sensors

Review
Recent Advancements and Challenges of AIoT Application in
Smart Agriculture: A Review
Hasyiya Karimah Adli 1 , Muhammad Akmal Remli 2 , Khairul Nizar Syazwan Wan Salihin Wong 2 ,
Nor Alina Ismail 1 , Alfonso González-Briones 3 , Juan Manuel Corchado 3 and Mohd Saberi Mohamad 4, *

1 Faculty of Data Science & Computing, University Malaysia Kelantan, City Campus,
Kota Bharu 16100, Kelantan, Malaysia; [email protected] (H.K.A.)
2 Institute for Artificial Intelligence and Big Data, Universiti Malaysia Kelantan, City Campus,
Kota Bharu 16100, Kelantan, Malaysia
3 Grupo de Investigación BISITE, Departamento de Informática y Automática, Facultad de Ciencias,
University of Salamanca, Instituto de Investigación Biomédica de Salamanca, Calle Espejo 2,
37007 Salamanca, Spain
4 Health Data Science Lab, Department of Genetics and Genomics, College of Medical and Health Sciences,
United Arab Emirates University, Al Ain 17666, United Arab Emirates
* Correspondence: [email protected]

Abstract: As the most popular technologies of the 21st century, artificial intelligence (AI) and the
internet of things (IoT) are the most effective paradigms that have played a vital role in transforming
the agricultural industry during the pandemic. The convergence of AI and IoT has sparked a recent
wave of interest in artificial intelligence of things (AIoT). An IoT system provides data flow to
AI techniques for data integration and interpretation as well as for the performance of automatic
image analysis and data prediction. The adoption of AIoT technology significantly transforms the
traditional agriculture scenario by addressing numerous challenges, including pest management
and post-harvest management issues. Although AIoT is an essential driving force for smart
agriculture, there are still some barriers that must be overcome. In this paper, a systematic literature
review of AIoT is presented to highlight the current progress, its applications, and its advantages.
Citation: Adli, H.K.; Remli, M.A.; The AIoT concept, from smart devices in IoT systems to the adoption of AI techniques, is discussed.
Wan Salihin Wong, K.N.S.; Ismail, The increasing trend in article publication regarding to AIoT topics is presented based on a database
N.A.; González-Briones, A.; search process. Lastly, the challenges to the adoption of AIoT technology in modern agriculture are
Corchado, J.M.; Mohamad, M.S.
also discussed.
Recent Advancements and
Challenges of AIoT Application in
Keywords: artificial intelligence of things; smart agriculture; internet of things; artificial intelligence;
Smart Agriculture: A Review. Sensors
post-pandemic; crop diseases
2023, 23, 3752. https://doi.org/
10.3390/s23073752

Academic Editor: Yong He

Received: 15 January 2023

1. Introduction
Revised: 10 March 2023 COVID-19 left a visible and evident economic impact on the agriculture sector.
Accepted: 28 March 2023 Issues such as pest attacks and bacterial infections resulted in large-scale crop
Published: 5 April 2023 diseases [1,2]. The traditional approaches must be more effective in resolving the
issues, forcing researchers to consider using IR4.0 technology to tackle the problems and
transform the agricultural industry.
In the post-pandemic era, smart agriculture presents itself as an appropriate solution
Copyright: © 2023 by the authors.
for labor shortages and a continuous food supply chain [3,4]. Only some countries, such
Licensee MDPI, Basel, Switzerland.
as the U.S. and South Korea, have established holistic visions and frameworks for smart
This article is an open access article
agriculture solutions to achieve their sustainable development goals [5]. Artificial intelli-
distributed under the terms and
gence (AI) and internet of things (IoT) technologies were the most influential paradigms
conditions of the Creative Commons
that played a vital role during the COVID-19 pandemic [6]. As the most popular set of
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
technology for the 21st century, this integrated system can perform data informatization [7],
4.0/).
efficient remote-control monitoring in real-time [8,9], and intelligent management [10].

Sensors 2023, 23, 3752. https://doi.org/10.3390/s23073752 https://www.mdpi.com/journal/sensors

Sensors 2023, 23, 3752 2 of 22

A typical IoT system consists of wireless sensor networks (WSN) installed in different
venues to collect spatial and temporal data. Such a system includes a global visualization
consisting of a collection of “things” or objects or devices connected with micro-controllers,
embedded intelligence, communication means, sensing properties, and actuation. All
these objects are connected via internet protocol (IP) [11,12]. In industrial IoT applications,
AI functions as the technology enabler for hundreds of devices and a wide range of
applications, including environmental monitoring, agriculture [13,14], smart grids [15],
cities [16], buildings [17], homes [18], transportation [19], and healthcare [20].
AI supports the application, implementation, infrastructure, and management of
IoT technology. The application of AI in IoT environments boosts operational efficiency,
provides better risk management, triggers enhanced products/services, and increases the
scalability of IoT [21,22]. Fujitsu has developed a technology for estimating human body
postures using millimeter wave sensors and cloud data [23]. AI techniques include image
optimization, sensor processing, data transmission, and decision-making. The processes
involved require high-speed data streams, low-latency communications, fast processing
time, and time-sensitive actions on the IoT system [24]. Based on reinforcement learning
models, the model works in the compact memory space of IoT devices [25].
Paramount quantities of IoT data require powerful AI techniques for pre-processing
and preparing data to reduce noise, minimize dimensionality, and remove possible redun-
dancies [26]. In most reports, AI techniques such as artificial neural networks, fuzzy logic,
and evolutionary computation are mainly used for heterogeneous purposes [27], including
classification, regression, signal processing, forecasting, decision support, and data trans-
mission. Benefitting from their learning capabilities, various deep learning methods are
frequently used in the development of intrusion detection systems [28].

Contributions of This Study

There are still limited numbers of review articles regarding AIoT in smart agriculture,
although it is widely reported for other applications. Here, we specifically focus on the
AIoT concept by exploring AI and IoT as an overview. The advantages and potential
challenges through a systematic literature review are also included. The contributions of
this study can be summarized as follows.
(1) We discuss the AIoT concept, from smart devices in IoT systems to the adoption of
AI techniques.
(2) We present a systematic literature review to highlight the increasing trend in the
article publication regarding AIoT in different applications . . . and the progress of AI
and IoT.
(3) We summarize a few promising applications of AIoT and other AI/IoT-
enabling technologies.
(4) We highlight the challenges of AIoT adoption.

2. Artificial Intelligence of Things (AIoT)

Concept
The rapid evolution of AI, IoT sensors, and 5G infrastructures into robust technologies
has led to their intersection under the paradigm of AIoT (artificial intelligence of things) [29].
Despite AIoT still being in its infant stage, the applications and trends that it encompasses
are reshaping the future of enormous business potential. Overall, the IoT system provides
data flows and is further utilized with AI techniques to integrate the data, interpret the
data, perform automatic image analysis and data prediction, etc. Figure 1 illustrates the
integration of AI with IoT for better user outcomes [30].
In agricultural applications, the integration of AIoT mostly regards controlling har-
vests, greenhouse parameters, and smart fertigation to induce reactions to any change
in external conditions. For instance, convolutional neural networks are adopted to pre-
dict and detect possible diseases in applications at any scale on the basis of the collected
IoT data [31].
Sensors 2023, 23, x FOR PEER REVIEW 4 of 23
Sensors 2023, 23, 3752 3 of 22

Figure 1. From
Figure smartsmart
1. From “Things” in IoT
“Things” in systems to the
IoT systems to adoption of AI
the adoption of techniques.
AI techniques.

AIoT makes
In agricultural the operation
applications, the and management
integration of AIoTofmostly
agriculture easier
regards to access,harvests,
controlling effective,
and autonomous for the users. AIoT will become one of the main driving forces for smart
greenhouse parameters, and smart fertigation to induce reactions to any change in external
agriculture. However, some barriers still must be overcome, such as the cost factors and
conditions. For instance, convolutional neural networks are adopted to predict and detect pos-
the readiness for technology adoption as standard practice.
sible diseases in applications at any scale on the basis of the collected IoT data [31].
Several promising AIoT applications are discussed thoroughly in this paper. The
AIoT makes the operation and management of agriculture easier to access, effective,
motivation is triggered by existing proven applications of AI/IoT, such as computer vision,
and autonomous for the
video surveillance users. AIoT
systems, will become
and infrared oneThe
cameras. of the mainpropounds
review driving forces
ideasfor
onsmart
further
agriculture.
improving However, some barriers
AIoT applications andstill must be overcome,
advantages that benefitsuch as the cost
the users. factors andin
The challenges
the readiness
technology foradoption
technology adoption
are also as standard practice.
discussed.
Several promising AIoT applications are discussed thoroughly in this paper. The mo-
3. Methodology
tivation is triggered by existing proven applications of AI/IoT, such as computer vision,
video surveillance
A systematicsystems, and infrared
literature cameras.
review (SLR) The review
approach propounds
was applied ideas review,
to select, on further
and
improving AIoT
critically applications
appraise existing and advantages
published articlesthat
withbenefit
similarthe users.problems
research The challenges in
before sum-
technology
marizing adoption
selectedare also discussed.
primary articles. The overall stages of the SLR consist of planning an
Sensors 2023, 23, x FOR PEER REVIEW 5 of 23
reviewing, a similar approach to that reported previously. Figure 2 presents the step-by-step
of the SLR conducted in this study; a similar approach was applied previously [32].
3. Methodology
A systematic literature review (SLR) approach was applied to select, review, and crit-
ically appraise existing published articles with similar research problems before summa-
rizing selected primary articles. The overall stages of the SLR consist of planning an re-
viewing, a similar approach to that reported previously. Figure 2 presents the step-by-
step of the SLR conducted in this study; a similar approach was applied previously [32].

Figure
Figure2.2.Systematic
Systematicliterature
literature review approach[32].
review approach [32].

At the planning stage, the research objective and questions were identified and final-
ized before developing and evaluating the review protocol. Then, the identification of
main articles using search strategies, the selection of articles, and the extraction of the data
Sensors 2023, 23, 3752 4 of 22

At the planning stage, the research objective and questions were identified and final-
ized before developing and evaluating the review protocol. Then, the identification of main
articles using search strategies, the selection of articles, and the extraction of the data were
progressed during the review stage. Finally, the summary and report were made after the
data were synthesized and interpreted.

3.1. Objective
The objective of this study is to review the current progress of AIoT adoption in smart
agriculture and to identify the research gap and potentials related to AIoT technology. To
the best of our knowledge, few articles related to AIoT applications in smart agriculture
have been published, hence the scope of the study focusing on the technology based on
deep learning, computer vision, and intelligent video surveillance systems. Table 1 presents
the research questions of this study.

