Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)

IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

Development of a Multi-Sensor Fusion Framework

2023 2nd International Conference on Automation, Computing and Renewable Systems (ICACRS) | 979-8-3503-4023-5/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICACRS58579.2023.10404410

for Early Detection and Monitoring of Corn Plant

Diseases
Navaneethan S Sampath Jeyaseelan.L Sangavaram Sai kiran
Department of Electronics and Department of Electronics and Department of Electronics and
Communication Engineering Communication Engineering Communication Engineering
Saveetha Engineering College Saveetha Engineering College Saveetha Engineering College
Chennai, India Chennai, India Chennai, India
[email protected] [email protected] [email protected]

Abstract-- The accurate and timely identification of plant Detection and Monitoring of Corn Plant Diseases. The
diseases is critical for ensuring global food supplies and proposed approach intends to revolutionize the detection of
agricultural sustainability. Traditional approaches for diseases in corn crops by integrating multiple sensor
identifying and dealing with corn plant diseases are time- modalities and powerful machine learning algorithms. The
consuming, labor-intensive, and frequently result in late
diagnoses. In regard, the proposed work provides a unique introduction provides an outline of the research work,
multi-sensor fusion system to revolutionize the detection of emphasizing the need for innovative methods in agriculture
corn plant disease. The proposed system combines data from and plant pathology, as well as an overview of the benefits
visible and near-infrared cameras, thermal imaging, of the proposed system. The world's population continues to
hyperspectral imaging, and environmental sensors grow, necessitating an increase in agricultural output to
implementing an integrated approach. The multi -modal data fulfill rising food demand [4, 5]. Corn plant diseases, such
fusion enables advanced machine learning models to identify as fungal and bacterial infections, pose a danger to corn
and classify diseases with greater accuracy. The output stability [6]. Early and accurate disease identification
demonstrated results indicate the system's excellent is critical for minimizing crop losses and resource waste [8].
performance, with an accuracy of 95%, sensitivity of 96%, and
specificity of 94%, exceeding previous methods significantly. Traditional approaches, which rely on human inspection, are
Furthermore, the system exhibits consistent accuracy across naturally prone to errors and fail to give early diagnoses,
several growth phases, ensuring effective disease management. rendering them unsuitable for modern agricultural concerns
Real-time monitoring metrics demonstrate a 5-minute alert [9, 10].
reaction time and a 20% reduction in disease management
costs. The proposed methodology is a game changer,
The proposed study proposes a complete multi-sensor
overcoming the drawbacks of existing methodologies and
offering a solid framework for improving crop health, yield, fusion system as a foundation of a novel technique to
and resource allocation. detecting corn plant disease. The system includes a variety
of sensors strategically positioned in corn fields, such as
visible and near-infrared cameras, thermal imaging devices,
Keywords—Corn Plant Disease, Multi-Sensor Fusion System,
Visible and Near-Infrared Cameras, Thermal Imaging, hyperspectral cameras, and environmental sensors. All of
Hyperspectral Imaging, Environmental Sensors. these sensors work together to collect data on the health of
corn plants and the surrounding environment. A
revolutionary data fusion method combines the input from
I. INTRODUCTION these sensors, improving the accuracy and dependability of
the system. For data analysis, advanced machine learning
Corn is a global agricultural staple, providing a key models, particularly Convolutional Neural Networks for
source of food, feed, and industrial products. However, the image data and Recurrent Neural Networks for time series
health of corn crops is constantly challenged by numerous data, are used. These models are trained on a carefully
diseases, resulting in yield losses and economic difficulties curated dataset that contains images, temperature data,
for farmers [1]. The timely and precise detection of these hyperspectral data, and environmental variables, allowing
diseases is critical for effective disease management, food the system to spot disease patterns and categorize plant
security, and resource allocation [2]. Traditional methods of health [11]. The multi-sensor fusion system has several
visual inspection and physical disease diagnosis are not o nly distinct advantages. It enables early and accurate disease
time-consuming, but frequently result in delayed diagnoses, identification, continuous disease monitoring, and resource-
making urgent interventions impossible [3]. To address efficient disease management. Real-time monitoring enables
these issues, the proposed study presents a ground-breaking rapid responses to disease diagnosis, minimizing crop losses
solution, a Multi-Sensor Fusion Framework for Early and disease management expenses. Finally, the suggested

