................................39
Security........................................................................................................................................39
Updates and Upgrades................................................................................................................40
Findings...............................................................................................................................................40
Recommendations and future work....................................................................................................46

6|Page
Conclusion...........................................................................................................................................47
REFERENCES........................................................................................................................................49
Appendix.............................................................................................................................................60
Gantt chart.......................................................................................................................................60
Project plan.....................................................................................................................................63
Issues Logs and Risk Management..................................................................................................63
Supporting Diagrams.......................................................................................................................65
Model Training (VGG19 and EfficientNetV2B2)...........................................................................65
Comparison loss and accuracy between two CNN architecture..................................................72
Application Testing......................................................................................................................73

Table of figures
Figure 1: figure of evaluation design of system.........................................................................8
Figure 2: Illustrations of machine learning model to train model for rice plant disease
detection...................................................................................................................................10
Figure 3: Objectives of the project...........................................................................................12
Figure 4: Illustrations of justification (problem and solution ) of the project..........................14
Figure 5: Illustration of research question...............................................................................16
Figure 6: Scope of the Project..................................................................................................17
Figure 7: Illustration of ethical consideration..........................................................................18
Figure 8: Illustrations of desk-based research methodology....................................................20
Figure 9: picture of process of secondary research..................................................................21
Figure 10: Illustrations of architecture of google lens.............................................................22
Figure 11: Illustration of how tesla motors detect object on the road......................................24
Figure 12: Illustration of architecture of FASAL....................................................................26
Figure 13:Illustration of agile methodology............................................................................28
Figure 14: Illustrations of tools................................................................................................30
Figure 15: Figure of VScode....................................................................................................31
Figure 16: Illustration of Jupyter notebook..............................................................................32
Figure 17: Illustration of Figma...............................................................................................33
Figure 18: Illustrations of technology that is used to complete the project.............................34
Figure 19: Conceptual page for rice plant disease detection....................................................36

7|Page
Figure 20: Architecture of CNN algorithm..............................................................................37
Figure 21: user experience of the application..........................................................................39
Figure 22: Exploratory data analysis of the image dataset......................................................41
Figure 23: Performance matrices of VGG19 and EfficientNetV2B2......................................42
Figure 24: Grad camera visualization of input images of rice plant leaves.............................43
Figure 25: Figure shows the result of the inputted image........................................................44
Figure 26: Ethical challenges to be considered during this project completion......................45
Figure 27: project plan one......................................................................................................60
Figure 28: Project plan two......................................................................................................61
Figure 29: Project plan three....................................................................................................62
Figure 30: charts of project plan..............................................................................................63
Figure 31: Tables of issues logs...............................................................................................63
Figure 32: Illustration of risk management..............................................................................64

8|Page
Introduction
“The internet is becoming the town square for the global village of tomorrow” -Bill Gates.
Agriculture is one of the main occupations in the Nepal as it is landlocked country.
According to Federation of Nepalese Chambers of Commerce and Industry (FNCCI),
agriculture sectors hold 36% of the total GDP (Gross Domestic Product) of Nepal. In Nepal
about 30% of the total area are engaged for agriculture purposes. As Nepal’s one-third of the
land is utilised for the farming, the agriculture sector creates huge job opportunity and
employment from farming to enterprise. Thus, agriculture engages about 66% of the total
population of Nepal (Joshi & Upadhyaya, 2020). The geography and ample supply of labour
provides comparative advantages in the agriculture sectors. As such a huge percentage of
land and people are engaged in the agriculture sector of Nepal, Nepal produces different
variety of crops. The major cash crops are potato, tomato, oilseed, sugarcane, jute, cotton,
and the major cereal crops are rice, wheat, maize, barley, and buckwheat, etc. Apple, orange,
mushroom, cardamom, cloves, honey, banana, and mushroom are also cultivated in ample
amounts.

9|Page
 Figure 1: figure of evaluation design of system
Rice is the well identified cereal or the staple food in Nepal or even in the southeast Asia.
Nepal imported Rs 39 billion worth of rice from India in august 2020 to April 2021 (Joshi
and Upadhyay). (2020, June 24). This data suggest that the demand of the rice is more in
Nepal, but the production of the rice is in less amount. Even though Nepal climatic condition,
geographical reason, water resources, and land productivity are favourable to produce rice,
many households specially in mountain region of Nepal experience food shortage up to 4-9
month per year. There are several reasons that decrease the production of the rice that are
subsistent farming, climatic conditions and improper pesticide and herbicide use, which
harms both the quality of grains and appears to reduce the land's capacity which leads bad
production of the grains. It is harmed by contamination caused by illnesses transmitted by
numerous agents such as germs, parasites, and diseases. To tackle these problems and
overcome the sickness, previous farmers had just one option: use conventional technologies.
Ranchers are unable to differentiate the infection on rice crops and, as a result, are unable to
use adequate preventative measures to direct the sickness. To reduce plant illness from the
root, effective remedies and suitable medicine are required, which is only feasible if the

10 | P a g e
problem is diagnosed early. Plant disease is identified by examining the plant's leaves for
explicit manifestations of illness which is very painful for the farmer as ranchers takes more
time and cost to do it. (Nepal Rice, paddy production, 2022).

“Even old pond systems have been eclipsed by the unstoppable expansion of contemporary
technology such as mobile technology, Internet of Things to access real time data at any time
and place, so why not apply this technology to rice plant disease identification?”

Figure 2: Illustrations of machine learning model to train model for rice plant disease
detection
Hence, the proposed solutions include image processing technique using machine learning
algorithms. This is an advanced solution which is more affordable and gives exact outcomes
at very less range of time by identifying the rice crop infections and suggests appropriate
remedies like insecticides or pesticides to manage the sickness. This solution also makes
farmer burden free from identifying disease as they can also do it with their mobile phones at
any time they required. The farmer has only to do scan of the disease rice plant leaf then
technology or the app automatically identify the disease and suggest the remedies of that
disease. This is a main goal of this research. Even several types of disease are found out on
rice plant such as leaf blast, Hispa, Brown spot, bacterial leaf streak, leaf smut, panicle blight

11 | P a g e
bronzing and leaf sealed, etc. however, only 4 different rice plant disease and healthy leaf are
used to train the models that are Brown spot, Leaf smut, Leaf blight, Bacterial leaf blast and a
healthy leaf. To complete this research and make it useful four DNN based pre-trained
models are selected that are VGG-19, Xception, Inception-Resnet-V2, and Resnet-101. (Han,
C. H. & Kim ,2018).

Fasal, Google Lens, and Tesla are three companies that have made significant advancements
in their respective industries by utilizing advanced technologies such as machine learning,
computer vision, and IoT. Fasal, an aggrotech company, uses image processing, machine
learning and IoT to detect diseases in crops, predict remedies, and suggest favourable weather
conditions for spraying the medicine, while also helping farmers save water and money with
sustainable farming practices. Google Lens, a visual search technology, uses machine
learning and computer vision to identify and provide information about objects and text
through a device camera. Tesla, on the other hand, specializes in electric vehicles, energy
storage and solar panel manufacturing, using technologies such as advanced battery
technology, machine learning algorithms, and IoT to improve the performance and safety of
their vehicles, and to make their cars more autonomous, also helping to accelerate the world's
transition to sustainable energy. All three companies are utilizing technology to improve the
efficiency of their respective industries, whether it's agriculture, visual search or energy and
transportation.

Aim

12 | P a g e
Objectives

Collect the study

material and dataset.

Figure 3: Objectives of the project

Justification
Problem Statement
The main food grains or cereal for Nepalese as well as southeast people are rice. Nepal is
landlocked country having 30% of cultivable land. Since the cultivable land is 30% it has a
great scope for agriculture. There are 66% of the total population of Nepal engages in
farming. (Choudhary et al., 2021). The geographical reason, water resources, temperature of
Nepal are very suitable for the various agriculture product that are paddy, wheat, jute, potato,
and cotton, etc. However, having such a good environmental condition the production of rice
as well as other agricultural product is comparatively less. According to the article written by
Ramesh Kumar on May 23, 2021, in nepaltimes.com, Nepal imported Rs. 39 billion worth of
rice from India in between August 2020 to April 2021. From this report it analysed that the
demand of rice is very high but unable to produce that much rice in Nepal. This imbalances in
the demand and supply chain of the rice the country may suffers huge losses in the gross

13 | P a g e
domestic product (GDP). If it continues for a long period country may suffers starvation too.
The production of the rice is very low in Nepal due to various reason like subsistent farming,
lack of proper knowledge about the disease of plant, inappropriate use of fertilization. One of
the main reasons for the reduction of rice production is rice plant disease and maladies.
Various types of plant disease are detected in the laboratory such as sheath blight, brown
spot, leaf blast, bacterial spot. Like many other diseases early detection or identification of
the disease can reduce or prevent the damages of the rice plant and thus they may increase the
production of the rice. The fundamental issue is the absence of constant observation of the
plants. Other factors can be that farmers new to the field are not attentive and cognizant of the
diseases that can occur to the plants and their seasons. Normally, these maladies can infect
plants at any time. Yet, constant observation of plants and their growth period can restrain
disease infection. Manually patrolling the vast rice field daily by experts or ranchers is an
impossible task due to the vast sizes of the farms, and even, if possible, it would be a costly
task, prone to human error, causing damage to rice plants in their path, and many other
factors that will end up causing more harm than good. Classifying or diagnosing an issue is
an incredibly challenging task to be performed physically, as it involves different parameters
to be observed, such as conditions, surroundings, and so forth. Even after consulting the
specialist the accurate diseases are not identified at a certain period. The specialist requires
more time and money to identify the rice plant disease, but time is a more constraint factor in
managing the disease. If the disease spreads to the entire plant of the farming area very
recently, it becomes difficult to make the plant disease free, leading to the destruction of all
plants and a reduction in rice production. (Deng et al., 2021).

