CROP DISEASE DETECTION AND RECOMMENDATION

OF PESTICIDES AND SECONDARY CROPS USING

CONVOLUTIONAL NEURAL NETWORK

Benita J Shyamsundar Reddy G Venkata Rao K

Department of Computer Science and Department of Computer Science and Department of Computer Science and
Engineering Engineering Engineering
Kalasalingam Academy of Researchand
Kalasalingam Academy of Researchand Kalasalingam Academy of Researchand
Education
Education Education
Krishnankoil, Virudhunagar, India
Krishnan koil, Virudhunagar, India Krishnan koil, Virudhunagar, India
[email protected] [email protected] [email protected]

Bharath Kumar V Anil Kumar p

Department of computer Science and Engineering Department of computer Science and Engineering
Kalasalingam Academy of Research and Education Kalasalingam Academy of Research and Education
Krishnankoil, Virudhunagar, India Krishnankoil, Virudhunagar, India
[email protected] [email protected]

Abstract- I. INTRODUCTION
A sizable section of the populace in India works in In India, agriculture acts as the backbone of the
agriculture. Treating crop illnesses is crucial for maintaining the economy, contributing to 16% of the GDP and 10% of exports,
quantity and quality of agricultural produce, since crop disease with about 60% of the population engaged in agricultural
detection is becoming an increasingly important topic of research. In
activities either directly or indirectly. This sector, being a
this study, we train a dataset that has been specially selected for crop
disease identification using Deep Learning techniques, specifically primary source of food, fuel, and fodder, occupies over 60% of
Convolutional Neural Networks (CNNs). Prior to illness India's land area, ranking it second globally in arable land.
identification, the method entails image preprocessing to improve the Despite this vast expanse, agricultural productivity faces decline
image quality. The goal of using CNNs in this project is to increase due to numerous diseases, often not detectable in their initial
accuracy levels. Images of diseased crop leaves are one of the system's stages, leading to significant crop loss. This scenario
inputs; these images are preprocessed to make disease identification underscores the critical need for automated crop disease
easier. When a disease is detected, the system not only names the identification, capable of discerning diseases through symptoms
condition but also suggests suitable chemicals and pesticides to lessen on leaves. Crop diseases not only reduce the quantity and quality
its effects to increase the yield of crops. We also suggest that farmers
of agricultural output but also pose a risk of starvation. Accurate
be introduced to alternate crops. These crops are chosen for their
special qualities that complement the growth and productivity of the disease detection and classification are paramount for enhancing
major crops. They are chosen for their capacity to increase crop productivity and economic stability. Traditional manual
production, improve soil fertility, and more successfully manage inspections are inadequate for the vast number of cases, thus
pests. This project involves training a dataset on crop diseases image processing techniques, combined with machine learning
using CNN and applying image preprocessing techniques to algorithms, are employed to identify disease patterns on plants
enhance disease detection accuracy. By inputting images of and recommend appropriate treatments, offering a scalable and
diseased crop leaves into the system, it identifies the presence of efficient solution to this pressing agricultural challenge.
diseases and specifies their types. Following disease Currently, a critical study the field is the automation of crop
identification, the system recommends appropriate pesticides and
disease detection, which is capable of detecting illnesses from
chemicals to mitigate the impact on crop yield. Moreover, it
proposes alternative crops that offer benefits like improved yield, symptoms on leaves. Crop diseases could cause famine by
enhanced soil fertility, and better pest management, contributing significantly decreasing agricultural productivity. Precise
to the overall health and productivity of primary crops. disease diagnosis and timely pesticide application are essential
components for successful disease management. Crop diseases
Key Words: CNN, Pesticides, Secondary Crops, Pre-processing change the size, shape, and color of crops, which makes human
inspection ineffective and inadequate.
 of the system. The research report proposes that further
By examining variations in color, texture, or shape, image investigations could broaden the scope of this technology's use
processing techniques enable the identification of illness in plant to other plant species, evaluating the CNN model's flexibility
roots and leaves. Furthermore, the use of algorithms based on and efficacy in varying crop types in playhouses environments.
machine learning makes it possible to classify these illnesses and
provide suitable treatments, making them an essential tool for A technique using Convolutional Neural
boosting agricultural output and promoting economic expansion. Networks (CNNs) for the classification and prognosis of rice
plant illnesses is presented in the study by Sagarika, G. K.,
II. REVIEW OF LITERATURE RESEARCH Prasad, S. K., and Kumar, S. M. et al. [5], Using a camera, the
system takes pictures of rice plants, which are subsequently
The cited paper by Pranali K. Kosamkar et al. [1], processed by the CNN algorithm to ascertain the plants' current
addresses the important problem of leaf disease identification in state of health. Moreover, the program forecasts the degree of
agriculture by using Convolutional Neural Networks (CNNs) to any illnesses found.
do this. The study emphasizes how deep learning affects Using a dataset of photos of both healthy and sick paddy plants,
agriculture, especially in terms of how it might improve crop this method was thoroughly tested. With an astounding 95%
health and boost yield. It improves on earlier work to use CNNs accuracy rate, the system showed that it was very capable of
more effectively for diagnosing illnesses from photos of plant categorizing the plant conditions properly. Furthermore, it has
leaves. The scientists also recognize the ethical issues an 85% accuracy rate in predicting the severity of plant diseases.
surrounding the use of pesticides and the need for strong and These results demonstrate the possibility of CNNs as a
trustworthy datasets in order to properly train these networks— dependable instrument in the agriculture industry, particularly
both of which are essential for additional research in the field. for more efficient monitoring and management of crop health.