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 856

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

method contributes not only to higher crop yields but als o to preprocessing were all used in the proposed design. It used
global food security, highlighting its importance in the Otsu's Morphology and Canny Edge detection to segment
context of modern agriculture and the growing demand for images. The use OG histograms extract features, and DCNN
sustainable food production [12]. The study dives into the to classify the disease were discussed. C. Ashwini et al [17]
system's development, methodology, results, and future proposed the Farmers are having difficulty diagnosing
effect, establishing it as a pioneering breakthrough in plant maize crop diseases during farming and manufacturing. The
pathology and precision agriculture. proposed paper discussed the use of grouping and an
improved DL algorithm to detect three common maize leaf
II. LITERATURE REVIEW diseases, including grey spot, rusty, and leaf blight. Disease
prediction and classification require time due to low
foreground and background intensity. According to the
N. Fu et al [13] proposed a plant disease detection and study, the approach had the largest identification impact on
diagnosis using camera images has become possible because corn disease, rust, and grey spot disease observations.
to advances in deep learning-based computer vision. The ResNet18 and VGG-16 provide the best outcomes with
study provided an excellent method for detecting multiple medium diagnostic accuracy. The three corn diseases were
diseases in various plant species. Six vital plants diagnosed with high accuracy in the investigation. It can
diseases can be detected using the technology. It can detect better safeguard agricultural crops and diagnose than other
a wide range of plant diseases. In Using an enormous ways. R. Singh et al [18] explained the Corn leaf disease has
number of images of healthy and diseased leaves of plants, an impact on the food industry and corn crop yields since
system has been trained on deep learning models to detect corn is essential for human nutrition, particularly for
and differentiate plant diseases and their absence. The vegetarians and vegans. To maximize corn quality, it must
trained model has good accuracy and can identify and be protected from a variety of diseases. As a result, early
recognize plant kinds and diseases. Y. Zhang et al [14] plant disease detection and diagnosis are critical. The
explained that the Changes in agricultural ecological proposed article discusses the transfer learning model for
environment, farming system, planting structure and detecting corn plant disease using plant images. In real-time
epidemic conditions have changed as a result of climate images, The trained algorithm can effectively recognize and
warming, creating an ecological setting ideal for harmful categorize maize leaf illnesses. The model is capable of
microorganisms to accumulate, and corn diseases are identifying all four maize leaf diseases with high
growing year by year. The recognition model of corn gray recall, precision and F1-Scores, which is highly promising.
spot and brown spot has been developed using computer The proposed implications for agricultural management and
vision and SVM to target common illnesses and insect pests, disease control could be significant.
and its generalizability is discussed. The simulation results
indicate that the model can identify gray spot and brown Finally, the literature study gave an in-depth look at the
spot of corn, and the disease recognition rate is moderate, current state of research in corn plant disease detection. It
which can be utilized as a prerequisite for corn diseases and underlines the limitations of existing methodologies and
insect pests, giving a technical reference and practical lays the groundwork for the proposed novel multi-sensor
application value. T. Priyaradhikadevi et al [15] proposed fusion system, which promises more accurate and timely
that agriculture is vital in developing countries such as disease management in corn crops.
India. Agriculture accounts for over 60% of rural India's
livelihood intervention. Plant diseases affect plant quality III. PROPOSED SYSTEM
and yield, resulting in food scarcity. Thus, the classification
of plant diseases is critical in agriculture. The The existing method of detecting corn plant diseases
proposed approach discussed the use of image processing
mostly depends on human visual inspections and inadequate
and machine learning to categorize plant leaf disease
environmental monitoring. These procedures are time-
recognition in order to reduce crop production losses. The consuming and labor-intensive and frequently result in late
proposed approach makes use of the Plant Village dataset of
disease identification, which hinders efficient disease
four different categories of healthy and affected leaves.
management. Furthermore, it is susceptible to human error
Plant leaf texture and form were classified using decision and subjectivity, lowering disease identification accuracy
trees and gradient boosting methods. R. K. Vh et al [16]
and reliability. Traditional systems lack the capability of
described that agriculture is a major contributor to India's
continuous real-time monitoring, making rapid intervention
rising GDP. Corn is an important food source as well as a difficult. In the proposed approach, presented a novel multi-
raw ingredient for many industries. Plant diseases reduce
sensor fusion framework for early detection and continuous
production, which is problematic because demand
monitoring of corn plant diseases. The system's ability to
outnumbers supply. The research concentrated on DL-based aggregate data from several sensors, including visible and
Image Processing architecture. Noise reduction,
near-infrared cameras, thermal imaging, hyperspectral
Unsupervised Wiener filtering, and Entropy image
imaging, and environmental sensors, is its major

