Journal of Integrative Agriculture

Integrating artificial intelligence and high-throughput phenotyping for crop

improvement
Mansoor SHEIKH1, Farooq IQRA2, Hamadani AMBREEN3, Kumar A PRAVIN2, Manzoor IKRA3,
Yong Suk CHUNG1#
1 Phenomics Laboratory, Department of Plant Resources and Environment, Jeju National University,
Jeju 63234, Republic of Korea
2
Council of Scientific & Industrial Research, Indian Institute of Integrative Medicine, Field Station
Pulwama, J&K 192301, India
3
Animal and Dairy Science, University of Wisconsin, Madison WI 530706, United States of America
4
Division of Fruit Science, Faculty of Horticulture, Sher e Kashmir University of Agricultural Sciences
and Technology of Kashmir, Srinagar 190025, India

Abstract Crop improvement is crucial for addressing the global challenges of food security and
sustainable agriculture. Recent advancements in high-throughput phenotyping technologies and
artificial intelligence (AI) have revolutionized the field, enabling rapid and accurate assessment of crop
traits on a large scale. The integration of AI and machine learning algorithms with high-throughput
phenotyping data has unlocked new opportunities for crop improvement. AI algorithms can analyze
and interpret large datasets, extracting meaningful patterns and correlations between phenotypic traits
and genetic factors. These technologies have the potential to revolutionize plant breeding programs by
providing breeders with efficient and accurate tools for trait selection, reducing the time and cost
required for variety development. However, further research and collaborations are needed to
overcome the challenges and fully unlock the power of high-throughput phenotyping and AI in crop
improvement. By leveraging AI algorithms, researchers can efficiently analyze phenotypic data,
uncover complex patterns, and establish predictive models that enable precise trait selection and crop
breeding. The aim of this review is to explore the transformative potential of integrating high-throughput
phenotyping and AI in crop improvement. The review will encompass an in-depth analysis of recent
advancements and applications, highlighting the numerous benefits and challenges associated with
high-throughput phenotyping and intelligence.
Keywords: artificial Intelligence, crop improvement, data analysis, high-throughput phenotyping,
machine learning, precision agriculture, trait selection1

1. Introduction
High-throughput phenotyping is a revolutionary approach in plant science that aims to rapidly and
accurately assess the phenotypic traits of plants on a large scale. It involves the integration of
advanced technologies, data analytics, and automation to capture detailed information about various
traits, such as growth, yield, stress tolerance, and disease resistance (Shakoor et al. 2017; Mir et al.
2019). This approach has significantly accelerated the progress in plant breeding, crop improvement,
and understanding plant-environment interactions. The traditional methods of phenotyping, which rely
on manual measurements and observations, are time-consuming, labor-intensive, and often subject to
human errors. High-throughput phenotyping addresses these challenges by leveraging cutting-edge
technologies like remote sensing, imaging, robotics, and sensor networks to collect data from large
numbers of plants simultaneously. These technologies enable researchers to capture phenotypic data
at multiple scales, ranging from whole plant to cellular and molecular levels, with unprecedented speed
and precision (Fiorani and Schurr 2013; Fahlgren et al. 2015). Plant phenotyping is a crucial aspect of
studying the interaction between plants and their environment, with practical implications in crop
management and breeding.
One of the key technologies used in high-throughput phenotyping is remote sensing, which involves
the use of aerial or satellite-based platforms to capture plant-related data, such as canopy temperature,
chlorophyll content, and water status. These remote sensing techniques provide a non-destructive and

Correspondence Yong Suk CHUNG, E-mail: [email protected]

