NOCL Based Automatic Leaf Disease Detection with High Accuracy

Sindhuja V, Latha B, Dhanushiya S, Dharani A C, Swetha D and Udhika V

Department of ECE, K.S.R. College of Engineering, Tiruchengode, Namakkal, Tamil Nadu, India

Keywords: NOCL, Plant Leaf Disease, Neural Optimization, Classification, Crop Disease, Convolutional Neural

Network, CNN.

Abstract: Aim: The study introduces the Neural Optimization and Classification Logic (NOCL) method, a new deep

learning method to directly diagnose plant leaf diseases, which aims to reduce the mortality rate of plants

and improve agricultural productivity, as well as promote sustainable agriculture. Materials and Methods:

The proposed Neural Optimization and Classification Logic (NOCL) model was implemented using Python

and TensorFlow designs and tested on a dataset with diseases and healthy leaves. Group 1 refers to the

NOCL method that integrates advanced optimization techniques and classification logic with 1000 testing

counts. Group 2 refers to the traditional CNN method. The NOCL-based architecture shows direct detection

capability, providing a 25% improvement in classification accuracy and a 35% reduction in computational

complexity, compared to the CNN method. Result: The proposed method uses the Ka15e leaf disease

dataset, which contains a total of 78,456 images, which includes 75% as training data and 25% testing data.

The NOCL method achieves 96% of accuracy, F1 score and recall and performs better than traditional

methods. It classifies healthy and diseased specimens by examining signs such as black dots, mosaics also

greenish ventral patterns on the leaves. For comparison, the CNN model with 7 convolution layers was

used, which showed an accuracy of 90.26% to 92.16% whereas, F1 score and recall less than the proposed

method. The NOCL model is implemented through the Python language and is efficient in training and

validation with a significance of p < 0.05. Conclusion: The recommended NOCL based strategy provides

an efficient and reliable method for detection of plant diseases, which enables farmers to take measures for

early control of diseases and ensure sustainable agriculture through robust crop production practices.

1 INTRODUCTION

Neural optimization and classification logic is one of

the methods of transfer learning. One of India's main

sources of income is agriculture, and the amount of

it produced has a significant influence on the entire

nation. However, pests, fungi, and disease factors

are being introduced, and the effects of climate

change are having a growing impact on crops

(Hassan, Sk Mahmudul, 2021). In this context, early

detection and prevention of crop leaf diseases

becomes essential as it helps in preventing economic

losses and reduction in production. Although the

current method, the Convolutional Neural Network

(CNN) method, is capable of detecting leaf diseases,

it has some limitations, especially computational

crises and identity defects (Tugrul, 2022.). Thus, a

new deep learning system Neural Optimization and

Classification Logic (NOCL) has been introduced

with an aim to improve crop health. The NOCL

system can provide an elegant prognosis of diseases

(Alzubaidi, 2023). Also, recommended fertilizer and

preventive practices help reduce diseases and

improves quality of crops. It includes the importance

of agriculture and provides solutions for sustainable

agriculture (Krichen, 2023).

2 RELATED WORKS

Various fields have tracked down applications for

artificial intelligence, including medical services,

correspondence, object ID, and following. The

world's most significant yield, maize, is powerless to

various ailments that diminish both result and

quality. In this review, we presume that the proposed

model gives 90.2% of accuracy. As to

(Responsiveness) assessment, MobileNetV2

V, S., B, L., S, D., A C, D., D, S. and V, U.

NOCL Based Automatic Leaf Disease Detection with High Accuracy.

DOI: 10.5220/0013876400004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 2, pages

38-44

ISBN: 978-989-758-777-1

displayed fruitful execution in Earthy colored Spots,

Blended, White Scale (Zhang, et al., 2022.). PSO

strategy utilized for tweaking technique Thangaraj,

Rajasekaran, 2020. This model is the most amazing

models out of the relative multitude of applied

designs in the underlying demonstrating tests using

the informational index created for the order of plant

species. Through trial, different group of models

proposed are ideal geographies in every order. As

per the group model outcomes, each model

effectively breezes through the assessment with an

exactness of 91.60% (Demilie, 2024).

