Automatic Leaf Health Monitoring with an IoT Camera System based

on Computer Vision and Segmentation for Disease Detection

RICARDO YAURI, ANTERO CASTRO, RAFAEL ESPINO

Facultad de Ingeniería,

Universidad Tecnológica del Perú,

Lima,

PERÚ

Abstract: - Manual identification of diseases in crops is costly and subjective, driving the need for automated

systems for accurate detection in the field. This requires the use of technologies based on the integration of IoT

and deep learning models to improve the assessment capacity of crop health and leaf disease, with continuous

monitoring. The literature review highlights technological solutions that include weed and disease detection

using artificial intelligence and autonomous systems, as well as semantic segmentation algorithms to locate

diseases in field images whose processes can be improved with systems based on microcontrollers and sensors.

This research implements a leaf health monitoring system using IoT and AI technologies, with the development

of an IoT device with a camera, the configuration of an MQTT broker in NODE-Red, and the implementation

of a script in Python for leaf instance segmentation and image display. As a result, it is highlighted that image

analysis, with the Python tool, allowed obtaining valuable information for precision agriculture, while the

visualization or messaging interface allows health monitoring and management of crops. In conclusion, the

System adequately performs image capture, processing, and transmission, being a contributes to precision

agriculture solutions, considering that this can be improved with the integration of more complex deep learning

algorithms to increase precision.

Key-Words: - Computer vision, Segmentation, ESP32CAM, leaf health, Precision agriculture, IoT, Node-RED.

Received: April 25, 2024. Revised: October 28, 2024. Accepted: November 13, 2024. Published: December 16, 2024.

1 Introduction

Currently, processes related to the area of smart

agriculture are integrating technologies such as the

Internet of Things (IoT) and Deep Learning Models

(DLMs) to develop solutions that help determine the

health status of crops, [1], [2] and level. of leaf

disease in the fields efficiently, thus improving the

quality of agricultural production, [3], [4]. On the

other hand, the growth of industrialization and

urbanization has led to a decrease in agricultural

land worldwide, which makes it important to use

advanced techniques that optimize resource

consumption compared to traditional agricultural

systems, [5], [6]. For this reason, continuous

monitoring is necessary to detect the evolution of

crop leaf diseases by performing their early

classification, which is essential to promote healthy

agricultural production, determine the appropriate

use of water, and control weeds and diseases, as

well as decision-making, [7], [8].

The traditional identification of crop diseases

through visual inspection is complex because it

consumes human resources that have a degree of

subjectivity and inaccuracy, making manual

monitoring a process that demands significant

effort, [3], [9]. On the other hand, the automated

detection of diseases using sensors is a contribution

to monitoring which is strengthened by taking

images in the field, where one of the challenges is

determining the precise location of the diseased

areas on the leaves of the crops through image

segmentation, [7], [10], [11]. Furthermore, there is a

problem in simultaneously monitoring numerous

parameters and diagnosing plants, which makes it

necessary to use assistance systems in a smart

agriculture that integrate IoT, [12], and AI, [4], [5]

technologies, to recognize how environmental

variables affect crops, plants, and diseases, [8], [13],

Although technologies strengthen new plant

cultivation and breeding practices, they could be

expensive and require highly skilled labor, making

their adoption by small farmers difficult, [6], [14],

[15].

2 Literature Review

In the literature review, research is identified that

overcomes technological problems by integrating

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

148

Volume 15, 2024

techniques for object detection, artificial intelligence

and autonomous systems aimed at identifying

weeds, [1], [16], [17] through devices that

incorporate automatic identification technologies,

[18], [19], adaptation of machine learning models

for execution on devices with limited resources,

[20], [21] and use of data obtained by IoT sensors

and UAVs (Unmanned Aerial Vehicle), [22]. On the

other hand, in the case of field images, segmentation

algorithms are integrated for disease localization,

[3], [10], along with the development of techniques

using images with complex backgrounds and

Generative Adversarial Neural Networks (GAN),

[11]. The integration of the aforementioned

technologies allows automatically detecting

anomalous patterns in real time, improving

agricultural management, [9], [23], where it is

common to use systems based on ESP32

microcontrollers, temperature and humidity sensors,

[24]. Among the important limitations of the

reviewed literature is its application in controlled

environments, which reduces its usefulness in real

conditions. In addition, the adaptation of models for

devices with limited resources sacrifices precision,

and the use of GANs implies a high computational

demand. On the other hand, traditional IoT sensors

and UAVs present latency problems and energy

inefficiency for image acquisition and analysis, so

this work focuses on analyzing these deficiencies.