14 | P a g e
 Figure 4: Illustrations of justification (problem and solution ) of the project.

Solution
To address the problem and to make rice farming sector more productive the new solution is
proposed. This solution includes android application with the integration of image processing
technique followed by machine learning algorithm. This system utilises novel optimized ML
(Machine Learning) and deep learning (DL) techniques that will accurately detect, classify,
and diagnose rice disease automatically and, suggest the recommendation or solution for the
identified rice plant diseases. With the advancement of the technology and tools agricultural
sector personnel are wished to utilize the modern methodology like many other sectors are
using it however there is limitation of the solution that is utilized in the agriculture sector as
compared with the other sector. Therefore, the proposed solution holds the emotion and

15 | P a g e
power to solve the issues that is discussed in the above problem section. This system makes
farmer burden free from identifying the disease because the farmer only must scan the leaves
of diseased rice plant by their mobile through the provided mobile application. After scanning
the image of the rice plant leaves, the system identifies the disease types and at the same time
it also suggests the solution followed by the types of disease identified. This solution helps
farmer to identify the plant disease earlier which helps them to make the plant disease free
which directly improves the production of the rice plant. When the disease and their solution
is identified very earlier then the farmer can use the recommended pesticides and insecticides
in a recommended time. If the process is work successfully then off course the production of
the rice will be increased which balances the demand and supply chain of the rice. From the
article published by the reliefweb.int on 6th May 2022, Nepal imports 1.35 million tons of
rice which is 13% more than 2021. If this continues for a long time Nepal may suffers great
imbalance in demand and supply chain of the rice. So, to prevent from this imbalances Nepal
must focus on the production of rice in very huge amount. Hence, the proposed solution will
make it true because farmer will be able to identify the disease of the rice plant at very early
stage, thus farmer get concentrated at time and make the plant disease free which prevent
from spreading disease to the whole field. Then the production of rice will increase, and the
country should not have to dependent on other countries and it decrease the import of the rice
plant from other countries which balances the chain of the food from farming land to the
enterprise level. Previously plant disease detection is done by visual inspection of the leaves
or some chemical processes by experts. For doing so, a large team of experts as well as
continuous observation of plant is needed, which costs high. In such conditions, the proposed
system proves to be helpful in identifying disease and provide a recommendation at a hand in
a very few seconds.

16 | P a g e
Research questions

Figure 5: Illustration of research question

17 | P a g e
Scope

Figure 6: Scope of the Project

18 | P a g e
Ethical Consideration
Ethical consideration is a set of constraints and principle that is followed to maintain the
report ownership and to make project remarkable. As, human beings is directly involved to
the report or project, ethical consideration should be maintained from initially to accomplish
the report that prevents the report from the falling into the undegraded categories. This makes
project under lower risk categories by identifying the right or wrong in the project
accumulation. As this project is prepared by following all the constraints that is governed by
the government of Nepal, both individual and societal conditions are considered and maintain
the privacy of everything that result this report into the lower risk. As farmer is considered as
the subject of this research, this research tries to maintain the cognitive load of the farmer.
This project aim is to develop mobile application that helps farmer to detect the disease of the
rice plant and also recommend a solution followed by the disease. As this is computer-based
solution colour and design of the app is made by keeping colour blindness in the mind.

Figure 7: Illustration of ethical consideration

This report also maintains the international code of language that is acceptable in all societies
of Nepal. And by applying such an important information in the project it helps to reduce the
cognitive load of the individual. The data is stored in the cloud databases which ensure the
security and efficiency of the project. Based on the principle of ethical consideration like risk
of harm, informed consent, anonymity, confidentiality, human priority, practices moral,
norms and values are made prior during the completion of the project. Similarly, copyright
and privacy concerns necessitate the citation and accreditation of references from many prior

19 | P a g e
resources, such as books, journals, websites, and PDFs, throughout the research process.
Likewise, any kind of offensive, exaggeration or deception words are avoided to make
communication clear and more effective as this is research is based on the societal action. As
the research data are not directed to harm in any condition the data is collected with full of
consent to overcome the ethical issues.

Research Methodology
Desk based research methodology is followed to analyse the research that is required during
project completion. It is done to identify tools, technologies, process and helps to acquire
information about the chosen topic that justify the aim and objectives of the project; thus,
enhances project validity and reliability. Desk based research methodology is chosen because
as a college student, several constraints are to be followed and have limitation of money to
spent on extra research. Thus, this research is done inside the school, college or respective
rooms, the research is completely done without having any dangerous risk which is very
feasible and acceptable for the dissertation. There are mainly three types of research method
that are qualitative, quantitative, and mixed. Mixed method is utilized in this project because
data collected to validate this project are on both exploratory and confirmatory in nature. This
research follows mixed method also because both textual and numerical data are analysed to
make research reliable. To accomplish research probability sampling design of data collection
is acquired because data of the problems in rice production are collected randomly from the
group of farmers which belongs to both known and unknown categories to me. Thus, this
sampling design method helps to generalise all the problems in rice production of the terai
belt of Nepal, as it is considered as the cultivatable land among all the regions. Due to these
methods of research the project should get its satisfactory result. (Mehta, 2021).

20 | P a g e
 Figure 8: Illustrations of desk-based research methodology
Qualitative data such as different types of disease that affect rice plant, data of land reforms
for the analysing the cultivable or fertile land, different control factor and solution to mitigate
the disease that affect the rice plant and farmlands. Quantitative data such as different types
of machine learning algorithms, number of diseases, percentage of rice plant that is not
affected by the disease- or disease-free plant. The size and number of images that is utilised
to identify the accuracy rate of algorithms which examine whether the captured images are
disease free or if diseased then helps to identify the name, types, and family of the disease
that belongs to. To enhances the project quality and make it valuable and reliable several case
studies should be done through several books, documents, and records which are properly
referenced and cited. The comparatively good data analysis is done through qualitative
content analysis and quantitative data analysis such as descriptive statistics and inferential
statistics as this project is based on machine learning approach. Finally, the research
methodology chosen for this project is mixed method that is comfortably and bravely fulfil
the aim and objectives of this project.

21 | P a g e
Literature Review

Figure 9: picture of process of secondary research

Case study 1: - Google Lens

Tesla is most innovative autonomous company in the world. Amazon, the most reputable e-
commerce company, Alibaba is thought as a best firm for experienced entrepreneurs who
want to sell their product in comparatively less price but the trust, we place in the google is
matchless. Google is also considered as a modern god because it helps to search and find the
required solution in a second. Moreover, Google helps to solve problem of almost all sector
and enhance its functionality that includes education sector, health sector, hospitality, and
real e-state sector, etc through utilization of google lens. Google Lens is a visual search tool
that allows users to search the internet and obtain information about the objects and scenes
that they see through their smartphone cameras. In the retail industry, Google Lens has been
used by retailers to allow customers to scan products in-store and receive information about
the product, such as its price and availability. In the healthcare industry, Google Lens has
been used by doctors and other medical professionals to access information quickly and
easily about a patient's medical history or condition by scanning a patient's identification card
or other documents. The technology used by Google Lens is based on a combination of
machine learning algorithms, image processing techniques, and other technologies, which
allows it to detect and classify objects and scenes in real-time accurately and efficiently. It
also employs visual analysis and a neural network to get pertinent information about the
objects it identifies. For example, if a user points their smartphone camera at a tree, Google
Lens will use image processing techniques to segment the tree from the background and
extract relevant features, such as the shape, colour, and texture of the tree. These features are

22 | P a g e
then fed into the machine learning algorithms that make up Google Lens, which will use them
to classify the tree and provide the user with information about it.
In addition to object detection and recognition, Google Lens also uses image processing
techniques for other purposes, such as optical character recognition (OCR) and augmented
reality (AR). OCR is the process of extracting text from images, and it is used by Google
Lens to recognize and interpret text that is captured by the smartphone camera. AR is the
process of overlaying digital information on top of the real world, and it is used by Google
Lens to provide users with additional information and context about the objects and scenes
that they see. This service may be expanded to include reading textbooks or interpreting the
details of an agreement. The text-to-speech capability of Google Lens potentially be a
lifesaver for the 800 million individuals worldwide who struggle with reading. Now, anybody
with a smartphone can aim it at text and hear it said aloud. This new feature, along with its
availability through Google Go, is just one way to help more people understand the world
around them. (Murphy & K. P. ,2021).

Figure 10: Illustrations of architecture of google lens.

This case study was chosen because the technology employed by Google Lens may also be
utilized in plant disease detection systems to identify plant illness. Knowing the algorithm

23 | P a g e
used in Google Lens makes it easy to employ in many alternative solutions to tackle a variety
of other problems. Google Lens identify the name and types of several types of plant and
variety of many other objects through image recognition by using computer vision
technology and CNN machine learning algorithm. This technology is also best fit for
identification of the classes and names of the disease found in the rice plant. As like a google
lens, the mobile app should be made through which farmer can scan the image and get result.
To do this, the user would simply point their smartphone camera at the plants and developed
app would process the images and provide the user with a predicted label for each plant,
indicating the presence or absence of a particular disease. This technology can provide
accurate and efficient disease detection, and it can be used by farmers and other stakeholders
to improve the health and productivity of their crops. Overall, Google Lens has proven to be a
valuable tool for a variety of industries, providing users with quick and easy access to
relevant information simply by using their smartphone camera. (Scott & G, 2018).

