3rd Presentation Final

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 22

Plant Disease Classification and Detection

SUPERVISOR
Dr. Asifullah Khan

Dr. Tauseef Jamal

Noman Ahmed (Phd Scholar)


Final Year Project (FYP) Presentation - 2023

PRESENTED BY
Ahmed Faraz & Usama Hafeez
Department of Computer and Information Sciences
CONTENTS
1. Problem Statement 12. Inception Model Architecture & Graph
2. Introduction 13. Tools and Technologies
3. Motivation 14. Web Application Detail
4. Project Workflow 15. Mobile Application Detail
5. Dataset 16. Project Milestones
6. Apple Plant Diseases 17. Results & Conclusion
7. Literature Review 18. Future Work
8. Implementation of CNN Model
9. ResNet Results
10. VGG16 Results
11. Inception_v3 Results

Monitoring and Controlling of LAB Environment using IoT Enabled Devices – FYP 2023
Slide |1
1. PROBLEM STATEMENT

● Crop diseases mostly harm the leaves, resulting in loss in agricultural


output’s quality and quantity.
● The detection and identification of the disease is a tedious task in
agriculture because it is expensive, time-consuming, and requires
experts that are not always available.

Plant Disease Classification and Detection – FYP 2023 Slide |2


2. INTRODUCTION

●18.9% of Pakistan GDP consists of Agriculture.

●42% labour force is employed because of agriculture.

●Our agriculture sector is still under performed.

●Potential of at least 40% of GDP.

Plant Disease Classification and Detection – FYP 2023 Slide |3


3. MOTIVATION

Increasing global smartphone penetration and recent advances in

machine learning, especially deep convolutional learning made it

possible to develop smartphone-assisted disease diagnosis systems.

Plant Disease Classification and Detection – FYP 2023 Slide |4


4. PROJECT WORKFLOW
Implementation of CNN Development of Mobile App Thesis Writeup
Implemented the Convolutional Developed Native Mobile Thesis and Paper Writing with
Neural Network CNN (Better Application to capture the Results
for Images Related Problems) diseased plan

Dataset Selection Implemented Default CNN Integration


We have initially selected the We have used different models Development of Web interface
apple dataset (Resnet18, VGG16) for our to integrate CNN model with
problem. Mobile App

Plant Disease Classification and Detection – FYP 2023 Slide |5


5. DATASET

Dataset Detail
• Total 4 Classes
• Healthy Class (Apple)
• 3 Diseased Classes (Apple Scrab, Black Rot, Cedar apple rust)
• Total images in data set that we collected are 6000.

Pre-Processing
• We changed the dimensions of each image to (224 x 224)
• We also separated images for validation & testing

Plant Disease Classification and Detection – FYP 2023 Slide |6


6. APPLE PLANT DISEASES

Healthy Apple Scrab Cedar apple rust


Black Rot

Plant Disease Classification and Detection – FYP 2023 Slide |7


7. LITERATURE REVIEW

• FF-PCA-LDA: Intelligent Feature Fusion Based PCA-LDA Classification System for Plant Leaf Diseases

• A Deep Learning - based Approach for Tomato Plant Leaf Disease Detection and Classification

• Deep learning models for plant disease detection and diagnosis

Plant Disease Classification and Detection – FYP 2023 Slide |8


8. IMPLEMENTATION OF DIFFERENT MODELS

Three architectures were explored:

• Resnet18
• VGG16
• Inception_v3.

The goal of the experiments was to achieve high accuracy on the validation & testing set.

Plant Disease Classification and Detection – FYP 2023 Slide |9


9. RESNET RESULTS

Plant Disease Classification and Detection – FYP 2023 Slide |10


10. VGG16 RESULTS

Plant Disease Classification and Detection – FYP 2023 Slide |11


11. INCEPTION_V3 RESULTS

Plant Disease Classification and Detection – FYP 2023 Slide |12


12. INCEPTION MODEL ARCHITECTURE & GRAPH

Plant Disease Classification and Detection – FYP 2023 Slide |13


13. Results

Results:

• The ResNet model achieved an acc of 78.22% on validation set & on testing set it was 67.2%.

• The VGG16 model achieved an acc of 81.11% on the validation set & on testing set it was 73.1%.

• The Inception v3 model achieved an accuracy of 84.91% on the validation set & on testing set it was 76%.

• These results were better than those of previous models & customized cnn.

Plant Disease Classification and Detection – FYP 2023 Slide |14


14.Conclusion

Conclusion:
All three of the deep neural networks explored in this study were able to achieve high accuracy on the validation
set, & testing set demonstrating their effectiveness for this classification task. Further work could include fine-
tuning the models or evaluating them on a different dataset.

Plant Disease Classification and Detection – FYP 2023 Slide |15


15. Tools and Technologies

CNN Model Development


• PyTorch
• NumPy
• Pandas
• TensorFlow & Keras
App Development
• Native technologies for Mobile app
• MERN stack for web development
• Node.js & Express (Backend)
• MongoDB or SQL (Database)

Plant Disease Classification and Detection – FYP 2023 Slide |16


16. Web Application

Application
● HTML, CSS, JavaScript, React (Front End)
● MongoDB or SQL ( Database)
● Node.js & Express (Backend)

Cost Effectiveness & Scalability


● Containerization
● Docker
● Microservices

Plant Disease Classification and Detection – FYP 2023 Slide |17


17. Mobile Application

● Easy UI / UX ( Uneducated Farmers)


● Native Application
● Java (Backend)
● Node.js API
● Scalable

Plant Disease Classification and Detection – FYP 2023 Slide |18


18. Milestones
2022 2023
Task
Sep Oct Nov Dec Jan Feb Mar Apr May Jun

Literature survey and Problem


formulation 100%

Technologies (Self Learning) 100%


Gathering & Analysis of Dataset 100%
Model Implementation 100%
Web and Mobile Application 0%
Publication 0%
Documentation and Thesis write-up 0%

Future Tasks Overall progress = 55%


Progress in the current semester
Plant Disease Classification and Detection – FYP 2023 Slide |19
19. Future Work

•Development of Backend API

•Web Application Frontend

•Web Application Backend

•Mobile Application Development

•Integration

•Testing on Live Data

Plant Disease Classification and Detection – FYP 2023 Slide |20

You might also like