The methodologies found in the research are

aimed at using architectures based on convolutional

neural networks (CNN), Multi-Model Fusion

Networks (MMF-Net) to classify diseases of corn

leaves in the field of precision agriculture, [1], [7],

or tomato crops [5]. Other articles describe the use

of cameras integrated into embedded systems such

as CanopyCAM, which can continuously and

accurately monitor canopy cover in crops, using

image processing algorithms and IoT technology,

followed by field tests and comparison. Of the

results with conventional methods, [4], [8], [25],

[26]. On the other hand, some research integrates

the use of IoT systems that use semantic

segmentation methods to identify diseased parts in

leaf crops, and compare their performance with

other methods [27], using various methods such as

FCN-8s, SegNet, DeepLabv3, and U-Net,

evaluating performance, [1], [3].

The research results show that the proposed

systems, such as MMF-Net, CanopyCAM, and AI-

SHES with IoT, allow the classification and

detection of diseases in agricultural crops with an

accuracy greater than 90%, [5], [7], [8].

Furthermore, it is mentioned that future research

should focus on optimizing image processing

algorithms for greater accuracy under different

lighting and environmental conditions, [1], [3], [8].

Furthermore, continued exploration is required to

further improve the accuracy and efficiency of these

proposed systems, as well as their implementation

and validation in real agricultural environments, [4],

[5], [7].

Therefore, we have the following research

question: How can a leaf health monitoring system

be implemented by integrating IoT and AI

technologies for disease detection? To answer the

question, the objective is to develop an automated

system to monitor the health of leaves using an IoT

device with an integrated camera, using computer

vision and segmentation techniques to detect

diseases. For this, it is necessary to carry out the

following activities: Design an IoT device with a

camera based on the ESP32CAM module, configure

an MQTT (Message Queuing Telemetry Transport)

broker in NODE-Red to store the captured images,

implement a Python script to process the images and

perform the segmentation of the leaves to determine

the percentage of area affected by diseases and

visualize the images in the NODE-Red graphical

interface. Furthermore, the motivation of the

proposal is aimed at having an automated and low-

cost process to contribute to the productivity and

quality of crops that can be used by farmers and

people interested in more effective management.

The value of the research is in the integration of

IoT communication technologies based on MQTT

and image instance segmentation processes to detect

the evolution of diseased areas on the leaves. The

paper is organized into 5 sections as follows:

Section 1 presents the introduction, section 2 with

the methodological process, and section 3 shows the

results. The paper ends with section 4 of the

discussions and section 5 of conclusions.

3 Methodology

The proposed system uses the camera integrated

into the IoT device to acquire images, which are

then transmitted to a centralizing processing and

management team, which identifies the diseased

regions and calculates the affected percentage.

These images are sent via instant messaging service

applications for manual monitoring (Figure 1).

Finally, the results are visualized in a graphical

interface integrating image segmentation processing

techniques with deep learning.

The prototype is scalable to broad and diverse

agricultural environments thanks to its

implementation through a web API, which allows

integration of multiple IoT devices for real-time

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

149

Volume 15, 2024

monitoring. Its model based on instance

segmentation can adapt to different crops with

minimal adjustments, optimizing large-scale

analysis and generating customized agricultural

strategies to detect diseases in varied conditions.

3.1 Architecture

The components and techniques that make up the

system correspond to the determination of the

architecture. The proposed system is composed of:

a) an IoT device with an integrated camera for the

periodic sending of images obtained in the field.

