Scribd
Upload
2 views21 pages

Project Review 3

The document presents a CNN-based model for early identification of five common rice leaf diseases, achieving a test accuracy of 95.67%. It emphasizes the importance of early detection for crop yield and food security, and compares favorably against traditional methods and existing models. The system is designed for practical deployment to assist farmers in timely disease intervention.

Uploaded by

Boya Raghuveera
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
2 views21 pages

Project Review 3

The document presents a CNN-based model for early identification of five common rice leaf diseases, achieving a test accuracy of 95.67%. It emphasizes the importance of early detection for crop yield and food security, and compares favorably against traditional methods and existing models. The system is designed for practical deployment to assist farmers in timely disease intervention.

Uploaded by

Boya Raghuveera
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 21

HIGH-PRECISION CNN-BASED

MODEL FOR EARLY STAGE


IDENTIFICATION OF RICE LEAF
DISORDERS

Boya Raghuveera ,Nakkinti Ramyasree,


Mude Revathi, Chinnuru Saiprakash
Dept. of CSE, MITS
Guide: Mrs. Geetha Vadnala
ABSTRACT
The proposed system leverages Convolutional Neural
Networks (CNN) to identify rice leaf diseases at early stages.
Targets five common rice diseases: Brown Spot, Leaf Smut,
Bacterial Blight, Blast, and Tungro.
The CNN model is trained on a curated, augmented dataset
to enhance prediction accuracy.
The system is designed to assist farmers in timely
intervention and reduce crop losses.
Offers a deployable solution through a lightweight, real-time
prediction interface.
INTRODUCTION
 Rice is a staple crop globally, particularly in Asia, including
India.
Leaf diseases like Brown Spot, Leaf Smut, Blast, Bacterial
Blight, and Tungro severely affect crop yield.
Early and accurate detection of these diseases is crucial to
ensure food security and sustainable farming.
Traditional methods of disease identification are time-
consuming and error-prone.
Deep learning, especially Convolutional Neural Networks
(CNNs), provides promising results in image-based disease
detection.
LITERATURE SURVEY ON
THIS TOPIC
Year Author(s) Approach Key Contribution

First deep learning


CNN (AlexNet,
2016 Mohanty et al approach to plant disease
GoogleNet)
detection.

Effective classification of
2020 Sun et al Custom CNN
rice leaf diseases.
Transfer Learning (VGG, Improved accuracy with
2021 Khan et al ResNet) fewer images using
pretrained models.

2023 Sun et al CNN + Data Augmentation Boosted generalization in


diverse field conditions.

2024 Reddy et al. (hypothetical) CNN + IoT Edge Real-time disease detection
Deployment using CNN on field-edge
devices.
LITERATURE SURVEY ON
THIS TOPIC
Traditional disease identification methods depend on expert
observation—slow and inconsistent.
Machine learning models like SVM and Random Forest have
been used but rely on handcrafted features.
CNNs outperform traditional models by learning complex
image features automatically.
Pretrained CNN models (VGG16, ResNet) have shown
promising results in rice disease detection.
UAV + CNN integration enables large-scale, real-time
monitoring.
LITERATURE SURVEY ON
METHODOLOGY
• Manual methods are time-consuming, error-prone, and
subjective.
• Early ML models like Random Forest and SVM lacked
generalization.
• CNNs outperform traditional models in scalability and
feature extraction.
• Transfer learning reduces data dependency using pretrained
models (e.g., VGG16, ResNet).
• UAV + CNN models show potential for large-scale field-
level detection.
• Lightweight CNNs like MobileNet explored for mobile
deployment.
RESEARCH GAP IN
LITERATURE REVIEW
Lack of large, diverse rice leaf disease datasets.
Existing models fail at early-stage symptom detection.
High visual similarity causes misclassification.
Poor real-world performance due to limited generalization.
Few models are optimized for real-time/mobile deployment.
ARCHITECTURE
DIAGRAM
DATASET
 Source: Kaggle and self-collected rice
leaf images.
 Classes: Brown Spot, Leaf Smut,
Blast, Bacterial Blight, and Tungro.
 Image Format: RGB images, resized
to 128×128 pixels.
 Total Samples: ~5,000 images after
augmentation.
 Preprocessing: Normalization, noise
reduction, and augmentation (rotation,
flipping, scaling).
 Split Ratio: 70% training, 20%
validation, 10% testing.
INPUT AND RESULTS
 Input: RGB image of a rice leaf (128×128 resolution)
 Preprocessing: Normalization, resizing, augmentation
 Model Used: CNN with convolutional and pooling layers
 Output: Predicted rice leaf disease class (e.g., Blast, Brown Spot)
METHODOLOGY OF THE
PROPOSED WORK IN
DETAIL
 Collection of rice leaf images (from
Kaggle and self-collected)
 Image preprocessing: resizing
(128×128), normalization,
augmentation
 Dataset split into train, validation, and
test sets
 Custom CNN architecture built and
trained on the dataset
 Model optimized using Adam
optimizer and categorical cross-
entropy
 Evaluation using accuracy, precision,
recall, and F1-score
ALGORITHM USED IN
DETAIL
ALGORITHM USED IN
DETAIL
Custom Convolutional Neural Network (CNN) architecture