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 857

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

distinguishing feature. The Data fusion combines be sent to farmers or appropriate authorities, facilitating
information from several sensors, then fused data is prompt action. The implementation of the proposed system
processed by machine learning algorithms to identify has several important benefits over existing systems. The
disease patterns and environmental impacts, enabling for proposed technology detects corn plant diseases early by
early disease identification. While the feature extraction combining data from various sensors and utilizing advanced
decreases data dimensionality and extracts useful machine learning models, allowing for prompt intervention
information from sensor data. By collecting a holistic and preventing disease spread. The system's real-time
perspective of corn plant health, these techniques improve monitoring capacity ensures that corn fields are always
disease detection accuracy and reliability. The proposed monitored, enabling continuous disease management
approach is shown in Figure 1 and an intended for real-time throughout the growing period. The combination of sensor
monitoring, delivering a constant stream of data and data with machine learning models improves disease
enabling early intervention to reduce disease transmission detection accuracy and reliability while minimizing false
and crop loss. positives and negatives. Farmers may more effectively
utilize resources by focus ing on disease-infected areas,
minimizing the need for widespread application of
pesticides and saving both time and resources. Timely
disease management and enhanced disease identification
accuracy result in higher crop yields, which contribute to
food security and economic stability. Finally, the proposed
multi-sensor fusion framework for corn plant disease
identification offers an immense advancement in agriculture
and plant pathology. The proposed model improves
precision agriculture by giving real-time data on the health
and disease status of corn plants. Through early disease
detection and precise plant health assessment, it enables
timely interventions, optimizes resource allocation, and
ultimately enhances agricultural productivity and
management.