Fig. 1: Scheme for the system proposal

b) The data management system implemented with

the Node-Red software platform and an image

reception service using the MQTT protocol and a

communication broker called Mosquitto.

c) A Python script, which is executed to process

the received images applying semantic

segmentation model (IFast) and is trained in the

Roboflow Web tool using the COCO (Common

Objects in Context) checkpoint, which allows

identifying the diseased regions and calculating

the affected percentage.

d) Display and messaging interface for monitoring

using Node-RED and sending messages via

Telegram. These integrated components are seen

in Figure 2, improving the accuracy and efficiency

of disease detection.

3.2 IoT Device for Image Capture

The main component of the system hardware is the

IoT device composed of the ESP32-CAM module,

which is a low-power consumption device that

integrates a camera. The integration of the ESP32-

CAM with the OV2640 camera allows images to be

captured at a resolution of 1600x1200 pixels and

supports various formats such as JPEG and BMP.

Fig. 2: Architecture of the proposed system

Configuration and communication with the

camera are conducted with the I2C communication

protocol (inter circuit communication). After

capturing the image, it is sent to the ESP32 using

the DVP digital interface, synchronizing with the

HSYNC pins (to determine the start of a new image

line) and the PCLK (which provides the clock signal

to transmit the pixels). Finally, the ESP32 performs

preprocessing before displaying the image (Figure

3).

Fig. 3: ESP32CAM IoT Device

The device control algorithm initializes the I2C

communication protocol with the camera and sets

the Wi-Fi communication type to client mode. Then

the access credentials to the HiveMQ MQTT

Broker, [28] and the publication topic that has the

name generated for the project called:

“esp32cam/pic”. The camera captures images using

the Jpeg format and then a timer runs the streaming

function. Captured images are encrypted before

being published to an MQTT server. Finally,

mechanisms are used to manage errors and

reconnect with the MQTT Broker in case of

transmission failures (Figure 4).

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

150

Volume 15, 2024

Fig. 4: State diagram of the capture and

transmission algorithm

3.3 Segmentation Model for Leaf Disease

To use the segmentation model, it is necessary to

conduct a series of processes to identify the areas

affected by diseases on the leaves. The steps to

implement this technique are as follows:

 Instance segmentation: To develop the disease

detection model in plant leaves, the instance

segmentation technique was used, which

differentiates between classes of objects and

objects within each class. This technique first

performs object detection to find all bounding

boxes in an image. Semantic segmentation is then

applied within each rectangle that classifies each

pixel into a class. In this way, it is possible to

differentiate elements and their limits by detecting

the areas affected by diseases on the leaves (Figure

5).

Fig. 5: Instance segmentation

 Use of Roboflow: To implement the model, the

Roboflow platform was used, which has a set of

images of diseased leaves already labeled with the

affected areas. Then, these images are used to

generate the instance segmentation model, using

the Roboflow 3.0 model type (machine learning

framework version) and the COCO checkpoint

that allows efficient learning from the pre-existing

data set (Figure 6). This model will be accessible

through a web communication API for the

inference process.

To deploy the instance segmentation model to

detect disease-affected areas on leaves, a process

was followed that guarantees accuracy and

reproducibility. The Roboflow platform was key in

the implementation, working with a set of

previously labeled images, where the affected

areas were already delimited and there is a pre-

created model accessible with a web API KEY.

The model already existing on the platform has an

average accuracy (mAP) of 92%, showing a high

capacity to detect and segment affected areas even

in images with complex backgrounds. For

inference, the model was deployed in a web API,

which facilitates its integration into IoT systems

and its real-time use in agricultural applications.

 Python script for segmentation: it is used to import

the model, which integrates image processing,

segmentation, and confidence percentage. This

script reads the images stored by the Node-RED

application. Processing is performed using an

HTTP client to make inferences through the

RoboFlow API. Then, for each prediction, a

polygon is drawn over the image, and class and

confidence labels are added. Finally, the image is

stored with a detailed analysis of the affected

areas, performing the process every 15 seconds

(Figure 7).

Fig. 6: Model creation process with RoboFlow

3.4 IoT Data Management and Processing

Platform

The components necessary for data management

and image processing include the MQTT broker

HiveMQ and Node-RED software. In addition,

connection services with the instant messaging

platform Telegram are used to send the processed

image. The activities in this stage are the following:

 Node-RED is responsible for receiving the images

and data sent from the ESP32CAM module

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

151

Volume 15, 2024

through the MQTT broker. Subsequently, this data

is stored on the hard drive.