Many techniques have been proposed for plant

sickness identification, and profound learning has

turned into the favoured strategy in view of its

awesome achievement. In this review, U2-Net was

utilized to eliminate the undesirable foundation of an

info picture by choosing multiscale highlights. This

work proposes a cardamom plant sickness location

approach utilizing the EfficientNetV2 model. An

extensive arrangement of trials was done to

determine the presentation of the proposed approach

and contrast it and different models like EfficientNet

and Convolutional Brain Organization (CNN).

Inception design include learning that further develop

the data extravagance, that is particularly useful to

fine-grained highlight learning (Pradhan, 2022). The

NOCL method gets increased exactness compared to

the past convolution and Vi-based models. The

analysis shows dominance on the current models

(Bangari et al., n.d.). In the analyses, both the models

with and without LBP attributes are utilized. The

recommended ANFIS CNN model really does

astoundingly well in the two arrangements of tests.

Barring LBP qualities yields accuracy(0.8953),

recall(0.9045), and F1 score (0.8478)( Nandhini, S.,

and K. Ashokkumar. 2022). In the wake of

consolidating LBP highlights, the proposed model

accomplished F1 scores (above91%), review

(above92%), exactness (above90%), and accuracy

(above91%). Broad correlations with latest

techniques further show the proposed method's

greatness. The outcomes were likewise checked for

unwavering quality and vigour utilizing cross-

approval. Further developed precision and

productivity in true applications are guaranteed by

this procedure, which addresses a significant

forward-moving step in rural sickness identification.

From the above findings, it is concluded that the

accuracy of CNN is less. The prediction rate is a

crucial element to consider while detecting leaf

disease. The focus of this study is to improve the

accuracy using Neural optimization and classification

logic in comparison with Convolutional neural

networks.

3 MATERIALS AND METHODS

This review zeroed in on the assessment of the

precision and computational capacity of the Brain

Streamlining and Order Rationale (NOCL) strategy

in base leaf illness determination contrasted with the

conventional CNN technique (Elhassouny,). The

dataset utilized is taken from Kaggle did in the past

in plant sickness finding. The main characteristics

examined were accuracy of classification,

computational complexity, rate of diagnosis,

processing time and power output. The experimental

setup was designed and implemented using the

Python latest version on the Anaconda navigator GUI.

Group 1 refers to the recommended NOCL-based

translational leaf disease detection method, which

includes areas such as image pre-processing, feature

extraction, and disease classification where 100 to

1000 testing sample counts are taken. This method

was practiced on kal5e datasets that included leaf

images and tested to predict the presence of the

disease with greater accuracy. The NOCL model was

implemented using Python, TensorFlow and Keras

libraries.

Group 2, similarly, refers to the CNN based

method, which is the traditional method that uses

predetermined partial noise elimination and

classification techniques. This method uses

predefined rules and segment-based models for

classification.

Figure 1: Workflow of Neural optimization and

classification logic system.

Figure 1 shows the data collection for a Plant

Leaf Disease Detection system using NOCL

involves obtaining, preprocessing, and organizing a

NOCL Based Automatic Leaf Disease Detection with High Accuracy

dataset suitable for training and testing the NOCL

model. This approach accumulates data from a well-

known opensource data repository called Kaggle.

Before training a NOCL model to identify plant

diseases, data must be pre-processed. Before feeding

the images into the neural network, pre-process them

to make sure they are consistent. Data augmentation,

scaling, and normalizing are some of the methods

that may be used to make the dataset more diverse.

𝑌 =𝑊1𝑅𝑒𝑠𝑖𝑧𝑒(𝑋,(𝐻,𝑊)) +𝑊2(𝑋/255) +𝑊3𝑇(𝑋) (1)

Where:

‘X’ denotes Raw input data ‘Y’ denotes

Preprocessed data

‘H’ denotes the height of the resized image

‘W’ denotes the width of the resized image

‘T’ represents transformations

‘W1’,’W2’,’W3’ Weights representing the

importance or influence of each preprocessing step.

Equation (1) gives the pre-processed data with

the help of input data, weights and transformations.

Ensure that there are a variety of healthy and

diseased plant leaf images in the collection. Get rid

of any low-quality, irrelevant, or duplicate

photographs from the collection. Validate that the

labels or annotations are accurate before using them

for classification. Be certain that all of the images are

the same size so that the NOCL model can use them.

To enhance convergence when training, normalize

the pixel values. Enhance images with modifications

to avoid overfitting and increase generalization.

