Journal of Neonatal Surgery

ISSN(Online): 2226-0439
Vol. 14, Issue 15s (2025)
[Link]

A Deep Learning Based Model for Plant Disease Detection

Mohd Haris1, Ashish Tripathi2, Sudhans Shekhar Pandey3, Himanshu Tiwari4, Ashar Ahmad Ansari5
1
SCSE, Galgotias University, Greater Noida. {hs4658419, ashish.mnnit44, sudhansh4u, himanshutiwari73026,
asharahmad2255}@[Link]
*Corresponding Author:
Ashish Tripathi,
Email ID: ashish.mnnit44@[Link]

Cite this paper as: Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar Ahmad Ansari, (2025)
00

A Deep Learning Based Model for Plant Disease Detection. Journal of Neonatal Surgery, 14 (15s), 913-925.

ABSTRACT
Plant diseases present increasing threats to agriculture which require precise early disease detection systems and powerful
management solutions to preserve crop health and maximize productivity. This research studies the application of CNNs and
combines it with ML and DL while investigating their capability to identify plant diseases accurately. Through precise
detection capability the system incorporates Artificial Intelligence (AI)-based treatment recommendations which suggest
optimal solutions. These innovative technologies enable better agricultural decisions which result in higher crop production
with lower losses. Nothing works as effectively toward plant disease management as the early and accurate identification of
diseases. Traditional expansive plant disease assessment requires specialized personnel and operates at slow speeds which
restricts its practical use on extensive areas. This research develops a computing solution which combines image processing
techniques and machine learning for plant disease detection automation. The system performs feature extraction through
texture analysis alongside color attribute examination and shape descriptor evaluation applied to images containing healthy
and infected plant leaves. A dataset made up of healthy and diseased plant leaf images supports the proposed system that
utilizes feature extraction through texture analysis with color properties combined with shape descriptors. A proposed CNN
is used on plant leaves to identify healthy versus diseased classes, producing an average accuracy of 94.65% on 14 types of
plants having 38 types of diseases.

Keywords: Detection of Plant Diseases, Convolutional Neural Networks, Machine Learning, Deep Learning

1. INTRODUCTION
Agriculture serves as a key element of global food security yet faces raising demands for performance improvement
combined with resource management requirements. The sector deals with various challenges yet disease detection and plant
management emerge as a critical component which directly affects agricultural yield and crop quality performances [1].
Standard methods of detecting plant diseases through manual expert inspection prove both slow and expensive because they
follow time-consuming human-based procedures. Advanced plant disease detection approaches became feasible because of
modern technological developments [2]. Convolutional Neural Networks (CNNs), Machine Learning (ML) and Deep
Learning (DL) create powerful tools for automated disease identification because they surpass traditional approaches by
providing both improved precision and swifter processing speeds. Scientists have thoroughly examined these new
technologies across numerous investigations which produced favorable results. Our investigation extends previous research
through the combination of CNNs for precise disease identification and artificial intelligence-based disease management
recommendations [3]. Throughout the generations farmers have employed manual evaluation to detect plant diseases in their
fields. The practice of traditional plant disease detection through manual inspection continues from past generations but
creates specific problems [4]. The practice of manual inspection requires both substantial physical labor and extended time
to undergo assessment while human error frequently occurs throughout extensive crop supervisors.
Plant symptoms in early disease phases remain hard to observe for humans resulting in delayed treatment that leads to higher
crop damage.
Modern technological advances have transformed plant disease detection by delivering advanced solutions to address
existing detection challenges [5]. The agricultural sector benefits significantly from the recent emergence of artificial
intelligence applications and deep learning as its dominant technology element. Convectional Neural Networks serving as
AI-based systems revealed outstanding capabilities to detect and classify plant diseases from visual images.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 913
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
Using 87900 RGB images this approach successfully achieved 94.65% accuracy in recognizing 38 distinct diseases of 14
plants. The proposed method extends beyond traditional detection research by adding custom treatment suggestions, which
leads to heightened decisions among farmers. This predictive and recommendation platform represents a fundamental step
forward in agricultural technology development.
The proposed work is further organized into seven sections, section 2 covers the problem statement and research contribution
which includes problems with plants, objective, scope, and contribution. An overview of the convolutional neural network
is given in section 3. Section 4 covers literature review. Complete methodology is given in section 5. Results & discussion
and comparative analysis with other state-of-the-art algorithms are shown in sections 6 and 7 respectively. Finally, the paper
is concluded in section 8.