 Visualization of the original and processed images

in the NODE-Red graphical interface, which

facilitates making informed decisions about plant

management.

 Integration and sending of notifications using

Telegram by detecting an information request

command “/annotated” (Figure 8).

Fig. 7: Python script to import the model

Fig. 8: Program by node flow in Node-RED

4 Results

The model is already available on the Roboflow

platform, however, for future implementations, a

proprietary segmentation model can be developed to

optimize the adaptation of the system to other

scenarios. On the other hand, although exhaustive

evaluations of metrics such as processing time and

hardware performance were not conducted, this is

because the main objective is to build a prototype

that would demonstrate the technical viability of the

system.

The ESP32 IoT device was installed on a tripod

mechanical structure which is protected by an IP65

protection box for external environments and then

placed in a field environment for pilot testing. In

this case, it powered up and began capturing images

of a bush and transmitting them over Wi-Fi using

the MQTT protocol (Figure 9).

The Python script was evaluated by generating

images representing: the image of the bush, the

image with the detected regions of disease

highlighted in black, and a third image showing the

confidence data about the detected diseased areas

(Figure 10).

The working Node-RED application is shown in

Figure 11. This shows previous images received in

the node streams. In addition, a screenshot of the

Node-RED GUI visual interface is observed that

shows both the original image received and the

segmented image with the detected disease.

Fig. 9: Pilot test of the IoT Device

Fig. 10: Images of diseased leaves detected in

python

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

152

Volume 15, 2024

Fig. 11: GUI with the original leaf image and

detected diseased section

(a)

(b)

Fig. 12: Telegram app for (a) chatbot setup and (b)

status response

Figure 12 demonstrates the operation of

communication with the Telegram instant

messaging application. This application receives

images of the detected diseased region when the

user sends a command. For this, a chatbot, created

with BotFather in Telegram, was used to automate

this process, allowing users to send a specific

command (/annotated) and receive the processed

image.

5 Discussion

The integration of the IoT device into the

mechanical structure shows the practical application

of image transmission to a computer vision stage,

intended for monitoring and analyzing vegetation in

natural environments. The ESP32CAM module is

integrated with remote applications efficiently using

the MQTT protocol for high-resolution image

transmissions of the bush. This allows this data to be

processed in real-time to segment and classify

different elements within the framework with a

confidence level assigned to each class.

Running the Python script demonstrates the

system's ability to identify and highlight disease-

affected areas. This script analyzes the captured

images, segments the diseased areas, and overlays

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

153

Volume 15, 2024

the confidence information on the original image,

providing a clear and detailed visualization of the

health status of the shrub. This result is crucial for

applications in precision agriculture, allowing

farmers to make informed decisions about crop

management.

The operation of the GUI highlights the efficient

integration between the IoT device and the Node-

RED platform, facilitating the visualization and

analysis in real-time of images of diseased leaves

and their evolution. The connection with the

Telegram application shows the effectiveness of

using the disease detection system, providing an

accessible and easy-to-use tool for monitoring.

Automation through chatbots improves response

capacity and decision-making.

The proposed solution is distinguished from

other methods described in the literature because

while some have developed systems such as MMF-

Net, CanopyCAM, or GAN-based models for

disease segmentation and detection, these present

challenges in real environments due to their high

computational demand, lack of adaptation to

uncontrolled conditions and latency problems. The

developed prototype uses an IoT device integrated

into a structure for external environments and

optimized to transmit images using the MQTT

protocol. Although the prototype does not yet

include exhaustive evaluations of processing time

and hardware performance, its modular design can

be scaled for future optimizations.

Hardware optimizations have been left as future

work to consolidate the solution in operational

environments and ensure its maximum efficiency.

This planning allows current efforts to focus on the

initial validation of the approach and to

progressively move towards comprehensive

improvements.