The dataset contains the three major sections:

training, validation, and testing. For multiclass

classification, use categorical cross-entropy; for

binary situations, use binary cross-entropy. Train the

model through its training process with a set number

of epochs while keeping an eye on parameters like

loss and accuracy. In order to train the model,

Equation (2) gives the epochs by equating with input

data, output from CNN model f, parameterized by θ,

loss function and regulation coefficient.

𝜃𝑡 + 1 = 𝜃𝑡 − 𝜂 ⋅ 𝛻𝜃 







∑ℓ

(

𝑓

(

𝑋𝑖;𝜃

)

,𝑌𝑖

)

+ 𝜆

(2)

Where:

‘X’ denotes Input data

‘Y’ denotes True labels for input data

‘𝑓(𝑋;𝜃)’ denotes Predicted output from the CNN

model f, parameterized by θ

‘ℓ(⋅)’ denotes Loss function

‘N’ denotes Number of samples in the dataset

‘λ’ denotes Regularization coefficient

‘η’ denotes Learning rate for gradient descent

The assortment of measurements used to assess a

prepared model which incorporates TP, TN, FN, and

FNP decides the model's presentation and the

outcomes it produces. By applying the prepared

model to the testing set, you might decide its

precision in separating among sick and solid leaves.

Assessing the model's presentation might be finished

utilizing measurements, for example, F1 score,

exactness, accuracy, and review.

4 DESCRIPTIVE ANALYSIS

SPSS version 26 is used for the descriptive analysis

based on the statisticaldatas collected. The features

and model outputs of the database are analysed using

TensorFlow and SPSS. Colour information such as

RGB and HSV values, appearance features (contrast,

homogeneity, energy), and shape features (area, edge

shapes) are considered independent variables. At the

same time, variables are considered depending on the

type of disease (for example: healthy, leaf irritation,

powdery mildew) and the level of severity (mild,

moderate, severe). The independent t-test and

statistical measurements were calculated in SPSS and

validated the relationships, while TensorFlow was

used for training and evaluating the model(Shoaib,

Muhammad et. Al., 2023).

Table 1: To analyze the prediction accuracy for 10 testing count across two methods.

Testing

count

CNN (%) NOCL (%)

Accuracy

Score

Recall Accuracy F1 Score Recall

100 91.56 90.00 89.50 96.54 96.00 95.30

200 91.64 90.80 90.00 96.87 96.80 96.20

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

300 90.35 88.90 88.30 95.89 95.00 94.00

400 92.16 92.00 91.50 96.37 96.00 95.50

500 91.54 90.50 89.80 96.79 96.80 96.00

600 91.59 91.00 90.30 96.49 96.00 95.30

700 90.47 88.80 88.00 96.49 96.00 95.00

800 91.69 90.60 89.90 96.49 96.00 95.50

900 90.26 88.20 87.50 96.49 95.50 94.50

1000 91.79 91.00 90.20 96.49 96.00 95.50

Table 2: Group statistics [n, mean, standard deviation, standard error mean].

N Mean

Std.

Deviation

Std. Error

Mean

CNN

Accurac

10 91.305 0.677 0.214

F1 Score 10 90.180 1.193 0.377

Recall 10 89.500 1.216 0.384

NOCL

Accurac

10 96.491 0.260 0.082

F1 Score 10 96.010 0.530 0.167

Recall 10 95.280 0.652 0.206

Table 3: An independent sample t-test was conducted to compare the Area (LUT) and Total Power (Watts) values between

the Pre-scaled method and Adaptive method.

Levene’s test for

equality of

variances

Independent samples test

F Sig t df Sig(2-tailed)