2. PROBLEM STATEMENT AND RESEARCH CONTRIBUTION

2.1 Problems with Plants:
 Plant diseases pose a major threat to global food security and agricultural productivity and impose considerable
economic losses each year [6].
 Early detection of disease is important to control spread of disease, reduce damage to crops and reduce the need
for chemical treatments. However, currently available methods are somehow deficient, often failing to yield timely
and accurate results.
2.2 Objective:
 This research aims to develop an automated system to identify plant diseases using learning technique based on
machine and image process. With the help of sophisticated algorithms, these leaf images will be classified and
diseases diagnosed with great precision and speed [7].
 A real time indicator with visualities like discoloration, spots or lesions will be proposed to be identified as replies
to farmers and allied experts.
2.3 Scope:
 The leaf images will be collected and preprocessed, features extracted, model trained and evaluated in this study.
 The research will use various machine learning models, such as Convolutional Neural Networks (CNN’s) and
evaluate their ability to distinguish between ‘healthy’ and ‘diseased’ plants.
 The outcomes of the paper will provide support for sustainable crop management practices like providing early
disease detection and the promotion of precision agriculture.
2.4 Contribution:
 We introduce this system that predicts more plant diseases than previous work typically focuses on such a small
set of diseases.
 In the proposed research a very large collection of 87900 images of healthy and diseased crop leaves from multiple
classes have been used. The dataset is split into training set (70,295 images), validation set (17,572 images),
allowing 80/20 as split ratio for training and validation respectively.

Figure 1. Leaf Dataset

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 914
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
Figure 1 presents a complete dataset that shows data from Apple, Corn, Grapes, Potato and Tomato plants. Plants in this
dataset possess multiple diseases including healthy alongside Scab & black rot, cedar apple rust, leaf blight and early blight
and Late Blight and Common rust and Bacterial Spot.

3. CNN (CONVOLUTIONAL NEURAL NETWORK)

The deep learning framework known as Convolutional Neural Network (CNN) possesses multiple layers that extract input
data features automatically and create output predictions. The system architecture contains multiple processing levels that
consist of convolutional and pooling along with flatten and dense layers.
3.1 Convolutional layer
A CNN relies on its convolutional layer functionality to extract vital features across input image areas. Each section of the
image receives dedicated filters to extract main components which form the basis of pattern detection.
3.2 Pooling layer
Through the pooling layer the model reduces processing complexity along with parameter numbers. Two main pooling
techniques exist to lower feature maps scale: max pooling and global pooling which retain fundamental information.
3.3 Dense Layer
In a CNN the dense layer performs two functions: neuronal connections between prior layers before producing a final output
by processing data. As a fully connected final layer it integrates data from multiple sources to assist decision- making by
using the learned features.