Table 1. Research question and justification.

ID Research Question Justification

What are common subject areas of AIoT application that Providing the context and analyzing the current trends of
RQ1
have been reported recently? subject area.
Which deep learning (DL) models are widely used in Identifying common DL models used for the same scope
RQ2
AIoT applications in smart agriculture? of study and assessing the comparison of few DL models.
What are common technologies reported for AIoT
RQ3 application in agriculture? Is there any potential Providing the review of current related technology.
technology to explore for the same application?
What is the current trend in terms of the number of Providing the context related to the trend of publication
RQ4
studies according to similar work? number over few years.

3.2. Process
The automatic database search was carried out on 12 December 2022 with the search
keywords “Artificial Internet of Things (AIoT)” and “AIoT in agriculture”, as shown
in Table 2.

Table 2. The details for database search process.

Database Search Keywords Number of Articles URL

Artificial Internet of Things (AIoT) 157 http://www.isiknowledge.com
WOS (accessed on 4 March 2023)
AIoT in agriculture 18
Artificial Internet of Things (AIoT) 133 http://www.sciencedirect.com
SCD (accessed on 4 March 2023)
AIoT in agriculture 64
Artificial Internet of Things (AIoT) 174 https://ieeexplore.ieee.org
IEEE (accessed on 4 March 2023)
AIoT in agriculture 14

3.3. Article Selection Criteria

This section focuses on the selection process of the articles and specific criteria applied
to the available filters of each of the databases. Table 3 presents the summary of inclusion
and exclusion criteria. The contents of each article were read, reviewed, and considered
during the process. A total of 11 articles remained, all of which were in the context of AIoT
applied in smart agriculture.
Sensors 2023, 23, 3752 5 of 22

Table 3. Inclusion and exclusion criteria.

No. Criteria Database

Inclusion
1. Articles published between 2017 and 2022. WOS, SCD, IEEE
2. Peer-reviewed primary articles. WOS, SCD, IEEE
Studies within the context of AIoT applied in smart agriculture based
3. WOS, SCD, IEEE
on research scope established.
4. Articles published in English. WOS
Exclusion
Secondary or tertiary studies within the context of AIoT applied in
1. -
smart agriculture based on research scope established.
2. Other document types, such as short articles, proceedings, books etc. WOS, SCD, IEEE
3. Redundant studies by the same researcher. WOS, SCD
4. Articles published prior to 2017. WOS, SCD, IEEE

3.4. Quality Assessment and Data Extract

After articles were identified, quality assessment was carried out to ensure the article’s
quality and its relevancy to this study. Table 4 shows five questions to comply in order to
assess the quality of the articles based on specific scores: yes (1.0), partially (0.5) and no (0).

Table 4. Quality assessment on the articles based on the questionnaire.

ID Question
Q1 Are the objectives of the study clearly defined?
Q2 Are the research questions clearly answered?
Q3 Did the study use DL algorithm in its research scope?
Q4 Did the study report a well-described experiment?
Q5 Does the finding of the study prove the validity which relevant to it?

The total score for each article from the sum of the values obtained from answers. A
score of 1.0 indicated well-matched with this study, and 0.0 indicated otherwise. A cut-off
score of 0.5 was defined as moderate, and only articles with a score greater than 0.5 were
considered for this study.

Table 5. Quality evaluation.

ID Research Scope Author Final Score

A1 Smart aquaculture farm management system [33] 3.5
A2 Plant growth monitoring and environmental [34] 4.0
A3 Algae culture monitor [35] 5.0
A4 Agriculture 4.0 [36] 4.5
A5 Pest Detection [31] 3.5
A6 Smart Livestock Surveillance [37] 3.5
A7 Empowering Things with Intelligence: A Survey [29] 3.0
A8 AI-Rooted IoT System Design Automation [38] 3.0

All primary articles that passed inclusion and exclusion criteria in previous stage
(Section 3.3) were read and evaluated to acquire the score. Table 5 shows the summary of
scores from the quality assessment of selected 11 articles as the input set, of which 3 articles
received scores less than or equal to 0.5.
 A4 Agriculture 4.0 [36] 4.5
A5 Pest Detection [31] 3.5
A6 Smart Livestock Surveillance [37] 3.5
A7 Empowering Things with Intelligence: A Survey [29] 3.0
Sensors 2023, 23, 3752 6 of 22
A8 AI-Rooted IoT System Design Automation [38] 3.0

4. Results
4. Results
At the primary stage, the initial amount of five hundred and sixty articles was found
from At threethedatabases.
primary stage, the articles
Eleven initial amount of five hundred
were considered andprimary
as final sixty articles wasafter
articles foundcon-
from three databases. Eleven articles were considered as final primary articles after con-
sidering inclusion and exclusion criteria. It cannot be neglected that the biases during the
sidering inclusion and exclusion criteria. It cannot be neglected that the biases during the
initial stage of primary article selection due to the small number of articles were published
initial stage of primary article selection due to the small number of articles were published
that according to similar scope of study.
that according to similar scope of study.
This
This may
may also
also imply that the
imply that the possibility
possibilityofofirrelevance
irrelevanceofofthe thearticles
articlestoto answer
answer thethe
research questions. Additionally, the choice of databases also cannot
research questions. Additionally, the choice of databases also cannot be considered canbe considered can
represent
represent thethe completeness
completeness of of the
the studies.
studies. Other
Otherreasons
reasonsthat
thatcan
canalso
alsobebeconsidered
considered are:
are:
(a) the limited AIoT applications in agriculture have been reported; (b)
(a) the limited AIoT applications in agriculture have been reported; (b) related policies related policies
and
andmechanisms
mechanisms also also might
mightbebeunder-developed
under-developed to to push
push as the
as the assessment.
assessment. However,
However, it is it
isworth
worth noting
noting that
that thethe number
number of articles
of articles reporting
reporting on AIoT
on AIoT adaption
adaption in smart
in smart agricul-
agriculture
ture
has has increased
increased massively
massively since
since 2021,
2021, fromfrom
lessless
than than
5 in5 2017
in 2017
to atoprojection
a projection of of close
close to to
5050papers
papers in in 2023
2023 (Figure 3).
3).

Figure 3. Number of articles published since 2017 related to this scope of study.

4.1. Current Applications of AIoT in Smart Agriculture

The articles presented in the next subtopics have been contributed to the massive
research development of AI and IoT in smart agriculture and have potential for AIoT
application in the future. To date, 37,230 articles have been published regarding the scope
of AI and IoT in agriculture (Table 6).

Table 6. Number of articles reported for AI and IoT in smart agriculture.

Database Search Keywords Number of Related Articles URL

Artificial Intelligence in agriculture 1237 http://www.isiknowledge.com
WOS (accessed on 4 March 2023)
IoT in agriculture 319
Artificial Intelligence in agriculture 21,309 http://www.sciencedirect.com
SCD (accessed on 4 March 2023)
IoT in agriculture 7045
Artificial Intelligence in agriculture 4690 https://ieeexplore.ieee.org
IEEE (accessed on 4 March 2023)
IoT in agriculture 2630

4.1.1. Deep Learning Methods

Deep learning offers many opportunities in various agricultural stages, and its ap-
plication has increased substantially in recent years, with numerous studies providing
innovative insights into this topic. Table 7 shows the recent work on deep learning, and
Table 8 shows that for traditional machine learning, both for the agriculture application.
Sensors 2023, 23, 3752 7 of 22

Most proposed works focus on predicting and classifying crop production, disease detec-
tion, and automating irrigation systems. The data consists of various sources, including
irrigation systems, temperature, humidity, water, weeds and crops, disease, and fruit
grading. Combining IoT data with advanced deep learning algorithms has brought vast
development and continuous breakthroughs in agriculture, AI, and deep learning fields.

Table 7. Deep learning for smart agriculture.

Method Description Data Disadvantages Advantages References

More accurate
Gated recurrent unit A hybrid deep learning
prediction on humidity,
(GRU) and reversible method embedded in Humidity, wind, Short time data period
temperature, and wind [39]
automatic selection IoT system and temperature. for prediction
speed compared to
normalization (RASN) as a predictor.
other methods
Deep learning Efficient resource
Deep learning No report on
algorithm that Smart agriculture utilization and
algorithm on fog prediction and [40]
performs some DL application able to reduce
nodes (DLEFN) algorithm performance
tasks on fog nodes network congestion
K-nearest neighbors
Most of the proposed
algorithm (KNN), Proposed method for
method and algorithms
artificial neural system monitoring of Growth of strawberries,
There is no standard report competitive
network (ANN), long fruits, smart agriculture weeds, crops, lettuce
procedure to conduct performance in terms
short-term memory of UAVs, crop disease production, plant
verification and of classification [41–50]
(LSTM), recurrent prediction, leaf disease, bird species,
validation of the accuracy, identification,
neural network (RNN), identification and apple leaf diseases, and
proposed method. and prediction, which
and ensemble classification in agricultural machinery
applied in
subspace discriminate real-time.
computer vision.
analysis (ESDA)
Efficient integrated
Deep reinforcement system incorporating
Deep reinforcement No validation and
learning in Water consumption cloud computing [51]
learning experiment report
smart systems to improve
food production

Table 8. Previous reports on traditional machine learning methods in smart agriculture.

Method Description Data Disadvantages Advantages References

Small-scale machine Real time classification
Only focus on
Fuzzy logic controller intelligence for Cultivated data with 90% and [52]
irrigation system
smart agriculture higher accuracy
Application of computer Provide direction and
Neural network No report on the
vision and crop farming Fruit grading and trend on using various
(ANN) and neural performance based on [53,54]
algorithms for spraying system methods in
network (CNN) ANN, CNN, and SVM
smart farming smart agriculture.
An algorithm for Limited amount of data Classification
Supervised learning detecting and preventing for the trained model and performance achieved
High-yield crop [55]
algorithms spread of diseases to the challenging to be used in 98% with adequate
whole crop real-world practice computation time.
Focusing security IDS on
Support vector,
IDS system in smart Uses standard machine smart farming and
machine, random Field’s water
agriculture and learning algorithms, and compare with several
forest, neural network, supply and [56–58]
Intelligent biochemical the performance is not algorithms (SVM
and K-nearest crop/weed
spraying tested with deep learning. achieve higher
neighbors (KNN)
accuracy).
WPART able to achieve
No time series analysis for
Machine learning high accuracy with 90%
Wrapper PART prediction future value
algorithm for Crop productivity and above in crop [59]
(WPART) from previously
smart farming productivity and
observed values
drought prediction
Sensors 2023, 23, 3752 8 of 22

Table 8. Cont.