The task of forecasting plant leaf illnesses is The paper by Bhagat, M., Kumar, D., Mahmood,
tackled with Convolutional Neural Networks (CNNs) in the R., Pati, B., and Kumar, M. et al. [6], investigates the use of
paper "Plant Leaf Disease Prediction" by Vaishnavi Monigari, convolutional neural networks (CNNs) in the disease
G. Khyathi Sri, and T. Prathima et al. [2]. The CNN is trained categorization of bell pepper leaves. This method uses a camera
using a sizable and varied dataset of plant leaf photos by the to take pictures of the leaves, which are then analyzed by a CNN
researchers, which helps the network make precise disease to determine the health of the leaves. A web application is then
predictions. The research's contribution to agriculture is clearly used to present the categorization findings, which indicate
significant; it offers a useful tool for the early diagnosis of plant whether the leaves are infected or healthy. The excellent 96%
diseases, which is essential for safeguarding crop production and classification accuracy rate of this system highlights its efficacy
guaranteeing food security. Artificial intelligence has the and highlights the promise of CNNs in automating and
potential to improve agricultural techniques and maintain food improving the precision of plant disease diagnosis.
supplies, as seen by CNN's impressive accuracy in identifying
diseases. In 2021, Swathika, S., Srinidhi, N., Radha, and
Sowmya, K. et al. [7], conducted a study that highlights the
The authors Habiba, S. U., Islam, M. K. et al. [3], significance of using Convolutional Neural Networks (CNNs), a
and colleagues have made a valuable contribution to the field of fundamental deep learning approach, to diagnose illnesses in
agricultural research by presenting a deep learning technique rice leaves. The implementation of CNNs for the accurate
that uses leaf image analysis to diagnose diseases in tomato diagnosis of paddy leaf diseases in this research constitutes a
plants. Convolutional Neural Networks (CNNs) were utilized to substantial contribution that is expected to outperform more
extract features from the leaf images. These features were then conventional diagnostic methods in terms of accuracy and
divided into four groups using a Support Vector Machine efficiency. This work is a part of a bigger movement that
(SVM): healthy leaves, bacterial spot, early blight, and late addresses important sector concerns by integrating cutting-edge
blight. A collection of 1,500–3,000 photos of tomato leaves was technology into agriculture. It emphasizes the importance of
used to validate their methods, and they achieved an astounding multidisciplinary approaches by fusing computer vision and
accuracy rate of 95.6%. This result clearly outperforms earlier agricultural knowledge to enhance precision farming advances,
approaches, which generally achieved accuracy rates of about a step toward more cutting-edge and effective agricultural
80%, and shows how CNN and SVM may be combined to management techniques.
effectively diagnose plant diseases.
Using a publically accessible database, S.P.
The study by Radha, N., and Swathika, R. et Mohanty et al. [8] trained deep convolutional neural networks
al. [4], introduces a system using Convolutional Neural extensively to identify 26 illnesses in 14 different crop species
Networks (CNNs) for disease diagnosis and plant monitoring in over a collection of 54,306 photos. Their model's astounding
a playhouses, a controlled environment. By letting more heat and 99.35% accuracy rate was attained. Nevertheless, the model's
sunlight to the plants, playhouses—greenhouses with translucent performance decreased to 31.4% when it was tested using web
coverings—create an ideal environment for plant growth and pictures instead of photographs from the original dataset,
may increase agricultural yields. The suggested method suggesting problems with generalizing to new, undiscovered
incorporates a camera to take pictures of the plants, which are images. In parallel work, Ferentinos investigated multiple CNN
subsequently processed by a CNN to ascertain each plant's level architectures to identify between 58 different plant and disease
of health. In reaction to CNN's forecasts, a control system is combinations utilizing an open database, including popular
triggered if a disease is found. To improve its accuracy, a set of frameworks like AlexNet, VGG, and Google Network. With a
photos featuring both healthy and sick plant leaves is used to success percentage of 99.53%, the VGG convolutional neural
train the CNN. By means of analysis and expanding the image network architecture was shown to be the most efficient in these
dataset, the researchers want to strengthen the predicting powers studies. This study highlights CNNs' potential for use in
agricultural applications, especially disease detection, and The flowchart you've provided depicts the steps involved in a