Figure 2. Working of CNN

4. LITERATURE REVIEW
Rangarajan et al. working on both AlexNet and VGG16 models, VGG16 achieving an accuracy of 96.19%, while AlexNet
achieved a slightly lower accuracy of 95.81% [8].
Muhammad E.H Chowdhury et al. emphasize the necessity for automatic recognition systems that cater to the diverse needs
of mobile phone users, highlighting the challenge of accurately extracting and identifying leaf region from complex
backgrounds. This task becomes more difficult due to varying background patterns. To address this, the system collects
18,161 raw and segmented images of tomato leaves, which are then used in a deep learning architecture based on EfficientNet
[9].
S. P. Mohanty et al. proposed a deep Convolutional Neural Network approach, which achieved an accuracy of 96.3%. The
advantages of this method include improved classification accuracy. However, one limitation is that fine-tuning and data
augmentation are necessary to further enhance the model's performance [10].
S. Ishak et al. conducted research aimed at classifying leaf diseases using artificial neural networks (ANN). The study focused
on collecting and analyzing leaf images to identify healthy and diseased leaves from medicinal plants through image
processing methods. Feature extraction was performed on the images, followed by the use of ANN for analysis [11].
Emanuel Cortes conducted research that uses a deep neural network in combination with semi-supervised methods to classify
crop species and diseases of 57 different classes. The study uses a dataset containing 86,147 images of diseased and healthy
plants [12].
S. A. Wallelign et al. uses the Convolutional Neural Networks (CNNs) for detecting diseases in soybean crops. The study
uses a dataset of 12,673 images containing leaf images of four classes, including the healthy leaf images. The model achieved
an accuracy of 99.21% [13].
G. Latif et al. uses the CNN model for detection of rice plant diseases. The study uses a dataset of 75000 collected images,
15000 of the five different varieties of rice [14].

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 915
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
M. Ahmad et al. uses the CNN for plant disease detection in an imbalanced dataset. It uses datasets of 54,309 images. 39,218
images belong to 5 major classes of diseases whereas 15,085 belong to healthy plant images [15].
S. M. Hassan et al. uses the Novel Convolutional neural Network for plant disease identification. The model has been trained
and tested on three different plant diseases datasets. The performance accuracy obtained on plant dataset is 99.39%, on the
rice disease dataset is 99.66%, and on the cassava, dataset is 76.59% [16].

5. METHODOLOGY
5.1 Overview of the Proposed Work
The primary objective of the proposed work is to develop automated systems for identifying plant diseases, classifying them
into various categories to facilitate easier understanding. This section outlines the operation of the model across different
stages. Figure 3 provides a concise summary of the proposed system, along with a detailed explanation of each component
involved in the plant disease detection process. This system is broadly organized into four major parts starting from capturing
image to image pre- and post-processing followed by leaf disease identification and classification.

Figure 3. Steps for working model.

5.2 Data Gathering

Building a detailed and diverse dataset which includes healthy and infected leaf images for direct comparison forms the core
goal of data collection among different plant species. An extensive collection of healthy and unhealthy leaf images draws
primarily from Kaggle in addition to other real-world image libraries [17]. Figure 4 shows a sample of leaf images. In this
work there are a total of 87900 RGB images of 14 different plants having 38 different types of diseases have been used.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 916
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Figure 4. Sample image of dataset

5.3 Data Pre-Processing

Data pre-processing transforms raw datasets into a training state by removing unwanted noise then strengthening data
integrity before CNN Convolutional Neural Network (CNN) consumption.

5.4 Image Resizing

The altered leaf image collection requires uniform size adjustments to create standardized dimensions. Processing the data
delivers equalized results while adapting it to CNN operation. Image normalization creates standard data values which
normalize mean values to zero units and contribute to rapid network convergence while boosting CNN performance.
5.5 Balanced Data
Dataset consists of imbalanced distribution of classes so one type of image occurs often compared to others this can result in
model bias which prioritizes the class with more occurrences. This problem has been solved by applying under sampling the
majority class and oversampling the minority class. Rearranging the data frequency maintains a model that lends itself toward
improved performance during training.
5.6 Data Splitting
The next step requires splitting the available data into independent training and validation and testing data sections.
 Training Data: A significant part of the available data serves as training material for the CNN model because the
training process demands the majority allocation. For training purposes 70295 images are used.
 Validation Data: During training the model performs assessments using this data to determine its overall
effectiveness. Validation uses 17572 images.
 Test Data: The model assessment occurs on this segment following its training operation concludes. The testing is
done on 33 images.
5.7 Model Training
Convolutional Neural Networks (CNNs) are utilized to identify diseased leaves. During training, CNN learns to recognize
patterns and features in leaf images, which it uses to classify new images as either healthy or diseased. The network is trained
on an extensive dataset containing both healthy and diseased leaves, allowing it to differentiate between the two categories.
The architecture generally consists of several convolutional and pooling layers, culminating in a fully connected layer that
outputs the final classification.
5.8 Algorithm for Plant Disease Detection System
Step 1: Import Required Libraries
IMPORT streamlit as st
IMPORT tensorflow as tf
IMPORT numpy as np

