Plant Disease Detection and Crop

Recommendation using Deep Learning

Batch - 9

Team Members :
Guided By :
Dr. M. V. Ganeswara Rao • 18B01A0476 — N. Madhu Sri Sushmitha
Associate Professor • 19B05A0404 — B. V. Prathima
• 18B01A0404 — A. Shilpa
• 18B01A04B4 — V. Sravanthi

DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING

SHRI VISHNU ENGINEERING COLLEGE FOR WOMEN:: Bhimavaram
(AUTONOMOUS)
Abstract :
Farming is one of the major sectors that influences a country’s economic growth. In countries like
India, the majority of the population is dependent on agriculture for their livelihood. But now-a-days
farmers lack information about what kind of crop they need to grow in the climatic conditions of their
locality or what kind of fertilizer, pesticide they need to use for a particular disease.

This project helped us to find solutions for the above problem by implementing Crop
recommendation, Plant disease detection to improve crop productivity. In the crop recommendation
model, the user should provide the soil data from their side and the Random Forest with Decision tree
as base classifier predicts which crop the user should grow. For the plant disease detection model, the
user can input an image of a diseased plant leaf, and the trained VGG19 model predicts what disease
it is and will also give a little background about the disease and suggestions to cure it.

Finally, this project is helpful in agriculture. The used Python libraries are Numpy, Pandas,
Matplotlib, Scikit Learn, PyTorch. This project helps farmers to get suggestions, which makes it
easier for them to grow and maximize their yield.
Motivation :
• Farming is one of the major sectors that influences a country’s
economic growth.
• In countries like India, the majority of population depends on
agriculture for their livelihood.
• So, We thought of implementing certain ways to make the farming
easy and innovative using Machine Learning.
• Using this we can easily recommend a certain crop to certain
environmental conditions.
• It also makes us easy to detect the type of plant disease and type of
fertilizer to be used to get rid of the disease.
• So, this can help us in implementing a smart way of monitoring the
crop based on weather conditions, detect disease, fertilizers used.
Objectives :

• To develop a model that is capable of detecting and identifying the type of

plant disease.
• To develop Crop recommendation model depending on soil parameters and
climatic conditions.
Project Background :
• Crop diseases are a major threat to food security, but their rapid identification
remains difficult in many parts of the world due to the lack of the necessary
infrastructure.
• The combination of increasing global smartphone penetration and recent
advances in computer vision made possible by deep learning has paved the way
for smart-assisted ways of disease diagnosis.
• Crop Recommendation model can be implemented by taking the parameters like
N, P, K, average temperature of the area, average humidity in the locality, pH of
soil, average rainfall
• By using these parameters as inputs and we can create a model using Random
Forests to recommend the type of Crop to a particular location.
Project Background :
• Plant disease detection can be done using some pre-trained models, this helps us
to get high accuracy during prediction and reduces complexity.
• The random forest algorithm is an extension of the bagging method as it utilizes
both bagging and feature randomness to create an uncorrelated forest of decision
trees and used for classification type problems like crop recommendation.
• The Key difference between decision trees and random forests is decision trees
consider all the possible feature splits, random forests only select a subset of
those features.
• Plant diseases are the major issues that can affect the cultivation of rice, tomato,
potato, and pepper plants. Hence, a more serious diagnosis and proper handling
of crops on time are needed to prevent them from heavy losses.
Literature Survey :

AUTHOR & PUBLICATION METHODOLOGY OBSERVATIONS

Uses snake segmentation, here Hu’s

P. R. Rothe and R. V. Kshirsagar introduced a moments are used as distinctive attribute,
Active contour model used to limit the
Cotton Leaf Disease Identification using BPNN classifier tackles the numerous class
vitality inside the infection spot
Pattern Recognition Techniques problems. The average classification is found
to be 85.52%.
It uses deep artificial neural network (ANN)
Aakanksha Rastogi, Ritika Arora and Shanu K-means clustering used to segment the
as a classifier which mainly helps to check
Sharma, Leaf Disease Detection and Grading defected area; GLCM is used for the
the severity of the diseased leaf and increases
using Computer Vision Technology &Fuzzy extraction of texture features, Fuzzy logic is
complexity of model.
Logic used for disease grading.