Mean

differen

Std.error

difference

95% confidence

interval of the

difference

lower upper

Accura

Equal variance

assume

10 0.000 426.255 9 0.000 91.305 0.214 90.82 91.78

Equal variance

not assume

0.000 1170.805 9 0.000 96.491 0.082 96.30 96.67

Score

Equal variance

assume

10 0.000 238.977 9 0.000 90.180 0.377 89.32 91.03

Equal variance

not assume

0.000 572.748 9 0.000 96.010 0.167 95.63 96.38

Recall

Equal variance

assume

10 0.000 232.644 9 0.000 89.500 0. 384 88.62 90.37

Equal variance

not assume

0.000 461.513 9 0.000 95.280 0.206 94.81 95.74

Statistical experiments carried out by SPSS

confirmed important positive differences between

CNN accuracy and NOCL accuracy when the p-

value is less than 0.05. However, it can be seen in

Table 1 that no standard sigma, variance or fine

differences in t-test results were observed between

the systems. CNN accuracy for 10 test counts ranged

in values from 90.26% to 92.16%, whereas NOCL

accuracy had a high and consistent range from

95.89% to 96.87% (found in Table 1). In the T-test

comparison, CNN had an average accuracy of

91.25% (SD = 0.677) and NOCL had an even higher

accuracy of 96.49% (SD = 0.260) (shown in Table

2). This reveals the quality of NOCL accuracy under

varying test conditions and in Table 3. shows there

is a significant difference between the two groups p

< 0.05.

5 RESULTS

The NOCL framework accomplished a normal

exactness of 96.49% with a SD of 0.260 which is

NOCL Based Automatic Leaf Disease Detection with High Accuracy

high compared to the ongoing CNN framework that

has a SD of 0.677. The typical precision of the CNN

framework was just 91.25%. While the test count

varied from 100 to 1000, the NOCL method resulted

in accuracy values ranging from 95.89% to 96.87%,

and demonstrated reliability and consistency with a

low data inaccuracy ratio of 0.092. In contrast, the

CNN method only revealed accuracy values ranging

from 90.26% to 92.16%. These results emphasize

the improved accuracy and consistency of the NOCL

system over the previous CNN system. The plot in

Figure 3 shows the comparison of accuracy of

NOCL and CNN with testing count. CNN has

obtained a sample mean accuracy of 91.25%, which

varies from 90.26% to 92.16%. On the other hand,

the NOCL model has a stable and high specificity

ranging from 96% to 97%. This shows that the

function of the CNN model varies according to the

experiment counts, but the NOCL model works with

great consistency and reliability

Figure 4 shows the F1 Score for the proposed

NOCL technique is 96.08%, with a standard

deviation of 0.530, which is a lot higher than the

current CNN framework's score of 90.58%. CNN

system standard deviation was 1.193. When the test

counts increased from 100 to 1000, the NOCL

model was able to maintain an F1 score from 95.0%

to 96.8%, which indicates a balanced recall and

precision tradeoff. CNN has low classification

stability, achieving an F1 score of only 88.2% to

92.0%. The increased stability and higher overall F1

score demonstrate that NOCL can deliver more

reliable and accurate classifications, which makes it

a better option for the task of leaf disease detection.

The graph in Fig.5 shows the comparison of F1

score of NOCL and CNN with testing count. The

CNN model's F1-score varies from 88% to 92%,

which is consistent with its referral and recall values.

Whereas, the NOCL model has an F1-score of about

96%, which balances the accuracy and recoil

perfectly.

The NOCL model has shown an average recall of

95.23 percent with a standard deviation of 0.652

while outperforming the CNN system, which had an

average recall of 89.87 percent but a higher standard

deviation of 1.216 in Figure 3. From a range of

testing 100 to 1000, Recall for NOCL was

maintained between 94.0 and 96.2, thus maximizing

the efficiency of false negative responses and disease

detection. Contrary to this, the CNN, while

monitoring recall values, reported values from 87.5

to 91.5 which indicate a higher degree of variability

in the detection of diseased leaves. Less deviation

and high recall with true NOCL positive values

increased the model’s practicality in the field of

agriculture, making NOCL more advantageous than

its counterpart. The plot in Fig.4 depicts comparison

of recall of NOCL and CNN with testing count. In

the recall value, the CNN sample varies from 88% to

91%, indicating moderate changes. Instead, the

NOCL sample receives a constant and high recall

value of 94% to 96%. This shows that the NOCL

model is advanced in its ability to correctly identify

true positive types. Figure 2 shows the comparison

of NOCL and CNN accuracy.

CNN has obtained a sample mean accuracy of

91.25%, which varies from 90.26% to 92.16%. On

the other hand, the NOCL model has a stable and

high specificity ranging from 96% to 97%. This

shows that the function of the CNN model varies

according to the experiment counts, but the NOCL

model works with great consistency and reliability.

Figure 2: Comparison of accuracy of NOCL and CNN with

testing count.

Figure 3: Comparison of recall of NOCL and CNN with

testing count.