#
 Journal of Integrative Agriculture

efficient way to monitor plant growth and health over large areas, enabling researchers to detect stress
responses, nutrient deficiencies, and disease outbreaks at an early stage. Another important tool in
high-throughput phenotyping is imaging, particularly with techniques like hyperspectral imaging and 3D
imaging (Ghimire et al. 2023; Karunathilake et al. 2023). Hyperspectral imaging allows the
measurement of a wide range of spectral signatures from plants, providing detailed information about
their biochemical and physiological properties. On the other hand, 3D imaging techniques enable the
reconstruction of plant architecture and can be used to quantify traits related to plant structure, such as
leaf area, leaf angle, and plant height. Advancements in robotics and automation have also played a
crucial role in high-throughput phenotyping. Automated systems equipped with robotic arms, conveyor
belts, and high-resolution cameras can perform tasks like plant watering, fertilization, and image
capture with minimal human intervention. These robotic platforms can handle large numbers of plants
simultaneously, enabling high-speed phenotyping and reducing the labor required for data collection
(Fig. 1) (Liu et al. 2020; Sarić et al. 2022; Xu and Li 2022; Ghimire et al. 2023). Machine Vision is also
playing a crucial role automation of production (Tian et al. 2022).
Leveraging ML algorithms in plant breeding will provide breeders with streamlined and powerful tools
to expedite the creation of novel plant varieties and enhance the overall efficiency of the breeding
workflow. These advancements are crucial for addressing the agricultural challenges brought about by
the realities of climate change (Najafabadi-Yoosefzadeh et al. 2023). Advancements in plant
phenotyping aim to meet breeding objectives by offering benefits in terms of throughput, applicability in
field conditions, and usefulness in the breeding process. The prediction of breeding values using
machine learning could lead to the practice becoming more popular due to the requirement of less
computational power as well as expertise for running deployed models (Hamadani et al. 2022). Over
the past two decades, significant progress has been made in addressing these challenges through the
development of novel sensors, automation technologies, and quantitative data analysis methods.
These advancements have expanded the scope of analysis, allowing for higher capacity and
throughput. Phenotyping can now be conducted on various plants, from model species to crops and
forests, at different scales and time scales. This has facilitated the study of plant development across
different levels, from cellular to canopy and seasonal growth. Overall, these advancements have paved
the way for a deeper understanding of plant-environment interactions and improved crop management
and breeding practices (Furbank and Tester 2011; Busby et al. 2017; Rouphael et al. 2018; Pieruschka
and Schurr 2019). High-throughput phenotyping has gained considerable attention and adoption in the
plant science community, offering great potential for accelerating crop improvement, optimizing
resource usage, and developing resilient agricultural systems. Despite its promise, there are
challenges to overcome, including protocol standardization, data integration, and analytical frameworks'
development. Nevertheless, high-throughput phenotyping is set to revolutionize plant science and
contribute to global food security amidst changing climates and increasing population demands.
However, the field continues to evolve rapidly, necessitating increased phenotyping capacity, seamless
integration of new technologies, improved access for researchers, and robust data management
systems.
AI algorithms can analyze images acquired from high-throughput imaging systems and extract
important features and patterns. This capability is particularly valuable in the field of plant phenotyping,
where AI can identify leaf shapes, measure plant growth parameters, and detect disease symptoms. In
high-throughput phenotyping, multiple types of data, such as imaging, genomic data, and environmental
data, are often involved. AI techniques can integrate these diverse datasets and reveal intricate
relationships between genotype, phenotype, and environment. By employing AI algorithms, predictive
models can be constructed to establish connections between phenotypic data and other factors like
genotypes or environmental conditions (Xu et al. 2022). These models have the potential to predict
phenotypic outcomes and provide guidance for breeding programs or experimental designs.
Additionally, AI techniques enable the exploration of extensive phenotypic databases to uncover new
associations, correlations, or causal relationships that might have been overlooked through traditional
analysis methods (Tripodi et al. 2022). Therefore, this review aims to provide a comprehensive
overview of the power of high-throughput phenotyping and AI in crop improvement. By harnessing the
synergy between these technologies, breeders, and researchers can enhance the efficiency and
precision of crop breeding programs, leading to the development of resilient, high-yielding varieties that
can address the global challenges of food security and sustainability.
 Journal of Integrative Agriculture

Fig. 1 Various phenotyping platforms utilised for in the crop improvement.

2. Understanding machine learning

ML or Machine learning is a combination of advanced modelling approaches that are capable of
drawing patterns from data which are not possible using conventional techniques. Machine learning
generally utilizes experiences or case scenarios and projects them to discover underlying patterns in
the data to explain or classify a new experience with minimal error. Thus most machine learning
techniques have the ability to generalize trends and patterns after learning from scenarios or data that
was previously available. There is a plethora of machine learning tools available each has its pros and
cons. The choice of a technique depends on the conditions at hand and the experts' discretion. Broadly,
ML techniques are capable of solving four possible problems, identification/detection, classification,
quantification/estimation, and prediction. In addition, all data is first pre-processed before it can be
input into any technique. Pre-processing essentially cleans the data and removes any noise, outliers or
unreliable data. It also makes the data more specific to the task at hand. Popular pre-processing
methods include dimension reduction, image cropping, clustering, contrast enhancement, de-noising,
segmentation etc. These ensure that the outputs obtained are reliable.