Color features are represented in RGB to

uan Tian, Chunjiang Zhao, Shenglian Lu and This method underperforms in cases where
HIS, by using GLCM, seven invariant
Xinyu Guo, SVM-based Multiple Classifier the number of features for each data point
moment are taken as shape parameter. They
System for Recognition of Wheat Leaf exceeds the number of training data samples,
used SVM classifier which has MCS, used
Diseases the SVM will underperform.
for detecting disease in wheat plant offline
 AUTHOR & PUBLICATION METHODOLOGY OBSERVATIONS

Color histograms are extracted and

transformed from RGB to HSV, RGB to
It requires much computational power as
Godliver Owomugisha, John A. Quinn, L*a*b. Peak components are used to create
well as resources as it builds numerous trees
Ernest Mwebaze and James Lwasa, max tree, five shape attributes are used and
to combine their outputs. It also requires
proposed Automated Vision-Based area under the curve analysis is used for
much time for training as it combines a lot
Diagnosis of Banana Bacterial Wilt Disease classification. In seven classifiers extremely,
of decision trees to determine the class.
and Black Sigatoka Disease randomized trees yield a very high score,
provide real time information provide
flexibility to the application
It mainly works on the method of
Segmenting the defected area, color and
S. S. Sannakki and V. S. Rajpurohit, texture are used as the features. Here neural
proposed a Classification of Pomegranate network classifier is used for the The main disadvantage is that it is used only
Diseases Based on Back Propagation Neural classification. The main advantage is it for the limited crops.
Network Converts to L*a*b to extract chromaticity
layers of the image and Categorization is
found to be 97.30% accurate.
Problem Identification and Proposed Solution :

• Farmers face a lot of problem since they are not getting proper guidance of
what kind of crop need to grown and type of fertilizer to be used to increase the
productivity and detect the plant disease before it spreads to all plants.
• So, by using this project we can easily guide the farmers to increase the yields.
• This can be done by using a pre trained deep learning models and different
classifiers so as to get the accurate prediction.
• By using this project, We can help farmers to know about the details of what
kind of crop can be grown in their fields.
• And also detect the type of disease in the initial stages in order to stop
spreading of disease to all the plants in the fields.
Project Implementation :

• For plant disease detection, Plant Village Dataset can be used and It consists of 38
classes and deep CNN can be used to predict the type of disease effected out of 38
classes.
• VGG19 is an advanced CNN with pre-trained layers and a great understanding of what
defines an image in terms of shape, color, and structure. VGG19 is very deep and has
been trained on millions of diverse images with complex classification tasks.
• So, VGG19 Architecture can give accurate results for Plant disease prediction.
• For Crop recommendation, a model using Random forests with decision tree classifier
can perform well.
Project Block Diagram :

Plant Disease
Detection

Output

VGG 19
Sub-System and Design of Project :
VGG-19 is a convolutional neural network that is 19 layers deep. This helps us to
load a pre-trained version of the network trained on more than a million images
from the ImageNet database. The pre-trained network can classify images into 1000
object categories, such as keyboard, mouse, pencil, and many animals. As a result,
the network has learned rich feature representations for a wide range of images. The
network has an image input size of 224-by-224.

Important Layers of VGG 19 are

• Convolution Layer
• MaxPooling
• Fully Connected Layer
• Soft Max
Sub-System and Design of Project :

• Convolution : A convolution is the simple application of a filter to an input that

results in an activation. Repeated application of the same filter to an input results
in a map of activations called a feature map, indicating the locations and strength
of a detected feature in an input, such as an image. Getting our data in the right
dimensions is extremely important for any learning algorithm for giving good
results. VGG 19 has 19 convolution layers.
• (ReLU) : ReLU for short is a piecewise linear function that will output the input
directly if it is positive, otherwise, it will output zero. The rectified linear
activation function overcomes the vanishing gradient problem, allowing models to
learn faster and perform better.
Sub-System and Design of Project :