6 Conclusion

In this research, an automated system was

developed to monitor the health of leaves,

demonstrating its proper integration of an IoT

camera module with the ESP32CAM, using

computer vision and segmentation techniques to

detect diseases. The implementation of the IoT

device, the MQTT broker in Node-RED, and the

Python script were properly integrated for the

capture, processing, and real-time visualization of

images, contributing to precision agriculture

solutions.

The construction of the prototype demonstrated

the technical feasibility of the system for disease

detection in leaves, but the integration of additional

environmental data (such as humidity and

temperature), models such as transformers, native

mobile interfaces, or extensive field testing would

have diverted the focus to complex aspects that are

outside the initial scope of the project. These

activities are considered as future research to

optimize, validate, and scale up the system in real

environments and improve its accessibility.

The IoT module allows the automatic capture of

images in the field, with the use of the MQTT

protocol being an appropriate form of transmission.

In the case of using Node-RED with MQTT, the

storage and management of the images captured by

the ESP32CAM module was facilitated, showing

the system's ability to manage and provide

information in agriculture.

The use of Python allows for a flexible and

simple stage for identifying areas affected by

diseases on leaves using segmentation services in

the cloud, overlaying trusted information. In

addition, its proper functioning with the Node-RED

platform was demonstrated by writing and reading

images. Connecting the system with the Telegram

application is an effective way to automate the

monitoring process, as it is conducted at the user's

request but can easily be updated to become

automatic.

As future improvements, more complex deep

learning algorithms can be integrated to increase the

accuracy of detecting disease types. Additionally,

the integration of additional sensors, such as soil

moisture and temperature, provides a more complete

analysis of environmental conditions and their

impact on plant health through multimodal analysis.

It would also be beneficial to develop a friendlier

user interface to make it easier for farmers to use the

system. Finally, testing is suggested in different

agricultural environments to validate the robustness

of the system.

Acknowledgement:

This research was supported and funded by

Universidad Tecnológica del Perú

References:

[1] T. Nakabayashi, K. Yamagishi, and T. Suzuki,

“Automated Weeding Systems for Weed

Detection and Removal in Garlic / Ginger

Fields,” Int. J. Adv. Comput. Sci. Appl., vol.

15, no. 4, pp. 103–109, 2024, doi:

10.14569/IJACSA.2024.0150411.

[2] Q. Dong, R. Gu, S. Chen, and J. Zhu, “Apple

Leaf Disease Diagnosis Based on Knowledge

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

154

Volume 15, 2024

Distillation and Attention Mechanism,” IEEE

Access, vol. 12, pp. 65154–65165, Jan. 2024,

doi: 10.1109/ACCESS.2024.3397329.

[3] G. Balram and K. K. Kumar, “Crop Field

Monitoring and Disease Detection of Plants in

Smart Agriculture using Internet of Things,”

Int. J. Adv. Comput. Sci. Appl., vol. 13, no. 7,

pp. 819–826, 2022, doi:

10.14569/IJACSA.2022.0130795.

[4] K. K. Sarma, K. K. Das, V. Mishra, S.

Bhuiya, and D. Kaplun, “Learning Aided

System for Agriculture Monitoring Designed

Using Image Processing and IoT-CNN,” IEEE

Access, vol. 10, no. 1, pp. 41525–41536,

2022, doi: 10.1109/ACCESS.2022.3167061.

[5] S. V. S. Ramakrishnam Raju, B. Dappuri, P.

Ravi Kiran Varma, M. Yachamaneni, D. M.

G. Verghese, and M. K. Mishra, “Design and

Implementation of Smart Hydroponics

Farming Using IoT-Based AI Controller with

Mobile Application System,” J. Nanomater.,

vol. 2022, no. 1, p. 4435591, 2022, doi:

10.1155/2022/4435591.

[6] N. Fatima, S. A. Siddiqui, and A. Ahmad,

“IoT-based Smart Greenhouse with Disease

Prediction using Deep Learning,” Int. J. Adv.

Comput. Sci. Appl., vol. 12, no. 7, pp. 113–

121, Sep. 2021, doi:

10.14569/IJACSA.2021.0120713.

[7] R. Rashid, W. Aslam, R. Aziz, and G.