Method Description Data Disadvantages Advantages References

Application and Challenging to
Direct recommendation
suggestion in smart non-technical people,
Various machine Cultivate crops to individual and farmer
agriculture using specifically when [60]
learning algorithms productivity in managing their
various machine introduced to machine
crop cultivation
learning algorithms learning algorithms

A recent study used a hybrid method based on empirical mode decomposition

(EMD) to decompose the data. At the same time, a gated recurrent unit (GRU) was
utilized as a predictor to predict various climate data, including temperature, humidity,
and wind speed. The performance of the proposed method is superior and can produce
the highest accuracy in precision agriculture production. Another predictor model for
automating normalization using reservable automatic selection normalization (RASN)
based on indoor data from the smart greenhouses is presented in [39]. The authors
have introduced scaling and translating techniques with parameters that can be learned
to improve prediction and adaptability in innovative agriculture systems. Combining
IoT devices with the DL algorithm is a challenging problem due to device limitations,
as DL requires high computational resources to process the IoT data. To solve this
issue, ref. [40] proposed the deep learning based on fog nodes (DLEFN) algorithm and
experimented with it in order to determine the optimal layers in DL for execution in each
fog node. The availability of device capacity and bandwidth influences the results. The
output shows that the proposed algorithm can reduce network congestion and utilize
resources efficiently.
Several CNN algorithms have been improved and utilized in computer vision for
smart agriculture. For example, ref. [41] introduced a computer vision monitoring
system to monitor tomato growth stages using a regional-based convolution neural
network (R-CNN). Other utilized algorithms include artificial neural networks (ANN),
KNN, and support vector machine (SVM). The proposed system shows the highest
accuracy in predicting flower and fruit and in maturity grading. The RCNN was also
used to detect leaf disease in smart agriculture [42]. The authors combined R-CNN
with MASKRCNN to detect the infected disease in apple leaves. Transfer learning was
used to extract features from the pre-trained CNN model. Results confirmed that the
proposed framework was superior compared to other frameworks. Aside from that, long
short-term memory (LSTM) is a recent DL technique that has had remarkable results.
On the other hand, auto-encoder (AE) is an artificial neural network that shows
excellent performance in learning codings for unlabeled data. Both LSTM- and AE-based
techniques have been used in smart agriculture. In a study [43], a deep neural network
based on LSTM was used in an IoT irrigation system for precision agriculture (DLiSA).
In their work, LSTM was utilized to predict several irrigation conditions including
soil moisture, irrigation time, and water amount needed to spray the arable land. The
simulation of the proposed system showed that it was efficient in water usage compared
to other state-of-the-art irrigation systems. The authors of [44] focused on an unmanned
aerial vehicle (UAV) framework in which several techniques—including blockchain for
data authentication, sparse auto-encoder (SAE) for data transformation, and stacked long
short-term memory (SLSTM)—were combined for training, and they evaluated the result.
The experiments demonstrated that the framework outperformed other state-of-the-art
block chain and non-block chain frameworks. Deep reinforcement learning is another
advanced deep learning model used in smart agriculture [51]. Specifically, the model
was combined with cloud computing technology, resulting in an intelligent model that
can determine the amount of water needed for irrigation. The work resulted in increased
production and an improved crop growth environment.
 Sensors 2023, 23, 3752 9 of 22

4.1.2. Computer Vision

A smart agriculture system using DL-based computer vision is promising due to
the massive growth of agriculture data commonly collected from IoT sensors. Other
researchers also studied and applied traditional machine learning (also in combination
with DL) techniques, such as fuzzy logic, SVM, supervised learning, decision tree, linear
regression, and KNN [52–60]. The authors of [52] proposed a small-scale agriculture
machine to irrigate and weed automatically in the cultivated area. In their work, the fuzzy
logic controller was applied to provide data on the wet distribution area of surface soil
(WDAS). The results showed that the machine achieved an accuracy of at least 90% in
weeding and watering the deep soil. In recent works from [53,54], the combination of
ANN, Gaussian curve fitting, and CNN was proposed in crop farming specifically for
fruit-grading systems. The SVM classifier is mainly used for various classifications in
smart farming. In [55], the authors proposed an IoT framework using machine learning
algorithms, as illustrated in Figure 4. The primary purpose of the framework was to
measure and predict crops using supervised learning. The dataset consists of 200 instances
and eight attributes. The experiments showed that the proposed model could predict the
high-yield crop in precision agriculture with an accuracy of 98%. Other frameworks have
Sensors 2023, 23, x FOR PEER REVIEW been developed to detect and automate agriculture, including water supply and crop/weed11 of 23
spraying.

Figure 4. A framework for smart irrigation [55].

Figure 4. A framework for smart irrigation [55].

The
Theframework
framework commonly
commonly collects primary data
collects primary datafrom
fromscalable
scalableIoT IoT sensors
sensors [56–58].
[56–58].
Some
Someworks
worksfocused
focused on
on the
the security aspectof
security aspect ofIoT
IoTsystems,
systems,specifically
specificallyin in performing
performing smart
smart
agriculture—for
agriculture—forexample,
example,ref.ref.[56]
[56]studied
studiedrisk
risk mitigation
mitigation in in smart
smart irrigation
irrigation based
based on an
intrusion detection
an intrusion system
detection (IDS).
system The The
(IDS). authors proposed
authors proposeda framework
a framework using SVM
using as aas
SVM clas-
a classifier,
sifier, linear regression,
linear regression, and random
and random foreston
forest based based on the
the NSL KDD NSL KDD
data set.data
The set. The
framework
framework performance
performance was compared was
to compared to state-of-the-art
state-of-the-art machine learningmachine learningbased
algorithms algorithms
on recall
based on recall and precision. Another work also used a random forest classifier
and precision. Another work also used a random forest classifier in detecting crops/weeds in detecting
crops/weeds
for for real-time
real-time variable-rate variable-rate
spraying spraying
[57]. Their work [57].
aimsTheir work aims
to improve sprayto improve
yield andspray
protect
yield and protect crops from disease according to a specific amount of agrochemicals
crops from disease according to a specific amount of agrochemicals and field/crop require- and
field/crop requirements. The random forest classifier detected and classified various weeds
ments. The random forest classifier detected and classified various weeds and crops. The
and crops. The desired amounts of agrochemicals were then sprayed using a vision-based
desired amounts of agrochemicals were then sprayed using a vision-based feedback system.
feedback system. The simulation result shows that the proposed method is effective.
The simulation result shows that the proposed method is effective.
Handling enormous amounts of data is a daunting task and results in higher costs,
especially if the data come from heterogeneous information. To cater to this challenge,
[58] introduced a method of managing heterogeneous agriculture datasets (biological, sen-
sory, and physical values). Various machine learning techniques have been utilized to sug-
gest the best effort and investments in data management. Another work proposed an
 Sensors 2023, 23, 3752 10 of 22

Handling enormous amounts of data is a daunting task and results in higher costs,
especially if the data come from heterogeneous information. To cater to this challenge,
ref. [58] introduced a method of managing heterogeneous agriculture datasets (biological,
sensory, and physical values). Various machine learning techniques have been utilized to
suggest the best effort and investments in data management. Another work proposed an
intelligent method based on wrapper feature selection combined with the PART classifi-
cation technique (WPART) to improve crop productivity and drought prediction [59]. In
their experiments, five data sets were used to train the model. Using productivity crop data
including soybean and sugarcane, the proposed method could obtain high accuracy and
outperform other machine learning algorithms in crop productivity and drought prediction.
Machine learning and deep learning algorithms consist of various techniques and methods.
Applying all these algorithms to perform prediction and classification in smart agriculture
is time-consuming, challenging, and requires enormous cost. To facilitate the selection of
Sensors 2023, 23, x FOR PEER REVIEW
the best machine learning and deep learning algorithms, ref. [60] studied various algo-
rithms to be applied in smart farming. Their work guides other researchers and farmers
in cultivating crops efficiently and helps improve productivity while keeping costs low.
Furthermore, the work also helps farmers better manage their crops and suggests a smarter
learning
harvesting[61,62].
process. However, due to the heterogeneous and variety data sources fro
It should
riculture field, be noted that the and
verification methods presented
validation ofrely
theon machinelearning
machine learning and deep must b
system
learning algorithms, which have seen substantial improvement and breakthroughs in
analyzed. This area
smart agriculture. is alsodeep
Although known as the
learning hasformal
receivedmethod, whichinisrecent
great attention a field of study th
years,
ines
machineandlearning
strictlymethods
verifiesbased
machine learning
on basic systems
algorithms, suchboth in hardware
as random and software
forest, support
Several approaches
vector machine, have been
and K-nearest proposed
neighbors, are stillto provide
useful verification
because framework
of the labelled data and for th
supervised learning [61,62]. However, due to the heterogeneous
mensionality, complexity, and uncertainty of machine learning algorithms and variety data sources [63,64]
from the agriculture field, verification and validation of the machine learning system must
be further analyzed. This area is also known as the formal method, which is a field of study
4.1.3. Intelligent
that examines Video
and strictly Surveillance
verifies Systems
machine learning systems both in hardware and software
systems.
Intelligent video surveillance systemsprovide
Several approaches have been proposed to verification
are widely usedframework
in manyfor the
applications
high dimensionality, complexity, and uncertainty of machine learning algorithms [63,64].
crime prevention, security, monitoring and controlling essential infrastructures
smart agriculture
4.1.3. Intelligent Videoindustry [63,64].
Surveillance From 2010 until 2019, there were 220 video sur
Systems
system (VSS)video
Intelligent studies, whichsystems
surveillance highlights the continuous
are widely used in many relevance
applications,ofsuch
research
as in
crime prevention, security, monitoring and controlling essential infrastructures,
[64]. VSS has garnered great interest in the past decades, especially with the in and the
smart agriculture industry [63,64]. From 2010 until 2019, there were 220 video surveillance
of computer vision, image processing, and artificial intelligence capabilities.
system (VSS) studies, which highlights the continuous relevance of research in this field [64].
shows
VSS has the VSS great
garnered architecture
interest inthat includes
the past components
decades, such
especially with theas sensors,ofservers,
integration
work
computer types [65].
vision, The
image analytics
processing, andcover the
artificial part of capabilities.
intelligence processingFigure algorithms
5 shows to ens
the VSS architecture
surveillance. that includes components such as sensors, servers, and network
types [65]. The analytics cover the part of processing algorithms to ensure good surveillance.