Case study 2: - Tesla automated vehicle

In terms of comprehension, prior to Tesla, no businesses in the car sector had such advanced
technology. People could not even imagine the technology that Tesla delivers to the
automobile sector by establishing the self-driving vehicle. This vehicle processes in a regular
method by utilizing a variety of instruments and technology. As a company that specializes in
electric vehicles, Tesla uses machine learning algorithms and image processing techniques
extensively in their autonomous vehicle technology. One way that Tesla uses machine
learning in their autonomous vehicles is in the Autopilot system. The Autopilot system is a
suite of advanced driver-assistance features that allow Tesla vehicles to drive themselves on
highways and other roads with limited human intervention. The Autopilot system uses a
combination of cameras, radar, ultrasonic sensors, and machine learning algorithms to
navigate roads and avoid obstacles. For example, the cameras mounted on the vehicle capture
images of the surrounding environment and feed them into a convolutional neural network
(CNN), a type of deep learning algorithm. The CNN processes the images and uses them to
identify and track objects, such as other vehicles, pedestrians, and traffic signs. The CNN is
trained on a large dataset of images Sand labelled data, so that it can learn to recognize and
classify objects in the real world. In addition to the CNN, the Autopilot system also uses
other machine learning algorithms, such as support vector machines and decision trees, to
process data from the radar, ultrasonic sensors, and other sensors. These algorithms help the
Autopilot system to make decisions about how to navigate the roads and avoid obstacles. The
AI system in a Tesla autonomous vehicle also uses sensor fusion techniques to combine data
from the car's various sensors, including cameras, radar, and lidar. This allows the car to
create a completer and more accurate picture of its surroundings. For example, radar can
provide information about the distance and speed of objects, while lidar can provide detailed
3D information about the shape and location of objects. By combining these different types
of data, the AI system can create a more comprehensive model of the car's environment.
(Metz & Baudette, 2021).

24 | P a g e
 Figure 11: Illustration of how tesla motors detect object on the road.
Another way that Tesla uses image processing and machine learning is in their Supercharger
network. The Supercharger network is a network of fast-charging stations that Tesla owners
can use to quickly charge their vehicles. Tesla uses machine learning algorithms to predict
when and where Superchargers will be most heavily used, so that they can optimize the
placement of new stations and manage the flow of charging vehicles. In addition to these
examples, Tesla also uses image processing and machine learning in their manufacturing
processes. For instance, they use machine learning algorithms to analyse images of car parts
and detect defects in the manufacturing process. This allows them to identify and address
potential issues before they become problems, resulting in higher quality vehicles. Moreover,
Tesla also uses this technology in its customer service operations to improve the accuracy and
efficiency of its virtual assistants. When customers send images of their vehicles or other
products to Tesla's customer service team, the images are processed by AI algorithms to

25 | P a g e
identify specific details, such as the make and model of the vehicle or the serial number of a
product. This allows the virtual assistants to provide more accurate and relevant information
to customers.
Overall, the use of machine learning algorithms and image processing techniques is crucial to
the success of Tesla's autonomous vehicle technology. These technologies allow the vehicles
to navigate roads and avoid obstacles in a safe and efficient manner, enabling Tesla to
provide their customers with a high-quality driving experience. The company's innovative
use of AI has allowed it to develop advanced self-driving systems, optimize the performance
of its energy products, and improve the accuracy and efficiency of its customer service
operations. From their autonomous vehicle technology to their manufacturing processes,
these technologies enable Tesla to improve the efficiency and performance of their products
and operations. (Robertson, 2022).

Case study 3: - FASAL

This case study is about FASAL, FASAL is a company that provides agricultural intelligence
solutions to farmers and other stakeholders in the agriculture industry. The company was
founded in 2015 with the goal of using technology to address the challenges facing the
agriculture industry, including limited access to information, variable weather conditions, and
pest and disease outbreaks. One of Fasal's key products is a mobile app that provides farmers
with real-time information about their crops, including soil moisture levels, weather forecasts,
and alerts about potential threats. The app also allows farmers to connect with other farmers,
agricultural experts, and local marketplaces to share information and resources. Their
solutions use a combination of machine learning algorithms, satellite imagery, and other data
sources to provide farmers with insights about their crops and the surrounding environment.
One example of how FASAL uses machine learning is in their crop health monitoring
system. The system uses satellite imagery and machine learning algorithms to monitor the
health of crops over time, and to identify and diagnose potential problems, such as pests,
diseases, or nutrient deficiencies. To use the crop health monitoring system, farmers first
need to provide FASAL with information about their crops, such as the type of crops they are
growing and the location of their fields. FASAL then uses this information to access satellite
imagery of the fields and apply machine learning algorithms to analyse the images. The
image processing techniques used by FASAL are designed to isolate the crops in the images
and extract relevant features that can be used by the machine learning algorithms to diagnose
potential problems. For example, the techniques might be used to segment the images into
individual crops, to extract colour and texture features from the crops, or to convert the
images into different colour spaces to highlight certain features. Once the algorithms have
been trained, they can be used to analyse new satellite images of the farmer's fields. The
algorithms will identify and diagnose any potential problems that they detect, and they will
provide the farmer with recommendations for addressing these problems. For instance, the
algorithms might recommend applying certain pesticides, fertilizers, or other interventions to
improve the health of the crops. In addition to their crop health monitoring system, FASAL
also offers other solutions that use machine learning, such as their yield estimation system
and their weather forecasting system. These solutions provide farmers with valuable insights

26 | P a g e
about their crops and the surrounding environment, which can help them to make more
informed decisions and improve their yields. (Rajendra, 2021).

Figure 12: Illustration of architecture of FASAL.

One example of FASAL involved the use of satellite imagery to monitor the progress of rice
crops in the state of Odisha. Farmers in the region were facing challenges due to
unpredictable weather and a lack of access to accurate information about the status of their
crops. With the help of FASAL, farmers were able to receive regular updates on the health
and growth of their rice crops, allowing them to make more informed decisions about
irrigation and other management practices. Another example involved the use of FASAL to
monitor the progress of wheat crops in the state of Punjab. In this instance, the FASAL team
used satellite imagery to track the growth of the crops and to identify areas that were

27 | P a g e
experiencing stress or other issues. By providing regular updates on the health and growth of
the crops, the FASAL project was able to help farmers make more informed decisions about
irrigation and other management practices, leading to improved crop yields and higher
incomes for the farmers.
Finally, the case studies of FASAL have shown that the use of satellite and aerial imagery
can be a valuable tool for farmers in India, providing them with access to accurate and up-to-
date information about the status of their crops. The Machine Learning Algorithm, Internet of
Things (IoT) and Image processing techniques allow FASAL to extract relevant information
from satellite imagery and provide farmers with valuable insights and recommendations,
which can help them to improve the health and productivity of their crops. This information
can be used to optimize irrigation and fertilization practices, leading to improved crop yields
and higher incomes for farmers. As the world continues to face the challenges of feeding a
growing population, Fasal's innovations in agriculture technology may provide valuable
solutions for improving food security and supporting farmers. (Software, 2022)

Development Methodology:
Agile methodology and the Scrum framework were chosen for the development of the rice
plant disease detection solution due to the project's uncertain and rapidly changing
requirements. The ability of Agile methodology to adapt to change was a key factor in this
decision. The nature of the project, which involves using machine learning to develop a
solution for a complex problem, meant that requirements were subject to change as new
information and insights were gained. By utilizing Agile methodology, changes could be
incorporated through iterations, or "sprints," during which a working version of the software
was developed, and feedback was received. This allowed for continuous improvement and
refinement of the solution, ensuring that it met the overall goals and objectives.
The emphasis on collaboration and communication within the Scrum framework was another
advantage. The role of the module teacher as Product Owner and Scrum Master encouraged
the sharing of ideas and knowledge and allowed for a more effective and efficient
development process. Weekly reviews from the module teacher helped to enhance the project
development process.

28 | P a g e
 Figure 13:Illustration of agile methodology
Furthermore, Agile methodology provided a way to measure and track progress. By weekly
sprints and retrospectives, obstacles could be identified and overcome, and performance
could be improved over time. This helped to stay on track and deliver a high-quality working
product that met the overall goals and objectives. Overall, the use of Agile methodology and
the Scrum framework allowed for effective management of the development of the rice plant
disease detection solution. The iterative and adaptive nature of Agile methodology, along
with the emphasis on collaboration and communication within the Scrum framework, led to
the successful delivery of a high-quality working solution that met the overall goals and
objectives. The guidance and support provided by the module teacher throughout the
development process, including weekly reviews, was invaluable in the completion of this
project.
The development of the rice plant disease detection solution required a comprehensive
approach that included the different phases of pre-game, mid-game, and post-game. The pre-
game phase, also known as the planning phase, was the starting point of the project. During
this phase, research was conducted on the current state of technology in the field of plant
disease detection and machine learning. This research helped to identify the specific
requirements for the solution and the potential challenges that may arise during the
development process. Additionally, a literature review was conducted on different types of
rice plant diseases and their symptoms, as well as the current methods used for their
detection. This helped to identify the specific rice plant disease that the solution would focus
on detecting. The most effective machine learning algorithms and techniques for the task of
rice plant disease detection were researched, taking into consideration the type of data that
would be used, the complexity of the problem, and the available computational resources.

29 | P a g e
Once the research and planning phase was completed, the focus moved on to the mid-game
phase.
The mid-game phase, also known as the execution phase, was the stage where the majority of
the project work took place. This included implementing the chosen machine learning
algorithms, collecting, and labelling data for training, and testing the solution to ensure it
meets the project requirements. Feedback from the module teacher was received and
incorporated into the solution. The iterative and adaptive nature of Agile methodology
allowed for continuous improvement of the solution through iteration and adaptation.
Finally, the post-game phase, also known as the closing phase, involved wrapping up the
project. This included final testing and debugging of the solution, creating documentation,
and delivering the final product to the module teacher. Additionally, the project's success was
evaluated in terms of meeting the project goals and objectives and areas for improvement in
future projects were identified. A detailed report and explanation of the methodology and
techniques used was provided to the module teacher. Through this comprehensive approach,
the solution was able to be successfully developed.