IV. METHODOLOGY

Fig.1: Flowchart of the proposed system The methodology proposed herein offers an innovative
and comprehensive approach for detecting and monitoring
The proposed system's main implementation processes corn plant diseases early and continuously. The proposed
are Sensor Selection, Data Collection, Data Fusion technology intends to improve disease management in corn
Algorithm, Machine Learning Models, and Real-time fields by integrating varied sensors, data fusion algorithms,
Monitoring and Alerts. The first step is to carefully choose and advanced machine learning techniques, ultimately
and integrate the sensors into the system. The improving agricultural yields and global food security.
system comprises the installation of visible and near-
infrared cameras, thermal imaging devices, hyperspectral A. Sensor Selection and Installation
cameras, and environmental sensors including humidity, First, a variety of sensor types are chosen in order to
solar radiation and rain gauge sensor in corn fields. collect various data about the corn plants and their
Throughout the growth cycle, the system continuously surroundings. The system comprises visible and near-
collects data from the sensors, capturing numerous features infrared (NIR) cameras, thermal imaging devices,
of the corn plants and their environment. Data is collected hyperspectral imaging cameras, and environmental sensors
on a regular basis, resulting in a rich and dynamic including humidity, solar radiation and rain gauge sensor.
collection. To merge the data from the various sensors , a To achieve complete coverage, these sensors are
novel data fusion technique is developed. By combining the strategically placed in corn fields. These sensors are
unique features of each sensor modality, the fusion process calibrated using established procedures for each sensor type
improves the system's ability to recognize disease-related to ensure reliable data capture.
patterns. To discover disease patterns and classify the health
state of corn plants, machine learning models such as B. Data Collection and Preprocessing
CNN and RNN are trained on fused data. The equipment to Data collection is an ongoing procedure that takes
monitor the corn fields in real time is in field. It place throughout the growth season. Sensors collect data at
continuously examines data and broadcasts timely alerts regular intervals, generating a wealth of data. Preprocessing
when it identifies disease symptoms. These notifications can data is critical to ensuring data quality. For visible and near-

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 858

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

infrared cameras, well-established image processing Machine learning models, notably CNN for image data
techniques are used to correct lighting and atmospheric and RNN for time series data, are used for disease detection
conditions. Thermal imaging data is preprocessed in order to and plant health classification. These models were chosen
remove noise and calibrate temperature values. Standard for their ability to handle the complicated, multi-modal
calibration techniques are used to convert hyperspectral data nature of the fused data. To ensure high-quality,
into reflectance values. Data from environmental sensors is standardized data for machine learning model training, the
recorded without any preprocessing. Wireless connections, datasets are formulated by collecting various sensor data,
such as Wi-Fi and cellular connectivity, enable data manually labeling it for disease detection, augmenting and
transmission from sensors. Following that, data from many balancing the data, splitting it into training, validation, and
sensors is gathered in a central server as shown figure 2, test sets, and applying feature extraction and normalization
which can be on-site or remotely situated, depending on the techniques. The system is trained on a labeled dataset that
scope and requirements of the proposed work. includes images, temperature data, hyperspectral data, and
environmental variables. The training dataset contains
samples of healthy and diseased corn plants at various
phases of growth. Transfer learning is used by initializing
the models with pre-trained architectures, ResNet for CNN
and LSTM-based RNN, to ensure model convergence and
efficiency.
E. Disease Detection and Classification
The trained machine learning models are utilized to
detect and classify diseases. Images from visible and near-
infrared cameras are inputted into CNN, which learns to
recognize disease-related patterns and predicts the health
status of the plant. RNNs are used to process data on
temperature, hyperspectral data, and environmental
variables. These data are then fed into RNN models, which
predict the progression of disease and the influence of the
environment. The predictions of both CNN and RNN are
combined in order to provide a more complete analysis of
the state of the plant's health.
F. Real-time Monitoring and Alerts
The Proposed system is utilized to monitor corn fields
in real time. The sensors' data are continually fed into th e
machine learning models. When the system detects
symptoms of disease or substantial changes in
environmental conditions, alert notifications are sent via
Fig.2: Visual representation of proposed system SMS and email. When the system identifies disease or
environmental changes, it notifies farmers and monitoring
C. Data Fusion Algorithm Development authorities in real time, enabling immediate and effective
The data fusion algorithm is the heart of the disease management decisions.
proposed methodology. The proposed approach involves G. Evaluation and Validation
developing a novel algorithm that efficiently combines data To demonstrate the efficacy of the proposed approach,
from many sensor modalities. Data from visible and near- extensive field testing in real corn fields is done for a period
infrared cameras, thermal imaging, hyperspectral imaging, of 90 days from seeding to yield. To evaluate the system's
and environmental sensors are combined by the algorithm. performance against traditional disease detection
The algorithm utilizes mathematical approaches such as methods, accuracy, sensitivity, and specificity are utilized.
weighted averaging and principal component analysis In addition, the system's ability to detect disease early and
(PCA), the unique data fusion program that merges data its influence on crop health and productivity are also
from many sensor modalities. Weighted averaging evaluated.
prioritizes the data from each sensor, enhancing the fusion
process. PCA decreases the dimensionality of data while By using the comprehensive approach, the proposed
retaining important information, allowing for successful system can enhance accuracy and detect corn diseases
integration. These algorithms improve the ability to detect earlier by utilizing cutting-edge machine learning algorithms
disease patterns and environmental elements that are in addition to utilizing the benefits of many sensors and data
important for plant health evaluation. fusion. The proposed method provides a solid solution to the
D. Machine Learning Model Selection and Training issues that present disease detection systems encounter,