The modelling objective is generally either generative or discriminative (Hassan et al. 2022). If the
model being trained is working to distinguish two different data patterns, the model is said to be
discriminative. Sometimes the model is supposed to generate patterns that it learnt synthetically. In
that case, it is said to be generative. An illustration to explain the core difference between the two is
shown in Fig. 2. The use of the model would also depend on the task at hand e.g., if the amount of
data is very large, discriminative models would function better than generative ones in simple
classification tasks like comparing rice and wheat plants. With low training data having latent features,
generative models would be more robust as they would help in overcoming overfitting.

Fig. 2 Illustration of generative and discriminative models.

 Journal of Integrative Agriculture

2.1. Machine learning for high throughput phenotyping in plants

Machine learning (ML) algorithms aim to enhance task performance by leveraging examples or
historical data. Specifically, ML can establish effective connections between data inputs and build a
coherent knowledge framework. Within this data-centric approach, the greater the amount of data
utilized, the more effectively ML operates. For this reason, machine learning is gaining popularity for
thigh throughput phenotyping in plants (Benos et al. 2021). The adoption of high-throughput
phenotyping (HTP) has opened up novel opportunities for non-destructive field-based phenotyping.
Autonomous, semi-autonomous, or manually operated platforms equipped with single or multiple
sensors gather spatial and temporal data, leading to the accumulation of extensive data sets for
analysis and storage (Singh et al. 2016). Additionally, the integration of Unmanned Aerial Vehicles
(UAVs) and other self-operating machinesonfarmsholds significant promise for enhancing the precision
(Hamadani and Khan 2015) and effectiveness of crop monitoring processes (Teshome et al. 2023).

Also, with the rapid expansion of computing capabilities and the accessibility of cloud computing
resources, automated machine learning (AutoML) has gained significant traction in both industry and
academia. AutoML has emerged as an appealing substitute for traditional manual ML practices. It
holds the potential to provide high-performance, comprehensive ML pipelines encompassing various
stages such as data preparation (cleaning and preprocessing), feature engineering (extraction,
selection, and construction), model generation (selection and hyperparameter tuning), and model
evaluation. Remarkably, AutoML aims to achieve all these steps with minimal user effort or intervention
(Koh et al. 2021). With the increasing capabilities of the machine to process and analyse images,
computer vision is also leading to high throughput phenotyping (Tsaftaris et al. 2016).

A large number of ML algorithms have been used successfully in plants for disease diagnosis, yield
prediction etc. To decipher the intricate link between genotype and phenotype, the fields of systems
and network biology have found applications. Network biology, as previously elaborated, has shed light
on the intricacies of molecular events within biological systems. Concurrently, machine learning has
played a pivotal role in predicting novel components within plants and pathogens. A promising hybrid
approach involves the amalgamation of deep learning and network centrality features extracted from
multidimensional -omics data. This innovative fusion holds the potential to unveil unexplored domains
within the realm ofplant–microbe interactions (Mishra et al. 2019). Image processing to the disease
identification problem, thermal and stereo visible light, remote sensing, Kinect RGB depth images,
visible and thermal images, hyperspectral images, fluorescence imaging spectroscopy, UAV-based
RGB images and multispectral images etc have successfully been used as data sources.

Agricultural and allied sectors have great potential which can be realized by aligning our data
analysis tools with the 4th industrial revolution. There are several popular Machine Learning
algorithms which have been used in plants and animals for high-throughput phenotyping (Hamadani
and Ganai 2022). These include algorithms like SVM, SAM, Gaussian processes classifier (GPC),
Bayes factor, DAR, OBIA-based classification, KNN, quadratic discriminant analysis (QDA), linear
discriminant analysis (LDA), Naïve Bayes (NB), simple logistic (SL), LibSVM (SVM), LibLINEAR (LINE),
MLP (BNN), functional trees (FT), random forests (RF) for plant species identification and classification
in tomato, sugarbeet, apple, spinach, barley etc. (Singh et al. 2016; Ataş et al. 2022; Hassan et al. 2022)
for multiple traits and diseases. Recent advances in computation have made it possible to use ML for
all four stages of ICQP. Today it is possible to use hybrid data that includes phenotypic and genotypic
data like QTL9 (Hassan et al. 2022), and GWAS as well (Cobb et al. 2013).

ML algorithms are capable of handling time-series phenotypic data with time-series gene expression
to get a steady insight into the complex molecular mechanisms of disease resistance and
spatiotemporal difference in the expression of stress.

Although the algorithms used complex computations at the back end, plant scientists can still take
advantage of them due to the availability of tools packaged in the graphical user interfaces (GUI). They
thus do not have to worry about the underlying mathematical and computational complexities. A major
contributor to the seamless integration of ML with conventional phenotyping techniques is the data
collection and curation pipeline. Such ecosystems would help in solving the basic problems in
agriculture and ensure food security.

2.2 AI and ML future in phenotyping

 Journal of Integrative Agriculture

The future of AI for high throughput phenotyping in plants holds tremendous potential for advancements
in plant breeding, crop management, and agricultural research. Deep learning techniques, such as
deep neural networks, are continuously evolving and have the potential to further enhance the accuracy
and efficiency of plant phenotyping. These advanced models can provide more accurate predictions
and a better understanding of complex plant traits. Advanced fusion techniques, such as deep
multimodal learning and graph-based models, will enable researchers to uncover hidden patterns and
relationships between different data sources. Real-time phenotyping allows continuous monitoring of
plant traits throughout the growth cycle, providing valuable insights into dynamic responses to
environmental conditions it will enable rapid decision-making and adaptive management strategies in
agriculture, optimizing resource allocation, and improving crop yield and quality. Machine learning
techniques will facilitate the discovery of genotype-phenotype associations and help develop predictive
models for trait selection and breeding strategies. Advanced algorithms, such as causal inference
methods and Bayesian networks, will aid in unravelling the complex genetic architecture underlying
plant traits. Machine learning will drive the development of automated phenotyping platforms that can
efficiently collect and analyze plant phenotypic data at large scales. These platforms will combine
robotics, sensors, and imaging technologies with advanced machine-learning algorithms to enable high-
throughput phenotyping in controlled environments or the field. Such platforms will revolutionize the
breeding process, accelerate trait selection, and enhance crop improvement efforts (Fig. 3).

Fig. 3 Workflow illustrating the synergy between high-throughput techniques, AI, machine learning, and deep learning in
the context of precise plant phenotyping in a comprehensive and dynamic workflow. It begins with the acquisition of
diverse plant images using high-throughput systems, followed by preprocessing, advanced segmentation, and the
training of machine learning models, particularly deep learning architectures. These models then exhibit the capacity to
predict intricate phenotypic traits accurately, ultimately leading to a nuanced and precise quantification of various plant
characteristics.

3. Artificial Intelligence in HTP and its advantages in crop breeding

High throughput phenotyping (HTP) methods with their ability to monitor and measure multiple
phenotypic traits related to aspects of plant growth yield as well as their adaptability to stress is a major
scientific stride. HTP unlocks new prospects for phenotypic with the many advantages it offers. It is
non-destructive and collects massive amounts of data through sensors both spatially and temporally
(Pabuayon et al. 2019).

The utilization of automated systems for data collection is leading to the creation of extensive
volumes of data and images, destined for storage, processing, and analysis purposes. This
phenomenon qualifies as big data due to its substantial volume, diverse range, and high speed of
 Journal of Integrative Agriculture

generation. Consequently, drawing inferences or interpreting this data becomes a challenging task
(Younas 2019). In response to this situation, artificial intelligence (AI) is increasingly garnering attention
within the field. AI’s capacity to provide rapid, effective, and enhanced data analytics is making
significant strides in the realm of biology. Leveraging machine learning (ML) tools has been shown by
many researchers to be advantageous (Hassan et al. 2022) since such toolsenable data assimilation,
feature identification, prediction etc. ML is capable of handling all four stages of plant phenotyping viz.
identification, classification, quantification, and prediction (ICQP). Some notable studies in Artificial
Intelligence in high-throughput phenotyping (HTP) for crop breeding are detailed in Table 1.

ML also handles multidimensional data obtained from diverse sources as well as complex
mathematical models. This is especially useful for plant phenotyping for stress and yield which are
challenging to model efficiently and holistically given the many factors involved viz. genetic, economic,
agronomic, meteorological, and even human inputs.HTP platforms have been utilized across various
crops such as cotton (Gossypium hirsutum L.), triticale (× TriticosecaleWittmack L.), and maize (Zea
mays L.). Recent strides have been made in plant imaging sensors, encompassing a spectrum from
remote sensing, including techniques like spectroradiometry and light detection and ranging (LIDAR), to
technologies such as hyperspectral, thermal, fluorescence, and 3D laser scanning. These innovations,
combined with advancements in autonomous vehicle technology, have paved the way for high-
throughput stress phenotyping (HTSP) approaches (Hassan et al. 2022). For example, many variants
of Artificial Neural Networks have been for predicting the effects ofmorphological responses of
melatonin on citrus due to drought stress (Jafari and Shahsavar 2020).
Table 1 Notable studies in the area of Artificial Intelligence in high-throughput phenotyping (HTP)

Focus Methodology Key Findings

Plant Phenotyping with Deep Image analysis+CNN Accurate leaf segmentation and trait
Learning quantification. (Minervini et al. 2015)
Automated High-Throughput Hyperspectral imaging+Machine Rapid assessment of crop
Phenotyping Learning performance (Montes 2007).
Disease Detection using AI UAV imagery+Deep Learning Early identification of disease in wheat
(Singh et al. 2010)

Genomic Selection in Plant Genomic data+Machine Learning Enhanced prediction of breeding

Breeding outcomes (Crossa et al. 2017).

Optimizing Crop Breeding with AI Bayesian optimization+Genomic Improved selection of high-yielding

prediction genotypes (Merrick et al. 2022).

Weed Detection Using Computer UAV imagery+Convolutional Neural Efficient identification of weed
Vision Networks presence (Haq 2022).

High-throughput field phenotyping, substantiated by AI technology, has been beneficial for precision
agriculture and plant breeding. Recent developments in data analysis, sensors, and the advancement
of robots have enabled quick, efficient, non-destructive, non-invasive, measurable, repeatable, and
objective phenotyping (Chawade et al. 2019). The study of identifying crop phenotypes on a wide scale,
at cheap cost with greater precision than human vision is made possible by computer vision, while
machine learning helps researchers to identify significant trends in plant data. Combining, contrasting,
and visualizing huge datasets from various sources requires big data management and IoT. Effective
data management systems play a vital role in facilitating data exchange and knowledge sharing across
different installations, locations, and experiments. Developing reliable systems for data storage,
sharing, and analysis is crucial for harnessing the vast amount of phenotypic data generated by
different phenotyping facilities and experiments. This entails addressing issues of data standardization,
interoperability, and privacy while promoting data-driven discoveries and advancements in the field
(Arend et al. 2022; Morisse et al. 2022; Nagasubramanian et al. 2022). These developments in
envirotyping data and AI technologies open up new possibilities for breeding in the future (van Dijk et al.
2021; Karunathilake et al. 2023). Various aspects of the use of AI in plant breeding are represented in
Fig. 4. Through the analysis of vast volumes of complicated genetic data, machine learning h`as
 Journal of Integrative Agriculture

completely changed plant breeding. Big data and machine learning in plant breeding have the potential
to revolutionise the sector and increase food security. In the past five years, a number of machine
learning techniques have been proposed, highlighting both their advantages and disadvantages
(Veeragandham and Santhi 2020; van Dijk et al. 2021; Varshney 2021). For instance, the productivity
of crops like oil palm has grown as a result of machine learning approaches (Latif et al. 2021). The
classification and regression algorithm methods used for agricultural enhancement produce crop type
recommendations and yield projections, respectively (Sundari et al. 2022). Decision tree regression
(DTR) and random forest regression (RFR) can accurately predict wheat productivity using datasets for
training and testing, with evaluation metrics such as R2, RMSE, information criterion (AIC) with weights
(AICW), evidence ratio (E.R), and decompositions of prediction error (Islam and Shehzad 2022).
Classifiers such as k-Nearest Neighbor (kNN), Naive Bayes (NB), Decision Tree (DT), Support Vector
Machines (SVM), Random Forests (RF), and Bagging are used to recommend the most suited
cultivable crop(s) for a particular piece of land, depending on soil and environmental variables
(Ganesan et al. 2022). A machine learning technique called ensemble learning combines several basic
models to improve prediction accuracy (Najafabadi-Yoosefzadeh et al. 2023). These algorithms are
employed in a variety of agricultural applications, such as crop recommendation and yield forecasting,
determining whether a location is suitable for agricultural cultivation.