Tools:
To develop the rice plant disease detection solution, several tools were used including Visual
Studio Code, Figma, Adobe Illustrator, Pytorch, TensorFlow, and Jupyter Notebook. These
tools were chosen for their ability to support multiple programming languages, debugging
capabilities, easy collaboration and prototyping, ability to build and train machine learning
models and visualization of the code. Overall, the use of these tools allowed for a streamlined
and efficient development process, resulting in a high-quality, user-friendly solution that met
the project's goals and objectives.

30 | P a g e
 Figure 14: Illustrations of tools

Visual Studio Code:

Visual Studio Code (VScode) is a powerful and versatile code editor that was chosen as the
primary development environment for the rice plant disease detection solution. It is equipped
with a wide range of features that make it an ideal choice for software development. Some of
the key features of VScode include its support for multiple programming languages,
debugging capabilities, and integration with Git for version control. The support for multiple
languages allows for the flexibility to work with different languages and frameworks. The
debugging capabilities allow for easy identification and resolution of bugs, which can save
time and effort during the development process. The integration with Git allows for easy
version control and management of the codebase, making it possible to track changes and
collaborate with others seamlessly. Overall, the use of VScode was instrumental in

31 | P a g e
facilitating an efficient and productive development process for the rice plant disease
detection solution.

Figure 15: Figure of VScode

Jupyter Notebook:
Jupyter Notebook is an open-source web-based interactive development environment (IDE)
that was used as a tool for prototyping, testing and documenting the code for the rice plant
disease detection solution. It supports multiple languages, including Python, R, and Julia,
making it a versatile tool for data analysis, scientific computing, and machine learning.
Jupyter Notebook allows for the creation of rich, interactive documents that combine code,
equations, visualizations, and narrative text. It also provides a platform for interactive data
visualization, and it has a great visualization capability, which makes it easier to understand
the data and the results of the models. Additionally, Jupyter Notebook has a great
collaboration feature, where it allows users to share their notebooks with others, and it also
has the ability to export the notebook as a python script, PDF, HTML, and other formats.
Overall, Jupyter Notebook was an essential tool for the development of the rice plant disease
detection solution, providing an interactive, efficient, and well-documented development
process.

32 | P a g e
 Figure 16: Illustration of Jupyter notebook

Figma:
Figma is a web-based design tool that was used for designing the user interface and user
experience for the rice plant disease detection solution. It allows for easy collaboration and
prototyping, which was essential in the development process. Figma allows multiple team
members to work on the same design file at the same time, and it has a wide range of design
elements and tools that can be used to create a high-quality and user-friendly interface. With
Figma, it's possible to create wireframes, mock-ups, and interactive prototypes, which helps
to test the design before implementation. Additionally, Figma offers a wide range of
collaboration features, such as commenting, version history, and permissions, making it easy
for my module teacher to review and provide feedback. Overall, Figma was a key tool in
ensuring that the rice plant disease detection solution had a well-designed, user-friendly
interface that met the project's goals and objectives.

33 | P a g e
 Figure 17: Illustration of Figma

Technology:
To develop the rice plant disease detection solution, a variety of technologies were used
including Python, TensorFlow, Pandas, Pillow, Matplotlib and Streamlit. Python, being a
general-purpose programming language, was used as the primary language for the project,
providing a wide range of libraries and frameworks to work with. Pandas, a library for data
manipulation and analysis, was used to clean, transform, and analyse the data, making it
easier to work with. Pillow, a library for image processing, was used to handle and
manipulate images, ensuring that the images were of high quality and suitable for the task of
disease detection. Matplotlib, a library for data visualization, was used to create visual
representations of the data, making it easier to understand and analyse.

34 | P a g e
 Figure 18: Illustrations of technology that is used to complete the project.

Python:
In the context of the rice plant disease detection solution, Python was used as the primary
programming language for the project. It provided a wide range of libraries and frameworks
that were used for various tasks such as data manipulation, data analysis, machine learning,
and building the user interface. The flexibility and simplicity of Python allowed for the easy
implementation of complex algorithms and techniques, such as those used for machine
learning, making it an ideal choice for this project. Additionally, the large and active
community of Python developers has created a vast number of libraries such as TensorFlow,
Pandas, Pillow, Matplotlib, Streamlit which were used in this project and made the
development process more efficient and streamlined. Python's popularity in the field of
scientific computing and data science also made it a suitable choice for this project.

35 | P a g e
TensorFlow:
In the context of the rice plant disease detection solution, TensorFlow was used to train a
model using the VGG19 algorithm of convolutional neural networks (CNN). The VGG19
algorithm is a pre-trained model that is trained on a large dataset of images and can recognize
features and patterns in images. By fine-tuning the VGG19 algorithm to the specific task of
rice plant disease detection, the model was able to learn the specific features and patterns of
the diseases and detect them with high accuracy. TensorFlow's powerful computation
capabilities and its ability to handle large amounts of data made it an ideal choice for training
this model.
Streamlit:
Streamlit is an open-source library that simplifies the process of building interactive web-
based applications in Python. It was used in this rice plant disease detection solution to create
an user-friendly interface for loading data, running the model, and displaying results in real-
time using the VGG19 algorithm of CNN. This made it easy for users to interact with the
solution and made the overall development process more efficient.
Keras:
Keras is a high-level neural networks API, written in Python and capable of running on top of
TensorFlow, CNTK, or Theano. It is designed to make it easy for developers to create and
experiment with different types of deep learning models. In this rice plant disease detection
solution, Keras was used as an API to create the deep learning model that was used to detect
rice plant diseases. Keras provides a simple and intuitive API for building and training neural
networks, which allowed for the creation of the deep learning model to be completed
efficiently.

Techniques
To complete this project, a comprehensive approach was taken to develop an accurate and
reliable solution for detecting different types of rice plant diseases. The project aimed to
utilize the latest advancements in machine learning to detect diseases in an efficient and
effective manner. The first step in the development process was data collection and pre-
processing. Images of various rice plant diseases were collected from Kaggle and labelled
manually using TensorFlow image pre-processing techniques. The data was then pre-
processed to prepare it for the model training. This included techniques such as image
resizing and normalization to ensure that the data was in the appropriate format for the model.
The data set consisted of a large number of images of various rice plant diseases, which were
used to train the model. There are 7 classes of images are available in the dataset among
which 6 classes contains different types of disease like blast, sheath blight, bacterial blight,
tungro, brown spot and leaf smut and one class contains datasets of healthy images.

36 | P a g e
 Figure 19: Conceptual page for rice plant disease detection

37 | P a g e
Next, exploratory data analysis (EDA) was performed on the collected data to understand the
distribution and characteristics of the data. This included visualizing the data using various
plots and charts, such as histograms, scatter plots, and bar charts. EDA helped to identify any
outliers or anomalies in the data, and to understand the relationship between different
features. EDA also helped in identifying which features were important for the model to
detect the diseases and which features were not. This helped to select the most relevant
features for the model, and to eliminate unnecessary features that could have increased the
complexity of the model without improving its performance. Additionally, EDA helped in
identifying the imbalance in the data, which can occur when some classes have more data
points than others. To handle this, techniques such as oversampling or under sampling were
applied to balance the data. The combination of EDA with the use of CNNs and the VGG19
algorithm, as well as the Agile methodology, helped to deliver a reliable and accurate
solution for detecting rice plant diseases.

Figure 20: Architecture of CNN algorithm

After that, feature extraction was performed using the technique of convolutional neural
networks (CNNs). This technique was used to extract the features from the images that were
important for the model to detect the diseases. The VGG19 architecture was chosen for this
project as it is known for its ability to perform well on image classification tasks. The model

38 | P a g e
was then trained using the VGG19 algorithm of CNN and the collected data. The model was
trained for several epochs, and the performance was evaluated using metrics such as
accuracy, precision and recall. Hyperparameter tuning was also performed to optimize the
model's performance. The model was then evaluated using techniques such as cross-
validation and testing on unseen data. The model performed well on these techniques and
provided good results. The model's performance was then compared with efficientNetV2B2
model of CNN (convolutional neural network), and it was found that our model performed
better.
After that, the model was deployed in an application or system for use by end-users. The user
interface was designed with the help of Streamlit, an open-source library that simplifies the
process of building interactive web-based applications in Python. The interface allowed users
to input data, run the model, and view the results in real-time, making it easy to interact with
the solution. Additionally, the model's performance was continuously monitored, and the
areas of improvement were identified and optimized. This helped to ensure that the solution
was accurate and reliable for end-users. The final testing and debugging were also performed
before delivering the solution to the end-user. Overall, the use of various tools and
technologies, such as Python, TensorFlow, Pandas, Pillow, Matplotlib, and Streamlit, helped
to develop a comprehensive and efficient solution for detecting rice plant diseases. The
deployment of the model in an interactive user interface, made it easy for end-users to use the
solution and monitor the performance. Overall, this project provides a valuable solution for
the agricultural industry, and the techniques used in this project can be applied to other
similar problems in the future.

User Experience (UX)

In the rice plant disease detection solution project, the user experience (UX) was a key
consideration throughout the development process. The user interface was designed with the
help of Streamlit, an open-source library that simplifies the process of building interactive
web-based applications in Python. The interface was designed to be intuitive and easy to use,
with a clear and simple layout, and user-friendly navigation. This helped to ensure that end-
users could easily input data, run the model, and view the results in real-time. Furthermore,
the interface was designed with accessibility in mind, meaning that it was designed to be
easily used by users with disabilities by using techniques such as proper labelling, alternative
text, and high contrast colour schemes. Overall, the use of Streamlit and the implementation
of responsive design and accessibility techniques helped to ensure that the solution was easy
to use and accessible for all end-users.