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 917
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
Step 2: Define Function for Model Prediction
DEFINE FUNCTION model_prediction(test_image):
LOAD trained model ("trained_plant_disease_model.keras")
LOAD test_image and RESIZE to (128,128)
CONVERT image to array
EXPAND dimensions to create batch
PREDICT using model
RETURN index of highest probability prediction
Step 3: Create Streamlit Sidebar
[Link]("Dashboard")
app_mode ← SELECT from ["Home", "About", "Disease Recognition"]
Step 4: Define Home Page
IF app_mode == "Home":
DISPLAY HEADER: "PLANT DISEASE RECOGNITION SYSTEM"
DISPLAY Home Page Image
DISPLAY Description about the project:
- Upload an image
- System detects plant disease
- Provides results & recommendations
Step 5: Define About Page
ELSE IF app_mode == "About":
DISPLAY HEADER: "About"
DISPLAY Dataset Information:
- 87900 images in 38 classes
- Training: 70295, Validation: 17572, Test: 33
Step 6: Define Disease Recognition Page

ELSE IF app_mode == "Disease Recognition":

DISPLAY HEADER: "Disease Recognition"
test_image ← UPLOAD Image
IF "Show Image" button is clicked:
DISPLAY Uploaded Image
IF "Predict" button is clicked:
DISPLAY Snow Animation
DISPLAY "Our Prediction"
result_index ← CALL model_prediction(test_image)

DEFINE class_names = [List of 38 disease classes]

DISPLAY Predicted Disease using class_names[result_index]

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 918
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
6. RESULTS AND DISCUSSION

Figure 5. Model Summary

Figure 5 provides a summary of the model details and all relevant parameters. The number of parameters is 7,842,762, with
all of these parameters being trainable, as there are no non-trainable parameters in the model.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 919
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Figure 6. Graphs for accuracy

Figure 6 gives information about how visualization of result accuracy through evaluated the images internally up to 10
epochs.

Figure 7. Graphs for Loss

Figure 7 gives information about visualization of result accuracy through evaluating the images internally up to 10 epochs.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 920
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Figure 8. Evaluation Metrices

Figure 8 outlines the evaluation metrics used for detecting leaf diseases under different conditions like Apple scab, healthy
Apple, Grape black rot, Potato early blight, and Peach bacterial spot etc. The metrics include precision, recall, and F1-score,
which evaluate the model's performance and accuracy in disease detection. Figure 8 also provides a summary of the model's
effectiveness, including overall accuracy, macro and weighted averages, along with the results for leaf disease detection.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 921
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Figure 9. Confusion Matrix

Figure 9 shows the performance of the CNN model for different disease types. The confusion matrix presents actual classes
on its rows and predicted classes on its columns. The evaluation matrix presents total data about model performance in
dealing with the dataset before evaluating metrics. The matrix shows how predictions match with actual classes for better
disease classification assessmen