In the recall value, the CNN sample varies from

88% to 91%, indicating moderate changes. Instead,

the NOCL sample receives a constant and high recall

value of 94% to 96%. This shows that the NOCL

model is advanced in its ability to correctly identify

true positive types.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

Figure 4: Comparison of F1 score of NOCL and CNN with

testing count.

The CNN model's F1-score varies from 88% to

92%, which is consistent with its referral and recall

values. Whereas, the NOCL model has an F1-score

of about 96%, which balances the accuracy and recoil

perfectly.

6 DISCUSSION

The proposed NOCL (Neural Optimization and

Classification Logic) method has significantly better

accuracy and recall than the existing CNN method in

the testing count ranges from 100 to 1000 using

sample test. The output obtained here is having a

high gain as compared with previous studies. The

proposed method consists of 78,456 images in a

collection which includes both testing (25%) and

training images (75%) that improve the disease

detection accuracy. The results obtained in the

research are having a high accuracy, F1 score and

recall in comparing with previous studies.

The maximum accuracy obtained for the NOCL

method is 96.49% and for the CNN method is

91.49%. The overall accuracy of 96% is achieved.

The maximum F1 Score obtained for the NOCL is

96.80% and for CNN is 92%. Similarly, the

maximum recall obtained for the NOCL is 91.50%

and for CNN is 96.20%.

A methodology for leaf infection location

utilizing Brain Improvement and Grouping

Rationale is proposed to upgrade the exactness and

decrease the computational intricacy for ongoing

farming applications (Bhattacharjee et. al., 2020).

The results of this method reveal an average

accuracy of 90.06% (with a low data inaccuracy rate

of 0.062), obtained by pre-processing the data group

and using a hyperparameter precision system

(Ahadian, Krisnanda, 2024). The CNN method has

better classification ability and higher accuracy than

traditional methods such as SVM (Support Vector

Machines). The proposed CNN system allows you to

accurately summarize the main features of the

pathology and accurately diagnose various leaf

diseases (Uddin et.al., 2022). It provides more

efficiency with less computational resources. In test

counts from 100 to 500, the CNN method provides

stable and improved accuracy values from 90.89%

to 91.16%, whereas the SVM method reveals data

only from 88.26% to 90.26%. Further, the CNN

method is designed to provide greater public utility

for direct application in field conditions and

precision solutions for agricultural purposes (He

et.al., 2021). This method reveals improved results

in accuracy, robustness, and computational

efficiency over current SVM systems (Savaş and

Serkan. 2024).

The limitations of this method include the

reliance on a fixed dataset for training the proposed

NOCL model, which may impair the technique's

capability to adapt to new or rare leaf diseases.

Acquiring high-quality images would overburden

the computer resources and could lead to increased

latency in performing live operations. However, this

method can be applied on a larger scale for precision

farming owing to its increased accuracy, reduced

error rate, and extension possibilities. This method is

also particularly suited to smart agriculture and can

be used to diagnose issues in different crop types. In

future studies, advanced methods and inventive

ways may be employed in leaf disease detection.

7 CONCLUSIONS

The proposed model NOCL (Neural Optimization

and Classification Logic) for the leaf diseases

identification is proven to be a significant

improvement over the existing method of CNN

(Convolutional Neural Networks). The NOCL

model performed exceptionally well, achieving

superior accuracy with percentages ranging from

95.89% to 96.87% compared to the 90.26% to

92.16% range of the CNN model, whereas F1 score

of NOCL ranging from 95% to 96.8% compared to

the CNN ranging from 88.2% to 92%. Similarly, the

recall percentages of NOCL ranging from 94.50% to

96% compared to the percentages of CNN from

87.5% to 91.5%. The average accuracy achieved

with the CNN method was 91.25%. However, the

NOCL method demonstrated a significantly higher

mean accuracy of 96.49%. Likewise, the average F1

score and recall of NOCL is higher than CNN.

These statistics display greater dependability and

NOCL Based Automatic Leaf Disease Detection with High Accuracy

accuracy of the model, hence prove that the NOCL

method is more efficient in classification

performance than the others. These statistics display

greater reliability of the model. NOCL clearly

proves to be superior in disease detection in these

scenarios and emphasizes the practicality of the

technique in agriculture.

REFERENCES

Hassan, Sk Mahmudul, 2021. “Identification of Plant-Leaf

Diseases Using CNN and Transfer-Learning

Approach.” Electronics 10 (12): 1388.