Aldehim, “An Early and Smart Detection of

Corn Plant Leaf Diseases Using IoT and Deep

Learning Multi-Models,” IEEE Access, vol.

12, no. 1, pp. 23149–23162, 2024, doi:

10.1109/ACCESS.2024.3357099.

[8] W. zhen Liang, J. Oboamah, X. Qiao, Y. Ge,

and B. Harveson, “CanopyCAM – an edge-

computing sensing unit for continuous

measurement of canopy cover percentage of

dry edible beans,” Comput. Electron. Agric.,

vol. 204, no. 1, p. 107498, Jan. 2023, doi:

10.1016/J.COMPAG.2022.107498.

[9] J. Feng, W. E. Ong, W. C. Teh, and R. Zhang,

“Enhanced Crop Disease Detection with

EfficientNet Convolutional Group-Wise

Transformer,” IEEE Access, vol. 12, no. 1, pp.

44147–44162, 2024, doi:

10.1109/ACCESS.2024.3379303.

[10] N. G. Rezk, A. F. Attia, M. A. El-Rashidy, A.

El-Sayed, and E. E. D. Hemdan, “An Efficient

Plant Disease Recognition System Using

Hybrid Convolutional Neural Networks

(CNNs) and Conditional Random Fields

(CRFs) for Smart IoT Applications in

Agriculture,” Int. J. Comput. Intell. Syst., vol.

15, no. 1, pp. 1–21, Dec. 2022, doi:

10.1007/s44196-022-00129-x.

[11] R. Rathinam, P. Kasinathan, U. Govindarajan,

V. K. Ramachandaramurthy, U.

Subramaniam, and S. Garrido, “Cybernetics

approaches in intelligent systems for crops

disease detection with the aid of IoT,” Int. J.

Intell. Syst., vol. 36, no. 11, pp. 6550–6580,

Nov. 2021, doi: 10.1002/INT.22560.

[12] Y. Zhang, “IoT Agricultural Pest

Identification Based on Multiple

Convolutional Models,” J. Internet Technol.,

vol. 24, no. 4, pp. 905–913, Jul. 2023, doi:

10.53106/160792642023072404008.

[13] H. Design, J. García Morillo, E. Camacho

Poyato, J. Manuel Díaz-Cabrera, and C.-L.

Chang, “Design and Implementation of

Artificial Intelligence of Things for Tea

(Camellia sinensis L.) Grown in a Plant

Factory,” Agronomy, vol. 12, no. 10, p. 2384,

Oct. 2022, doi:

10.3390/AGRONOMY12102384.

[14] T. J. Maginga, P. Bakunzibake, E. Masabo, D.

P. Massawe, P. R. Agbedanu, and J. Nsenga,

“Design and implementation of IoT sensors

for nonvisual symptoms detection on maize

inoculated with Exserohilum turcicum,”

Smart Agric. Technol., vol. 5, no. 1, p.

100260, Oct. 2023, doi:

10.1016/J.ATECH.2023.100260.

[15] A. Barriga, J. A. Barriga, M. J. Moñino, and

P. J. Clemente, “IoT-based expert system for

fault detection in Japanese Plum leaf-turgor

pressure WSN,” Internet of Things

(Netherlands), vol. 23, no. 1, p. 100829, Oct.

2023, doi: 10.1016/J.IOT.2023.100829.

[16] A. Etienne, A. Ahmad, V. Aggarwal, and D.

Saraswat, “Deep learning-based object

detection system for identifying weeds using

uas imagery,” Remote Sens., vol. 13, no. 24, p.

5182, Dec. 2021, doi: 10.3390/rs13245182.

[17] R. Yauri, B. Guzman, A. Hinostroza, and V.

Gamero, “Weed Identification Technique in

Basil Crops using Computer Vision,” WSEAS

Trans. Syst., vol. 22, no. 1, pp. 636–644,

2023,

https://doi.org/10.37394/23202.2023.22.64.

[18] L. Yang, L. S. Nasrat, M. E. Badawy, and D.

E. M. Wapet, “A new automatic sugarcane

seed cutting machine based on internet of

things technology and RGB color sensor,”

PLoS One, vol. 19, no. 3, pp. 1–24, Mar.