The
Figure 5.5.
Figure component
The and functionality
component of VSS. of VSS.
and functionality

With traditional video, human energy and input are required to watch and
the video multiple times, which is tedious and time-consuming. Many academic
proposed various intelligent surveillance methods that can recognize human acti
 Sensors 2023, 23, 3752 11 of 22

With traditional video, human energy and input are required to watch and analyze
the video multiple times, which is tedious and time-consuming. Many academics recently
proposed various intelligent surveillance methods that can recognize human action based
on high accuracy and efficiency. Intelligent VSS may consist of a few stages, such as video
preprocessing, object detection, activity detection, recognition, and action classification.
The combination of AI and IoT sensors makes video surveillance smarter and recog-
nizes an object automatically [66]. The intelligent VSS system operates without human
control or within autonomous systems; it detects, analyzes, and anticipates events or be-
haviors of things that trigger a warning. The system can perform face recognition, identify
intrusion and anomalies and send alerts to farmers to take appropriate action. The integra-
tion with the IoT system effectively manages all devices such as cameras, smoke detectors,
and audio sensors in the monitoring area [66]. In the future, AIoT will allow all devices,
especially VSS, to think and decide on their own if any unexpected events occur on the farm.
With the fast evolution of IR4.0 Technology, video surveillance systems are also undergoing
continuous development in terms of the devices and analysis based on AI algorithms.
Among various AI models used for surveillance analysis are CNN, auto-encoders, and
sors 2023, 23, x FOR PEER REVIEW their combination. The author summarized AI methods most commonly applied13 in of 23
visual
surveillance systems: deep learning, Gaussian, support vector machine (SVM), fuzzy logic,
and nearest neighbor [65].
Meanwhile, the capabilities of smartphone technology also revolutionize farming
the-art of current publication on the application of smartphone sensors in agricul-
activity. Current smartphones are equipped with various in-built sensors, such as GPS,
ture. With users of smartphones
temperature, accelerometers,being everywhere
and light and
sensors. The widely
authors connected
of [67] reportedthrough net-
on the state-of-
works, there is of
the-art ancurrent
increasing opportunity
publication in the development
on the application of smartphone-based
of smartphone sensor
sensors in agriculture. With
systems users
for agriculture. As shown in Figure 6, the sensors of smartphones monitor
of smartphones being everywhere and widely connected through networks, there the
environment and collect
is an increasing and process
opportunity thedevelopment
in the data either of
in smartphone-based
the smartphone itselfsensororsystems
on local for
machines or remoteAs
agriculture. cloud servers
shown [68].6, the sensors of smartphones monitor the environment
in Figure
and collect and process the data either in the smartphone itself or on local machines or
remote cloud servers [68].

Figure 6. Smartphone for monitoring.

Figure 6. Smartphone for monitoring.
The main advantage of a smartphone-based sensor system (SBSS) over a typical
wireless sensor network is its low cost of equipment; the built-in sensors on smartphones
The main advantage of a smartphone-based sensor system (SBSS) over a typical wire-
reduce the cost of software and central equipment [69]. For light use applications, smart-
less sensor network is its low cost of equipment; the built-in sensors on smartphones re-
phones can partially process the data with free and powerful development tools. For
duce the cost ofAndroid
instance, software and central
smartphone equipment
users may program[69]. For
in Java on light useStudio
Android applications,
Integrated
smartphones can partially process the data with free and powerful development
Development Environment (IDE) [70]. Smartphones have become powerful tools in tools. For
the
instance,agricultural
Android smartphone users may program in Java on Android Studio Integrated
sector [71], and recently, a mobile vision system was reported for identifying
Development Environment (IDE) [70]. Smartphones have become powerful tools in the
agricultural sector [71], and recently, a mobile vision system was reported for identifying
early-stage diseases in plants [72]. By taking photographs of the parts of the plant to ana-
lyze for the infection, these photographs are then pre-processed and transferred to distant
Sensors 2023, 23, 3752 12 of 22

early-stage diseases in plants [72]. By taking photographs of the parts of the plant to
analyze for the infection, these photographs are then pre-processed and transferred to
distant labs. Additionally, a color estimator on the application of smartphone is reported
for chlorophyll estimation [73].
However, there are still some limitations on the use of smartphones. For example,
the sensor-collected data might not be complete or uniform. In addition, the quality of
smartphones also varies with different levels of data accuracy are provided. The privacy of
smartphones users also contributes to the need for proper protection mechanisms in order
to avoid the leaking of the user’s personal data.
The use of smartphones helps farmers to be more productive and helps better decision-
making after obtaining useful insight from agriculture data. The use of AI and IoT in
agriculture technology has proven to be a significant improvement in farming efficiency,
cost, and crop yields. However, further research and development in AI and IoT towards
AIoT should be accelerated to fully optimize the future of agriculture technologies. In the
future, the increasing number of smartphones and their sensor capacities will be improved
and should contribute to faster adoption of AIoT in real life. To achieve this goal, it is
necessary to deepen the research on whole system components improvement, sensors, data
exchange and gathering, and data security [74].

4.2. Advantages of AIoT in Agriculture

According to past studies, about 75–80% of people live in rural areas and still depend
on agriculture. In the context of AIoT, the technology can serve as a solution to help farmers
increase yield and productivity exponentially. It uses various sensors connected to the
internet and integrated into the satellites.
Smart agriculture, or smart farming, is the concept of maintaining and monitoring
farms using modern technologies to increase the quantity and quality of products. With
the advancement of sensor technologies, miniaturization, and cost reduction, farmers have
access to navigation, soil scanning, data management, and pest detection technologies.
AIoT-enabled devices, real-time data collection, and automation can vastly improve the
smart agriculture industry. Smart farming is not the same as embedded systems, which
allow AIoT to enable devices and take advantage of AI technology. Considering the benefits
of IoT and AI, better possibilities could be expected from AIoT in performing better than
previous technology. The subtopics that follow describe the four critical advantages that
AIoT is expected to contribute to the agricultural sector.

4.2.1. Disease Identification and Plant Monitoring

Farmers and researchers may use crop imagery analysis to spot early infection, en-
abling them to take action before the illness becomes a significant issue. One of the main
advantages of employing deep learning is the ability to detect infections in their early
stages. It can take a lot of time and effort to discover disease using traditional approaches,
such as manual examinations by trained workers. Additionally, these techniques might
only sometimes be reliable because it can be challenging for human inspectors to spot
minute evidence of infection. On the other hand, deep learning systems can swiftly and
reliably assess photos of crops, enabling the early diagnosis of illnesses. The system may be
trained to accurately recognize infection symptoms using a sizable photo collection. This
can assist farmers in acting quickly to stop the virus from spreading in order to safeguard
their crops.
Crop diseases represent a significant threat to agricultural output. Frameworks for
deep learning utilize sick leaves’ unique characteristics to detect different diseases. The
paper employed complete deep learning (CDL) architecture to offer a multi-crop disease
detection model capable of categorizing crop diseases irrespective of crop type [75].
Agricultural progress is constantly aided by early disease diagnosis, classification,
and analysis of ill leaves as well as the identification of viable treatments. The authors
concentrated on detecting, classifying, and predicting several plant diseases, especially in
 Sensors 2023, 23, 3752 13 of 22

PEER REVIEW 15 of 23
tomatoes and grapes [76]. An algorithm based on deep learning was used to extract visual
features to discriminate sick leaves from healthy ones, as illustrated in Figure 7.

Figure 7. Identification of plant

Figure leaf diseases
7. Identification of plantusing CNNusing
leaf diseases [77].CNN [77].

Identifying plant diseases is highly significant in agriculture to boost crop output. Due
Identifying plant diseases
to recent is highly
breakthroughs significant
in imaging, in agriculture
visual plant to boost
disease analysis croptooutput.
is now used address
Due to recent breakthroughs in imaging,
this issue. Authors examined visual plant disease
the difficulties analysis
in visually is now
recognizing plantused to ad-
illnesses for
disease diagnosis
dress this issue. Authors examined [78].the difficulties in visually recognizing plant illnesses
Plant disease images are more likely to have randomly dispersed lesions, various
for disease diagnosis [78].
symptoms, and complex backgrounds than other typical types of photography, making it
Plant disease more
images are more
challenging likely
to capture to have randomly
distinguishing information. dispersed lesions, various
symptoms, and complex backgrounds than other typical types
The behavior of the soil varies due to alternating of photography,
climatic conditions. Pests making
are anotherit
major worry. Image processing has evolved into a valuable instrument for the early
more challenging to capture distinguishing information.
detection and diagnosis of plant diseases. Several methods have been employed to identify
The behavior diseases
of the soil varies
in their due
earliest to alternating
stages, climatic
resulting in little crop lossconditions.
and good crop Pests are The
quality. an-
other major worry.authors’
Imagestudy
processing
on banana has evolved
crop intotheir
diseases and a valuable instrument
potential solutions for the
aids faster early
detection
and diagnosis [79]. Thanks to an IoT system, agricultural
detection and diagnosis of plant diseases. Several methods have been employed to iden- data can be collected while AI
mechanisms train and automatically analyze data in real time.
tify diseases in their earliest stages, resulting in little crop loss and good crop quality. The
4.2.2. Intelligent
authors’ study on banana Farm Machinery
crop diseases and Crop
and their Management
potential solutions aids faster detection
and diagnosis [79]. ThanksComplex to mathematical
an IoT system, models called “deepdata
agricultural learning”
can are
be modelled
collectedtowhileresembleAI
the structure and operation of the human brain. These neural networks are trained on
mechanisms train and automatically analyze data in real time.
enormous datasets and can learn and make predictions or judgments based on the data.
Additionally, deep learning may be utilized to create autonomous agricultural equip-
4.2.2. Intelligent Farm
mentMachinery and
that can plant, Crop
weed, andManagement
harvest crops autonomously, without human assistance.
This may lower labor expenses and boost overall effectiveness. Deep learning algorithms
Complex mathematical models called “deep learning” are modelled to resemble the
structure and operation of the human brain. These neural networks are trained on enor-
mous datasets and can learn and make predictions or judgments based on the data.
Additionally, deep learning may be utilized to create autonomous agricultural equip-
Sensors 2023, 23, 3752 14 of 22