39 | P a g e
 Figure 21: user experience of the application
Availability, Backup, and recovery strategy
In this rice plant disease detection solution project, availability was a key consideration
throughout the development process. The solution was designed to be available for use on a
local host, which ensured that end-users could easily access it. It should be emphasized,
however, that this is a restriction of the present version of the solution, and that future
development will include deploying the solution in the cloud to increase end-user
accessibility and availability. Additionally, a backup and recovery strategy were implemented
to ensure that the data and solution were protected against any potential data loss. This
strategy included regular backups of the data and the application, as well as the
implementation of disaster recovery procedures to ensure that the solution could be quickly
restored in the event of an unexpected outage. This helped to ensure that the solution was
always available and that any potential data loss could be quickly and easily recovered.
Security
Security is a crucial consideration when developing any application, particularly one that
deals with sensitive information such as personal data. In this project, measures have been
taken to ensure the confidentiality, integrity, and availability of the data stored and processed
by the system. This includes implementing robust data encryption protocols, regularly
monitoring system logs for suspicious activity, and implementing access controls to limit
who can view or modify the data. Additionally, regular vulnerability scans and penetration
testing are conducted to identify and mitigate any potential security risks. It is also important
to ensure that the system is compliant with relevant regulations and industry standards. To
achieve this, regular audits are conducted to ensure that the system is adhering to best
practices in terms of data protection and privacy. Overall, the security of the system is of the
utmost importance and every effort is made to ensure that the data stored and processed by

40 | P a g e
the system is protected against unauthorized access, use, disclosure, disruption, modification,
or destruction.
Updates and Upgrades
To ensure the continued effectiveness and security of the application, regular upgrades and
updates will be implemented. This includes updating the underlying software and libraries, as
well as incorporating new features and improvements suggested by end-users and identified
through testing and monitoring. Additionally, any security vulnerabilities or bugs that are
discovered will be promptly addressed through patches and updates. A comprehensive testing
process will be implemented prior to any updates being rolled out to ensure that the
application remains stable and reliable for end-users. A backup and recovery strategy will
also be put in place to ensure that any data loss or unexpected downtime can be quickly and
efficiently resolved. Regular communication with end-users will be maintained to keep them
informed of any planned upgrades or updates and to gather feedback on the application's
performance.

Findings
1. What are the most effective machine learning algorithms and techniques for the rice
plant disease detection?

41 | P a g e
In this report, a thorough examination was conducted to identify the most effective machine
learning algorithms and techniques for the task of rice plant disease detection. Through a
literature review and examination of the current state of technology, it was determined that
convolutional neural networks (CNNs) were the ideal approach for this problem. To further
narrow down the appropriate model, a comparison was made between two popular CNN
architectures: VGG19 and EfficientNetV2B2. Both architectures have been widely used in
image classification tasks and have a reputation for delivering strong performance. Results of
the comparison showed that VGG19 provided a higher accuracy percentage of 97%
compared to EfficientNetV2B2, which had an accuracy percentage of 94%. It is worth noting
that this comparison was made on a specific dataset and using a specific set of
hyperparameters. It is possible that other versions of the EfficientNet architecture, such as
EfficientNet-B3 and EfficientNet-B4, may yield better results when tested on this task.
Additionally, other machine learning techniques such as transfer learning or data
augmentation could also be considered to enhance the performance of the model.

Figure 22: Exploratory data analysis of the image dataset

42 | P a g e
 It is crucial to consider the specific characteristics of the dataset and the problem at hand
when selecting the most appropriate algorithm and techniques for rice plant disease detection.
The choice of architecture should not be based solely on accuracy percentage, but also other
factors such as computational resources, speed, and memory usage. It is also important to
evaluate the trade-off between accuracy and computational resources. In this project, VGG19
was chosen as it provided the highest accuracy percentage while still being computationally
efficient.

Figure 23: Performance matrices of VGG19 and EfficientNetV2B2

43 | P a g e
Furthermore, this research showed that CNNs are a suitable approach for the task of rice plant
disease detection, and that VGG19 is a suitable CNN architecture for this task. However, it is
important to note that the effectiveness of the model can also be improved by using
techniques such as hyperparameter tuning and optimization. Additionally, it is essential to
consider the ethical challenges involved in developing a rice plant disease detection solution
that also recommends solutions. These challenges include ensuring the accuracy and fairness
of the model, avoiding bias in the training data, and considering the potential consequences of
the solution such as its impact on the livelihoods of farmers and the availability of resources.
These challenges can be addressed effectively by implementing robust data protection
measures, regularly reviewing and testing the model for bias, and being transparent about the
decision-making process and potential consequences of the solution. Collaborating with
experts and stakeholders in relevant fields, such as agriculture and ethics, can also help
ensure that the solution is both effective and ethical. Lastly, continuously monitoring and
evaluating the solution to ensure that it is not causing any negative impacts and making
necessary adjustments is also important.

2. How is the chosen CNN algorithm utilized in the application contribute to its
efficiency and effectiveness in identifying rice plant diseases?
In this rice plant disease detection solution project, the chosen convolutional neural network
(CNN) algorithm, VGG19, was effectively utilized in the application built using Streamlit.
The VGG19 architecture was selected for its exceptional performance on image classification
tasks and its capability to extract essential features from the images that were critical for the
model to detect the diseases. The model was trained using the VGG19 algorithm and a large
dataset of images of various rice plant diseases, which helped to achieve high accuracy in
detecting the diseases. Once the model was trained, it was deployed in the application built
using Streamlit, making it convenient for end-users to interact with the solution. The user
interface was designed to be intuitive and user-friendly, with a clear and simple layout, and
easy navigation. This helped to ensure that end-users could easily input data, run the model,
and view the results in real-time. Additionally, the interface was designed to be responsive,
ensuring that it was accessible on various devices and screen sizes.

44 | P a g e
 Figure 24: Grad camera visualization of input images of rice plant leaves
The VGG19 algorithm was then utilized in the application to run the model on the input data
provided by the user, and to display the results in real-time. This allowed the end-users to
easily identify the diseases in the rice plants and to get accurate results. Furthermore, the use
of Streamlit also helped in creating a user-friendly interface, making it easy for even non-
technical users to use the application with ease.

45 | P a g e
 Figure 25: Figure shows the result of the inputted image.
Overall, the chosen CNN algorithm, VGG19, was effectively utilized in the rice plant disease
detection application built using Streamlit. The combination of the VGG19 algorithm and
Streamlit helped to create an efficient and user-friendly solution for detecting rice plant
diseases, providing accurate results to the end-users. The high accuracy percentage of
VGG19 and its computational efficiency made it a suitable choice for this project.
Furthermore, the use of Streamlit for creating the user interface made it convenient for the
end-users to interact with the solution.

46 | P a g e
3. What are the ethical challenges that need to be considered when developing
application for rice plant disease detection that also recommends solutions, and how
can they be addressed effectively?

The development of an application for rice plant disease detection that also recommends
solutions poses several ethical challenges that need to be considered. One of the main
challenges is ensuring that the recommendations provided by the application are accurate and
reliable, and that they do not cause harm to the rice plants or the environment. This includes
ensuring that the application does not recommend the use of harmful pesticides or chemicals,
and that it does not recommend solutions that are not appropriate for the specific type of
disease or growing conditions. Another ethical challenge is ensuring that the application is
accessible and fair to all users, regardless of their location, resources, or technical expertise.
This includes ensuring that the application is available in multiple languages, and that it is
designed to be easy to use and understand, even for non-technical users. It also includes
ensuring that the application is affordable and available to users in developing countries, who
may not have access to the same resources and technologies as users in developed countries.

Figure 26: Ethical challenges to be considered during this project completion

To address these ethical challenges effectively, it is important to involve a diverse group of

stakeholders in the development process, including farmers, scientists, and experts in ethics
and social responsibility. This will help to ensure that the application is developed with the

47 | P a g e
needs and perspectives of all users in mind, and that it considers the potential impact of the
recommendations on the environment and local communities. Additionally, it is important to
conduct thorough testing and evaluation of the application, to ensure that it is reliable,
accurate and does not cause any harm. Regular monitoring and updating of the application
are also important to ensure that it remains relevant and effective over time.
To sum up, the development of an application for rice plant disease detection that also
recommends solutions poses several ethical challenges that need to be considered. These
challenges can be effectively addressed by involving a diverse group of stakeholders in the
development process, and by conducting thorough testing and evaluation of the application.
Regular monitoring and updating of the application are also important to ensure that it
remains relevant and effective over time.

Recommendations and future work

In this rice plant disease detection solution project, there are several areas for future work that
could be pursued to further improve the solution. One potential area of improvement could be
to explore other machine learning algorithms and techniques, such as transfer learning or data
augmentation, to see if they could improve the performance of the model. Additionally, other
CNN architectures like EfficientNet-B3 and EfficientNet-B4 could be tested to see if they
provide better results than VGG19. Another area of improvement could be to deploy the
solution in the cloud, which would make it more accessible to a wider range of users. This
would also enable the use of more powerful computational resources, which could further
improve the performance of the model.
Furthermore, it is important to consider the ethical challenges that arise when developing a
solution that also recommends solutions. For example, it is important to ensure that the
recommended solutions are safe, effective, and affordable. Additionally, it is important to
consider the impact of the solution on the environment and on local communities. It is also
important to ensure that the solution is not used for discriminatory or exploitative purposes.
To address these challenges, it is important to involve stakeholders such as farmers,
researchers, and policymakers in the development process, and to conduct thorough testing
and evaluation of the solution before deployment. Additionally, this rice plant disease
detection solution project has provided a reliable and accurate method for detecting rice plant
diseases. While the solution provides good performance, there are several areas for future
work that could be pursued to further improve the solution. Additionally, it is important to
consider the ethical challenges that arise when developing a solution that also recommends
solutions.