Tugrul, 2022. “Convolutional Neural Networks in

Detection of Plant Leaf Diseases: A Review.”

Agriculture 12 (8): 1192.

Alzubaidi, 2023. “A Survey on Deep Learning Tools

Dealing with Data Scarcity: Definitions, Challenges,

Solutions, Tips, and Applications.” Journal of Big

Data 10 (1): 1–82.

Krichen, 2023. “Convolutional Neural Networks: A

Survey.” Computers 12 (8): 151.

Zhang, Zhao, Paulo Flores, Andrew Friskop, Zhaohui Liu,

C. Igathinathane, X. Han, H. J. Kim, N. Jahan, J.

Mathew, and S. Shreya. 2022. “Enhancing Wheat

Disease Diagnosis in a Greenhouse Using Image Deep

Features and Parallel Feature Fusion.” Frontiers in

Plant Science 13 (March):834447.

Thangaraj, Rajasekaran, 2020. “Automated Tomato Leaf

Disease Classification Using Transfer Learning-Based

Deep Convolution Neural Network.” Journal of Plant

Diseases and Protection 128 (1): 73–86.

Demilie, 2024. “Plant Disease Detection and

Classification Techniques: A Comparative Study of

the Performances.” Journal of Big Data 11 (1): 1–43.

Pradhan, 2022. “Comparison of Various Deep

Convolutional Neural Network Models to Discriminat

e Apple Leaf Diseases Using Transfer Learning.” Jou-

rnal of Plant Diseases and Protection 129 (6): 1461–

73.

Nandhini, S., and K. Ashokkumar. 2022. “An Automatic

Plant Leaf Disease Identification Using DenseNet-121

Architecture with a Mutation-Based Henry Gas

Solubility Optimization Algorithm.” Neural Computin

g and Applications 34 (7): 5513–34. “IJSREM –

International Journal of Scientific Research in

Engineering and Management.” n.d. Accessed

December 29, 2024. www.ijsrem.com.

Elhassouny, Smart Mobile Application to Recognize

Tomato Leaf Diseases Using Convolutional Neural

Networks. Infinite Study.

Shoaib, Muhammad, Babar Shah, Shaker Ei-Sappagh,

Akhtar Ali, Asad Ullah, Fayadh Alenezi,

TsankoGechev, Tariq Hussain, and Farman Ali. 2023.

“Corrigendum: An Advanced Deep Learning Models-

Based Plant Disease Detection: A Review of Recent

Research.” Frontiers in Plant Science 14

(September):1282443.

Bhattacharjee, Arup, Samir Kr Borgohain, Badal Soni,

Gyanendra Verma, and Xiao-Zhi Gao. 2020. Machine

Learning, Image Processing, Network Security and

Data Sciences: Second International Conference,

MIND 2020, Silchar, India, July 30 - 31, 2020,

Proceedings, Part II. Springer Nature.

Ahadian, Krisnanda, 2024. “Maize Disease Classification

Using Transfer Learning and Convolutional Neural

Network with Weighted Loss.” Heliyon 10 (21):

e39569.

Uddin, Mohammad Shorif, and Jagdish Chand Bansal.

2022. Computer Vision and Machine Learning in

Agriculture, Volume 2. Springer Nature.

He, Yong, Pengcheng Nie, Qin Zhang, and Fei Liu. 2021.

Agricultural Internet of Things: Technologies and

Applications.

Springer Nature.

Savaş, Serkan. 2024. “Application of Deep Ensemble

Learning for Palm Disease Detection in Smart

Agriculture.” Heliyon 10 (17): e37141.

Bangari, Sindhuja, “A Survey on Disease Detection of a

Potato Leaf Using CNN.” Accessed December 29,

2024.https://ieeexplore.ieee.org/abstract/document/974

2963.

Shi, Yinghuan, 2018. Machine Learning in Medical

Imaging: 9th International Workshop, MLMI 2018,

Held in Conjunction with MICCAI 2018, Granada,

Spain, September 16, 2018, Proceedings. Springer.

M. Mohana Soundarya, S. Jayachitra, “An Efficient Code

Compression Technique for ECG Signal Application

Using Xilinx Software”, International Journal of

Scientific & Technology Research Volume 8, Issue 09,

September 2019 ISSN 2277-8616

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES