See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/379810822

Machine vision and artificial intelligence for plant growth stress detection
and monitoring: A review

Article in Precision Agriculture Science and Technology · April 2024

DOI: 10.12972/pastj.20240003

CITATIONS READS

3 1,401

8 authors, including:

Sumaiya Islam Md Nasim Reza

Chungnam National University Chungnam National University
31 PUBLICATIONS 151 CITATIONS 74 PUBLICATIONS 390 CITATIONS

SEE PROFILE SEE PROFILE

Samsuzzaman Samsuzzaman Shahriar Ahmed

Chungnam National University Chungnam National University
11 PUBLICATIONS 43 CITATIONS 17 PUBLICATIONS 18 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Sumaiya Islam on 14 April 2024.

The user has requested enhancement of the downloaded file.

 pISSN : 2672-0086
eISSN : 2713-5632

REVIEW ARTICLE

Machine vision and artificial intelligence for

plant growth stress detection and monitoring:
A review
Sumaiya Islam1, Md Nasim Reza1,2, Samsuzzaman2, Shahriar Ahmed2, Yeon Jin Cho3, Dong Hee Noh4,
Sun-Ok Chung1,2,†*, Soon Jung Hong5,†*

Department of Smart Agricultural Systems, Graduate School, Chungnam National University, Daejeon
1

34134, Republic of Korea

2
Department of Agricultural Machinery Engineering, Graduate School, Chungnam National University,
Daejeon 34134, Republic of Korea
3
Jeonnam Agricultural Research Extension Services, Naju 58213, Republic of Korea
4
Jeonbuk Regional Branch, Korea Electronics Technology Institute (KETI), Jeonju 54853, Republic of
Korea
5
Korea National University of Agriculture and Fisheries, Jeonju 54874, Republic of Korea

*Corresponding author: [email protected]; [email protected]

†
These authors equally contributed to this study as corresponding author.

Abstract
The agricultural sector faces increasing challenges in ensuring food security and optimizing
crop yield, necessitating innovative solutions for early detection and mitigation of plant
growth stress. The integration of advanced imaging technologies with artificial intelligence (AI)
has emerged as a powerful tool for non-invasive, real-time monitoring of plant health. The
objective of this paper was to review the application of machine vision and AI in identifying
and classifying plant growth stress, with a focus on stressors, datasets, and the use of
intelligent algorithms. The significance of plant growth stress induced by environmental
variables, including temperature, light, nutrient deficiencies, and water supply were addressed
and the conventional stress detection methodologies, underscores their inherent limitations,
and establishes the groundwork for the exploration of state-of-the-art technologies in stress
OPEN ACCESS assessment. Various sensor technologies were explored, encompassing traditional RGB
DOI: https://doi.org/10.12972/pastj.20240003 cameras, multispectral and hyperspectral sensors, and thermal imaging, each capable of
capturing distinct stress signatures. Machine vision, leveraging high-resolution imaging and
Received: February 22, 2024
spectroscopy, offers detailed insights into plant physiological responses. Coupled with AI
Revised: April 05, 2024
approaches such as deep learning, neural networks, and pattern recognition, machine vision
Accepted: April 05, 2024
enables the automated analysis of vast datasets, enhancing the accuracy and speed of stress
Copyright: © 2024 Korean Society of detection. The recent advancements in image processing techniques tailored for plant stress
Precision Agriculture
identification were focused and discussed the role of feature extraction, classification, and
This is an Open Access article
distributed under the terms of predictive modelling in achieving robust results. The potentials of AI in plant stress physiology
theCreativeCommonsAttributionNon-Commercial and its role in overcoming the limitations of traditional methods, and the use of unsupervised
License (http: //creativecommons.org/licenses/by- identification of visual symptoms to quantify stress severity, allowing for the identification
nc/4.0/)whichpermitsunrestrictednon-commercial
of different types of plant stress were studied. Moreover, the potentials of machine vision
use, distribution, and reproduction in any medium,
technology and AI for real-time monitoring and decision support systems in precision
provided the original work is properly cited.

Precision Agriculture Science and Technology 6(1) March 2024

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

agriculture were discussed. The findings of this review would contribute to the growing field of agricultural
technology, offering insights into the development of automated tools that could aid farmers and researchers in
mitigating the impact of abiotic stressors on crop/plant health and productivity.

Keywords: Precision agriculture, abiotic stress, deep learning, image processing, stress detection

Introduction
In the current era, agriculture stands as a key contributor to the global economy. The expanding population, coupled with urbanization,
is progressively shrinking cultivable land, intensifying the challenges faced by the agricultural sector (Tian et al., 2020; Wang et al.,
2019). The increasing demand for efficient and safe food production methods necessitates a shift towards innovative sensing, driving
technologies, and enhanced information and communication technologies (Liakos et al., 2018). To address these challenges, traditional
agricultural management approaches need supplementation with advanced technologies to boost productivity accurately, fostering the
evolution of high-quality and high-yield agriculture (Balaska et al., 2023). Furthermore, there is a growing need to adopt sustainable
technologies that can improve agricultural efficiency and address environmental concerns.
Plant stress refers to external conditions that adversely affect the growth, development, or productivity of plants (Lichtenthaler, 1998).
Plant stress poses a major threat to crop yield and quality, making the rapid and robust detection and diagnosis of plant stress essential
for the application of precision agriculture (Gao et al., 2020). The effects of plant growth stress on crop yield can be significant, with
various factors such as temperature, light, nutrient, water and other environmental stressors leading to reduced productivity (Kopecká
et al. 2023; Fahad et al., 2017). Water supply and temperature stress can cause damage to the reproductive growth phase of crops,
resulting in a substantial decrease in yield (Seleiman et al., 2021). These stresses can also lead to damaged photosynthetic process,
oxidative damage, and membrane instability, further impacting crop productivity. For instance, temperature stress has been observed to
cause substantial yield reductions in crops such as common beans, peanut, and tomato (Hussain et al., 2018). Additionally, prolonged
exposure to water stress during the reproductive stage can significantly affect grain quality and yield.
In this context, optical imaging methods have gained prominence for detecting plant diseases and stressors. These advanced
techniques enable the rapid and non-contact measurement of physiological changes in plants caused by various stressors, including
abiotic and biotic factors. Commonly used imaging technologies for detecting crop stress include digital, fluorescence, thermography,
LIDAR, multispectral, and hyperspectral imaging techniques. Digital imaging sensors capture RGB coloured images, providing
valuable information on plant attributes like canopy vigour, leaf color, texture, size, and shape for disease detection (Mahlein et al.,
2016). Color and texture features, including RGB, LAB, YCBCR, HSV spaces, contrast, homogeneity, dissimilarity, energy, and
entropy aid in distinguishing between healthy and symptomatic plants (Gao et al., 2023). The high spatial resolution images provided
by RGB cameras make them a valuable tool for studying plant phenomena and assessing visual symptoms related to plant stress.
Thermal cameras detect radiation in the infrared wavelength range and produce measurements displayed as false-color images,
allowing for the visualization of temperature variations in plants (Pineda et al., 2020). These sensors have been employed to detect a
variety of plant growth stresses, such as pathogen infections, water scarcity, and nutrient deficiencies. Fluorescence imaging sensors
and multispectral sensors are valuable tools for plant stress detection. These sensors offer a comprehensive approach to assessing plant
health and detecting various types of stress, contributing to improved crop management and yield (Zubler et al., 2020). Fluorescence
imaging evaluates spatiotemporal changes in leaves and detecting plant stress. It can provide early-stage detection of both biotic and

Precision Agriculture Science and Technology 6(1) March 2024 34

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

abiotic stresses before visible symptoms appear (Kumar et al., 2021). Multispectral and hyperspectral imaging across visible, near-
infrared, and infrared wavelengths detect changes in leaf reflectance linked to plant stresses (Weng et al., 2023). Hyperspectral offers
finer spectral detail than multispectral, enhancing stress identification.
The integration of optical sensors and machine learning has shown promise in the proximal detection of plant stress, allowing for
the evaluation of plant health using various optical sensors. These sensors provide valuable data that can be analyzed using machine
learning algorithms to detect and assess plant stress. Machine learning techniques, such as artificial neural networks and support
vector machines, are employed for data processing and analysis, enabling the rapid and accurate detection of plant stress. Additionally,
machine vision technologies, including deep learning and image processing techniques, have been increasingly utilized for the
detection of stresses and diseases in precision agriculture. These advancements hold significant potential for enabling early and accurate
identification of plant stress, thereby contributing to improved agricultural efficiency and productivity. The objective of this paper was
to conduct an overview of the application of machine vision and AI for plant growth stress detection, and provide insights that can
inform future research, innovation, and practical applications in sustainable agriculture and food production.

Types of plant stress

Stress is a changed physical state brought on by biotic or abiotic causes that upset a plant's natural balance. In consequence
environmental stress causes massive agricultural losses (Bechtold et al., 2018). Plant stress encompasses adverse conditions that impact
a plant's metabolism, growth, and development (Gull et al., 2019). Diverse factors contribute to plant stress, including extremes in
temperature, light fluctuations, pathogenic infections, pest infestations, nutrient deficiencies, and stresses like salinity, drought, and
heavy metal exposure. These stressors can induce alterations in gene expression, disrupt cellular metabolism, affect growth rates, and
ultimately result in diminished crop yields. Plants have evolved an array of mechanisms to cope with these stresses, underscoring the
importance of comprehending their responses for devising effective strategies to mitigate their impact on agricultural production (Kollist
et al., 2019). Plant stress also leads to alteration of the leaf structure, which adjusts light reflection on the leaf surface and modifies the
color and texture of the leaf (Ghosal et al., 2018).
There are two types of plant growth stress, named biotic stress and abiotic stress. The term "biotic stress" refers to living organisms
that harm crops, such as fungi, bacteria, viruses, weeds, parasites, insects, and diseases (Dhingra et al., 2019). Abiotic stress is the result
of non-living elements including nutrition shortages, floods, and other environmental variables like drought, temperature, humidity,
light. Fig. 1 shows various types of stress on crop and plant growth significantly impact agricultural production. As a defence strategy
against the threat posed by the stresses they encounter, plant have special ways of reacting to stress (Czedik‐Eysenberg et al., 2018).
Plant senescence, leaf curling, and wilting caused by biotic stress result in decreased leaf area, which lowers harvest weight, increases
plant diameter, and produces fewer leaves (Saakre et al., 2017). Affected tissues or plants may experience abiotic or biotic stress due to
nutrition and water shortages, insect infestations, or cell architectures that prevent photosynthesis and transpiration (Virlet et al., 2016).
It is essential to prevent plant stress in crops that require substantial water for growth and yield. This stress can occur when there is
excessive water supply in the field, leading to issues such as waterlogging and clogging (Zhao et al., 2017). Water delivery to plants
during the period of growth and during the time when mature harvests are being harvested can also contribute to plant stress (Duarte-
Galvan et al., 2014). Lack or excess of nutrients leads to plant stress, which is brought on by a lack of nutrients required for plant growth
(Nachtigall et al., 2017).

Precision Agriculture Science and Technology 6(1) March 2024 35

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Fig. 1. Various forms of stress on crop and plant growth significantly impact agricultural production.

Numerous aspects of the effects of abiotic stress on plant performance are being investigated, including metabolic/physiological
reactions, molecular signalling pathways, ecophysiology, and crop breeding research. Abiotic stress significantly affects plant growth
and productivity. Various factors such as drought, salinity, extreme temperatures, and heavy metals can limit crop productivity and
lead to great yield losses (Islam et al., 2021a). Plants have developed potent adaptive tactics to combat these stresses, including cellular,
physiological, and morphological defences (He et al., 2018; Cramer et al., 2011). The effects of abiotic stress on plants are dynamic and
complex, impacting the synthesis, concentration, metabolism, transport, and storage of sugars, as well as causing ionic toxicity, osmotic
pressure, oxidative damage, and nutritional shortage (Cramer et al., 2011). As a result, it is imperative to develop crops with multi-stress
tolerance to alleviate the pressure of these adverse environmental conditions. In the majority of abiotic stress condition plant metabolism
is disrupted and accumulates reactive oxygen species (ROS) causing oxidation and dysfunction of cellular components and finally cell
death (Obata et al., 2012). Therefore, assuring optimal crop yields with minimum losses from plant environmental stressors including
light, temperature, nutrition, and water is essential for the agricultural industry's sustainability as the global population continues to grow.
For various plant parts like flowers, leaves, stems, fruits, canopies, and roots, plant phenotypic criteria include color, size, texture,
and shape (open to be observed) (Story et al., 2009). This information is useful for plant growth and stress monitoring in horticulture
systems since it indicates the growth stage and health levels of plants. In order to determine the growth rate or annual growth cycle of
plants cultivated under various or similar weather or nutrient circumstances, plant height is a quantitative morphological characteristic
(Chaudhury et al., 2018; Gupta et al., 2019). Fall foliage is a qualitative indicator of a plant's overall health (Kumar et al., 2019). The
fresh and dry weight of the plant may also be connected to the upper canopy area (Shimizu et al., 1995; Lee et al., 2018). The growth
and health status of the plants are reflected in this data. It is beneficial for horticulture systems management and decision-making.
The information was previously gathered via visual inspection or physical measurement techniques, which calls for a high degree
of experience from producers by measuring the size of seedling and determine whether it has grown to a sufficient level before
transplantation. Early plant stress detection is essential in order to reduce both acute and chronic productivity loss. When stress develops
in plants, quick diagnosis and decision-making are required to alleviate the stress and avoid financial loss.

Precision Agriculture Science and Technology 6(1) March 2024 36

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Quantification of plant growth stress symptom

When a plant grows under less-than-ideal conditions that put more demands on it, it is said to be under stress. The quantification of
plant growth stress symptoms involves measuring various physiological, morphological, and visual parameters to detect the effects
of stress on plants (Galieni et al., 2021). These parameters can include biomass production, growth, visible stress symptoms, root, and
shoot biomass, root-shoot ratio, leaf area, plant height, leaf number, wilting symptoms, and presence of flowers. Fig. 2 shows the
effects of abiotic stress on different types of seedlings. The key indicator to determine the type of stress cause the relative distribution
of symptoms at the leaf, branch, and crown levels. With the changes of environment, internal activities of plant metabolism changes
to survive in the new environment. The effects of abiotic stresses can occur either in molecular level or morphological level. The
physiological reaction to plant stress is leaf wilting, leaf abscission, decrease of leaf area etc. Environmental stress suppresses cell
elongation in consequence with decreasing leaf area, plant height, and other developments of crops owing to cell elongation.

Fig. 2. Visual effects of abiotic stresses on different types of seedlings.

Rafique et al. (2020) showed the response of different growth parameters (leaf area, plant height, and stem diameter) under drought
stress. Lack of water reduces the number of leaves on a plant, their size, and their lifespan. Temperature, acclimation, and leaf turgor
pressure are necessary for the creation and increase of the leaf area index. Drought reduces the leaf zone, which is thought to be caused
by the leaf extension being hidden, which lowers the photosynthetic rate. A common adverse effect of water stress on rice plants is
a reduction in both fresh and dry biomass (Zhang et al., 2006). Francesca et al. (2022) showed the difference of tomato plant height,
leaf number, leaf area, shoot fresh weight and shoot dry weight between drought and heat stressed and non-stressed plants. López-
Marín et al. (2013) exhibited pepper plant height, leaf area, leaf weight decreased under thermal stress condition than normal condition.
Temperature, light intensity, and photoperiod all affect plants actively; all three variables working together can boost biomass, promote
growth, and enhance plant morphology.
Islam et al. (2021b) proposed and image processing algorithm for ice-plant leaf area calculation under different light condition.
Different LED combination affects the growth and leaf size. The purpose of one study was to identify the ideal photo-thermal settings

Precision Agriculture Science and Technology 6(1) March 2024 37

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

and examine the impact of various photoperiods, light intensities, and day/night temperatures on the growth and quality of lettuce
cultivated in glasshouses (Iqbal et al., 2022). Study shows photosynthetic rate can suppress in high salt concentration can limit stomatal
activities resulting in available CO2 restriction with discoloration and freshness in lettuce plants (Neocleous et al., 2014). Abdullah et al.
(2018) showed impact of infrared lighting and water stress on growth and quality of cucumber. The obtained outcomes showed using
infrared light and any of the commonly used irrigation schedules (used 70, 85, or 100% of the water) improved every measurement
of vegetative development (plant height, number of leaves per plant, fresh and dry weight, whole plant). Chowdhury et al. (2023)
investigated the effects of light quality, photoperiod, and light intensity on Chinese kale growth and glucosinolate content during
different growth stages of cultivation in an aeroponics plant factory. The light significantly affect the growth and glucosinolate contents
of Chinese kale leaves.

Image-based plant growth stress symptom detection

Plant phenotyping refers to the systematic and comprehensive analysis of observable traits or characteristics (phenotypes) of plants
throughout their life cycle. Plant phenotyping is ideal in environmentally controlled greenhouses and glasshouses since they provide
a variety of automatic control techniques to monitor and record ambient parameters (temperature, relative humidity, light, and water
content) in plants (Basak et al., 2019). There are few research to investigate environmental stress detection in indoor farming such as
classifying water stress (Hendrawan et al., 2021), looking for tip-burn stress (Gozzovelli et al., 2021), grading lettuce light sensitivity
(Hao et al., 2020), and categorizing odd leaves (Wu et al., 2021). Nondestructive plant phenotypic image collections are available
using imaging and image processing techniques with illumination sources ranging from the visible to near-infrared. For modeling
and predicting plant growth and development, these techniques have improved precision, boosted throughput, and produced high-
dimensional phenotypic data (Tardieu and Tuberosa, 2010, Golzarian et al., 2011). Better performance and a novel viewpoint for
improving plant phenotypes have been attained through the employment of integrated image-based innovative tools and techniques
in phenomics and environmentally controlled platforms and facilities (Granier et al., 2014). In recent years, machine vision has
become widely used in plant industries' intelligent systems (Dewi et. Al., 2020, Schor et al., 2017, Bechar et al., 2010, Valente et al.,
2019). Almost every aspect of production is covered by this technology, including fruit grading and seedling raising, transplanting,
management, and harvesting. By mimicking human vision, it gathers information from images, evaluates it, and provides guidance for
actual production (Ren et al., 2007, Yang et al., 2019).
Unlike chemical and physical procedures, machine vision systems gather information about plants without causing harm to the plants.
It also operates continuously at a low cost and excellent efficiency. Machine vision systems, also known as computer vision systems,
have been increasingly used for non-invasive plant information gathering. These systems utilize various techniques such as image
processing, pattern recognition, and deep learning to identify plant species based on visual data (Oppong et al., 2022). For instance,
research has been conducted on the development of computer vision systems for automatic plant species identification using techniques
like deep learning, leaf image analysis, and feature extraction (Wäldchen et al., 2018). These systems allow for the non-destructive
collection of plant data, enabling tasks such as species identification and monitoring without causing harm to the plants (Oppong et al.,
2022). Machine vision applications have certain obstacles because the illumination environment varies based on the needs of the plant.
Instead of considering the needs of the plants, environmental factors including air temperature, relative humidity, and lighting intensity
were changed in a traditional greenhouse based on farmer observations or predetermined thresholds (Lakhiar et al., 2018; Kim et al.,
2010). Machine vision can assist with this limitation by providing the optimal method for plants to receive nutrition, water, temperature,

Precision Agriculture Science and Technology 6(1) March 2024 38

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

and light based on continuously recorded growth data. Fig. 3 shows different imaging systems for plant stress detection. Application
of genomics and cutting-edge technical advancements improve current plant breeding programs to create plants with larger biomass,
greater nutrition, and stronger tolerance to biotic and abiotic stressors. Significant breakthroughs in genetic tools and methodologies
have enabled the production of high-yielding, stress-tolerant. However, progress in this area is hampered by the difficulty in accessing
phenotypic data. In addition, the current phenotyping technologies are slow, expensive, labor-intensive, and primarily harmful (Furbank
et al., 2011).

Fig. 3. Different imaging systems for plant growth stress detection.

Utilizing various imaging methods and methodologies, symptoms can be efficiently monitored (Sankaran et al., 2015). Utilizing
sensors, static scanning, fisher discrimination analysis, and three-dimensional modeling are just a few examples of imaging technologies.
To detect phenotypic reactions during plant under stress condition magnetic (Gout et al., 2001), soft x-ray imaging (Numajiri et al.,
2021), and ultrasound (Yakupoğlu et al., 2023) have been used. Static scanning technologies have been used for drought stress detection
by collecting images and working on parameters using RGB mean value functions (Wu et al., 2021). Chen et al. (2014) used fisher
discrimination analysis for the identification nitrogen, phosphorus, potassium stress in plant (Chen et al., 2014). For the water stress
measurement of plant infrared thermal imaging has been commonly used in recent days (Vadivambal et al., 2011; Egea et al., 2017;
García-Tejero et al., 2016; García-Tejero et al., 2018). When water stress occurs, plants respond with a limited stomatal closure, limiting
stomatal conductance, restraining leaf transpiration, and aiding in the reduction of the evaporative chilling procedure, resulting in high
leaf temperature. Visible light imaging is also a simple and inexpensive imaging technique by using traditional color cameras with
wavelengths from 400 to 750nm in the electromagnetic spectrum. This imaging systems also can investigate the effects of plant under
stress condition before the appearance of visible symptoms (Raza et al., 2015). Visible imaging techniques in glasshouses, greenhouses,
and screen houses are very useful in assessing leaf biomass, crop traits, panicle traits, inhibition and growth rates, leaf physiology and

Precision Agriculture Science and Technology 6(1) March 2024 39

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

structure, seedling strength, coleoptile length, and biomass at kernel morphology, and root structural mechanism under controlled
conditions (Li et al., 2014).
Spectrometry (hyperspectral imaging) is a proximal and remote sensing method for determining crop status and health (Khan et
al., 2018). Multispectral or hyperspectral sensors, which can periodically scan wavelengths of interest, can be used to gather spectral
measurements of the electromagnetic spectrum (Fiorani et al., 2013). Absorption waveband in the infrared range from the multispectral
and hyperspectral measurements are frequently employed to explain various water statuses that measure the canopy water content
(Cabrera‐Bosquet et al., 2012). Susic et al. (2018) used hyperspectral imaging to discriminate between abiotic and biotic drought
stress in tomatoes. Positron emission tomography (PET) imaging technology has advanced and introduced to get exact visualization of
plants. By recognizing a pair of gamma rays emitted by a positron-emitting tracer injected into the plant, positron emission tomography
creates 3D images. It primarily focuses on physiological changes and is used to evaluate how well photosynthetic function and
environmental stress are occurring (Baker et al., 2008).
Chlorophyll a, a fluorophore that is sensitive to a variety of environmental stressors, is present in high concentrations in plants by
nature. Chlorophyll fluorescence is a tool that can measure photosynthetic efficiency and detect various types of stress that impact
the photosynthetic apparatus. Chlorophyll fluorescence is directly related to the efficiency of plants' photosynthetic processes as here
are three different outcomes for the energy of photons that have been absorbed by photosynthetic pigments. Energy from photons
can be transformed into heat, used to power the light processes of photosynthesis (electron transport), or reemitted as fluorescence by
chlorophyll a. The variability in chlorophyll fluorescence can thus be used as a measure of photosynthetic efficiency as well as stress
identification. Legendre et al., (2021) used Chlorophyll fluorescence imaging (CFI) to visualize stress induced by the photosystem II-
inhibitory herbicide atrazine. The reflectometer observations based on wavelengths with the highest (741.2 nm) and lowest variability
(548.9 nm) in response to atrazine damage were used to create a fluorescence-based stress index (Legendre et al., 2021).

Image processing techniques for plant stress symptom

identification
Pre-processing, feature extraction, selection, and classification are few of the steps involved in image processing for plant stress
detection. One crucial step in implementing image processing techniques is pre-processing image data. Pre-processing techniques that
range from simple background removal, contrast enhancement, and cropping to more complex procedures like clustering and principal
component analysis (PCA)-based dimensionality reduction that greatly increases the performance of computer vision based analysis.
Principle Component Analysis (PCA) is a statistical technique that turns correlated variable data into a set of linearly uncorrelated
variables known as principle components by the application of an orthogonal transformation. Unfold Principal Component Analysis
(UPCA) is a tool used to identify plant stress conditions and/or help farmers prevent or limit plant damage caused by stress on plants by
identifying these conditions in a timely manner.
In order to improve the signal-to-noise ratio and enable the machine learning model to more readily identify patterns and trends
that will help categorize the data into the right groups, pre-processing depends on the concentration of information. Pre-processing
helps identify pertinent image elements that are essential for model training by utilizing domain knowledge. In order to identify the
photograph, this involves removing any tags, dirt, or soil from the foreground image of either leaves or fruits. Image segmentation,
contrast enhancement, thresholding to binary data, image conversion from RGB to greyscale or RGB to hue saturation value (HSV)
are examples of pre-processing operations that can be applied. Other pre-processing operations include de-noising images using filters

Precision Agriculture Science and Technology 6(1) March 2024 40

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

such as band-pass, low-pass, and fast Fourier transform (FFT); feature extraction at different scales using image transforms such as FFT,
wavelet transforms, Hough transform, Harr transforms, and Radon transforms; pixel-based classification, image clustering into classes,
and dimensionality reduction of images. Using image data generator in Keras for pre-processing, the images were scaled and denoised
based on the observation of changes in the leaf and fruit surface response to the various biotic and abiotic stress (Li et al. 2020).
In earlier research on plant stress detection, Scale-invariant feature transform (SIFT), histogram of oriented gradients (HOG), and
local binary pattern (LBP) feature extractors were used for classifiers such as adaptive bosting (AdaBoost) (Schapire et al., 1999)
and support vector machine (SVM) (Iqbal, 2022). (Marceau et al., 1990) assessed the grey level co-occurrence matrix (GLCM)
for evaluating texture representations that can be examined in spatial attributes retrieved from digital images. This is the connection
between the grey levels that now exist in nearby pixels that affect the overall appearance of image. In order to characterize particular
textural aspects of the image or to generate a new texture image or band that may be included in spectral feature space for classification,
GLCM can be used to measure statistical metrics including homogeneity, contrast, and entropy (Marceau et al., 1990). During co-
occurrence computing, the GLCM technique makes judgments based on variables such as inter-pixel distance and angle, spectral and
quantization level of the picture, moving window size, and statistics utilized as texture measurements. Fig. 4 shows a detailed schematic
representation of machine vision and image processing fusion for comprehensive detection of plant growth abiotic stress symptoms.
Feature selection is another important part of image processing is essential for improving computational efficiency, reducing noise,
enhancing interpretability, and ensuring the effectiveness and generalization of image processing algorithms. Sun et el. (2019) used
sequential feature selection algorithm was employed to select chlorophyll features that could represent the responses of three genotypes
caused by drought stress for each day post drought stress affected plants.

Fig. 4. A detailed schematic representation of machine vision and image processing fusion for
comprehensive detection of plant growth abiotic stress symptoms.

Precision Agriculture Science and Technology 6(1) March 2024 41

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Classification in plant stress detection involves building models that can automatically categorize plants into different stress classes
based on various features extracted from images or sensor data. The goal is to develop robust and accurate algorithms that can assist in
identifying and managing plant stress conditions. Naik et al. (2017) utilized several classification algorithms, namely classification trees
(CT), random forests (RF), Naïve Bayes (NB), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), multi-class
support vector machines (SVM), k-nearest neighbors (KNN), and Gaussian mixture models (GMM). Building on the findings, they
applied the concept of hierarchical classification to create two more models using a combination of linear discriminant analysis (LDA)
and SVM algorithms.

Artificial intelligence application for plant stress symptom

detection
The technologies of artificial intelligence (AI), machine learning (ML), and deep learning (DL) have had a significant impact on the
topic of plant phenotyping. AI was first defined as the science and engineering of making intelligent machines, especially intelligent
computer programs (Buchanan et al., 2005). In recent years, the fusion of AI, ML, and DL technologies has revolutionized plant stress
symptom detection. Machine learning algorithms, a subset of AI, have been important in plant stress symptom detection. By analysing
vast datasets of plant images, ML models can identify precise indicators of stress, such as leaf discoloration or wilting, with remarkable
accuracy. These models learn patterns and features from labelled data, allowing them to classify plants based on their health status and
detect deviations from the norm that may signify stress.
Deep learning, a sophisticated branch of ML, has further enhanced the capabilities of plant stress detection systems. Deep neural
networks, inspired by the structure of the human brain, excel at processing complex data and extracting intricate features. In plant
phenotyping, convolutional neural networks (CNNs) have emerged as a powerful tool for image analysis. By traversing multiple
layers of abstraction, CNNs can discern nuanced variations in plant physiology, enabling precise identification of stress symptoms
even amidst diverse environmental conditions. Fig. 5 shows a schematic representation of a deep learning framework for plant stress
classification through image analysis and feature extraction.

Fig. 5. A schematic representation of a deep learning framework for plant stress classification through image
analysis and feature extraction.

Precision Agriculture Science and Technology 6(1) March 2024 42

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Recent developments in big data have accelerated the development of AI by combining vast data sets with intelligent algorithms,
enabling AI software to automatically learn from the patterns or features in the vast datasets. A form of artificial intelligence is machine
learning is used to carry out particular task by computer systems. It is typically divided into supervised and unsupervised learning
techniques. The expression of augmented learning techniques uses an input matrix with independent x and dependent y variables. When
dealing with classification problems, y is often a scalar to represent the category labels and a vector of continuous values when using
regression (Harrington, 2012). According to the current status of the research, there are not many datasets that support fine-grained
instance segmentation that are available for autonomous crop disease and pest detection. Fig. 6 shows the application of real-time
monitoring and detection of seedling stress symptoms using sensor fusion and CNN based algorithm. Afzaal et al. (2021) presented a
novel dataset consisting of 2500 photos representing seven different types of strawberry diseases. In order to enable the development
of deep learning-based autonomous detection systems capable of segmenting strawberry diseases under intricate background settings.
Khanna et al. (2019) contributed a generic framework for remote plant stress phenotyping. In this research, a machine learning-based
methodology for methodically inferring these stress conditions from the remotely measured data was presented, along with a dataset
containing spatiotemporal-spectral data following sugar-beet crop growth under optimal, drought, low and surplus nitrogen fertilization,
and weed stress conditions. Naik et al. (2017) investigated ten different classification strategies; a hierarchical classifier with a mean
per-class accuracy of approximately 96% was found to be the best classifier. By linking the automatically extracted features of canopy
traits with the plant stress severity rating, a population canopy graph" that has phenotypical significance. They integrated the image
capture, image processing, and classification workflow into a smartphone app to allow digital photographs of the canopy to be used for
automatic real-time iron deficiency chlorosis score evaluation.

Fig. 6. Application of CNN based algorithm for seedlings stress symptoms detection at the Agricultural
Production Machinery and Precision Agriculture Lab, Chungnam National University, Republic of Korea.

Precision Agriculture Science and Technology 6(1) March 2024 43

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Another effective supervised machine learning technique that may be used to identify people in high-dimensional space is the
Support Vector Machine (SVM) (Cortes et al., 1995). Moshou et al. (2014) discriminated between healthy and water stressed wheat
canopies under greenhouse conditions. They developed stress type detection algorithms based on the combination of least-squares
support vectors machine with sensor fusion. With the classification performance of 99% the model was developed for automated
recognition of different biotic and abiotic stress. Pujari et al. (2016) detected and quantified plant disease symptoms in the initial stage
and developed algorithms for color and textures features extraction from the sample images of plant disease and used to train support
vector machine (SVM) classifiers. The findings show that SVM classifier is more suited for identifying and categorizing plant diseases
that impact crops in horticulture and agriculture. Some researches were conducted for detecting and classifying drought stress (Behmann
et al., 2014), weed and nitrogen stress (Karimi et al., 2006), cadmium stress (Zhou et al., 2019), disease and insect stress (Zhao et al.,
2022) of crops by using hyperspectral images image based SVM classification model. For eliminating water stress discrimination
method for plant root zone water status in greenhouse was developed by Guo et al. (2017) by integrating phenotyping and machine
learning classification model Random Forest (RF), Neural network (NN), Support Vector machine (SVM). For the dimensionality
reduction of a high dimensional dataset, feature selection selects discriminative features to capture most valuable information with
minimum redundancy.
Moghimi et al. (2018) used six feature selection methods to provide remarkable results to observe health status of plants. Among
them the best results were achieved by ReliefF, SVM-RFE, and Random Forest model respectively. Another form of artificial
intelligence is Deep Learning (DL) which is a statistical learning technique that extracts features from unprocessed datasets (Sarker
2021). To achieve this goal a multi-layered artificial neural network with hidden layers stacked one after another is used by DL. Due
to its complex algorithms, DL calls for strong computational resources. There are three popular DL models such as Recurrent Neural
Networks (RNN) (Concepcion II et al., 2020), Generative Adversarial Networks (GNN) (Qi et al., 2019), Convolution Neural
Networks (CNN) (Yang et al., 2019). To achieve high accuracy in tasks like object detection (Oh et al., 2020; Lu et al., 2021), language
translation (He, 2020), speech recognition (Zheng et al., 2018) , as well as in measuring plant stress and other biotic and abiotic
conditions that affect plants, deep learning (DL) uses multi-layered artificial neural networks (ANN). There are five key artificial
neuron components make up the fundamental feed forward neural network. The input nodes, connections, weighted sum, transfer or
activation function, and output node are among them. The numerical value linked with the input node is often a real number. Weights
(w) are connected to connections, which can be any real number. The values and weights from the input nodes of the connections are
entered into the weighted sum. The transfer or activation function will take the weighted sum's output as an input.
CNNs are frequently used in Deep Learning for applications involving image processing or computer vision. The most well-known
technique among the deep learning methods is the CNN model, which has demonstrated outstanding success and performance in the
fields of computer vision and pattern recognition, designed for visual recognition, image annotation and retrieval as well as plant stress
detection (Azimi et al., 2021). Convolutional neural network (CNN) and long short-term memory (LSTM) network were utilized by
Rojanarungruengporn et al. (2021) to detect the early stage of nutrient deficiency-induced plant stress. They made use of sorghum
datasets, which are made up of more than 40,000 photographs of the crop in growth that were taken in three views at the phenotyping
facility. The use of deep learning is expanding in the field of precision agriculture rapidly (Tian et al., 2020, Li et al., 2021). Wang et
al. (2021) proposed an automatic extraction algorithm for crop images based on Mask RCNN. 360 datasets of several types of fruits
were pre-processed and additionally an improved Mask RCNN network model structure was established to the network design to
enhance functions and optimize region extraction network, and feature pyramid network. Finally, by using the Sobel operator to
forecast the target edge, adding the edge loss to the loss function, and adding a micro-fully connected layer to the mask branch of the

Precision Agriculture Science and Technology 6(1) March 2024 44

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

ROI output, the edge accuracy of the segmentation mask is further improved. The proposed algorithm demonstrated better precision,
recall, average precision, mean average precision, and F1 scores of crop image extraction. Chu et al. (2021) reported the creation
of Suppression Mask R-CNN, a revolutionary deep learning-based apple detection system. In order to suppress non-apple features
produced by the original network, a suppression branch was added to the conventional Mask R-CNN in their innovative suppression
Mask R-CNN for apple detection. The developed model showed a higher F1-score of 0.905 and a detection time of 0.25 second per
frame.
Jia et al. (2020) applied a model of harvesting robot vision detector based on Mask R-CNN, which shows good performance of
feature extraction and target detection applied in deep learning to fruit identification in orchards. To make the model better suited for
identifying and segmenting apples that overlap, it was updated. A random test set including 120 photos is used to evaluate the approach;
the precision rate is 97.31%, and the recall rate is 95.70%. Additionally, the identification speed is quicker, which satisfies the vision
system requirements of the apple harvesting robot. Yu et al. (2019) introduced Mask Region Convolutional Neural Network (Mask-
RCNN) to enhance machine vision performance in fruit detection for a strawberry harvesting robot. The method put forth shows
increased universality and resilience in a non-structural context, especially for overlapping and buried fruits as well as those under
different lighting conditions. Kaneda et al. (2017) proposed a novel multi-modal sliding window-based support vector regression (multi-
modal SW-SVR) method for accurate water stress prediction from environmental and plant image data. It included SW-SVR and deep
neural network (DNN) as a feature extractor, which performed better and features for precise and stable water stress prediction.
Weng et al. (2023) proposed a drought stress (DS) identification method using hyperspectral imaging (HSI) and subsample fusion.
Reflectance spectra from HSI images of young and mature leaves at different DS levels were extracted, and effective wavelengths (EWs)
were determined using a genetic algorithm. DS identification models were constructed using EW spectra and image features from RIS
by LeNet-5, support vector machine (SVM), random forest, and dense convolutional networks (DCN). SVM and subsample fusion
achieved global optimal classification performance, with a classification accuracy of 95.90% and 95.78% for calibration and prediction
sets, respectively. Tran et al. (2019) used used artificial neural network models to recognize, classify, and predict the nutritional
deficiencies occurring in tomato plants (Solanum lycopersicum L.). A total of 571 images were captured used and the accuracy of
87.273% and 79.091% for Inception-ResNet v2 and Autoencoder, respectively, and 91% validity using Ensemble Averaging were
observed.
Khatoon et al. (2021) proposed an AI-based system for real-time detection of diseases in tomato plants, utilizing various deep
learning models trained on a large dataset of plant images. ShallowNet, a shallow network trained from scratch, and DenseNet, a state-
of-the-art deep learning network fine-tuned via transfer learning, were compared. DenseNet consistently showed high performance,
with an accuracy of 95.31% on the test dataset. Taha et al. (2022) combined color imaging with deep convolutional neural networks
(DCNNs) to diagnose nutrient status in aquaponics grown lettuce. The method involved multi-stage processes, including plant
detection and nutrient deficiency classification. Using 3,000 lettuce images categorized into four nutrient classes, DCNNs outperformed
traditional machine learning methods, achieving 99.1% accuracy for segmentation and 96.5% for classification. Ruan et al. (2023)
introduced meta-learning to hyperspectral imaging and crop detection, utilizing 88 hyperspectral images of drought-stressed tomato
plants and 68 images of freeze-stressed tomato plants. Results showed that meta-learning, with a minimum of eight target domain
samples, achieved a 69.57% detection accuracy, 59.29% precision, 66.32% recall, and 62.61% F1-score for classifying frost stress
severity. Shomali et al. (2023) conducted a study comparing the chlorophyll fluorescence (ChlF) features of tomato plants under high
light (1200 μmol m−2 s−1) with control plants (300 μmol m−2 s−1). Three artificial neural networks (ANNs) based algorithms, Boruta,
Support Vector Machine (SVM), and Recursive Feature Elimination (RFE) were used to identify the most reliable ChlF features

Precision Agriculture Science and Technology 6(1) March 2024 45

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

which demonstrated the effectiveness of these features for phenotyping plants under high light stress. Long and Ma (2022) explored
a method for identifying drought stress in tomato seedlings by using chlorophyll fluorescence imaging. They collected chlorophyll
fluorescence parameters and corresponding images for four different drought stress levels. The features with high Pearson correlations
were input into three models–Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), and k-Nearest Neighbor (KNN)
to classify drought stress classes. The recognition accuracy rates were 86.8% for LDA, 87.1% for SVM, and 76.5% for KNN. Kuo et
al., (2023) developed the 1D-ResGC-Net, a convolutional neural network with an embedded residual global context (ResGC) block, to
analyze visible and near-infrared (Vis/NIR) spectroscopy data of tomato leaves for early drought stress detection, and utilizing gradient-
weighted class activation mapping to identify the most informative feature bands, showed an accuracy of 90%. Kumar et al. (2021)
demonstrated the effectiveness of image-based phenotyping for water-stress classification and identified genomic loci for horticultural
traits, chlorophyll fluorescence, and spectral vegetation indices. They found 25 quantitative trait loci (QTLs) for horticultural traits,
26 for chlorophyll fluorescence, and 34 for VI, explaining 6.41% to 19.5%, 6.93% to 13.26%, and 7.2% to 17.19% of phenotypic
variation, respectively. Yang et al. (2021) explored the potential of machine learning models, including Multiple Linear Regression
(MLR), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM), to detect yellow and rotten leaves in hydroponic lettuce.
They used one-way analysis of variance to select color features for training these models and applied it to reduce the number of RGB,
HSV, and Lab* features in hydroponic lettuce images. MLR and SVM provided the highest detection accuracies of 89.48% and
99.29%, and 98.33% and 97.91% for yellow and rotten leaves, respectively. Concepcion II et al. (2020) proposed advanced computer
vision and machine learning techniques to predict lettuce water stress indicators using 330 samples from an aquaponic system. Color
components, canopy area, and a thermal index were the key components for an RNN optimized by evolutionary strategy (ES) model,
which showed high prediction accuracy (R2 =0.81 to 0.82) with an efficient inference time of 7.5 seconds. Yee-Rendon et al. (2021)
used four well-known pre-trained deep-learning models, namely VGG-16, Xception, Inception v3, and MobileNet v2, to classify
jalapeño pepper leaves with two diseases: Tobacco Mosaic Virus (TMV) and Pepper Huasteco Yellow Vein Virus (PHYVV). They
introduced new vegetation indices, Normalized Red-Blue Vegetation Index (NRBVI) and Normalized Green-Blue Vegetation Index
(NGBVI), and employed data augmentation to enhance model accuracy. The Xception model using the NGBVI dataset exhibited the
highest performance, achieving an average accuracy of 98.3%. Table 1 shows the key indicators and detection algorithms for plant
growth stress detection using machine learning algorithms.

Precision Agriculture Science and Technology 6(1) March 2024 46

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Table 1. Key indicators and detection algorithms for plant growth stress detection.
Crop Key indicator Stress type Sensor AI Method Reference
Tomato Stem diameter Water RGB Multi-modal support vector Kaneda et al. (2017)
regression (SW-SVR)
Tomato Fruit and leaf (color) Nutrient RGB Deep convolutional neural network Tran et al. (2019)
(CNN)
Tomato Fruit and leaf (color) Nutrient RGB CNN Khatoon
et al. (2021)
Lettuce Leaf area Nutrient RGB Deep convolutional neural network Taha et al. (2022)
(CNN)
Lettuce Leaf Water RGB Multiple linear regression (MLR), Yang et al. (2023)
KNN, SVM
Pepper Leaf Water RGB VGG-16, Xception, Inception v3, Yee -Rendon et al. (2021)
MobileNet v2
Tomato Leaf Water Hype rspectral Support vector machine (SVM), Weng et al. (2023)
random forest (RF), dense
convolutional network
Tomato Canopy Tem perature Hype rspectral SVM, PLS-DA Ruan et al. (2023)
Tomato Leaf Water Mult ispectral 1D CNN Kuo et al. (2023)
Lettuce Leaf Water Thermal Recurrent neural network (RNN) Concepcion II et al. (2020)
Tomato Leaf Light Ch lorophyll ANN, SVM, Recursive Feature Shomali et al. (2023)
flu orescence Elimination (RFE)
Tomato Leaf Water Ch lorophyll LDA, SVM, KNN Long and Ma (2022)
flu orescence
Lettuce Leaf Water Ch lorophyll RF Kumar et al. (2021)
flu orescence

Challenges and limitations of using machine vision techniques

Using machine vision for plant stress detection comes with its own set of challenges and limitations. Fig. 7 shows the summary of
the characteristics and pros and cons of AI technologies for plant stress detection. While this technology has shown promising results
in various applications, there are certain issues that need to be addressed for its successful implementation in plant stress detection. As
the morphological characteristics of plant change with the growing time period it is challenging to detect stress symptoms in very
early stage with images due to the dynamic nature of plant phenotypes. Similar physiological changes of plants can be exhibited due
to different abiotic stresses which limits the machine vision systems to differentiate between different causes of plant abnormalities.
Creating a universal model that works well across diverse plant types and growing environments is challenging. In the domain of image
based plant stress detection systems, image data quality and variability in imaging techniques can affect the performance of machine
vision algorithms. Obtaining labeled datasets that cover a wide range of stress conditions can be challenging.
Additionally, data collection may be hindered by factors such as weather conditions or restricted access to certain plant locations.
Plant stress can manifest in subtle and complex ways, such as changes in color, texture, or morphology. Developing algorithms that
can accurately interpret these diverse symptoms is a complex task. Deploying machine vision systems on a large scale, especially in
extensive agricultural fields, can be logistically challenging and may require significant investment.

Precision Agriculture Science and Technology 6(1) March 2024 47

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Fig. 7. A schematic representation of the characteristics and pros and cons of AI technologies for plant stress
detection.

Future of machine vision in agriculture

The future of machine vision in plant stress detection holds several exciting possibilities, driven by advancements in technology,
research, and interdisciplinary collaboration. Continued development of high-resolution imaging technologies, including hyperspectral
and multispectral imaging, will provide more detailed information about plant characteristics. These technologies can capture a broader
range of wavelengths, enabling better discrimination of stress-related changes. Machine vision systems are likely to be integrated into
larger sensor networks. Combining data from various sensors, such as environmental sensors, drones, and ground-based sensors, will
enhance the accuracy of stress detection by providing a more comprehensive view of plant and environmental conditions. Further
advancements in deep learning techniques and neural network architectures will contribute to improved image recognition and feature
extraction. These advancements can enhance the ability of machine vision systems to accurately detect subtle and complex stress
symptoms.
The development of real-time decision support systems will enable quick responses to detected stress conditions. This is particularly
important in precision agriculture, where timely interventions can mitigate the impact of stress and optimize crop yields. Automation
in agriculture, including autonomous drones and robotic systems equipped with machine vision, will become more prevalent. These
systems can efficiently monitor large agricultural areas, identify stress conditions, and take appropriate actions without human
intervention. Collaboration between machine vision systems and human experts will continue to be crucial. Interactive systems that
allow plant scientists and farmers to provide feedback and guidance to machine learning models will contribute to more accurate and
interpretable results.

Precision Agriculture Science and Technology 6(1) March 2024 48

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Conclusions
Plant stress detection represents a critical aspect of modern agriculture, where early identification of stress factors can significantly
impact crop yield, quality, and overall plant health. The integration of image processing techniques for plant abiotic stress symptoms
detection represents a significant progress towards addressing the pressing challenges in modern agriculture. The advancements in
computer vision and machine learning have enabled the development of robust algorithms capable of extracting valuable information
from high-resolution images, facilitating the accurate identification of stress indicators. Despite the natural challenges associated with
the variability in plant responses to abiotic stressors, ongoing research in this domain holds promise for enhancing the precision and
adaptability of detection models across diverse plant species and environmental conditions. Furthermore, bridging the realms of plant
biology, computer vision, and data science, underscores the importance of collaborative efforts in addressing complex challenges
in agriculture. Moving forward, the continuous refinement of image processing techniques, coupled with advancements in sensor
technologies and artificial intelligence, holds the potential to revolutionize crop management practices, contributing to global food
security and the sustainable cultivation of crops in the face of changing environmental conditions.

Acknowledgement
This work was supported by the Korea Institute of Planning and Evaluation for Technology in Food, Agriculture and Forestry (IPET),
through Smart Farm Innovation Technology Development Program, funded by Ministry of Agriculture, Food and Rural Affairs
(MAFRA) (Project No. 421035-04), Republic of Korea.

References
Abdullah MA, Ibrahim HA, Elsawy AM. 2018. Impact of thermal performance of infrared lighting and water
stress on growth and quality of cucumber. Sciences 8(04): 1311-1324.

Afzaal U, Bhattarai B, Pandeya YR, Lee J. 2021. An instance segmentation model for strawberry diseases
based on mask R-CNN. Sensors 21(19): 6565.

Azimi S, Kaur T, Gandhi TK. 2021. A deep learning approach to measure stress level in plants due to
Nitrogen deficiency. Measurement 173: 108650.

Baker NR. 2008. Chlorophyll fluorescence: a probe of photosynthesis in vivo. Annu. Rev. Plant Biol. 59: 89-
113.

Balaska V, Adamidou Z, Vryzas Z, Gasteratos A. 2023. Sustainable crop protection via robotics and artificial
intelligence solutions. Machines 11(8): 774.

Basak JK, Qasim W, Okyere FG, Khan F, Lee YJ, Park J, Kim HT. 2019. Regression analysis to estimate
morphology parameters of pepper plant in a controlled greenhouse system. Journal of Biosystems
Engineering 44: 57-68.

Bechar I, Moisan S. 2010. On-line counting of pests in a greenhouse using computer vision. In VAIB
2010-Visual Observation and Analysis of Animal and Insect Behavior.

Precision Agriculture Science and Technology 6(1) March 2024 49

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Bechtold U, Field B. 2018. Molecular mechanisms controlling plant growth during abiotic stress. Journal
of Experimental Botany 69(11): 2753-2758.

Behmann J, Steinrücken J, Plümer L. 2014. Detection of early plant stress responses in hyperspectral
images. ISPRS Journal of Photogrammetry and Remote Sensing, 93, 98-111.

Buchanan BG. 2005. A (very) brief history of artificial intelligence. AI Magazine 26(4): 53-53.

Cabrera‐Bosquet L, Crossa J, von Zitzewitz J, Serret MD, Luis AJ. 2012. High‐throughput phenotyping
and genomic selection: The frontiers of crop breeding converge F. Journal of integrative plant biology
54(5): 312-320.

Cao Y, Zhang Z, Zhang T, You Z, Geng J, Wang Y, Hu T, Yang P. 2018. Overexpression of a zeaxanthin
epoxidase gene from Medicago sativa enhances the tolerance to low light in transgenic tobacco. Acta
Biochimica Polonica 65(3): 431-435.

Chaudhury A, Ward C, Talasaz A, Ivanov AG, Brophy M, Grodzinski B, Huner NP, Barron JL. 2018. Machine
vision system for 3D plant phenotyping. IEEE/ACM transactions on computational biology and
bioinformatics 16(6): 2009-2022.

Chowdhury M, Gulandaz MA, Islam S, Reza MN, Ali M, Islam MN, Park SU, Chung SO. 2023. Lighting
conditions affect the growth and glucosinolate contents of Chinese kale leaves grown in an aeroponic
plant factory. Horticulture, Environment, and Biotechnology 64(1): 97-113.

Chen L, Lin L, Cai G, Sun Y, Huang T, Wang K, Deng J. 2014. Identification of nitrogen, phosphorus, and
potassium deficiencies in rice based on static scanning technology and hierarchical identification
method. PloS one 9(11): e113200.

Chu P, Li Z, Lammers K, Lu R, Liu X. 2021. Deep learning-based apple detection using a suppression mask
R-CNN. Pattern Recognition Letters 147: 206-211.

Concepcion II R, Lauguico S, Almero VJ, Dadios E, Bandala A, Sybingco E. 2020. Lettuce leaf water
stress estimation based on thermo-visible signatures using recurrent neural network optimized
by evolutionary strategy. In IEEE 8th R10 Humanitarian Technology Conference (R10-HTC), Dec 1,
Kuching, Malaysiam, pp. 1-6.

Cortes C, Vapnik V. 1995. Support-vector networks. Machine learning 20: 273-297.

Cramer GR, Urano K, Delrot S, Pezzotti M, Shinozaki K. 2011. Effects of abiotic stress on plants: a systems
biology perspective. BMC plant biology 11(1): 1-14.

Czedik‐Eysenberg A, Seitner S, Güldener U, Koemeda S, Jez J, Colombini M, Djamei A. 2018. The

‘PhenoBox’, a flexible, automated, open‐source plant phenotyping solution. New Phytologist 219(2):
808-823.

Dewi T, Risma P, Oktarina Y. 2020. Fruit sorting robot based on color and size for an agricultural product
packaging system. Bulletin of Electrical Engineering and Informatics 9(4): 1438-1445.

Dhingra G, Kumar V, Joshi HD. 2019. A novel computer vision based neutrosophic approach for leaf
disease identification and classification. Measurement 135: 782-794.

Precision Agriculture Science and Technology 6(1) March 2024 50

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Ding X, Jiang Y, Zhao H, Guo D, He L, Liu F, Yu J. 2018. Electrical conductivity of nutrient solution
influenced photosynthesis, quality, and antioxidant enzyme activity of pakchoi (Brassica campestris L.
ssp. Chinensis) in a hydroponic system. PloS one 13(8): e0202090.

Egea G, Padilla-Díaz CM, Martinez-Guanter J, Fernández JE, Pérez-Ruiz M. 2017. Assessing a crop water
stress index derived from aerial thermal imaging and infrared thermometry in super-high density olive
orchards. Agricultural Water Management 187: 210-221.

Fahad S, Bajwa AA, Nazir U, Anjum SA, Farooq A, Zohaib A, Sadia S, Nasim W, Adkins S, Saud S, Ihsan
MZ.2017. Crop production under drought and heat stress: plant responses and management options.
Frontiers in plant science 2017: 1147.

Fiorani F, Schurr U. 2013. Future scenarios for plant phenotyping. Annual review of plant biology 64: 267-
291.

Francesca S, Najai S, Zhou R, Decros G, Cassan C, Delmas F, Rigano MM. 2022. Phenotyping to dissect the
biostimulant action of a protein hydrolysate in tomato plants under combined abiotic stress. Plant
Physiology and Biochemistry 179: 32-43.

Furbank RT, Tester M. 2011. Phenomics–technologies to relieve the phenotyping bottleneck. Trends in
plant science 16(12): 635-644.

Galieni A, D'Ascenzo N, Stagnari F, Pagnani G, Xie Q, Pisante M. 2021. Past and future of plant stress
detection: an overview from remote sensing to positron emission tomography. Frontiers in Plant
Science 11: 609155.

Gao Z, Luo Z, Zhang W, Lv Z, Xu Y. 2020. Deep learning application in plant stress imaging: a
review. AgriEngineering 2(3): 29.

García-Tejero IF, Costa JM, Egipto R, Durán-Zuazo VH, Lima RSN, Lopes CM, Chaves MM. 2016. Thermal
data to monitor crop-water status in irrigated Mediterranean viticulture. Agricultural Water
Management 176: 80-90.

García-Tejero IF, Rubio AE, Viñuela I, Hernández A, Gutiérrez-Gordillo S, Rodríguez-Pleguezuelo CR, Durán-
Zuazo VH. 2018. Thermal imaging at plant level to assess the crop-water status in almond trees (cv.
Guara) under deficit irrigation strategies. Agricultural water management 208: 176-186.

Ghosal S, Blystone D, Singh AK, Ganapathysubramanian B, Singh A, Sarkar S. 2018. An explainable deep
machine vision framework for plant stress phenotyping. Proceedings of the National Academy of
Sciences 115(18): 4613-4618.

Golzarian MR, Frick RA, Rajendran K, Berger B, Roy S, Tester M, Lun DS. 2011. Accurate inference of shoot
biomass from high-throughput images of cereal plants. Plant methods 7: 1-11.

Granier C, Vile D. 2014. Phenotyping and beyond: modelling the relationships between traits. Current
opinion in plant biology 18: 96-102.

Gout E, Boisson AM, Aubert S, Douce R, Bligny R. 2001. Origin of the cytoplasmic pH changes during
anaerobic stress in higher plant cells. Carbon-13 and phosphorous-31 nuclear magnetic resonance
studies. Plant Physiology 125(2): 912-925.

Precision Agriculture Science and Technology 6(1) March 2024 51

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Gozzovelli R, Franchetti B, Bekmurat M, Pirri F. 2021. Tip-burn stress detection of lettuce canopy grown
in Plant Factories. In Proceedings of the IEEE/CVF international conference on computer vision, 1259-
1268.

Gull A, Lone AA, Wani NUI. 2019. Biotic and abiotic stresses in plants. Abiotic and biotic stress in plants:
1-19.

Guo Q, Wang Y, Zhang H, Qu G, Wang T, Sun Q, Liang D. 2017. Alleviation of adverse effects of
drought stress on wheat seed germination using atmospheric dielectric barrier discharge plasma
treatment. Scientific reports 7(1): 16680.