emphasizes the value of a variety of training data sets for machine learning process tailored for crop disease
improving the model's capacity to generalize across various classification using a Convolutional Neural Network (CNN). It
image sources. starts with the initial step of uploading the crop dataset,
followed by preprocessing of the data. After preprocessing, the
The use of deep learning techniques with imaging data is split into training and testing sets. The model is then
for crop disease diagnosis and prediction was investigated by H. trained using the CNN algorithm. Finally, the trained model is
Park, J. S. Eun, and S. H. Kim et al. [9], and colleagues in their used to classify diseases, potentially along with
2017 paper. Their research was primarily concerned with recommendations for secondary pest management strategies.
illnesses that impact sugarcane leaves, an activity that is This structured process reflects the typical stages of a deep
typically carried out by professionals visually. But in large-scale learning project lifecycle, from data preparation to model
agricultural operations, where hiring experts for disease application.Our project uses a structured methodology that
identification can be unaffordable and unfeasible, this manual includes gathering a dataset of crop photos that are diseased,
method presents considerable obstacles, particularly in less preprocessing the images, training a CNN model to detect
developed nations like the Philippines. The author’s trained deep illnesses, and then evaluating the model's illness detection
learning models over sixty epochs to attain a validation accuracy accuracy. To suggest appropriate pesticides and substitute
of 95% in order to overcome these restrictions. This high degree crops, the system also compares sicknesses that have been
of accuracy indicates how well deep learning techniques can diagnosed with a database.
supplement or even replace conventional disease diagnosis
techniques agricultural practices, providing a scalable and IV. PROPOSED SYSTEM
economical answer for large-scale farms to manage crop health.
USER INPUT:
The authors of a 2017 paper by Khan Wahid, By the use of HTML, CSS, and JavaScript, we
Anh Dinh, Pankaj Bhowmik, Monzurul Islam et al.[10], titled create a user-focused front-end web portal for this project. It has
"Detection of Potato Diseases Using Segmentation of Images several components, such as a Home page, Sign-up, Sign-in,
and Multiple classes Support Vector Machine" present a novel Image Upload, and Live Prediction features. First, users must
method for diagnosing potato diseases from leaf images by register on the platform by entering specific data such as their
fusing image processing and machine learning. This study name, email address, phone number, password, and location.
makes use of the "Plant' Village" database, a publically After registering, customers use their email address and
accessible comprehensive resource of plant photos, to identify password to sign in. After logging in, users can submit photos of
diseases by applying sophisticated image segmentation their agricultural products for diagnosis of diseases by visiting
algorithms and a Multiple classes Support Vector Machine the image uploading section. The initiative also makes possible
(SVM). The implementation of state-of-the-art phenotyping and a live prediction function that lets users take and submit pictures
disease detection technologies to improve food security and straight from their cameras to detect diseases in real time.
sustainable farming practices is the relevance of this research.
The writers show a great degree of proficiency by using an
algorithm to analyze photos in order to detect potato infections DISEASE CLASSIFICATION USING CNN:
of precision, identifying 300 photos of sick leaves with a 95% At this point, a model developed by CNN that has
success rate. This research highlights the possibility of been trained on a dataset of photos of crop diseases is used to
combining agricultural science and machine learning to enhance compare an image that a user submits via the interface. The
disease identification, providing a scalable and effective means seven crops in our dataset—apples, grapes, tomatoes, corn,
of crop health monitoring. peaches, blueberries, and oranges—each with a spectrum of
disease expressions and benchmarks for healthy leaves. After
III. ARCHITECTURE DIAGRAM processing the uploaded image, the CNN model analyzes the
crop leaves' condition, determining whether a disease is present
and, if so, what kind of sickness it is. The user is then informed
of this information once more via the interface. To speed up and
simplify the training process, the CNN deep learning algorithm-
powered model, which was constructed in Python, must go
through a process of preprocessing on the dataset.