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 859

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

ultimately leading to more efficient disease management approaches for corn plant disease detection. The proposed
and improved crop yields. system outperforms all other approaches, achieving the
highest accuracy (95%) representing overall precision,
sensitivity (96%) representing disease detection capabilities,
and specificity (94%) representing healthy plant
V. RESULT AND ANALYSIS identification. These metrics evaluate the model's capacity
to identify and monitor corn plant diseases at an early stage.
The section delves into results and analysis of the This superior performance is attributed to the utilization of
proposed novel multi-sensor fusion system for detecting and sophisticated machine learning models and the fusion of
monitoring corn plant disease. The proposed technology data from multiple sensors, enabling more accurate and
outperforms previous methods in terms of accuracy, timely disease identification. The peer researcher discovered
sensitivity, and specificity, as demonstrated by rigorous SURF-BVWF Approach [3] and ML & DL Approach [2]
testing. And also look at how it performs at different stages exhibit moderate performance, while DL Approach [1], Pre-
of growth, emphasizing its adaptability. Furthermore, trained DL [7] and PCA & SVM Approach [5] have lower
the system's real-time monitoring capabilities illustrate rapid accuracy, sensitivity, and specificity, highlighting the
alert response times and significant cost savings, placing it superiority of the proposed multi-sensor fusion system in
as a vital innovation in precision agriculture and crop health enhancing corn plant disease detection.
management.
T ABLE II: DISEASE DET ECT ION BY GROWT H ST AGE
T ABLE I: PERFORMANCE COMPARISON
Growth Stage Detection Accuracy
Sensitivity Early Growth 92%
Approach Accuracy Specificity Mid-Growth 96%
(Recall)
Proposed approach 95% 96% 94% Late Growth 94%
DL [1] 86% 80% 84%
ML & DL [2] 90% 88% 88% Table II shows the system's disease detection accuracy
SURF-BVWF [3] 91% 87% 90% varying depending on the stage of growth. It has a high
PCA & SVM [5] 76% 83% 81% accuracy 96% at the mid-growth stage, ensuring effective
Pre-trained DL [7] 86% 84% 86% disease management during the critical stage of plant
development. It also retains high accuracy over the early and
late stages of growth, with 92% and 94%, respectively. The
proposed system's ability to adapt to different stages of
Pre-trained DL [7] growth is a significant advantage.

T ABLE III: REAL-T IME MONIT ORING MET RICS

PCA & SVM [5]

SURF-BVWF [3] Metric Value

Specificity Average Alert Response Time (minutes) 5
ML & DL [2] Sensitivity Disease Management Cost Savings (%) 20%

Accuracy
DL [1] Table III shows the proposed system's real-time
monitoring capacity is highly efficient, with an average alert
Proposed approach
response time of only 5 minutes. Because of the rapid
response time, farmers can take prompt action upon disease
diagnosis, contributing to effective disease management.
0% 100% 200% Furthermore, the technique results in a 20% cost savings in
disease management. The proposed strategy not only
increases crop productivity but also optimizes resource
allocation by eliminating the requirement for widespread
Fig.3: Graphical representation of performance comparison of
proposed vs various approaches
pesticide application and crop loss.