7. COMPARATIVE ANALYSIS
Table 1: Comparative analysis of different models
Author(s) Aim Dataset Methodology Accuracy
Rangarajan et al. [8] To train AlexNet and Dataset of 54,306 AlexNet and VGG16net 99.35% accuracy
VGG16n et for plant images. trained with varying batch
disease detection, sizes, learning rates for
focusing on batch size weight, and bias
impact.
Chowdhury, M.E. et Automatic recognition Dataset of 18,161 Efficient Net-based deep 99.89% accuracy
al. [9] and segmentation of leaf images. learning architecture to
regions from complex process segmented and
backgrounds. complex leaf images.
Mohanty, S.P. et al. To classify plant diseases Dataset of 14 types CNN methodology 96.35% accuracy
[10] using a deep of plants. trained with
Convolutional Neural augmentation; fine tuning
Network. required for
improvements.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 922
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Ishak, S. et al. [11] Classification of healthy Data extracted Multilayer Feed Forward 90.3% accuracy
or diseased medicinal using segmentation, Neural Networks:
plant leaves using ANN contrast adjustment, Multilayer Perceptron
and image processing. and feature (MLP) and Radial Basis
extraction Function (RBF).
algorithms.
Emanuel Cortes [12] Classification of plant Dataset of 86,147 Deep Neural Network, 80% accuracy
species and diseases using images. Semi-supervised
semi supervised learning algorithms rs-net with
and DNNs. unlabeled data for
experiments
Wallelign, S. et al. Soybean disease detection Dataset of 12,673 Soybean disease detection 99.21% accuracy
[13] using CNNs in natural images. using CNNs in natural on classification
environments. environments.
Latif, G. et al. [14] Rice plant disease Dataset of 75000 Using VGG19 based 96.08% accuracy
detection using DCNN. images. learning method for
detection accurately.
Ahmad, M. et al. [15] Classification of plant Dataset of 54,309 Using DL approach to 99.69% accuracy
disease symptoms using images. classify plant disease.
CNN.
Hassan, S.M. et al. Identification of plant The model has been Using CNN to classify 99.66% accuracy
[16] disease using novel trained and tested diseases in plants.
convolutional neural on only three plant
network. datasets.
Propose Model Plant Disease Detection The model has been Using CNN to identify The proposed
Using CNN trained on a large the disease in plants. model gives
dataset in which 94.65% average
87,900 RGB images accuracy on a
have been used, huge number of
which include 14 plants and their
different types of various types of
plants having 38 diseases as
different types of compared to
plant diseases. other existing
models.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 923
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari

Accuracy
99.35% 99.89% 96.35% 99.21% 96.08% 99.69% 99.66% 94.65%
100.00% 90.30%
90.00% 80%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%

Figure 10. Comparison of the accuracy of different models

Table 1 and Figure 10 shows the accuracy of the models of different authors. The proposed model achieved an average
accuracy of 94.65%. This model is capable of to detect the plant disease of 14 different plants. Also, this model is trained to
detect 38 types of plant diseases of 14 such plants. Also, this model can easily classify the healthy and diseases plants. The
highest accuracy of this model is 99.99% while the lowest accuracy is 84%. When compared with the other existing models
it is found that the size of the dataset that is used by other models is very small. Also, the number of plants and types of
diseases considered by the other models are very small compared to the proposed model.

8. CONCLUSION
The proposed plant disease detection system achieves an average accuracy of 94.65% on 38 different plant diseases of 14
plants with an F1 score of 94.61%, Recall of 94.66%, and Precision of 94.92% respectively.
 Efficiency: The proposed system runs efficiently with straightforward implementation that enables accessibility
directly to farmers through any available computer system. The affordable methodology remains essential in
agricultural settings since it enables early detection of diseases to affect crop production levels and financial
outcomes positively.
 Automated Monitoring: Automation of disease detection by the system helps reduce both the requirement for
manual labor and trained expert knowledge. Live monitoring of extensive agricultural regions becomes possible
through this method which lets interventions start immediately after detecting diseases.
 Versatility: Through its ability to detect numerous diseases in various plant types with one unified approach the
system showcases its broad application potential. This dataset of 87,000 leaf images supports the system's reliability
and robust design.