may be used to train autonomous farm machinery to navigate and operate in complex
agricultural situations while making judgments based on the information obtained from
sensors and other sources.
Numerous challenges remain, such as the shortage of workers in the agricultural sector
and the increased demand for newer high-tech advanced machinery. New technologies
within autonomous robotics are expanding in the agricultural industry. Huge investments
are being made to develop autonomous agricultural mobility robots; as a result, modern
farms have high prospects for increased productivity. Due to the complexity and diversity
of the agricultural work environment, it is difficult to overcome current obstacles using the
present machinery design. The paper examined a technique for creating and managing
a mobile robot platform that may handle these difficulties in a greenhouse [80]. The
developed platform has two driving wheels and four casters that could operate on a route
and a rail. In addition, it provides technology for a multi-AI deep learning system to
operate a robot, a robot-operating algorithm, and a VPN-based network and security
communication system [81].
Food shortages are expected to worsen because of the growing world population. Due
to the diversity of orchard conditions and tree types, fruit farming requires substantial
manual labor, causing mechanization and automation to fall behind [82]. Mechanization is
important to improve efficiency and reduce dependence on manual labor. A system for
automated fruit picking using robots fitted with robotic arms was proposed by the authors
of [83]. Before putting end-effectors into the fruit’s bottom half, the fruit-harvesting robot
uses sensors and computer vision to identify and estimate the fruit’s location. Experiments
demonstrated that this technology could detect pears and apples in the field and pick them
autonomously [84].
For agricultural automation, the exact distribution of liquid fertilizer and pesticides
to plants is a critical activity in precision agriculture. It provides a more cost-effective and
ecologically friendly alternative to conventional, non-selective spraying by identifying and
decreasing the number of chemicals used. Spraying with precision involves the detection
and tracking of each plant. Traditional detection or segmentation techniques lump all
plants inside an image collected by a robotic platform irrespective of the plant’s unique
identifier. In addition to recognizing each plant, it is vital to track each plant to administer
pesticides precisely once to each plant. The previous researcher proposed a multiple object
tracking (MOT) technology that recognizes and tracks lettuce concurrently, only spraying
plants that have never been treated before [85]. The approach leverages YOLO-V5 for
identifying lettuce and includes plant feature extraction and data association algorithms to
monitor all plants successfully.
The authors created a virtual simulation setting by fusing a robot operating system
(ROS) to illustrate the possibilities of a simulator channel to present a case study on indoor
robotic farming [86]. The paper developed a technique for evaluating the harvest of sweet
peppers. The method uses aerial robotics control and trajectory planning, followed by
deep learning-based recognition and a clustering algorithm for fruit counting. This case
study illustrated that a complex robotic system may be modeled by integrating real-time
modeling with almost practical rendering capacities.

4.2.3. Efficient Agricultural Data Analysis

A wide variety of variables crucial for effective crop management may be predicted
using deep learning. Deep learning algorithms may be used to assess sensor data and data
from other sources to forecast agricultural production, soil health, and other elements that
are crucial for effective crop management. This information may be utilized to optimize
irrigation, fertilization, and other farming practices to enhance crop health and yield. The
health and quality of crops may also be predicted using deep learning algorithms.
For instance, using information from cameras and other sensors, algorithms may be
taught to find illnesses or pests in crops. This data may be utilized to spot issues early on
and take appropriate actions before the problem causes serious harm to the crops. Deep
Sensors 2023, 23, 3752 15 of 22

learning algorithms may also forecast customer preferences and market circumstances.
Farmers and other agricultural stakeholders may utilize this information to better guide
their choices on what crops to grow, when to plant them, and how to market and sell them.
Deep learning may assist related stakeholders in making better decisions and increasing
the productivity and sustainability of their operations by enabling the construction of more
precise and complex prediction models.
The work outlined the construction of a low-cost and low-power wireless sensor
network (WSN) based on photovoltaic (PV) sensor nodes that can obtain ambient circum-
stances and soil data [87]. Soil moisture sensors are the most critical installed sensors due to
huge costs and difficulties in installation, reliability, and calibration. This article presented a
deep learning (DL) technique using long short-term memory (LSTM) networks to simulate
a soil moisture sensor using data collected from the other transducer mounted on the node.
To confirm the usefulness of the suggested soft sensing technique, this study evaluated the
performance of virtual sensors and compared it to other approaches.
Recent advancements in remote sensing using unmanned aerial vehicles (UAVs) for
precision agricultural operations have significantly enhanced crop health and management.
UAVs outfitted with sensors, cameras, LIDAR, and thermal cameras have been used for
crop remote sensing since they offer new methods and possibilities. The article examined
the use of UAVs for pest and disease control, yield estimation, phenotypic measurement
soil moisture assessment, and nutritional status evaluation in the sugarcane industry to
boost efficiency and maintain an ecological state [88].
Proper land use and crop maps obtained from remote sensing provide critical and
timely information for agricultural monitoring on a wide scale. Due to their limited
model transfer capabilities, the bulk of existing multi-crop products for complex agricul-
tural landscapes centered on standard machine learning approaches must be improved
for large-scale agricultural management. That is why developing a segmentation and
classification model with spatial and temporal transfer across regions and years is es-
sential. A study developed a deep learning technique for large-scale land use and crop
mapping by combining feature fusion with the up sampling of small data using the
UNet++ architecture. The method improved classification accuracy for datasets with
variables by analyzing the full confusion matrix [89].
In smart agriculture, computer vision and AI can optimize agricultural output while
minimizing resource use and improving environmental and economic results. This effort
aims to develop cutting-edge algorithms for image-based crop evaluation to help growers
make real-time choices. The previous paper made two substantial algorithmic advances.
First, the report devised a technique for segmenting cranberry instances which offers
a number of sun-exposed berries susceptible to warming [90]. The second algorithmic
contribution of an end-to-end differentiated network is a combined in-field prediction
of sun irradiance and berry temperature. The integrated system evaluates the risk of
overheating impacting irrigation decisions.
Developing and using new technology to solve agricultural issues and boost agricul-
tural productivity is necessary. The authors described heterogeneous data management for
agriculture to explore IoT [91]. The article proposed an IoT-based smart farming prediction
and intelligent agricultural analytics model, as well as a decision tendency that reliably
anticipates crop yield using a deep learning approach. Ensemble voting increases the
agricultural enterprise’s profitability, efficiency, and sustainability in this model.
Another author proposed a decision system able to predict the crop yield at the country
level [92]. The results calibrated and trained regression methods for the simulation model
using meteorological, soil, crop, and agro-management data.
The results show that the three proposed machine learning models fit well the crop
data with a high accuracy R2 and minimum values of the root mean square error (RMSE)
and mean absolute percentage error (MAPE) [92].
Sensors 2023, 23, 3752 16 of 22

4.2.4. Weather Forecasting System for Quality Production

Deep learning can be used to anticipate weather patterns and other meteorological
phenomena that may influence crops and farming activities. Deep learning algorithms
may be trained to create precise and trustworthy weather forecasts by studying data from
weather stations, satellites, and other sources. Farmers and other agricultural stakeholders
may utilize this information to guide their decisions regarding the best times to plant, water,
and harvest crops.
The chance of rainfall and the anticipated amount of precipitation may also be pre-
dicted using deep learning algorithms. This information may be utilized to improve
irrigation schedules and to minimize overwatering or underwatering of crops. Addi-
tionally, deep learning algorithms may be used to forecast the possibility of catastrophic
weather occurrences, such as hurricanes, floods, and droughts, which can significantly
influence crops and agricultural operations.
Additionally, forecasts concerning long-term weather patterns, such as trends in
temperature and precipitation, may be made using deep learning systems. Farmers and
other agricultural stakeholders may make plans and decisions about what crops to grow,
how to save water and other resources, and how to adapt to changing conditions using
this knowledge.
Variations in precipitation adversely affect agricultural yield and inflict climatic
severe conditions. Especially in rain-dominated countries, research concentrating on
climate swings, such as variations in rainfall and temperature, is essential. It is chal-
lenging to predict precipitation accurately due to its dynamic nature. Using 30 years of
climate data, the authors planned to develop a prediction model based on an optimized
GRU neural network [93]. The minimal loss reached by the model revealed the validity
of chosen elements to correctly predict precipitation irrespective of the volatility of
meteorological conditions.
The publication describes field-transportable rain forecasting equipment that can
identify the chance of precipitation by detecting relevant atmospheric variables, such
as temperature, humidity, and atmospheric pressure, as well as sky conditions [94].
In addition to the device’s portability, it can be used to predict the likelihood of
precipitation at a specific location by integrating the prediction model of cloud images
using deep learning [95].
The research by authors in the preprocessing phase employed wavelet decomposition.
The implementation of a long short-term memory (LSTM) network and of a suitable forecast
improvement phase was further optimized by algorithms for monthly rainfall forecast
modifications [96]. The approach was executed at four weather stations and compared to
transfer function models, multiple regression, and other prediction strategies.
One effective strategy for conserving water is utilizing rainwater to its maximum
capacity. It is possible to use weather forecasting to save irrigation water; however, unnec-
essary watering and yield loss should be avoided due to weather forecast unpredictability.
A deep Q-learning (DQN) irrigation decision-making system based on short-term weather
forecasts was established to determine the optimal irrigation method. The DQN irrigation
method showed strong generalization capability and may be used to make irrigation deci-
sions based on weather forecasts. The DQN irrigation strategy of learning from previous
irrigation experiences and uncertain weather forecasts mitigated the risks associated with
faulty weather forecasts [97].

4.3. Challenges in Technology Adoption

AIoT is very promising as it has an immense potential to transform society and
business positively. According to [98], the use of the newest technology with no experience
and digital skills is currently the biggest challenge in many industries since many companies
have started to use “things” to connect with others [99]. As “things” connected to the
internet are rapidly growing in number, it may lead to several issues in technology adoption.
Sensors 2023, 23, 3752 17 of 22

One of the issues most companies face is complexity. This complexity refers to
the interrelation of connected devices with other systems from a cyber-physical system
perspective [100]. In technology adoption, the major challenge faced by users is the
monetary budget for a given purpose, such as purchasing the tools and main training
the overall system. The limited knowledge and awareness also caused challenges in
technology adoption.
The other challenge is the need for more trust in AIoT technology. Lack of confidence
in AI and IoT also could delay technology adoption. It is expected that AIoT technology
is reliable and highly dependent on accurate knowledge in the context of the flexibility of
data handling. Farmers are not tech-savvy, so they may entirely depend on the experts to
understand and analyze the AIoT system. It can become challenging for them to learn and
adopt the technology. Additionally, the industry found it challenging to hire experts and
professionals with digital skills to implement new systems as well as operate and maintain
new technology operations.
Other than that, privacy and security issues also cause a delay in technology adoption.
The integration of AI and IoT can create new security risks, such as data breaches and
cyber-attacks. As IoT devices collect and transmit vast amounts of data, privacy concerns
are becoming more significant. Hence, organizations should be aware of the applicable
laws and regulations regarding the data preservation.
In many industries, technology infrastructure was found to be an essential factor to en-
suring the managerial success of the new technology [98]. Without a proper infrastructure,
the company can be addressed as having an outdated technology of AIoT.
The most challenging part of deploying an AIoT model in agriculture is ensuring it
maintains its performance in any environmental conditions or uncertainty factors, such as
rainfall, humidity, sunlight, temperature, and water availability.
AIoT systems require many data from agriculture to produce robust AI models. In the
case of agriculture, the data are usually based on seasons even though spatial data can be
collected in real-time. This limitation may make it challenging to achieve the accuracy of
AI models at a particular time. Moreover, an inaccurate prediction may entail costs. Errors
in making predictions and recommendations could mean losing crop production for an
entire year, affecting farmers’ lives and global food security. Hence, using a small portion
of the farmers’ land for data analysis based on AI systems may become a part of strategic
planning performed prior to deploying the AI model on the whole farm.