In conclusion, the performance comparison clearly

Table I and Figure 3 presents a comparative analysis of
the proposed multi-sensor fusion system and five alternative shows that the proposed multi-sensor fusion system
outperforms traditional methods in terms of accuracy, early

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 860

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Second International Conference on Automation, Computing and Renewable Systems (ICACRS-2023)
IEEE Xplore Part Number: CFP23CB5-ART; ISBN: 979-8-3503-4023-5

disease diagnosis, and disease detection throughout multiple Learning,” International Conference on Futuristic Technologies
(INCOFT), Nov. 2022, doi: 10.1109/incoft55651.2022.10094326.
growth stages. The system's capacity to give timely alerts
[3] R. Dijaya, N. Suciati, and A. Saikhu, “Corn Plant Disease
and cost savings is highlighted by the real-time monitoring Identification Using SURF-based Bag of Visual Words Feature,” 14th
data, making it a vital tool for precision agriculture and International Conference on Information Technology and Electrical
improved crop health management. Engineering (ICITEE), Oct. 2022.
[4] S. V. Militante, B. D. Gerardo, and N. V. Dionisio, “ Plant Leaf
Detection and Disease Recognition using Deep Learning,” 2019 IEEE
Eurasia Conference on IOT, Communication and Engineering
VI. CONCLUSION (ECICE), Oct. 2019, doi: 10.1109/ecice47484.2019.8942686.
[5] Z. Liu, Z. Du, Y. Peng, M. T ang, X. Liu, and W. Chen, “ Study on
corn Disease identification based on PCA and SVM,” 2020 IEEE 4th
In conclusion, a substantial improvement in the fields Information Technology, Networking, Electronic and Automation
of agriculture and plant pathology can be shown by the Control Conference (ITNEC), Jun. 2020.
proposed Multi-Sensor Fusion Framework for Early [6] Y. Cao, J. Macalisang, and A. A. Hernandez, “Multi-stage T ransfer
Learning for Corn Leaf Disease Classification,” IEEE International
Detection and Monitoring of Corn Plant Diseases. The Conference on Automatic Control and Intelligent Systems (I2CACIS),
limitations of traditional disease detection methods are Jun. 2023, doi: 10.1109/i2cacis57635.2023.10193168.
addressed by the proposed reliable and effective system, [7] Baliyan, V. Kukreja, V. Salonki, and K. S. Kaswan, “ Detection of
which have developed through the integration of many Corn Gray Leaf Spot Severity Levels using Deep Learning
Approach,” 9th International Conference on Reliability, Infocom
sensors and cutting-edge machine learning techniques. The
Technologies and Optimization (Trends and Future Directions)
results obtained demonstrate that the system outperforms (ICRITO), Sep. 2021, doi: 10.1109/icrito51393.2021.9596540.
traditional techniques, with an accuracy of 95%, sensitivity [8] P. Srivastava, S. P. Singh, A. Vishnoi, and V. Sapra, “ Classification
of 96%, and specificity of 94%. The utilization of cutting- of Corn Leaf Diseases using Various Pre-trained Deep Learning
edge machine learning models and multi-modal data fusion Networks and Performance Comparison,” International Conference
on Advances in Computing, Communication and Materials
are responsible for the exceptional performance because (ICACCM), Nov. 2022, doi: 10.1109/icaccm56405.2022.10009476.
these techniques enable continuous monitoring and early [9] P. Shenbagam and N. Sanjana, “Corn Leaf Disease Detection; A
disease identification regardless of the stage at which corn Survey,” 2022 International Conference on Inventive Computation
Technologies (ICICT), Jul. 2022.
crops are growing. Despite its impressive powers, the [10] S. Singh, A. Verma, V. Guleria, S. Yadav, and N. Singh, “ Deep
proposed technology has several limits. The deployment of Learning-Based Networks to Detect Leaf Disease in Maize and