PESTICIDES AND SECONDARY CROPS

RECOMMENDATIONS:
After the crop image given by the user is successfully used to
identify the disease, the system provides a list of efficient
pesticides that can either eradicate or lessen the condition. Along
with insecticides, the process suggests a range of secondary
crops. These suggestions are designed to help control pests that
spread the disease in addition to bettering the yield of the
primary crop. The suggested auxiliary crops are picked because
they complement the main crop well and work well together to
improve plant health and agricultural output.
Figure 1: System Architecture
 V. PROPOSED METHODOLOGY In order to process and learn from visual data, convolutional
This project's main goal is to develop a system that can precisely neural networks, or CNNs, are made up of many crucial
identify plant diseases from leaf photos, suggest the right layers:
pesticides to treat the diseases that have been found, and offer
recommendations for secondary crops that will increase yields Convolutional Layer: CNNs' core components are their
and resistance to disease. Training and testing are the two convolutional layers, which make use of learnable filters or
primary stages of the project's framework. kernels. In order to generate feature maps, these filters conduct
element-wise multiplications as they glide across the input
Training Phase: Convolutional neural network (CNN) models image in a window-like manner. These maps capture several
are used in the first phase to train the model. Getting a wide patterns seen in the image, including edges, textures, and shapes.
variety of leaf photos and preparing them is part of creating a
solid training dataset. Activation Functions: To add non-linearity to the network after
the convolutional layer, activation functions like as ReLU
Testing Phase: Using the trained model in real-world situations (Rectified Linear Unit) are used. CNNs can now recognize
is covered in phase two. After acquiring fresh photos, it intricate links and patterns in the data because to this.
preprocesses them before moving on to classify diseases. Based
on the disease and principal kind of crop, the model suggests Pooling Layers: These layers minimize the spatial dimensions
suitable pesticides and alternate crops after a disease has been of the feature maps, hence reducing the complexity of the
discovered. network while preserving crucial information. Maximum
pooling for illustration, down samples the feature map by
We used a dataset with over 50,000 photos of seven distinct crop choosing the greatest value inside a specified region.
types—oranges, tomatoes, blueberries, corn, grapes, apples, and
peaches—for our study. The CNN model can train and Completely linked Layers: Determining that the network's
subsequently identify a broad range of plant illnesses thanks to fully linked layers perform similarly to those of conventional
this extensive dataset, which paves the way for accurate tests and neural networks. They enable the network to produce predictions
recommendations during the testing stage. or classifications based on the learned features by combining the
high-dimensional feature data into a flat vector.
IMAGE PRE-PROCESSING:
Pre-processing is done on the images to make sure CNNs are especially well-suited for image identification tasks
they are compatible with the input requirements of the CNN like diagnosing crop diseases in agriculture because of its
before introducing them for training and testing. Each picture in architecture, which is made to automatically and adaptively
this project has been scaled to have a standardized size of 250 by learn spatial hierarchies of information from input images.
250 pixels. Color conversion doesn't need to be done because we
are working with color photographs, which makes the pre-
processing phase simpler. Images are immediately sent into the CNN's Training Process:
CNN algorithm for training and testing after they have been pre-
processed. Maintaining accuracy as well as uniformity in the Forward Propagation:
algorithm's performance demands this standardization. Input data is fed into the network during the training phase.
Based on the current state of the model's weights, predictions are
CONVOLUTIONAL NEURAL NETWORK: formed, and the error between these predictions and the true
After pre-processing, the CNN is used to train the model, labels is evaluated by the loss function, which is often cross-
producing an experienced, well-tuned model. The CNN entropy for classification tasks.
implementation is based on the robust open-source machine
learning software platform Tensor Flow. To determine if a crop Backpropagation:
leaf is healthy or diseased, the system assesses each pre- In backpropagation, gradient descent methods are used to
processed image using the trained model. The diseases that have modify the weights of the model based on errors computed
been identified are given appropriate names, and this data is during forward propagation. In order to reduce the total loss, this
uploaded to an online application interface. The web tool makes procedure efficiently distributes the error back through the
recommendations for appropriate secondary crops and targeted network by adjusting the weights in the filters and fully
insecticides based on these identifications. With the help of these connected layers.
instructions, farmers may decrease the incidence of diseases and
protect the health of their crops. Training Iterations:
To improve the model's ability to recognize patterns and
features, the network goes through several iterations of forward
and backpropagation, fine-tuning the weights. Training keeps
going until the loss function of the model approaches a minimal
or appropriate value, showing that the model has acquired the
ability to correctly understand the input data.

Feature Extraction:
The CNN model is skilled at removing features from photos after
training. When new crop photos are supplied through the user
interface, illness detection is based on these derived attributes.
Figure 2: Architecture of CNN
Pesticide and Secondary Crop Recommendations: VI. RESULT
The system utilizes cross-referencing with an existing database
to find pesticides and secondary crops that are most appropriate
for fighting the crop disease that has been identified. The names
of crops, related diseases, and appropriate treatments and
preventive measures are all included in this database. With the
help of the ensuing recommendations, farmers are able to plant
secondary crops that could enhance crop health and yield overall
and treat specific illnesses with the proper pesticides.

Figure 3: Training and Validation Accuracy graph

S. CROP PESTICIDES SECOND
NO NAME DISEASE NAME ARY The training accuracy is showing an upward trend on the graph,
S NAME CROP which suggests that the model is successfully assimilating the
Chlorothaloni training dataset. But the validation accuracy swings after
Herbs and initially increasing, indicating that the model might be
Apple l, Captan,
Vegetables overfitting to the training set or that further fine-tuning is needed
scab Miclobutanil
, to improve generalization on unobserved data. In order to
1 APPLE Captan,
Flowering indicate a well-generalized model, one often tries to have both
Miclobutanil,
plants, lines as close together as feasible with high accuracy values.
Apple Boscalid
grains
black rot
Grasses,
Malathion, Cover
BLUEBERR Blueberry Abamectin, crops,
2
Y healthy copper-based Small
fungicides fruits
Triazoles,
Strobilurnis,
Common Mixed-model Forage
rust fungicides crops,
Root
3 CORN Methyl
crops, Figure 4: Training and Validation Loss Graph
Northern Benzimidazol
Vegetables
leaf blight e Carbamates
(MBCs) The model is clearly learning from the training dataset as seen
Mancozeb,
Herbs and by the graph, which shows that the training loss dramatically
Black rot Strobilurins,
Vegetables drops after the first epoch. Nevertheless, as epochs go by, the
GRAPE Boscalid
, validation loss starts to plateau or slightly grow instead of
Grape Bordeaux
4 Flowering decreasing as much. The pattern's increasing tendency in
Esca(Blac mixture,
plants, validation loss shows the model may be beginning to over fit to
k Tebuconazole
grains the training set, which could signal the model's ability to
measles) generalize to new data may not be advancing.
Neonicotinoid Herbs and
Huanglon s, flowers,
ORANGE gbing Organophosp Grasses,
5 (Citrus hates, Legumes
greening) Bactericides
Streptomycin,
Grasses,
Bacterial Copper-based
Cover
spot sprays
crops,
6 PEACH Insecticides,
Small
Healthy Bactericides,
fruits
Fungicides
Copper-based
Bacterial sprays, Basil,
TOMATO spot Streptomycin Marigolds, 1/1 [==============================] - 0s
7 Azoxystrobi Lettuce, 369ms/step
Early n, Nasturtiu True label: Corn_(maize)
blight Mancozeb, ms Northern_Leaf_Blight Predicted label:
Chlorothalo Corn_(maize) Northern_Leaf_Blight
nil Figure 5: Testing image of CNN trained model
 Science and Engineering Technology, 9, 1295-1305.

[3] Habiba, S. U., & Islam, M. K. (2021, February). Tomato plant

diseases classification using deep learning based classifier from
leaves images. In 2021 international conference on information
and communication technology for sustainable development
(ICICT4SD) (pp. 82-86). IEEE.

[4] Radha, N., & Swathika, R. (2021, March). A polyhouse: plant

monitoring and diseases detection using CNN. In 2021
International conference on artificial intelligence and smart
systems (ICAIS) (pp. 966-971). IEEE.