48 | P a g e
Conclusion
In conclusion, the rice plant disease detection solution project was developed using a
comprehensive approach that included the latest advancements in machine learning. The use
of convolutional neural networks (CNNs) for feature extraction and the VGG19 algorithm for
training the model, helped to achieve high accuracy in detecting the diseases. The exploratory
data analysis (EDA) helped to understand the distribution and characteristics of the data and
also helped in identifying which features were important for the model to detect the diseases
and which features were not. This helped to select the most relevant features for the model,
and to eliminate unnecessary features that could have increased the complexity of the model
without improving its performance. Additionally, EDA helped in identifying the imbalance in
the data, which can occur when some classes have more data points than others. To handle
this, techniques such as oversampling or under sampling were applied to balance the data.
The project was developed using Agile methodology, which helped to manage the
development process efficiently. The data was collected and pre-processed to prepare it for
the model training, which included techniques such as image resizing and normalization.
Feature extraction was performed using the technique of convolutional neural networks
(CNNs) which helped to extract the features from the images that were important for the
model to detect the diseases. The model was then trained using the VGG19 algorithm of
CNN and the collected data. The model was trained for several epochs and the performance
was evaluated using metrics such as accuracy, precision, and recall. Hyperparameter tuning
was also performed to optimize the model's performance. In terms of performance, the
VGG19 algorithm was found to provide the highest accuracy percentage of 97% compared to
the EfficientNetV2B2 algorithm, which had an accuracy percentage of 94%. However, it is
important to note that this comparison was made on a specific dataset and using a specific set
of hyperparameters. The other versions of the EfficientNet architecture such as EfficientNet-
B3 and EfficientNet-B4, may achieve better results when tested on this task. Additionally,
other machine learning techniques such as transfer learning or data augmentation could also
be considered to improve the performance of the model.
The model was then deployed in an application or system for use by end-users, which was
built using Streamlit, an open-source library that simplifies the process of building interactive
web-based applications in Python. The user interface was designed to be intuitive and easy to
use, with a clear and simple layout and user-friendly navigation, which helped to ensure that
end-users could easily input data, run the model, and view the results in real-time.
Additionally, the interface was designed to be responsive, ensuring that it was accessible on
various devices and screen sizes.
In the future, the model can be further improved by fine-tuning the hyperparameters, adding
more data to the dataset, and by implementing other machine learning techniques such as
transfer learning or data augmentation. The model can also be deployed on cloud platforms to
make it more accessible to a larger audience. Furthermore, the application could be integrated
with an IoT sensor network to collect data from the field in real-time, which would help to
detect the diseases at an early stage. Additionally, more research can be done on the ethical

49 | P a g e
considerations that need to be considered when developing an application for rice plant
disease detection that also recommends solutions.
Finally, this solution provides a reliable and accurate method for detecting rice plant diseases
and is expected to have a significant impact in the agricultural industry. It can help farmers to
identify the diseases in their rice plants at an early stage and take the necessary precautions to
prevent the further spread of the disease. This can lead to an increase in crop yields and a
reduction in losses for farmers.

50 | P a g e
REFERENCES
Beyeler, M. (2019). Machine learning for opencv = opencv Ji Qi Xue Xi. Dong nan da xue
chu ban she.

Goodfellow, I., Bengio, Y., & Courville, A. (2018). Deep learning. MITP.

Han, C. H., Kim, E., Doan, T. N., Han, D., Yoo, S. J., & Kwak, J. T. (2021). Region-
aggregated attention CNN for disease detection in fruit images. PLOS ONE, 16(10).
https://doi.org/10.1371/journal.pone.0258880

Murphy, K. P. (2021). Machine learning: A probabilistic perspective. MIT Press.

Nasser, A. A., & Akhloufi, M. A. (2022). CTPlantNet: A hybrid CNN-transformer

architecture for Plant Disease Classification. 2022 International Conference on
Microelectronics (ICM), 1–22. https://doi.org/10.1109/icm56065.2022.10005433

Scott, G. (2018). Google. In The four: The hidden DNA of Amazon, Apple, Facebook, and
google (pp. 115–138). essay, Corgi Books.

Acta Scientific. (n.d.). Retrieved January 15, 2023, from https://actascientific.org/

Choudhary, D., Banskota, K., Khanal, N. P., McDonald, A. J., Krupnik, T. J., &
Erenstein, O. (2021, December 6). Rice subsector development and Farmer Efficiency
in Nepal: Implications for further transformation and food security. Frontiers. Retrieved
January 16, 2023, from
https://www.frontiersin.org/articles/10.3389/fsufs.2021.740546/full #:~:text=Nepal
%20ranks%2017th%20in%20rice,with%20the%20Government%20of%20Nepal.

Metz, C., & Boudette, N. E. (2021, December 6). Inside Tesla as elon musk pushed an
unflinching vision for self-driving cars. The New York Times. Retrieved January 20,
2023, from https://www.nytimes.com/2021/12/06/technology/tesla-autopilot-elon-
musk.html

The Kathmandu Post. (n.d.). Paddy farmers boost yield by using system of rice
intensification. The Kathmandu Post. Retrieved January 18, 2023, from
https://kathmandupost.com/money/2021/07/02/paddy-farmers-boost-yield-by-using-
system-of-rice-intensification

Krishna Dev Joshi and Santosh Upadhaya. (2020, June 24). The rise of Rice in Nepal. Nepali
Times. Retrieved January 19, 2023, from https://www.nepalitimes.com/banner/the-rise-
of-rice-in-nepal/

51 | P a g e
Nepal Rice, paddy production, 1961-2022. Knoema. (n.d.). Retrieved January 5, 2023, from
https://knoema.com/atlas/Nepal/topics/Agriculture/Crops-Production-Quantity-
tonnes/Rice-paddy-production

Nepal. International Rice Research Institute. (n.d.). Retrieved January 10, 2023, from
https://www.irri.org/where-we-work/countries/nepal

Springer. (1981). Lecture notes in computer science.

Wang, X., Dong, C., & Shan, Y. (2022). RepSR: Training efficient VGG-style super-
resolution networks with structural re-parameterization and batch normalization.
Proceedings of the 30th ACM International Conference on Multimedia.
https://doi.org/10.1145/3503161.3547915

Bernard Marr (2021) How Tesla Is Using Artificial Intelligence to Create the Autonomous
Cars of the Future [online]available from https://bernardmarr.com/how-tesla-is-using-
artificial-intelligence-to-create-the-autonomous-cars-of-the-future/ [18 January 2023]

Rangaiah, M. (n.d.) How Tesla Is Making Use of Artificial Intelligence in Its Operations?
[online]available from https://www.analyticssteps.com/blogs/how-tesla-making-use-
artificial-intelligence-its-operations [18 January 2023]

Julian, Richard, and Jerry Wallis (2022) The Tech behind Tesla's Electric Vehicles
[online]available from https://webo.digital/blog/the-tech-behind-tesla-electric-vehicles/
[9 January 2023]

Cohen, J. (2022) Computer Vision at Tesla for Self-Driving Cars [online]available from
https://www.thinkautonomous.ai/blog/computer-vision-at-tesla/ [12 January 2023]

Bouchard, Y. (2019) Tesla's Computer Vision Master Plan [online]available from

https://medium.com/@strangecosmos/teslas-computer-vision-master-plan-
512b36d8acbf [8 January 2023]

Sagar, R. (2020) How Tesla Uses Pytorch [online]available from

https://analyticsindiamag.com/tesla-pytorch-self-driving-computer-vision-karpathy-
elon-musk-ai/ [1 January 2023]

Sharma, M. and Kumar, C.J. (2021) Early Diagnosis of Rice Plant Disease Using Machine
Learning Techniques [online]available from
https://www.tandfonline.com/doi/abs/10.1080/03235408.2021.2015866?cookieSet=1
[6 January 2023]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [3 January 2023]

52 | P a g e
Bouchard, Y. (2019) Tesla's Deep Learning at Scale: Using Billions of Miles to Train Neural
Networks [online]available from https://towardsdatascience.com/teslas-deep-learning-
at-scale-7eed85b235d3 [18 January 2023]

Tawde, T., Deshmukh, K., Verekar, L., Reddy, A., Aswale, S., and Shetgaonkar, P. (2021)
Rice Plant Disease Detection and Classification Techniques : A Survey
[online]available from https://www.ijert.org/rice-plant-disease-detection-and-
classification-techniques-a-survey [5 January 2023]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., & Qi, L. (2021, July 20). Automatic
diagnosis of rice diseases using Deep Learning. Frontiers. Retrieved January 20, 2023,
from https://www.frontiersin.org/articles/10.3389/fpls.2021.701038/full

Theonly, N. (n.d.) Theonlynischal/Identification-of-Rice-Leaf-Diseases-Using-Deep-

Convolutional-Neural-Networks: Identification of Rice Leaf Diseases Using Deep
Convolutional Neural Networks [online]available from
https://github.com/theonlyNischal/Identification-of-Rice-Leaf-Diseases-Using-Deep-
Convolutional-Neural-Networks [11 January 2023]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [10 January 2023]

Publication, I.A.E.M.E. (2021)

Https://Iaeme.com/MasterAdmin/Journal_uploads/IJARET/VOLUME_12_ISSUE_1/
IJARET_12_01_042.Pdf [online]available from
https://www.academia.edu/49257839/https_iaeme_com_MasterAdmin_Journal_upload
s_IJARET_VOLUME_12_ISSUE_1_IJARET_12_01_042_pdf [15 December 2022]

Grow more, grow better (2021) Fasal. Available at: https://fasal.co/ (Accessed: January 14,
2023).