several sensors, as well as the requirement for a constant Corn,” International Conference on IoT, Communication and
power supply, can be resource-intensive, making it less cost- Automation Technology (ICICAT), Jun. 2023.
[11] Ahmad, A. E. Gamal, and D. Saraswat, “T oward generalization of
effective for small-scale farmers. Furthermore, the
Deep Learning-Based plant disease identification under controlled
performance of the system may be influenced by and field conditions,” IEEE Access, vol. 11, pp. 9042–9057, Jan.
environmental circumstances and the availability of training 2023, doi: 10.1109/access.2023.3240100.
data for specific disease kinds. Future research could focus [12] F. Aness, R. Fernandes, B. Shaikh, K. Shaikh, and N. Shaikh, “ Leaf
on improving the system for resource-constrained situations disease detection,” 2022 6th International Conference on Computing,
Communication, Control and Automation (ICCUBEA, Aug. 2022.
and building more robust disease detection models to [13] N. Fu, C. Wang, and X. Ji, “Study on Visual Detection Device of
increase the usability and accessibility of the proposed Plant Leaf Disease,” IEEE International Conference on Mechatronics
technology. Furthermore, broadening the system's reach to and Automation (ICMA), Aug. 2019.
[14] Y. Zhang, T . Ren, C. Liu, C. Wang, and J. T ian, “ Research on
include a greater range of crop kinds and diseases is a Recognition Method of Common Corn Diseases Based on Computer
promising direction. Given the sensitive nature of Vision,” 11th International Conference on Intelligent Human -
agricultural data, emphasizing data security and privacy Machine Systems and Cybernetics (IHMSC), Aug. 2019.
safeguards will also be critical. Furthermore, connecting the [15] T . Priyaradhikadevi, K. Seshadri, T . Ragupathi, S. Prasanna, K.
Madhan, and R. S. M. Ananthi, “Leaf Disease Detection Using
system with automated disease control system, such as
Machine Learning Algorithm,” Eighth International Conference on
robotic crop treatment, might give farmers with a Science Technology Engineering and Mathematics (ICONSTEM) ,
comprehensive end-to-end solution. Overall, the future of Apr. 2023, doi: 10.1109/iconstem56934.2023.10142564.
the proposed technology holds significant promise for [16] R. K. Vh and R. T hamizhamuthu, “A Combined Architecture of
transforming agricultural disease management, improving Image Processing T echniques and Deep Neural Network for the
Classification of Corn Plant Diseases,” Third International
food security, and contributing to sustainable crop Conference on Artificial Intelligence and Smart Energy (ICAIS), Feb.
production. 2023, doi: 10.1109/icais56108.2023.10073762.
[17] Ashwini and V. Sellam, “Analyzing T he Prediction Accuracy Of
Corn Leaf Diseases Using A Pre-T rained Network Model,”
REFERENCES International Conference on Advancement in Computation &
Computer Technologies (InCACCT), May 2023.
[1] V. K. R. Kokatam and A. S. A. Doss, “Prediction of Corn and Tomato
[18] R. Singh, N. Sharma, and R. Gupta, “Classification and Detection of
Plant Diseases Using Deep Learning Algorithm,” International
Corn Leaf Disease using ResNet 18 T ransfer Learning Model,” 8th
Conference on Advancements in Smart, Secure and Intelligent
International Conference on Communication and Electronics Systems
Computing (ASSIC), Nov. 2022.
(ICCES), Jun. 2023, doi: 10.1109/icces57224.2023.10192642.
[2] V. R. Kumar, V. Shrishti, and P. Sridhar, “Corn Plant Disease
Classification using a combination of Machine Learning and Deep

979-8-3503-4023-5/23/$31.00 ©2023 IEEE 861

Authorized licensed use limited to: ULAKBIM UASL ISTANBUL TEKNIK UNIV. Downloaded on December 09,2024 at 05:56:37 UTC from IEEE Xplore. Restrictions apply.