Gupta H, Pahuja R. 2019. Estimating morphological features of plant growth using machine vision.
International Journal of Agricultural and Environmental Information Systems (IJAEIS) 10(3): 30-53.

He S. 2020. Research of a sign language translation system based on deep learning. In IEEE International
conference on artificial intelligence and advanced manufacturing (AIAM), January 9, Dublin, Ireland,
pp. 392-396.

Hao X, Jia J, Gao W, Guo X, Zhang W, Zheng L, Wang M. 2020. MFC-CNN: An automatic grading scheme for
light stress levels of lettuce (Lactuca sativa L.) leaves. Computers and Electronics in Agriculture 179:
105847.

Harrington, P. 2012. Machine learning in action. Simon and Schuster.

He M, He CQ, Ding NZ. 2018. Abiotic stresses: general defenses of land plants and chances for engineering
multistress tolerance. Frontiers in plant science 9: 1771.

Hendrawan Y, Damayanti R, Al Riza DF, Hermanto MB. 2021. Classification of water stress in cultured
Sunagoke moss using deep learning. TELKOMNIKA (Telecommunication Computing Electronics and
Control) 19(5): 1594-1604.

Hussain HA, Hussain S, Khaliq A, Ashraf U, Anjum SA, Men S, Wang L. 2018. Chilling and drought stresses
in crop plants: implications, cross talk, and potential management opportunities. Frontiers in plant
science 9: 393.

Islam S, Reza MN, Chowdhury M, Chung SO, Choi IS. 2021a. A review on effect of ambient environment
factors and monitoring technology for plant factory. Precis. Agric. Sci. Technol. 3(3): 83-98.

Islam S, Reza MN, Chowdhury M, Islam MN, Ali M, Kiraga S, Chung SO. 2021b. Image processing algorithm
to estimate ice-plant leaf area from RGB images under different light conditions. In IOP Conference
Series: Earth and Environmental Science 924(1): 012013.

Iqbal Z, Munir M, Sattar MN. 2022. Morphological, biochemical, and physiological response of butterhead
lettuce to photo-thermal environments. Horticulturae 8(6): 515.

Jia W, Tian Y, Luo R, Zhang Z, Lian J, Zheng Y. 2020. Detection and segmentation of overlapped fruits
based on optimized mask R-CNN application in apple harvesting robot. Computers and Electronics in
Agriculture 172: 105380.

Kaneda Y, Shibata S, Mineno H. 2017. Multi-modal sliding window-based support vector regression for
predicting plant water stress. Knowledge-Based Systems 134: 135-148.

Precision Agriculture Science and Technology 6(1) March 2024 52

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Karimi Y, Prasher SO, Patel RM, Kim SH. 2006. Application of support vector machine technology for weed
and nitrogen stress detection in corn. Computers and electronics in agriculture 51(1-2): 99-109.

Khan, MJ, Khan HS, Yousaf A, Khurshid K, Abbas A. 2018. Modern trends in hyperspectral image analysis: A
review. IEEE Access 6: 14118-14129.

Khanna R, Schmid L, Walter A, Nieto J, Siegwart R, Liebisch F. 2019. A spatio temporal spectral framework
for plant stress phenotyping. Plant methods 15(1): 1-18.

Khatoon S, Hasan MM, Asif A, Alshmari M, Yap Y. 2021. Image-based automatic diagnostic system for
tomato plants using deep learning. Comput. Mater. Contin. 67(1): 595-612.

Kim JW. 2010. Trend and direction for plant factory system. Journal of plant biotechnology 37(4): 442-455.

Kollist H, Zandalinas SI, Sengupta S, Nuhkat M, Kangasjarvi J, Mittler R. 2019. Rapid responses to abiotic
stress: Priming the landscape for the signal transduction network. Trends Plant Sci. 24: 25–37. doi:
10.1016/j.tplants.2018.10.003.

Kopecká R, Kameniarová M, Černý M, Brzobohatý B, Novák J. 2023. Abiotic stress in crop production.
International Journal of Molecular Sciences 24(7): 6603.

Kumar M, Gupta S, Gao XZ, Singh, A. 2019. Plant species recognition using morphological features and
adaptive boosting methodology. IEEE Access 7: 163912-163918.

Kumar P, Eriksen RL, Simko I, Mou B. 2021. Molecular mapping of water-stress responsive genomic loci
in lettuce (Lactuca spp .) using kinetics chlorophyll fluorescence, hyperspectral imaging and machine
learning. Frontiers in Genetics 12: 634554.

Kuo CE, Tu YK, Fang SL, Huang YR, Chen HW, Yao MH, Kuo BJ. 2023. Early detection of drought
stress in tomato from spectroscopic data: A novel convolutional neural network with feature
selection. Chemometrics and Intelligent Laboratory Systems: 104869.

Lakhiar IA, Jianmin G, Syed TN, Chandio FA, Buttar NA, Qureshi WA. 2018. Monitoring and control systems
in agriculture using intelligent sensor techniques: A review of the aeroponic system. Journal of
Sensors 2018: 1-18.

Lee U, Chang S, Putra GA, Kim H, Kim DH. 2018. An automated, high-throughput plant phenotyping
system using machine learning-based plant segmentation and image analysis. PloS one 13(4):
e0196615.

Legendre R, Basinger NT, van Iersel MW. 2021. Low-cost chlorophyll fluorescence imaging for stress
detection. Sensors 21(6): 2055.

Li H, Zhang C, Zhang S, Ding X, Atkinson PM. 2021. Iterative Deep Learning (IDL) for agricultural landscape
classification using fine spatial resolution remotely sensed imagery. International Journal of Applied
Earth Observation and Geoinformation 102: 102437.

Li L, Zhang Q, Huang D. 2014. A review of imaging techniques for plant phenotyping. Sensors 14(11):
20078-20111.

Li Z, Guo R, Li M, Chen Y, Li G. 2020. A review of computer vision technologies for plant phenotyping.
Computers and Electronics in Agriculture 176: 105672.

Precision Agriculture Science and Technology 6(1) March 2024 53

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Liakos KG, Busato P, Moshou D, Pearson S, Bochtis D. 2018. Machine learning in agriculture: A review.
Sensors 18(8): 2674.

Lichtenthaler HK. 1998. The stress concept in plants: an introduction. Ann N Y Acad Sci. 30: 851:187-98.
doi: 10.1111/j.1749-6632.1998.tb08993.x.

Long Y, Ma M. 2022. Recognition of drought stress state of tomato seedling based on chlorophyll
fluorescence imaging. IEEE Access 10: 48633-48642.

López-Marín J, González A, Pérez-Alfocea F, Egea-Gilabert C, Fernández JA. 2013. Grafting is an efficient

alternative to shading screens to alleviate thermal stress in greenhouse-grown sweet pepper. Scientia
Horticulturae 149: 39-46.

Lu X, Ji J, Xing Z, Miao Q. 2021. Attention and feature fusion SSD for remote sensing object detection. IEEE
Transactions on Instrumentation and Measurement 70: 1-9.

Mahlein AK. 2016. Plant disease detection by imaging sensors–parallels and specific demands for
precision agriculture and plant phenotyping Plant Dis. 100: 241–251.

Marceau DJ, Howarth PJ, Dubois JMM, Gratton DJ. 1990. Evaluation of the grey-level co-occurrence matrix
method for land-cover classification using SPOT imagery. IEEE Transactions on Geoscience and
Remote Sensing 28(4): 513-519.

Moghimi A, Yang C, Marchetto PM. 2018. Ensemble feature selection for plant phenotyping: a journey from
hyperspectral to multispectral imaging. IEEE Access 6: 56870-56884.

Moshou D, Pantazi XE, Kateris D, Gravalos I. 2014. Water stress detection based on optical multisensor
fusion with a least squares support vector machine classifier. Biosystems Engineering 117: 15-22.

Nachtigall LG, Araujo RM, Nachtigall GR. 2017. Use of images of leaves and fruits of apple trees for
automatic identification of symptoms of diseases and nutritional disorders. International Journal of
Monitoring and Surveillance Technologies Research (IJMSTR) 5(2): 1-14.

Naik HS, Zhang J, Lofquist A, Assefa T, Sarkar S, Ackerman D, Singh A, Singh AK, Ganapathysubramanian
B. 2017. A real-time phenotyping framework using machine learning for plant stress severity rating in
soybean. Plant methods 13(1): 1-12.

Neocleous D, Koukounaras A, Siomos AS, Vasilakakis M. 2014. Changes in photosynthesis, yield, and
quality of baby lettuce under salinity stress. Journal of Agricultural Science and Technology (JAST)
16(6): 1335-1343.

Numajiri Y, Yoshino K, Teramoto S, Hayashi T, Kawakatsu T, Tanabata T, Uga Y. 2021. iPOTs: Internet of
Things‐based pot system controlling optional treatment of soil water condition for plant phenotyping
under drought stress. The Plant Journal 107(5): 1569-80.

Obata T, Fernie AR. 2012. The use of metabolomics to dissect plant responses to abiotic stresses. Cellular
and Molecular Life Sciences 69: 3225-3243.

Oh S, Chang A, Ashapure A, Jung J, Dube N, Maeda M, Gonzalez D, Landivar J. 2020. Plant counting of
cotton from UAS imagery using deep learning-based object detection framework. Remote Sensing
12(18): 2981.

Precision Agriculture Science and Technology 6(1) March 2024 54

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Pineda M, Barón M, Pérez-Bueno ML. 2020. Thermal imaging for plant stress detection and phenotyping.
Remote Sensing 13(1): 68.