2024, doi:

10.1371/JOURNAL.PONE.0301294.

[19] R. Yauri, M. Rios, and R. Acosta, “Network

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

155

Volume 15, 2024

monitoring environmental quality in

agriculture and pisciculture with low power

sensor nodes based on ZigBee and GPRS

technology,” in Proceedings of the 2016 IEEE

23rd International Congress on Electronics,

Electrical Engineering and Computing,

INTERCON 2016, Piura, Perú,Jan. 2017, pp.

1–6. doi: 10.1109/INTERCON.2016.7815578.

[20] T. M. Hoang, T. A. Pham, and V. N. Nguyen,

“A Novel Light-Weight DCNN Model for

Classifying Plant Diseases on Internet of

Things Edge Devices,” Mendel, vol. 28, no. 2,

pp. 41–48, Dec. 2022, doi:

10.13164/MENDEL.2022.2.041.

[21] T. Daniya and S. Vigneshwari, “Rider Water

Wave-enabled deep learning for disease

detection in rice plant,” Adv. Eng. Softw., vol.

182, no. 1, pp. 1–12, Aug. 2023, doi:

10.1016/J.ADVENGSOFT.2023.103472.

[22] G. Fevgas, T. Lagkas, V. Argyriou, and P.

Sarigiannidis, “Detection of Biotic or Abiotic

Stress in Vineyards Using Thermal and RGB

Images Captured via IoT Sensors,” IEEE

Access, vol. 11, no. 1, pp. 105902–105915,

2023, doi: 10.1109/ACCESS.2023.3320048.

[23] Kurnianingsih, R. Widyowati, A. F. Aji, E.

Sato-Shimokawara, T. Obo, and N. Kubota,

“Unsupervised Anomaly Detection for IoT-

Driven Multivariate Time Series on Moringa

Leaf Extraction,” Int. J. Autom. Technol., vol.

18, no. 2, pp. 302–315, Mar. 2024, doi:

10.20965/IJAT.2024.P0302.

[24] R. E. Putri, N. U. Lestari, Ifmalinda, F. Arlius,

I. Putri, and A. Hasan, “Monitoring and

Controlling System of Smart Mini

Greenhouse Based on Internet of Things (IoT)

for Spinach Plant (Amaranthus sp.),” Int. J.

Adv. Sci. Eng. Inf. Technol., vol. 14, no. 1, pp.

131–136, Feb. 2024, doi:

10.18517/IJASEIT.14.1.18408.

[25] V. K. Perumal, S. T, S. P, and D. S, “CNN

based plant disease identification using PYNQ

FPGA,” Syst. Soft Comput., vol. 6, no. 1, p.

200088, Dec. 2024, doi:

10.1016/j.sasc.2024.200088.

[26] R. Sai Sharvesh, K. Suresh Kumar, and C. J.

Raman, “An Accurate Plant Disease Detection

Technique Using Machine Learning,” EAI

Endorsed Trans. Internet Things, vol. 10, no.

1, pp. 1–9, Nov. 2024, doi:

10.4108/eetiot.4963.

[27] S. Raman, A. Chougule, and V. Chamola, “A

low power consumption mobile based IoT

framework for real-time classification and

segmentation for apple disease,”

Microprocess. Microsyst., vol. 94, no. 1, p.

104656, Oct. 2022, doi:

10.1016/j.micpro.2022.104656.

[28] HiveMQ, “HiveMQ Cloud – Fully-Managed

MQTT Platform on the Cloud, Free Version

Available,” 2024, [Online].

https://www.hivemq.com/products/mqtt-

cloud-broker/ (Accessed Date: December 1,

2024).

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

This research funded by the 2024 R&D Research

Projects Competition - Lima Region (Concurso de

Proyectos de Investigación I+D 2024 Región Lima)

of the Universidad Tecnológica del Perú.

Conflict of Interest

The authors have no conflicts of interest to declare.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US

WSEAS TRANSACTIONS on ELECTRONICS

DOI: 10.37394/232017.2024.15.17

Ricardo Yauri, Antero Castro, Rafael Espino

E-ISSN: 2415-1513

156

Volume 15, 2024