• Fully Connected Layers : Fully Connected Layer is simply, feed

forward neural networks. Fully Connected Layers form the last few
layers in the network. The input to the fully connected layer is the
output from the final Pooling or Convolutional Layer, which is
flattened and then fed into the fully connected layer.
• SoftMax : The softmax function is used as the activation function in
the output layer of neural network models for multi-class classification
problems. The softmax layer will output the value between 0 and
1 based on the confidence of the model that which class the images
belongs to.
Mathematical Background of Project :
ReLU : ReLU for short is a piecewise linear function that will output the input
directly if it is positive, otherwise, it will output zero.
Mathematical Background of Project :
Softmax function calculates the probabilities distribution of the event over ‘n’
different events. In general way of saying, this function will calculate the
probabilities of each target class over all possible target classes. These calculated
probabilities will be helpful for determining the target class for the given inputs.
Architecture of VGG19 :
• A fixed size of (224 * 224) RGB image was given as input to this network which
means that the matrix was of shape (224,224,3).
• The only preprocessing that was done is that they subtracted the mean RGB value from
each pixel, computed over the whole training set.
• Used kernels of (3 * 3) size with a stride size of 1 pixel, this enabled them to cover the
whole notion of the image.
• Spatial padding was used to preserve the spatial resolution of the image.
• Max pooling was performed over a 2 * 2 pixel windows with stride 2.
• This was followed by Rectified linear unit(ReLU) to introduce non-linearity to make
the model classify better and to improve computational time as the previous models
used tanh or sigmoid functions this proved much better than those.
• Implemented three fully connected layers from which first two were of size 4096 and
after that a layer with 1000 channels and the final layer is a Softmax function.
Flow Chart : Gather Dataset

Input data for preprocessing

Divide dataset to train and test

Training model using VGG19

Performance Assessment for high accuracy

Validate the results using test data

Predict the type of Disease

Tools Required :

• Anaconda Power shell

• Python
• Google Colab
• Personal Computer
Results :

Preprocessing
Results :
Results :
Objective 1 Results :
Objective 2 Results : Crop Recommendation
Results :
Results :
Results :
Results :
Results :
Results Comparisions :
Results :
Results :
Conclusion :
• Finally we were able to develop a model to help farmers to detect the
type of disease the plant is effected by.
• We are also able to develop crop recommendation model using
Random forest with decision tree as base classifier to tell what kind of
plant to be grown in particular area.
References :

[1] S. S. Sannakki and V. S. Rajpurohit, "Classification of Pomegranate Diseases Based on Back Propagation
Neural Network", International Research Journal of Engineering and Technology (IRJET), vol. 2, no. 02.
https://bit.ly/34mhOQU
[2] P. R. Rothe and R. V. Kshirsagar, "Cotton Leaf Disease Identification using Pattern Recognition
Techniques", International Conference on Pervasive Computing (ICPC), 2015.
https://ieeexplore.ieee.org/document/7086983
[3] Aakanksha Rastogi, Ritika Arora and Shanu Sharma, "Leaf Disease Detection and Grading using Computer
Vision Technology &Fuzzy Logic", International Conference on Signal Processing and Integrated Networks
(SPIN), 2015. https://ieeexplore.ieee.org/abstract/document/7095350
[4] Owomugisha Godliver, A. Quinn John, Ernest Mwebaze and James Lwasa, "Automated Vision-Based
Diagnosis of Banana Bacterial Wilt Disease and Black Sigatoka Disease", Preceding of the 1‘st international
conference on the use of mobile ICT in Africa, 2014. https://bit.ly/3rek6us
[5] uan Tian, Chunjiang Zhao, Shenglian Lu and Xinyu Guo, "SVM-based Multiple Classifier System for
Recognition of Wheat Leaf Diseases", Proceedings of 2010 Conference on Dependable Computing
(CDC‘2010), November 20–22, 2010. https://bit.ly/3AKHRgs
THANK YOU