5. Conclusions
This study presented a systematic literature review of AIoT studies to highlight the
increasing attention on this technology. It is worth noting that the number of articles
reporting on AIoT adaption in smart agriculture has been increased massively since 2021,
from less than 5 in 2017 to a projection of close to 50 papers in 2023. From IoT to the AI
techniques, this brings remarkable progress towards AIoT in various applications. AIoT-
enabled devices, real-time data collection, and automation are important criteria that can
improve the smart agriculture industry. The current application of AI/IoT technologies
with some advantages of AIoT was summarized in this paper. Several advantages of
AIoT in agriculture, such as disease identification, smart farm monitoring, and efficient
agricultural data analysis, were also highlighted. It can also be noted that there is some
limitation on technology adoption in the smart agriculture industry as related policies
and mechanisms are still under-developed to push as the assessment. Lastly, technology
complexity, privacy and security issues, and under-developed infrastructure were identified
as a few challenges of the application of AIoT in smart agriculture.

Author Contributions: Conceptualization, M.S.M. and H.K.A.; methodology, M.A.R.; writing—

original draft preparation, K.N.S.W.S.W.; writing—review and editing, H.K.A. and N.A.I.; fund-
ing acquisition, A.G.-B. and J.M.C. All authors have read and agreed to the published version of
the manuscript.
Sensors 2023, 23, 3752 18 of 22

Funding: This research was funded by Universiti Malaysia Kelantan with grant numbers: R/NICE
/B0500/00950B/001/2021/00916 (UMK-NICE) , R/FUND/A0100/01862A/001/2020/00819 (UMK Fund)
and United Arab Emirates University through Research Start-up Program (No. Vot: 12M109).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The authors would like to thank researchers from the Institute for Artificial
Intelligence and Big Data, Universiti Malaysia Kelantan (UMK), Malaysia for administrative and
technical support. Thank you also to several members of the Health Data Science Lab, Department
of Genetics and Genomics, College of Medicine and Health Sciences, UAE University for providing
ideas, comments, and improvements for this research.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Deutsch, C.A.; Tewksbury, J.J.; Tigchelaar, M.; Battisti, D.S.; Merrill, S.C.; Huey, R.B.; Naylor, R.L. Increase in crop losses to insect
pests in a warming climate. Science 2018, 361, 916–919. [CrossRef] [PubMed]
2. Food and Agriculture Organization of the United Nations FAOSTAT Pesticides Use. Available online: https://www.fao.org/fao-
stat/en/#data/RP/visualize (accessed on 10 January 2022).
3. Javaid, M.; Haleem, A.; Singh, R.P.; Suman, R. Enhancing smart farming through the applications of Agriculture 4.0 technolgies.
Int. J. Intell. Netw. 2022, 3, 150–164.
4. Pant, J.; Pant, R.P.; Singh, M.K.; Singh, D.P.; Pant, H. Analysis of agricultural crop yield prediction using statistical techniques of
machine learning. Mater. Today Proc. 2021, 46, 10922–10926. [CrossRef]
5. O’Shaughnessy, S.A.; Kim, M.; Lee, S.G.; Kim, Y.; Kim, H.; Shekailo, J. Towards smart farming solutions in the U.S. and South
Korea: A comparison of the current status. Geogr. Sustain. 2021, 2, 312–327.
6. Khan, J.I.; Khan, J.; Ali, F.; Ullah, F.; Bacha, J.; Lee, S. Artificial Intelligence and Internet of Things (AI-IoT) Technologies in
Response to COVID-19 Pandemic: A Systematic Review. IEEE Access 2022, 10, 62613–62660. [CrossRef]
7. Said, S.; Ihab, R.; Hesham, O.; Tabl, I.A.; Maged, N.; Youssef, S.; Elagamy, M. AIOT-Arch: Furthering Artificial Intelligence in Big
Data IoT Applications. IOP Conf. Ser. Mater. Sci. Eng. 2021, 1051, 012008. [CrossRef]
8. Zhang, H.; He, L.; Di Gioia, F.; Choi, D.; Elia, A.; Heinemann, P. LoRaWAN based internet of things (IoT) system for precision
irrigation in plasticulture fresh-market tomato. Smart Agric. Technol. 2022, 2, 100053. [CrossRef]
9. Zhang, T.; Zhao, Y.; Jia, W.; Chen, M.-Y. Collaborative algorithms that combine AI with IoT towards monitoring and controsystem.
Future Gener. Comput. Syst. 2021, 125, 677–686. [CrossRef]
10. Chen, S.W.; Gu, X.-W.; Wang, J.-J.; Zhu, H.-S. AIoT Used for COVID-19 Pandemic Prevention and Control. Contrast Media Mol.
Imaging 2021, 2021, 1–23. [CrossRef]
11. Lombardi, M.; Pascale, F.; Santaniello, D. Internet of Things: A General Overview between Architectures, Protocols and
Applications. Information 2021, 12, 87. [CrossRef]
12. Triantafyllou, A.; Sarigiannidis, P.; Lagkas, T.D. Network Protocols, Schemes, and Mechanisms for Internet of Things (IoT)
Features, Open Challenges, and Trends. Wirel. Commun. Mob. Comput. 2018, 2018, 5349894. [CrossRef]
13. Hasyiya, K.A.; Ku, A.K.H.; Nurul, H.M.H.; Muhammad, A.R.; Ferda, E.; Panji, W.W. Forecasting and Analysis of Solar Power
Output from Integrated Solar Energy and IoT System. In Proceedings of the 5th International Conference on Informatics and
Computational Sciences (ICICoS), Semarang, Indonesia, 24–25 November 2021; pp. 222–226.
14. Ku, A.K.H.; Norfatihah, M.A.; Mohd, Y.M.Y.; Khairul, N.S.W.S.W.; Siti, Z.M.H.; Hasyiya, K.A. Monitoring and Optimizing
Solar Power Generation of Flat-Fixed and Auto-Tracking Solar Panels with IoT System. IOP Conf. Ser. Mater. Sci. Eng.
2020, 1062, 012011.
15. Massaoudi, M.; Abu-Rub, H.; Refaat, S.S.; Chihi, I.; Oueslati, F.S. Deep Learning in Smart Grid Technology: A Review of Recent
Advancements and Future Prospects. IEEE Access 2021, 9, 54558–54578. [CrossRef]
16. Herath, H.M.K.K.M.B.; Mittal, M. Adoption of artificial intelligence in smart cities: A comprehensive review. Int. J. Inf. Manag.
Data Insights 2022, 2, 100076. [CrossRef]
17. Baduge, S.K.; Thilakarathna, S.; Perera, J.S.; Arashpour, M.; Sharafi, P.; Teodosio, B.; Shringi, A.; Mendis, P. Artificial intelligence
and smart vision for building and construction 4.0: Machine and deep learning methods and applications. Autom. Constr.
2022, 141, 1–26. [CrossRef]
18. Ahamad, S.; Shah, D.; Udhayakumar, R.; Rajeswari, T.S.; Khatiwada, P.; Alanya-Beltran, J. A Comprehensive Analysis of
SmarHome Applying Natural Language Processing Using Internet of Technologies. In Proceedings of the 2nd International
Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE), Greater Noida, India, 28–29 April 2022;
pp. 135–138.
Sensors 2023, 23, 3752 19 of 22