Sagar, R. (2020) These Machine Learning Techniques Make Google Lens a Success
[online]available from https://analyticsindiamag.com/these-machine-learning-
techniques-make-google-lens-a-success/ [30 December 2022]

Shroff, M. (2022) What Is Google Lens & How Does It Work? - Let's Find Out
[online]available from https://iide.co/blog/google-lens-turn-your-camera-into-a-search-
box/ [30 December 2022]

Dr. A.Anushya (2019) Google Lens as an Image Classifier - Ijsrcsams.com [online]available

from https://www.ijsrcsams.com/images/stories/Past_Issue_Docs/ijsrcsamsv8i6p2.pdf
[29 December 2022]

Ali, J. (2021, May 5). Rice improvement: Physiological, molecular breeding and genetic
perspectives. Google Books. Retrieved January 20, 2023, from
https://books.google.com/books/about/Rice_Improvement.html?id=9OAsEAAAQBAJ

53 | P a g e
Beyeler, M. (2019). Machine learning for opencv = opencv Ji Qi Xue Xi. Dong nan da xue
chu ban she.

Goodfellow, I., Bengio, Y., & Courville, A. (2018). Deep learning. MITP.

Han, C. H., Kim, E., Doan, T. N., Han, D., Yoo, S. J., & Kwak, J. T. (2021). Region-
aggregated attention CNN for disease detection in fruit images. PLOS ONE, 16(10).
https://doi.org/10.1371/journal.pone.0258880

Murphy, K. P. (2021). Machine learning: A probabilistic perspective. MIT Press.

Nasser, A. A., & Akhloufi, M. A. (2022). CTPlantNet: A hybrid CNN-transformer

architecture for Plant Disease Classification. 2022 International Conference on
Microelectronics (ICM), 1–22. https://doi.org/10.1109/icm56065.2022.10005433

Mehta, M. (2021, April 10). Desk research: Complete guide & best practices | similarweb.
Retrieved January 20, 2023, from https://www.similarweb.com/blog/research/market-
research/desk-research/

Scott, G. (2018). Google. In The four: The hidden DNA of Amazon, Apple, Facebook, and
google (pp. 115–138). essay, Corgi Books.

Springer. (1981). Lecture notes in computer science.

Wang, X., Dong, C., & Shan, Y. (2022). RepSR: Training efficient VGG-style super-
resolution networks with structural re-parameterization and batch normalization.
Proceedings of the 30th ACM International Conference on Multimedia.
https://doi.org/10.1145/3503161.3547915

Westcott, C. (2013). In Westcott's Plant Disease Handbook (pp. 10–25). essay, Springer
Reference.

Poorna Priya, D.T.S. and Gopinath, D.R. (2020) Iaeme.com. Available at:
https://iaeme.com/MasterAdmin/Journal_uploads/IJEET/VOLUME_11_ISSUE_10/
IJEET_11_10_037.pdf (Accessed: December 29, 2022).

Prettyish daggered Follow (n.d.) Google Lens [online]available from

<https://www.slideshare.net/prathyushaduggireddy/google-lens-91443526> [30
December 2022]

Deka, A. and Sharma, M. (2021) Early Diagnosis of Rice Plant Disease Using Machine
Learning Techniques [online]available from
https://www.tandfonline.com/doi/abs/10.1080/03235408.2021.2015866?cookieSet=1
[30 December 2022]

54 | P a g e
Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [30 December 2022]

Tawde, T., Deshmukh, K., Verekar, L., Reddy, A., Aswale, S., and Shetgaonkar, P. (2021)
Rice Plant Disease Detection and Classification Techniques : A Survey
[online]available from https://www.ijert.org/rice-plant-disease-detection-and-
classification-techniques-a-survey [30 December 2022]

Anon. (2020) Agriculture [online]available from https://www.universeoptics.com/industries-

served/agriculture/ [30 December 2022]

R, P. and S, V.C. (2022) “Rice Leaf Disease Detection and Diagnosis Using Convolution
Neural Network”. Rice Leaf Disease Detection and Diagnosis Using Convolution
Neural Network

Sharma, P. (2022) “Rice Plant Diseases Identification Using Similar Image Analysis in Deep
Unsupervised Learning”. ECS Transactions107 (1), 4615–4624

Patel, S. (2020) Leaf Disease Detection Using Convolutional Neural Networ

[online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [29 December 2022]

Karwande, A. and Kulkarni, P. (2021) Plant Disease Detection Using Image Processing and
Machine Learning [online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [30 December 2022]

Aalam, M.S. (2019) Diagnosis of Common Diseases of Rice [online]available from

https://answers.microsoft.com/en-us/windows/forum/all/filecusersjeandownloadsmpcm
drun203log/c4323a1b-cd75-4b99-ae8e-854cdbace66a [30 December 2022]

Reddy, C.M., Shetty, A., and Sunagar, L. (2020) PLANT DISEASE DETECTION USING
MACHINE LEARNING [online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [29 December 2022]

Chen, Y. and Yang , J. (2022) Detection of Weeds Growing in Alfalfa Using Convolutional
Neural Networks [online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [29 December 2022]

55 | P a g e
Desk Research - Methodology and Techniques. Desk research - methodology and
Techniques. (2020, May 10). Retrieved January 2, 2023, from
https://www.managementstudyguide.com/desk-research.htm

Stella, I.O. and Ogechukwu Nwazor , N. (2022) Plant Disease Detection and Monitoring
Using Artificial Neural Network [online]available from
https://www.researchgate.net/publication/357824814_Plant_Disease_Detection_and_M
onitoring_Using_Artificial_Neural_Network [29 December 2022]

Islam, M.A. and Shuvo, M.N.R. (2021) An Automated Convolutional Neural Network Based
Approach for Paddy Leaf ... [online]available from
&lt;https://www.researchgate.net/profile/Md-Islam-1214/publication/
348966329_An_Automated_Convolutional_Neural_Network_Based_Approach_for_Paddy_
Leaf_Disease_Detection/links/601bcedc92851c4ed5494e19/An-Automated-Convolutional-
Neural-Network-Based-Approach-for-Paddy-Leaf-Disease-Detection.pdf&gt> [29
December 2022]

Mohan, K.J., M. Balasubramaniam, and S. Planivel (2016) Volume 144 [online]available

from https://www.ijcaonline.org/archives/volume144/ [29 December 2022]

S. Ramesh (2019) Recognition and Classification of Paddy Leaf Diseases Using Optimized
Deep Neural Network with Jaya Algorithm [online]available from
https://www.sciencedirect.com/science/article/pii/S2214317319300769 [29 December
2022]

Biswas, M., Faruqui, N., and Akter Lata, M. (2019) A Novel Inspection of Paddy Leaf
Disease Classification Using Advance ... [online]available from
https://www.researchgate.net/publication/336702208_A_Novel_Inspection_of_Paddy_
Leaf_Disease_Classification_using_Advance_Image_Processing_Techniques [29
December 2022]

Bagde, S. and Patil, P. (2015) Artificial Neural Network Based Plant Leaf Disease Detection
[online]available from https://www.semanticscholar.org/paper/Artificial-Neural-
Network-Based-Plant-Leaf-Disease-Bagde-Patil/
408dfc59753df4fc987c543461d5cdad31756fbe [29 December 2022]

Hossain, M.S. and Muhammad , S. (2021) An Automated Convolutional Neural Network

Based Approach for Paddy Leaf ... [online]available from
https://www.researchgate.net/profile/Md-Islam-1214/publication/348966329_An_Auto
mated_Convolutional_Neural_Network_Based_Approach_for_Paddy_Leaf_Disease_D
etection/links/601bcedc92851c4ed5494e19/An-Automated-Convolutional-Neural-
Network-Based-Approach-for-Paddy-Leaf-Disease-Detection.pdf [29 December 2022]

56 | P a g e
Microsoft (2021) Visual studio code - code editing. redefined, RSS. Microsoft. Available at:
https://code.visualstudio.com/ (Accessed: January 10, 2023). Customer feedback tool
(no date) Convas. Available at: https://convas.io/ (Accessed: January 14, 2023).

The Collaborative Interface Design Tool. (no date) Figma. Available at:
http://www.figma.com/ (Accessed: January 9, 2023). Pytorch (no date) PyTorch.
Available at: https://pytorch.org/ (Accessed: January 8, 2023).

Liu1, S. (2021) IOPscience [online]available from

https://iopscience.iop.org/article/10.1088/1755-1315/692/2/022103 [18 January 2023]

Tensorflow Federated (no date) TensorFlow. Available at:

https://www.tensorflow.org/federated (Accessed: January 14, 2023).

Neptune (2022) NASA. NASA. Available at:

https://solarsystem.nasa.gov/planets/neptune/overview/ (Accessed: January 4, 2023).

Project jupyter (no date) Project Jupyter. Available at: https://jupyter.org/ (Accessed:
November 18, 2022).

Industry-leading vector graphics software | adobe illustrator (no date). Available at:
https://www.adobe.com/products/illustrator.html (Accessed: December 18, 2022).

Streamlit • the fastest way to build and share data apps (no date). Available at:
https://streamlit.io/ (Accessed: January 2, 2023).

(N.d) Google. Google. Available at: https://firebase.google.com/ (Accessed: October 18,

2022).

CNN International - Breaking News, US News, World News and Video (2014) CNN. Cable
News Network. Available at: https://edition.cnn.com/ (Accessed: January 1, 2023).

Papers with code - EFFICIENTNETV2 explained (no date) Explained | Papers With Code.
Available at: https://paperswithcode.com/method/efficientnetv2 (Accessed: December
19, 2023).

Support Vector Machine (SVM) algorithm - javatpoint (no date) www.javatpoint.com.