[5] Sagarika, G. K., Prasad, S. K., & Kumar, S. M. (2020,

November). Paddy Plant Disease Classification and Prediction
Using Convolutional Neural Network. In 2020 International
Conference on Recent Trends on Electronics, Information,
Communication & Technology (RTEICT) (pp. 208-214). IEEE.
Figure 6: Solutions Output
[6] Bhagat, M., Kumar, D., Mahmood, R., Pati, B., & Kumar, M.
(2020, February). Bell pepper leaf disease classification using
VII. CONCLUSION AND FUTURE WORK
CNN. In 2nd international conference on data, engineering and
In summary, diseases in the agriculture industry have applications (IDEA) (pp. 1-5). IEEE
been successfully identified by Convolutional Neural Networks
(CNNs). CNNs offer prompt treatments that reduce potential [7] Swathika, R., Srinidhi, S., Radha, N., & Sowmya, K. (2021,
agricultural losses by properly identifying and classifying a February). Disease Identification in paddy leaves using CNN
variety of crop ailments. CNNs' accuracy in disease prediction based Deep Learning. In 2021 Third International Conference on
provides farmers with insightful information that helps them Intelligent Communication Technologies and Virtual Mobile
make wise decisions about crop management. The advancement Networks (ICICV) (pp. 1004-1008).IEEE.
of technology in agriculture has the potential to transform
disease control strategies and enhance global food security.
[8] Yu, H., Liu, J., Chen, C., Heidari, A. A., Zhang, Q., Chen, H.,
& Turabieh, H. (2021). Cornleaf diseases diagnosis based on K-
It's critical to address any system problems that
means clustering and deep learning. IEEE Access, 9, 143824-
could result from an incomplete dataset in future work. The 143835.
accuracy and dependability of the CNN model will be improved
by increasing the variety and amount of crop disease data
[9] Mohanty, S. P., Hughes, D. P., & Salathé, M. (2016).
utilized for training. Furthermore, it is essential to maintain an
Using deep learning for image-basedplant disease detection.
up-to-date database of pesticides and secondary crops in order to Frontiers in plant science, 7, 1419.
guarantee that the recommendations remain applicable and
efficient in combating newly emerging strains of disease. .
[10] Sandika, B., Avil, S., Sanat, S., & Srinivasu, P. (2016,
Looking ahead, inadequate data could lead to potential November). Random forest based classification of diseases in
errors in disease detection. This problem can be solved by grapes from images captured in uncontrolled environments. In
continuously adding fresh and diverse datasets to the CNN 2016 IEEE 13th international conference on signal processing
(ICSP) (pp. 1775-1780). IEEE.
model in order to increase its accuracy. The database will be
updated with current data on pesticides and secondary crops that
match new disease profiles enhance the system's functionality. [11] Dwivedi, R., Dey, S., Chakraborty, C., & Tiwari, S. (2021).
Furthermore, combining the CNN model and user interface with Grape disease detection networkbased on multi-task learning and
solutions from a relational database will provide a more attention features. IEEE Sensors Journal, 21(16), 17573-17580.
powerful and all-encompassing tool for agricultural disease
control. [12] Sharma, R., Singh, A., Jhanjhi, N. Z., Masud, M., Jaha, E.
S., & Verma, S. (2022). Plant Disease Diagnosis and Image
Classification Using Deep Learning. Computers, Materials &
REFERENCES Continua, 71(2).

[1] Kosamkar, P. K., Kulkarni, V. Y., Mantri, K., Rudrawar, S., [13] Thind, T. S., Arora, J. K., Mohan, C., & Raj, P. (2004).
Salmpuria, S., & Gadekar, N. (2018, August). Leaf disease Epidemiology of powdery mildew, downy mildew and
detection and recommendation of pesticides using convolution anthracnose diseases of grapevine. Diseases of Fruits and
neural network. In 2018 fourth international conference on Vegetables VolumeI: Diagnosis and Management, 621-638.
computing communication control and automation (ICCUBEA)
(pp. 1-4). IEEE. [14] Harakannanavar, S. S., Rudagi, J. M., Puranikmath, V. I.,
Siddiqua, A., & Pramodhini, R. (2022). Plant leaf disease detection
[2] Monigari, V., Sri, G. K., & Prathima, T. (2021). Plant Leaf using computer vision and machine learning algorithms. Global
Disease Prediction. International Journal for Research in Applied Transitions Proceedings, 3(1), 305-310.