19. Olugbade, S.; Ojo, S.; Imoize, A.L.; Isabona, J.; Alaba, M.O. A Review of Artificial Intelligence and Machine Learning for Incident
Detectors in Road Transport Systems. Math. Comput. Appl. 2022, 27, 77. [CrossRef]
20. Ciecierski-Holmes, T.; Singh, R.; Axt, M.; Brenner, S.; Barteit, S. Artificial intelligence for strengthening healthcare systems in
low-and middle-income countries: A systematic scoping review. NPJ Digit. Med. 2022, 5, 162. [CrossRef]
21. Bhai, J.; Alam, F.; Dhafer, A.; Bojović, M.; Altowaijri, S.M.; Niazi, I.K.; Mehmood, R. Machine Learning-Enabled Internet of Things
(IoT): Data, Applications, and Industry Perspective. Electronics 2022, 11, 2676.
22. Fraga-Lamas, P.; Lopes, S.I.; Fernández-Caramés, T.M. Green IoT and Edge AI as Key Technological Enablers for a Sustainable
Digital Transition towards a Smart Circular Economy: An Industry 5.0 Use Case. Sensors 2021, 21, 5745. [CrossRef]
23. Fujitsu Develops New Technology for Accurately Estimating Postures of the Human Body from Point Cloud Data Using
Millimeter-Wave Sensor. Available online: https://www.fujitsu.com/global/about/resources/news/press-releases/2022/0706-
01.html (accessed on 11 December 2022).
24. Mukhopadhyay, S.C.; Tyagi, S.K.S.; Suryadevara, N.K.; Piuri, V.; Scotti, F.; Zeadally, S. Artificial Intelligence-Based Sensors for
Next Generation IoT Applications: A Review. IEEE Sens. J. 2021, 21, 24920–24932. [CrossRef]
25. Ssengonzi, C.; Kogeda, O.P.; Olwal, T.O. A survey of deep reinforcement learning application in 5G and beyond network slicing
and virtualization. Array 2022, 14, 100142. [CrossRef]
26. Alghamdi, T.A.; Javaid, N. A Survey of Preprocessing Methods Used for Analysis of Big Data Originated from Smart Grids.
IEEE Access 2022, 10, 29149–29171. [CrossRef]
27. Lan, T. Fuzzy Logic-Based Machine Learning Algorithm for Cultural and Creative Product Design. Comput. Intell. Neurosci.
2022, 2022, 7747192. [CrossRef] [PubMed]
28. Lampe, B.; Meng, W. A survey of deep learning-based intrusion detection in automotive applications. Expert Syst. Appl.
2023, 221, 119771. [CrossRef]
29. Zhang, J.; Tao, D. Empowering Things with Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial
Intelligence of Things. IEEE Internet Things J. 2021, 8, 7789–7817. [CrossRef]
30. Srivastava, P.; Mishra, K.; Awasthi, V.; Sahu, V.K.; Pal, P.K. Plant Disease Detection Using Convolutional Neural Network. Int. J.
Adv. Res. 2021, 9, 691–698. [CrossRef] [PubMed]
31. Chen, C.-J.; Huang, Y.-Y.; Li, Y.-S.; Chang, C.-Y.; Huang, Y.-M. An AIoT Based Smart Agricultural System for Pests Detection.
IEEE Access 2020, 8, 180750–180761. [CrossRef]
32. Kitchenham, B.; Brereton, O.P.; Budgen, D.; Turner, M.; Bailey, J.; Linkman, S. Systematic literature reviews in software
engineering—A systematic literature review. Inf. Softw. Technol. 2009, 51, 7–15. [CrossRef]
33. Chiu, M.C.; Yan, W.M.; Bhat, S.A.; Huang, N.-F. Development of smart aquaculture farm management system using IoT and
AI-based surrogate models. J. Agric. Food Res. 2022, 9, 100357. [CrossRef]
34. Chang, C.-L.; Huang, C.-C.; Chen, H.-W. Design and Implementation of Artificial Intelligence of Things for Tea (Camellia sinensis L.)
Grown in a Plant Factory. Agronomy 2022, 12, 2384. [CrossRef]
35. Liao, Y.; Yu, N.; Zhou, G.; Wu, Y.; Wang, C. A wireless multi-channel low-cost lab-on-chip algae culture monitor AIoT system for
algae farm. Comput. Electron. Agric. 2022, 193, 106647. [CrossRef]
36. da Silveira, F.; Lermen, F.H.; Amaral, F.G. An overview of agriculture 4.0 development: Systematic review of descriptions,
technologies, barriers, advantages, and disadvantages. Comput. Electron. Agric. 2021, 189, 106405. [CrossRef]
37. Su, W.-T.; Jiang, L.-Y.; O, T.-H.; Lin, Y.-C.; Hung, M.-H.; Chen, C.-C. AIoT-Cloud-Integrated Smart Livestock Surveillance via
Assembling Deep Networks with Considering Robustness and Semantics Availability. IEEE Robot. Autom. Lett. 2021, 6, 6140–6147.
[CrossRef]
38. Chakraborty, P.; Dizon-Paradis, R.N.; Bhunia, S. ARTS: A Framework for AI-Rooted IoT System Design Automation. IEEE Embed.
Syst. Lett. 2022, 14, 151–154. [CrossRef]
39. Jin, X.; Zhang, J.; Kong, J.; Su, T.; Bai, Y. A Reversible Automatic Selection Normalization (RASN) Deep Network for Predicting in
the Smart Agriculture System. Agronomy 2022, 12, 591. [CrossRef]
40. Lee, K.; Silva, B.N.; Han, K. Deep Learning Entrusted to Fog Nodes (DLEFN) Based Smart Agriculture. Appl. Sci. 2020, 10, 1544.
[CrossRef]
41. de Luna, R.G.; Dadios, E.P.; Bandala, A.A.; Vicerra, R.R.P. Tomato Growth Stage Monitoring for Smart Farm Using Deep Trans-fer
Learning with Machine Learning-Based Maturity Grading. Agrivita 2020, 42, 24–36. [CrossRef]
42. Khan, S.; Tufail, M.; Khan, M.T.; Khan, Z.A.; Anwar, S. Deep Learning-Based Identification System of Weeds and Crops in
Strawberry and Pea Fields for a Precision Agriculture Sprayer. Precis. Agric. 2021, 22, 1711–1727. [CrossRef]
43. Kumar, R.; Kumar, P.; Tripathi, R.; Gupta, G.P.; Gadekallu, T.R.; Srivastava, G. SP2F: A Secured Privacy-Preserving Framework for
Smart Agricultural Unmanned Aerial Vehicles. Comput. Netw. 2021, 187, 107819. [CrossRef]
44. Kashyap, P.K.; Kumar, S.; Jaiswal, A.; Prasad, M.; Gandomi, A.H. Towards Precision Agriculture: IoT-Enabled Intelligent
Irri-gation Systems Using Deep Learning Neural Network. IEEE Sens. J. 2021, 21, 17479–17491. [CrossRef]
45. Bauer, A.; Bostrom, A.G.; Ball, J.; Applegate, C.; Cheng, T.; Laycock, S.; Rojas, S.M.; Kirwan, J.; Zhou, J. Combining Computer
Vision and Deep Learning to Enable Ultra-Scale Aerial Phenotyping and Precision Agriculture: A Case Study of Lettuce
Pro-duction. Hortic. Res. 2019, 6, 70. [CrossRef]
Sensors 2023, 23, 3752 20 of 22

46. Pallagani, V.; Khandelwal, V.; Chandra, B.; Udutalapally, V.; Das, D.; Mohanty, S.P. DCrop: A Deep-Learning Based Framework
for Accurate Prediction of Diseases of Crops in Smart Agriculture. In Proceedings of the 2019 IEEE International Symposium on
Smart Electronic Systems (iSES) (Formerly iNiS), Rourkela, India, 1–18 December 2019.
47. Fiehn, H.B.; Schiebel, L.; Avila, A.F.; Miller, B.; Mickelson, A. Smart Agriculture System Based on Deep Learning. In Proceedings
of the Proceedings of the 2nd International Conference on Smart Digital Environment, Rabat, Morocco, 18–20 October 2018; ACM
International Conference Proceeding Series. Association for Computing Machinery: New York, NY, USA, 2018; pp. 158–165.
48. Kothadiya, D.; Chaudhari, A.; Macwan, R.; Patel, K.; Bhatt, C. The Convergence of Deep Learning and Computer Vision:
Smart City Applications and Research Challenges. In Proceedings of the 3rd International Conference on Integrated Intelligent
Computing Communication & Security (ICIIC 2021), Bangalore, India, 6–7 August 2021.
49. ur Rehman, Z.; Khan, M.A.; Ahmed, F.; Damaševičius, R.; Naqvi, S.R.; Nisar, W.; Javed, K. Recognizing Apple Leaf Diseases
Using a Novel Parallel Real-Time Processing Framework Based on MASK RCNN and Transfer Learning: An Application for
Smart Agriculture. IET Image Process. 2021, 15, 2157–2168. [CrossRef]
50. Thomas, G.; Balocco, S.; Mann, D.; Simundsson, A.; Khorasani, N. Intelligent Agricultural Machinery Using Deep Learning.
IEEE Instrum. Meas. Mag. 2021, 24, 93–100. [CrossRef]
51. Bu, F.; Wang, X. A Smart Agriculture IoT System Based on Deep Reinforcement Learning. Future Gener. Comput. Syst.
2019, 99, 500–507. [CrossRef]
52. Chang, C.L.; Lin, K.M. Smart Agricultural Machine with a Computer Vision-Based Weeding and Variable-Rate Irrigation Scheme.
Robotics 2018, 7, 38. [CrossRef]
53. Mavridou, E.; Vrochidou, E.; Papakostas, G.A.; Pachidis, T.; Kaburlasos, V.G. Machine Vision Systems in Precision Agriculture for
Crop Farming. J. Imaging 2019, 5, 89. [CrossRef]
54. Arumugam, K.; Swathi, Y.; Sanchez, D.T.; Mustafa, M.; Phoemchalard, C.; Phasinam, K.; Okoronkwo, E. Towards Applicability of
Machine Learning Techniques in Agriculture and Energy Sector. Proc. Mater. Today Proc. 2022, 51, 2260–2263. [CrossRef]
55. Bakthavatchalam, K.; Karthik, B.; Thiruvengadam, V.; Muthal, S.; Jose, D.; Kotecha, K.; Varadarajan, V. IoT Frame-work
for Measurement and Precision Agriculture: Predicting the Crop Using Machine Learning Algorithms. Technologies
2022, 10, 13. [CrossRef]
56. Raghuvanshi, A.; Singh, U.K.; Sajja, G.S.; Pallathadka, H.; Asenso, E.; Kamal, M.; Singh, A.; Phasinam, K. Intrusion Detection
Using Machine Learning for Risk Mitigation in IoT-Enabled Smart Irrigation in Smart Farming. J. Food Qual. 2022, 2022, 3955514.
[CrossRef]
57. Alam, M.; Alam, M.S.; Roman, M.; Tufail, M.; Khan, M.U.; Khan, M.T. Real-Time Machine-Learning Based Crop/Weed Detec-tion
and Classification for Variable-Rate Spraying in Precision Agriculture. In Proceedings of the 2020 7th International Confer-ence
on Electrical and Electronics Engineering, ICEEE 2020, Antalya, Turkey, 14–16 April 2020; IEEE: Piscataway, NJ, USA, 2020;
pp. 273–280.
58. Balducci, F.; Impedovo, D.; Pirlo, G. Machine Learning Applications on Agricultural Datasets for Smart Farm Enhancement.
Machines 2018, 6, 38. [CrossRef]
59. Rezk, N.G.; Hemdan, E.E.D.; Attia, A.F.; El-Sayed, A.; El-Rashidy, M.A. An Efficient IoT Based Smart Farming System Using
Machine Learning Algorithms. Multimed. Tools Appl. 2021, 80, 773–797. [CrossRef]
60. Durai, S.K.S.; Shamili, M.D. Smart Farming Using Machine Learning and Deep Learning Techniques. Decis. Anal. J.
2022, 3, 100041. [CrossRef]
61. Zanury, N.A.; Remli, M.A.; Adli, H.K.; Wong, K.N.S.W.S. Recent Developments of Deep Learning in Future Smart Cities:
A Review. In Machine Learning for Smart Environments/Cities: An IoT Approach; Springer: Berlin/Heidelberg, Germany, 2022;
pp. 199–212.
62. Mazlan, A.U.; Sahabudin, N.A.; Remli, M.A.; Ismail, N.S.N.; Mohamad, M.S.; Nies, H.W.; Abd Warif, N.B. A Review on
Recent Progress in Machine Learning and Deep Learning Methods for Cancer Classification on Gene Expression Data. Processes
2021, 9, 1466. [CrossRef]
63. Krichen, M.; Mihoub, A.; Alzahrani, M.Y.; Adoni, W.Y.H.; Nahhal, T. Are Formal Methods Applicable to Machine Learning And
Artificial Intelligence. In Proceedings of the 2022 2nd International Conference of Smart Systems and Emerging Technologies
(SMARTTECH), Riyadh, Saudi Arabia, 22–24 May 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 48–53.
64. Dreossi, T.; Fremont, D.J.; Ghosh, S.; Kim, E.; Ravanbakhsh, H.; Vazquez-Chanlatte, M.; Seshia, S.A. VerifAI: A Toolkit for
the Formal Design and Analysis of Artificial Intelligence-Based Systems. In Computer Aided Verification: 31st International
Conference, CAV 2019, New York City, NY, USA, July 15-18, 2019, Proceedings, Part I 31; Springer: Berlin/Heidelberg, Germany, 2019;
pp. 432–442.
65. Shidik, G.F.; Noersasongko, E.; Nugraha, A.; Andono, P.N.; Jumanto, J.; Kusuma, E.J. A Systematic Review of Intelligence Video
Surveillance: Trends, Techniques, Frameworks, and Datasets. IEEE Access 2019, 7, 170457–170473. [CrossRef]
66. Jouini, M. The impact of Artificial intelligence on surveillance camera system “Facial recognition growth”. Am. J. Eng. Res.
2020, 9, 146–151.
67. Priyadarshi, R.; Bhardwaj, P.; Gupta, P.; Nath, V. Utilization of Smartphone-Based Wireless Sensors in Agricultural Science:
A State of Art. In Microelectronics, Communication Systems, Machine Learning and Internet of Things; Lecture Notes in Electrical
Engineering; Nath, V., Mandal, J.K., Eds.; Springer: Singapore, 2023; p. 887.
68. Krichen, M. Anomalies Detection Through Smartphone Sensors: A Review. IEEE Sens. J. 2021, 216, 7207–7217. [CrossRef]
Sensors 2023, 23, 3752 21 of 22