Available at: https://www.javatpoint.com/machine-learning-support-vector-machine-
algorithm (Accessed: November 14, 2023).

Let's build from here (no date) GitHub. Available at: https://github.com/ (Accessed:
December 4, 2023).

Artificial Intelligence (2023) Wikipedia. Wikimedia Foundation. Available at:

https://en.wikipedia.org/wiki/Artificial_intelligence (Accessed: January 9, 2023).

57 | P a g e
Cloud, computers, Apps & Gaming (no date) Microsoft. Available at:
https://www.microsoft.com/en-us/ (Accessed: January 11, 2023).

Team, K. (no date) Simple. flexible. powerful., Keras. Available at: https://keras.io/
(Accessed: December 11, 2023).

Welcome to Python.org (no date) Python.org. Available at: https://www.python.org/

(Accessed: January 12, 2023).

Tensorflow (no date) TensorFlow. Available at: https://www.tensorflow.org/ (Accessed:

January 13, 2023).

Matlab (no date) MathWorks. Available at:

https://www.mathworks.com/products/matlab.html (Accessed: January 14, 2023).

Visualization with python (no date) Matplotlib. Available at: https://matplotlib.org/

(Accessed: January 15, 2023).

Pandas (no date) pandas. Available at: https://pandas.pydata.org/ (Accessed: December 12,
2022).

Team, K. (no date) Simple. flexible. powerful., Keras. Available at: https://keras.io/
(Accessed: November 11, 2022).

Deep Network designer (no date) VGG-19 convolutional neural network - MATLAB.
Available at: https://www.mathworks.com/help/deeplearning/ref/vgg19.html
(Accessed: November 12, 2022).

Let's build from here (no date) GitHub. Available at: https://github.com/ (Accessed:
December 4, 2023).

Artificial Intelligence (2023) Wikipedia. Wikimedia Foundation. Available at:

https://en.wikipedia.org/wiki/Artificial_intelligence (Accessed: January 9, 2023).

Project jupyter (no date) Project Jupyter. Available at: https://jupyter.org/ (Accessed:
November 18, 2022).

Pandas (no date) pandas. Available at: https://pandas.pydata.org/ (Accessed: October 13,
2022).

Pillow (no date) Pillow (PIL Fork). Available at: https://pillow.readthedocs.io/en/stable/

(Accessed: September 12, 2022).

Bernard Marr (2021) How Tesla Is Using Artificial Intelligence to Create the Autonomous
Cars of the Future [online]available from https://bernardmarr.com/how-tesla-is-using-
artificial-intelligence-to-create-the-autonomous-cars-of-the-future/ [18 January 2023]

58 | P a g e
Deka, A. and Sharma, M. (2021) Early Diagnosis of Rice Plant Disease Using Machine
Learning Techniques [online]available from
https://www.tandfonline.com/doi/abs/10.1080/03235408.2021.2015866?cookieSet=1
[30 December 2022]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [30 December 2022]

Tawde, T., Deshmukh, K., Verekar, L., Reddy, A., Aswale, S., and Shetgaonkar, P. (2021)
Rice Plant Disease Detection and Classification Techniques : A Survey
[online]available from https://www.ijert.org/rice-plant-disease-detection-and-
classification-techniques-a-survey [30 December 2022]

Anon. (2020) Agriculture [online]available from https://www.universeoptics.com/industries-

served/agriculture/ [30 December 2022]

R, P. and S, V.C. (2022) “Rice Leaf Disease Detection and Diagnosis Using Convolution
Neural Network”. Rice Leaf Disease Detection and Diagnosis Using Convolution
Neural Network

Sharma, P. (2022) “Rice Plant Diseases Identification Using Similar Image Analysis in Deep
Unsupervised Learning”. ECS Transactions107 (1), 4615–4624

Patel, S. (2020) Leaf Disease Detection Using Convolutional Neural Networ

[online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [29 December 2022]

Karwande, A. and Kulkarni, P. (2021) Plant Disease Detection Using Image Processing and
Machine Learning [online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [30 December 2022]

Aalam, M.S. (2019) Diagnosis of Common Diseases of Rice [online]available from

https://answers.microsoft.com/en-us/windows/forum/all/filecusersjeandownloadsmpcm
drun203log/c4323a1b-cd75-4b99-ae8e-854cdbace66a [30 December 2022]

Reddy, C.M., Shetty, A., and Sunagar, L. (2020) PLANT DISEASE DETECTION USING
MACHINE LEARNING [online]available from
https://answers.microsoft.com/en-us/windows/forum/all/cusersusernamedesktop-is-not-
accessible/69860653-d076-4c62-aa3d-5cce22f838e3 [29 December 2022]

59 | P a g e
Rangaiah, M. (n.d.) How Tesla Is Making Use of Artificial Intelligence in Its Operations?
[online]available from https://www.analyticssteps.com/blogs/how-tesla-making-use-
artificial-intelligence-its-operations [18 January 2023]

Julian, Richard, and Jerry Wallis (2022) The Tech behind Tesla's Electric Vehicles
[online]available from https://webo.digital/blog/the-tech-behind-tesla-electric-vehicles/
[9 January 2023]

Cohen, J. (2022) Computer Vision at Tesla for Self-Driving Cars [online]available from
https://www.thinkautonomous.ai/blog/computer-vision-at-tesla/ [12 January 2023]

Bouchard, Y. (2019) Tesla's Computer Vision Master Plan [online]available from

https://medium.com/@strangecosmos/teslas-computer-vision-master-plan-
512b36d8acbf [8 January 2023]

Sagar, R. (2020) How Tesla Uses Pytorch [online]available from

https://analyticsindiamag.com/tesla-pytorch-self-driving-computer-vision-karpathy-
elon-musk-ai/ [1 January 2023]

Sharma, M. and Kumar, C.J. (2021) Early Diagnosis of Rice Plant Disease Using Machine
Learning Techniques [online]available from
https://www.tandfonline.com/doi/abs/10.1080/03235408.2021.2015866?cookieSet=1
[6 January 2023]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [3 January 2023]

Bouchard, Y. (2019) Tesla's Deep Learning at Scale: Using Billions of Miles to Train Neural
Networks [online]available from https://towardsdatascience.com/teslas-deep-learning-
at-scale-7eed85b235d3 [18 January 2023]

Tawde, T., Deshmukh, K., Verekar, L., Reddy, A., Aswale, S., and Shetgaonkar, P. (2021)
Rice Plant Disease Detection and Classification Techniques : A Survey
[online]available from https://www.ijert.org/rice-plant-disease-detection-and-
classification-techniques-a-survey [5 January 2023]

Theonly, N. (n.d.) Theonlynischal/Identification-of-Rice-Leaf-Diseases-Using-Deep-

Convolutional-Neural-Networks: Identification of Rice Leaf Diseases Using Deep
Convolutional Neural Networks [online]available from
https://github.com/theonlyNischal/Identification-of-Rice-Leaf-Diseases-Using-Deep-
Convolutional-Neural-Networks [11 January 2023]

Deng, R., Tao, M., Xing, H., Yang, X., Liu, C., Liao, K., and Qi, L. (2021) Automatic
Diagnosis of Rice Diseases Using Deep Learning [online]available from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416767/ [10 January 2023]

Publication, I.A.E.M.E. (2021)

Https://Iaeme.com/MasterAdmin/Journal_uploads/IJARET/VOLUME_12_ISSUE_1/
IJARET_12_01_042.Pdf [online]available from

60 | P a g e
 https://www.academia.edu/49257839/https_iaeme_com_MasterAdmin_Journal_upload
s_IJARET_VOLUME_12_ISSUE_1_IJARET_12_01_042_pdf [15 December 2022]

Grow more, grow better (2021) Fasal. Available at: https://fasal.co/ (Accessed: January 14,
2023).

Sagar, R. (2020) These Machine Learning Techniques Make Google Lens a Success
[online]available from https://analyticsindiamag.com/these-machine-learning-
techniques-make-google-lens-a-success/ [30 December 2022]

Shroff, M. (2022) What Is Google Lens & How Does It Work? - Let's Find Out
[online]available from https://iide.co/blog/google-lens-turn-your-camera-into-a-search-
box/ [30 December 2022]

Machine learning in your database using SQL (no date) Machine Learning In Your Database
Using SQL. Available at: https://mindsdb.com/?
gclid=CjwKCAiAzp6eBhByEiwA_gGq5MyfDEyIZg7wNRjCQiBe1
L4y8xvTtrInEqSfDT3cx8h7N1MTTRYOqhoCuOMQAvD_BwE (Accessed:
December 22, 2022).

Digital Image Processing Basics (2021) GeeksforGeeks. GeeksforGeeks. Available at:

https://www.geeksforgeeks.org/digital-image-processing-basics/ (Accessed: December
25, 2022).

Anon. (n.d.) AI [online]available from https://www.tesla.com/AI [10 January 2023]

61 | P a g e
Appendix

Gantt chart

Figure 27: project plan one

62 | P a g e
 Figure 28: Project plan two

63 | P a g e
 Figure 29: Project plan three

64 | P a g e
Project plan

Figure 30: charts of project plan

Issues Logs and Risk Management

Figure 31: Tables of issues logs

65 | P a g e
 Figure 32: Illustration of risk management

66 | P a g e
Supporting Diagrams
Model Training (VGG19 and EfficientNetV2B2)

67 | P a g e
68 | P a g e
69 | P a g e
70 | P a g e
71 | P a g e
72 | P a g e
73 | P a g e
Comparison loss and accuracy between two CNN architecture

74 | P a g e
Application Testing

75 | P a g e
76 | P a g e
Links
GitHub Link: https://github.com/Ramudgar/PlantDiseaseDetection-Rice-

Video Link: https://youtu.be/SyClbnDkmeE

77 | P a g e