Input: 128×128 RGB rice leaf images

3 convolutional layers with ReLU activation

MaxPooling after each convolution

Dropout for regularization

Fully connected Dense layer for classification

Output layer with Softmax for multiclass prediction


RESULTS AND COMPARISION
 Dataset Used: Public rice leaf disease dataset (Brown Spot, Leaf Smut, Bacterial
Blight, Tungro, Healthy)
 Training Accuracy: 98.45%
 Validation Accuracy: 96.12%
 Test Accuracy: 95.67%
 Precision: 95.10%
 Recall: 95.00%
 F1-Score: 95.04%
 Comparison with Existing Models:
 SVM: 78.6%
 Random Forest: 83.2%
 MobileNet: 91.4%
 Proposed CNN Model: 95.67%
Project Design

Use Case Diagram:


A use case diagram in the
Unified Modeling
Language (UML) is a
type of behavioral
diagram defined by and
created from a Use-case
analysis.
Class Diagram : A class diagram
in UML depicts the system's
classes, their attributes, methods,
and relationships among classes,
illustrating the information
contained within each class.
Sequence Diagram:A
sequence diagram in UML
displays interactions
between processes and their
order, derived from
Message Sequence Charts,
also known as event
diagrams or timing
diagrams.
Activity Diagram: Activity
diagrams visually represent
workflows, detailing stepwise
activities with options for
choices, iterations, and
concurrent actions. They
provide a clear view of
business and operational
processes, illustrating the
overall control flow within a
system.
SUMMARY
A deep learning model using Convolutional Neural Network (CNN) was
proposed to detect early-stage rice leaf diseases.
The model successfully classifies five classes: Brown Spot, Leaf Smut, Bacterial
Blight, Tungro, and Healthy.
Achieved high accuracy (95.67%) on test data with robust performance metrics
(Precision, Recall, F1-score).
Dataset was preprocessed with image resizing, augmentation, and normalization
to improve learning.
Compared with existing models like SVM, Random Forest, and MobileNet, the
proposed CNN outperformed in accuracy and reliability.
The system helps farmers and agronomists identify diseases early, enabling
timely action to reduce crop loss.
The project lays a foundation for future research in AI-powered agricultural
disease diagnosis.
REFERENCES
1. [1] A. Asfarian , Y. Herdiyeni , A. Rauf , K.H. Mutaqin , A computer vision for rice disease
identification to support integrated pest management, Crop Prot. 61 (2014) 103–104 .
2. [2] J.G.A. Barbedo , Digital image processing techniques for detecting, quantifying and classifying
plant diseases, SpringerPlus 2 (1) (2013) 660–672 .
3. [3] Y. Bengio , Learning deep architectures for AI, Found. Trends Mach. Learn. 2 (1) (2009) 1–127 .
4. [4] J. Bouvrie, Notes on convolutional neural networks, Technical report, 2006.
5. [5] Y. Chai , X.D. Wang , Recognition of greenhouse tomato disease based on image processing
technology, Tech. Autom. Appl. 9 (2013) 83–89 .
6. [6] D.C. Ciresan , U. Meier , J. Masci , L.M. Gambardella , J. Schmidhuber , Flexible, high
performance convolutional neural networks for image classification, in: Proceedings of the
International Joint Conference on Artificial Intelligence (IJ- CAI), Barcelona, Spain, 22, 2011, pp.
1238–1242 .
7. [7] F.R. Chen , Rice Diseases and Insect Pests Atlas and Control Technology, China Agriculture
Science Press, Beijing, China, 2012 .
8. [8] H. Dong , Z. Wang , S.X. Ding , H. Gao , Finite-horizon estimation of randomly occurring faults
for a class of nonlinear time-varying systems, Automatica 50 (12) (2014) 3182–3189 .
9. [9] H. Dong , Z. Wang , F.E. Alsaadi , B. Ahmad , Event-triggered robust distributed state estimation
for sensor networks with state-dependent noises, Int. J. Gen. Syst. 44 (2) (2015) 254–266 .
THANK YOU

You might also like