69. Nguyen, V.K. Detection Aggregation Anomalies Data from Smartphone Sensors. Master’s Thesis, University of Paris-Saclay,
Gif-sur-Yvette, France, 2019.
70. Hagos, T. Android Studio IDE Quick Reference: A Pocket Guide to Android Studio Development; Apress: New York, NY, USA, 2019.
71. Pongnumkul, S.; Chaovalit, P.; Surasvadi, N. Applications of Smartphone-Based Sensors in Agriculture: A Systematic Review of
Research. J. Sens. 2015, 2015, 195308. [CrossRef]
72. Prasad, S.; Peddoju, S.K.; Ghosh, D. Energy efficient mobile vision system for plant leaf disease identification. In Proceedings of
the 2014 IEEE Wireless Communications and Networking Conference, Istanbul, Turkey, 6–9 April 2014; pp. 3314–3319.
73. Intaravanne, Y.; Sumriddetchkajorn, S. Baikhao (rice leaf) App: A mobile device-based application in analyzing the color level of
the rice leaf for nitrogen estimation. Proc. SPIE 2012, 8558, 96–102.
74. Krichen, M.; Lahami, M.; Cheikhrouhou, O.; Alroobaea, R.; Maâlej, A.J. Security testing of Internet of Things for smart city
applications: A formal approach. In Smart Infrastructure and Applications; Springer: Cham, Switzerland, 2020; pp. 629–653.
75. Arun, R.A.; Umamaheswari, S. Effective multi-crop disease detection using pruned complete concatenated deep learningmodel.
Expert Syst. Appl. 2013, 213, 118905. [CrossRef]
76. Paymode, A.S.; Malode, V.B. Transfer Learning for Multi-Crop Leaf Disease Image Classification using Convolutional Neural
Network VGG. Artif. Intell. Agric. 2022, 6, 23–33. [CrossRef]
77. Gajjar, R.; Gajjar, N.; Thakor, V.J.; Patel, N.P.; Ruparelia, S. Real-time detection and identification of plant leaf diseases using
convolutional neural networks on an embedded platform. Vis. Comput. 2020, 38, 2923–2938. [CrossRef]
78. Singh, A.K.; Ganapathysubramanian, B.; Sarkar, S.; Singh, A. Deep Learning for Plant Stress Phenotyping: Trends and Future
Perspectives. Trends Plant Sci. 2018, 23, 883–898. [CrossRef]
79. Gupta, M.; Jain, S. Banana Leaf Diseases and Machine Learning Algorithms Applied to Detect Diseases: A Study. In Emerging
Technologies in Data Mining and Information Security; Springer: Singapore, 2023; pp. 121–130.
80. Baek, E.T.; Im, D.Y. ROS-Based Unmanned Mobile Robot Platform for Agriculture. Appl. Sci. 2022, 12, 4335. [CrossRef]
81. Devanna, R.P.; Milella, A.; Marani, R.; Garofalo, S.P.; Vivaldi, G.A.; Pascuzzi, S.; Reina, G. In-field automatic identification of
pomegranates using a farmer robot. Sensors 2022, 22, 5821. [CrossRef] [PubMed]
82. Yoshida, T.; Onishi, Y.; Kawahara, T.; Fukao, T. Automated harvesting by a dual-arm fruit harvesting robot. ROBOMECH J.
2022, 9, 19. [CrossRef]
83. Moraitis, M.; Vaiopoulos, K.; Balafoutis, A.T. Design and Implementation of an Urban Farming Robot. Micromachines
2022, 13, 250. [CrossRef]
84. Kuska, M.T.; Heim, R.H.; Geedicke, I.; Gold, K.M.; Brugger, A.; Paulus, S. Digital plant pathology: A foundation and guide to
modern agriculture. J. Plant Dis. Prot. 2022, 129, 457–468. [CrossRef]
85. Hu, N.; Su, D.; Wang, S.; Nyamsuren, P.; Qiao, Y.; Jiang, Y.; Cai, Y. LettuceTrack: Detection and tracking of lettuce for robotic
precision spray in agriculture. Front. Plant Sci. 2022, 13, 1–16. [CrossRef]
86. Ivanovic, A.; Polic, M.; Tabak, J.; Orsag, M. Render-in-the-loop aerial robotics simulator: Case Study on Yield Estimation in Indoor
Agriculture. In Proceedings of the 2022 International Conference on Unmanned Aircraft Systems (ICUAS), Dubrovnik, Croatia,
21–24 June 2022.
87. Patrizi, G.; Bartolini, A.; Ciani, L.; Gallo, V.; Sommella, P.; Carratù, M. A Virtual Soil Moisture Sensor for Smart Farming Using
Deep Learning. IEEE Trans. Instrum. Meas. 2022, 71, 1–11. [CrossRef]
88. Amarasingam, N.; Salgadoe, S.; Powell, K.; Gonzalez, L.F.; Natarajan, S. A review of UAV platforms, sensors, and applications for
monitoring of sugarcane crops. Remote Sens. Appl. Soc. Environ. 2022, 26, 100712. [CrossRef]
89. Wang, L.; Wang, J.; Zhang, X.; Wang, L.; Qin, F. Deep segmentation and classification of complex crops using multi-feature
satellite imagery. Comput. Electron. Agric. 2022, 200, 107249. [CrossRef]
90. Akiva, P.; Planche, B.; Roy, A.; Oudemans, P.; Dana, K. Vision on the bog: Cranberry crop risk evaluation with deep learning.
Comput. Electron. Agric. 2022, 203, 107444. [CrossRef]
91. Apat, S.K.; Mishra, J.; Srujan Raju, K.; Padhy, N. State of the Art of Ensemble Learning Approach for Crop Prediction. In Next
Generation of Internet of Things; Springer: Singapore, 2023; pp. 675–685.
92. Aworka, R.; Cedric, L.S.; Adoni, W.Y.H.; Zoueu, J.T.; Mutombo, F.K.; Kimpolo, C.L.M.; Nahhal, T.; Krichen, M. Agricul-
tural decision system based on advanced machine learning models for yield prediction: Case of East African countries.
Smart Agric. Technol. 2022, 2, 100048. [CrossRef]
93. Fahad, S.; Su, F.; Khan, S.U.; Naeem, M.R.; Wei, K. Implementing a novel deep learning technique for rainfall forecasting via
climatic variables: An approach via hierarchical clustering analysis. Sci. Total Environ. 2023, 854, 158760. [CrossRef]
94. Ambildhuke, G.; Banik, B.G. IoT based Portable Weather Station for Irrigation Management using Real-Time Parameters.
Int. J. Adv. Comput. Sci. Appl. 2022, 13, 267–278. [CrossRef]
95. Ravi, S.S.; Sai, R.G.K. Design of Deep Learning Model for Predicting Rainfall. In Proceedings of the 2022 8th Interna-
tional Conference on Advanced Computing and Communication Systems (ICACCS), Coimbatore, India, 25–26 March 2022;
IEEE: Piscataway, NJ, USA, 2022; Volume 1, pp. 1343–1347.
96. Vivas, E.; de Guenni, L.B.; Allende-Cid, H.; Salas, R. Deep Lagged-Wavelet for monthly rainfall forecasting in a tropical region.
Stoch. Environ. Res. Risk Assess. 2022, 37, 831–848. [CrossRef]
97. Chen, M.; Cui, Y.; Wang, X.; Xie, H.; Liu, F.; Luo, T.; Luo, Y. A reinforcement learning approach to irrigation decision-making for
rice using weather forecasts. Agric. Water Manag. 2021, 250, 106838. [CrossRef]
Sensors 2023, 23, 3752 22 of 22

98. Ayaz, M.; Ammad-udin, M.; Sharif, Z.; Ali, M.; El-Hadi, M. Internet -of-Things (IoT) based Smart Agriculture: Towards Mak-ing
the Fields Talk. IEEE Access. 2019, 7, 135917–135932. [CrossRef]
99. Musab, S.; Al-Hadrusi Nabil, J. Efficient Control of PTZ Cameras in Automated Video Surveillance Systems. In Proceedings of
the 2012 IEEE International Symposium on Multimedia, Irvine, CA, USA, 10–12 December 2012; IEEE: Piscataway, NJ, USA, 2012;
pp. 356–359. [CrossRef]
100. Nozari, H.; Szmelter-Jarosz, A.; Ghahremani-Nahr, J. Analysis of the Challenges of Artificial Intelligence of Things (AIoT) for the
Smart Supply Chain (Case Study: FMCG Industries). Sensors 2022, 22, 2931. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.