Pujari D, Yakkundimath R, Byadgi AS. 2016. SVM and ANN based classification of plant diseases using
feature reduction technique. International Journal of Interactive Multimedia and Artificial Intelligence
(IJIMAI) 3(7): 6-14.

Qi M, Wang Y, Qin J, Li A, Luo J, Van Gool L. 2019. StagNet: An attentive semantic RNN for group activity
and individual action recognition. IEEE Transactions on Circuits and Systems for Video Technology
30(2): 549-565.

Rafique S, Abdin MZ, Alam W. 2020. Response of combined abiotic stresses on maize (Zea mays L.) inbred
lines and interaction among various stresses. Maydica 64(3): 8.

Raza SEA, Prince G, Clarkson JP, Rajpoot NM. 2015. Automatic detection of diseased tomato plants using
thermal and stereo visible light images. PloS one 10(4): e0123262.

Ren Y. 2007. Development of transplanting robot in facility agriculture based on machine vision (Doctoral
dissertation, Dissertation, Zhejiang University).

Rojanarungruengporn K, Pumrin S. 2021. Early Stress Detection in Plant Phenotyping using CNN and
LSTM Architecture. In IEEE 9th International Electrical Engineering Congress (iEECON), May 31, Pattaya,
Thailand, pp. 389-392.

Ruan S, Cang H, Chen H, Yan T, Tan F, Zhang Y, Duan L, Xing P, Guo L, Gao P, Xu W.2023. Hyperspectral
Classification of Frost Damage Stress in Tomato Plants Based on Few-Shot Learning. Agronomy 13(9):
2348.

Saakre M, Baburao TM, Salim AP, Ffancies RM, Achuthan VP, Thomas G, Sivarajan SR. 2017. Identification
and characterization of genes responsible for drought tolerance in rice mediated by Pseudomonas
fluorescens. Rice Science 24(5): 291-298.

Sankaran S, Khot LR, Espinoza CZ, Jarolmasjed S, Sathuvalli VR, Vandemark GJ, Miklas PN, Carter AH,
Pumphrey MO, Knowles NR, Pavek MJ. 2015 Low-altitude, high-resolution aerial imaging systems for
row and field crop phenotyping: A review. European Journal of Agronomy 70: 112-23.

Sarker IH. 2021. Deep learning: a comprehensive overview on techniques, taxonomy, applications and
research directions. SN Computer Science 2(6): 420.

Schor N, Berman S, Dombrovsky A, Elad Y, Ignat T, Bechar A. 2017. Development of a robotic detection
system for greenhouse pepper plant diseases. Precision Agriculture 18: 394-409.

Seleiman MF, Al-Suhaibani N, Ali N, Akmal M, Alotaibi M, Refay Y, Dindaroglu T, Abdul-Wajid HH, Battaglia
ML. 2021. Drought stress impacts on plants and different approaches to alleviate its adverse effects.
Plants 10(2): 259.

Shimizu H, Heins RD. 1995. Computer-vision-based system for plant growth analysis. Transactions of the
ASAE 38(3): 959-964.

Shomali A, Aliniaeifard S, Bakhtiarizadeh MR, Lotfi M, Mohammadian M, Sadi MSV, Rastogi A. 2023.
Artificial neural network (ANN)-based algorithms for high light stress phenotyping of tomato
genotypes using chlorophyll fluorescence features. Plant Physiology and Biochemistry 201: 107893.

Precision Agriculture Science and Technology 6(1) March 2024 55

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Story D, Kacira M, Kubota C, Akoglu A. 2009. Morphological and textural plant feature detection
using machine vision for intelligent plant health, growth and quality monitoring. In International
Symposium on High Technology for Greenhouse Systems: GreenSys 893: 299-306.

Sun D, Zhu Y, Xu H, He Y, Cen H. 2019. Time-series chlorophyll fluorescence imaging reveals dynamic
photosynthetic fingerprints of sos mutants to drought stress. Sensors 19(12): 2649.

Susič N, Žibrat U, Širca S, Strajnar P, Razinger J, Knapič M, Vončina A, Urek G, Stare BG. 2018. Discrimination
between abiotic and biotic drought stress in tomatoes using hyperspectral imaging. Sensors and
actuators B: Chemical 273: 842-52.

Taha MF, Abdalla A, ElMasry G, Gouda M, Zhou L, Zhao N, Liang N, Niu Z, Hassanein A, Al-Rejaie S, He Y.
2022. Using deep convolutional neural network for image-based diagnosis of nutrient deficiencies in
plants grown in aquaponics. Chemosensors 10(2): 45.

Tardieu F, Tuberosa R. 2010. Dissection and modelling of abiotic stress tolerance in plants. Current opinion
in plant biology 13(2): 206-212.

Tian H, Wang T, Liu Y, Qiao X, Li Y. 2020. Computer vision technology in agricultural automation– A
review. Information Processing in Agriculture 7(1): 1-19.

Tran TT, Choi JW, Le TTH, Kim JW. 2019. A comparative study of deep CNN in forecasting and classifying
the macronutrient deficiencies on development of tomato plant. Applied Sciences 9(8): 1601.

Vadivambal R, Jayas DS. 2011. Applications of thermal imaging in agriculture and food industry–a review.
Food and bioprocess technology 4: 186-199.

Valente J, Giuffrida MV. 2019. Leaf counting from uncontrolled acquired images from greenhouse workers.
Proceedings of the Computer Vision Problems in Plant Phenotyping (CVPPP), Long Beach, CA, USA.

Virlet N, Sabermanesh K, Sadeghi-Tehran P, Hawkesford MJ. 2016. Field Scanalyzer: An automated robotic
field phenotyping platform for detailed crop monitoring. Functional Plant Biology 44(1): 143-153.

Wang A, Zhang W, Wei X. 2019. A review on weed detection using ground-based machine vision and
image processing techniques. Computers and Electronics in Agriculture 158: 226-240.

Wang D, He D. 2022. Fusion of Mask RCNN and attention mechanism for instance segmentation of apples
under complex background. Computers and Electronics in Agriculture 196: 106864.

Weng S, Ma J, Tao W, Tan Y, Pan M, Zhang Z, Huang L, Zheng L, Zhao J. 2023. Drought stress identification
of tomato plant using multi-features of hyperspectral imaging and subsample fusion. Frontiers in
Plant Science 14: 1073530.

Wu X, Feng H, Wu D, Yan S, Zhang P, Wang W, Zhang J, Ye J, Dai G, Fan Y, Li W. 2021. Using high-throughput
multiple optical phenotyping to decipher the genetic architecture of maize drought tolerance.
Genome Biology 22(1):1-26.

Wu Z, Yang R, Gao F, Wang W, Fu L, Li R. 2021. Segmentation of abnormal leaves of hydroponic lettuce

based on DeepLabV3+ for robotic sorting. Computers and Electronics in Agriculture 190:106443.

Xu Y, Yuan Y, Du N, Wang Y, Shu S, Sun J, Guo S. 2018. Proteomic analysis of heat stress resistance of
cucumber leaves when grafted onto Momordica rootstock. Horticulture Research 5: 53.

Precision Agriculture Science and Technology 6(1) March 2024 56

 Machine vision and artificial intelligence for plant growth stress detection and monitoring: A review

Yakupoğlu G. 2023. Effects of Magnetic Field and Ultrasound Applications on Endogenous Melatonin
Content and Drought Stress Tolerance of Pepper Seedlings. Horticulturae 9(6): 704.

Yang S, Huang L, Zhang X. 2019. Research and application of machine vision in monitoring the growth of
facility seedling crops. Jiangsu Agric. Sci. 47: 179-187.

Yang R, Wu Z, Fang W, Zhang H, Wang W, Fu L, Majeed Y, Li R, Cui Y. 2023. Detection of abnormal

hydroponic lettuce leaves based on image processing and machine learning. Information Processing
in Agriculture 10(1): 1-10.

Yee-Rendon A, Torres-Pacheco I, Trujillo-Lopez AS, Romero-Bringas KP, Millan-Almaraz JR. 2021. Analysis
of New RGB Vegetation Indices for PHYVV and TMV Identification in Jalapeño Pepper (Capsicum
annum ) leaves using CNNs-based model. Plants 10(10): 1977.

Yu Y, Zhang K, Yang L, Zhang D. 2019. Fruit detection for strawberry harvesting robot in non-structural
environment based on Mask-RCNN. Computers and Electronics in Agriculture 163: 104846.

Zhang J, Jia W, Yang J, Ismail AM. 2006. Role of ABA in integrating plant responses to drought and salt
stresses. Field Crops Research 97(1) 111-119.

Zhao W, Li J, Yang R, Li Y. 2017. Crop yield and water productivity responses in management zones for
variable-rate irrigation based on available soil water holding capacity. Transactions of the ASABE 60(5):
1659-1667.

Zhao X, Zhang J, Huang Y, Tian Y, Yuan L. 2022. Detection and discrimination of disease and insect stress of
tea plants using hyperspectral imaging combined with wavelet analysis. Computers and Electronics in
Agriculture 193: 106717.

Zheng L, Li Q, Ban H, Liu S. 2018. Speech emotion recognition based on convolution neural network
combined with random forest. In IEEE Chinese control and decision conference (CCDC), July 9,
Shenyang, China, pp. 4143-4147.

Zubler AV, Yoon JY. 2020. Proximal methods for plant stress detection using optical sensors and machine
learning. Biosensors 10(12): 193.

Precision Agriculture Science and Technology 6(1) March 2024 57

View publication stats