REFERENCES
[1] Ngugi, H.N., Ezugwu, A.E., Akinyelu, A.A. and Abualigah, L., 2024. Revolutionizing crop disease detection
with computational deep learning: a comprehensive review. Environmental Monitoring and
Assessment, 196(3), p.302.
[2] Buja, I., Sabella, E., Monteduro, A.G., Chiriacò, M.S., De Bellis, L., Luvisi, A. and Maruccio, G., 2021.
Advances in plant disease detection and monitoring: From traditional assays to in-field
diagnostics. Sensors, 21(6), p.2129.
[3] Lebrini, Y. and Ayerdi Gotor, A., 2024. Crops Disease Detection, from Leaves to Field: What We Can Expect
from Artificial Intelligence. Agronomy, 14(11), p.2719.
[4] Ding, W., Abdel-Basset, M., Alrashdi, I. and Hawash, H., 2024. Next generation of computer vision for plant
disease monitoring in precision agriculture: A contemporary survey, taxonomy, experiments, and future
direction. Information Sciences, 665, p.120338.
[5] Sahoo, L., Mohapatra, D., Raghuvanshi, H.R., Kumar, S., Kaur, R., Chawla, R. and Afreen, N., 2024.

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 924
Mohd Haris, Ashish Tripathi, Sudhans Shekhar Pandey, Himanshu Tiwari, Ashar
Ahmad Ansari
Transforming Agriculture through Artificial Intelligence: Advancements in Plant Disease Detection,
Applications, and Challenges. Journal of Advances in Biology & Biotechnology, 27(5), pp.381-388.
[6] Oraon, S., Padamini, R., Shahni, Y.S., Das, N., Sinha, D., Sujatha, G., Singh, O.B. and Karanwal, R., 2024.
Impact of Emerging Pathogens in Crop Production. Microbiology Research Journal International, 34(7),
pp.80-92.
[7] Demilie, W.B., 2024. Plant disease detection and classification techniques: a comparative study of the
performances. Journal of Big Data, 11(1), p.5.
[8] Rangarajan, A.K., Purushothaman, R. and Ramesh, A., 2018. Tomato crop disease classification using pre-
trained deep learning algorithm. Procedia computer science, 133, pp.1040-1047.
[9] Chowdhury, M.E., Rahman, T., Khandakar, A., Ayari, M.A., Khan, A.U., Khan, M.S., Al-Emadi, N., Reaz,
M.B.I., Islam, M.T. and Ali, S.H.M., 2021. Automatic and reliable leaf disease detection using deep learning
techniques. AgriEngineering, 3(2), pp.294-312.
[10] Mohanty, S.P., Hughes, D.P. and Salathé, M., 2016. Using deep learning for image-based plant disease
detection. Frontiers in plant science, 7, p.1419.
[11] Ishak, S., Rahiman, M.H.F., Kanafiah, S.N.A.M. and Saad, H., 2015. Leaf disease classification using artificial
neural network. Jurnal Teknologi, 77(17).
[12] Cortes, E., 2017. Plant disease classification using convolutional networks and generative adverserial
networks. Stanford University Reports, Stanford.
[13] Wallelign, S., Polceanu, M. and Buche, C., 2018, May. Soybean plant disease identification using convolutional
neural network. In The thirty-first international flairs conference.
[14] Latif, G., Abdelhamid, S.E., M allouhy, R.E., Alghazo, J. and Kazimi, Z.A., 2022. Deep learning
utilization in agriculture: Detection of rice plant diseases using an improved CNN model. Plants, 11(17),
p.2230.
[15] Ahmad, M., Abdullah, M., Moon, H. and Han, D., 2021. Plant disease detection in imbalanced datasets using
efficient convolutional neural networks with stepwise transfer learning. IEEE Access, 9, pp.140565-140580.
[16] Hassan, S.M. and Maji, A.K., 2022. Plant disease identification using a novel convolutional neural network.
IEEE Access, 10, pp.5390-5401.
[17] Samir, B. (2019). New Plant Diseases Dataset [Data set]. Kaggle.
[Link]

Journal of Neonatal Surgery | Year: 2025 | Volume: 14 | Issue: 15s

pg. 925