In order to recognize changes in plants, computer vision algorithms are employed in plant
phenotyping to extract data from photos and videos. Machine learning, plant recognition, and computer
vision-based phenotyping are a few of the methods (Šulc and Matas 2017; Mochida et al. 2019). The
taxonomic hierarchical classification is frequently related to the fine-grained classification challenge of
plant recognition, which has substantial intra-class variability and frequently minor inter-class variations.
High-throughput plant phenotyping has benefited from recent developments in image analysis enabled
by machine learning-based techniques, such as convolutional neural network-based modeling (Mochida
et al. 2019). TasselNet, a computer vision-based system, counts maize tassels, while CycleGAN, an
algorithm for generating and transforming unpaired pictures through generative learning, was used to
partition tomato developing trusses (Jung et al. 2022). Deep learning has made challenging
phenotyping tasks achievable, and an image processing pipeline for high-throughput plant phenotyping
has been developed. Big data management is crucial for crop improvement and plant breeding since it
makes it possible to change fundamental plant traits. Plant breeders now rely on high-performance
computing, bioinformatics tools, and machine-learning techniques to examine the massive volumes of
genomic data produced by high-throughput omics technologies (Najafabadi-Yoosefzadeh et al. 2023).
The field could undergo a revolution thanks to this strategy, which would also improve food security.
Big data analysis and machine learning approaches have been found in studies to be effective at
solving agricultural problems (Parmley et al. 2019; Najafabadi-Yoosefzadeh et al. 2023). Envirotypic
data, in general, is a crucial part of big data in plant breeding because it can aid in predicting how
different plant varieties will behave under various environmental circumstances and direct the creation
of new varieties that are better adapted to certain environmental difficulties (Taranto et al. 2018). It is
possible to overcome the calibration difficulties when using near-earth hyperspectral data for agriculture.
For instance, a data-driven, fully automated calibration workflow with a suite of reliable algorithms for
radiometric calibration, bidirectional reflectance distribution function (BRDF) correction and
normalisation, soil and shadow masking, and image quality evaluations was developed (Sagan et al.
2021). By delivering real-time environmental data that affect crop yield and productivity, IoT technology
has the potential to enhance plant breeding and crop production. IoT, sensor networks, and data
analysis can be used to construct a smart agricultural system. The approach emphasizes crop
selection, nutrient content and humidity maintenance, effective fertilizer application, and quality control.
Agricultural IoT applications can provide cutting-edge solutions including remote monitoring, decision
support, automatic irrigation, frost protection, and fertilization (Mentsiev and Amirova 2020; Gowda et al.
2021; Sagan et al. 2021). In order to hasten the development of climate-resilient crops, boost pest
control, and improve plant phenotyping, artificial intelligence (AI) is being employed progressively in
plant breeding. For instance, speed breeding combined with AI is being utilised to generate crops that
are climate-smart (Rai 2022). Modernized plant breeding operations have the potential to benefit
greatly from AI since it offers a mechanism to quickly assess massive volumes of complicated data
produced by high-throughput omics technologies. High-throughput phenotyping and gene functional
analysis problems can be solved using AI, which can also be used to predict or even explain
phenotypes from the underlying genotypes in various contexts. However, in order to fully realize the
promise of AI in crop and agronomic research, data management issues must be resolved.
Applications of AI in plant breeding have the potential to enhance the effectiveness and precision of
breeding programmes as well as assist in addressing the issues of food security and climate change.
 Journal of Integrative Agriculture

The development of novel plant varieties with improved yield performance and increased resistance to
pests, diseases, and climate change can also be aided by AI, which will help small-scale farmers by
giving them access to more productive and resilient crops. In conclusion, by enhancing the
effectiveness and precision of breeding programmes and creating crops that are better adapted to local
conditions, the application of AI in plant breeding has the potential to strengthen food security and
enhance the livelihoods of small-scale farmers.
Plant high-throughput genotyping has the potential to quickly provide genetic data and to foster the
growth of large mapping populations and a variety of phenotyping lines (McMullen et al. 2009). For
enhancing photosynthesis, it can be useful to assess new crop genotypes and identify crucial genes
(Parry et al. 2011). Previous research has demonstrated that the interaction between a crop's genetic
make-up and environmental factors results in its phenotype (Yang et al. 2017). Traditional phenotyping
techniques are detrimental, time-consuming, inefficient and expensive. Plant high-throughput
phenotyping traits, such as yield, disease resistance, drought resistance, and salt resistance, have
been studied using new techniques like machine vision, imaging spectroscopy, and thermal infrared
imaging (Li et al. 2014). In crop breeding and phenotyping, it's critical to collect data in extremely high
resolution in order to assess and choose the cultivars that perform the best. Plant breeding greatly
benefits from quick methods to assess crop response. High-throughput remote sensing could offer a
great way to quickly identify crop response. This means that the genotype to phenotype relationship
could be successfully established using a high-throughput phenotyping platform. Utilizing handheld and
ground-mounted sensors for plant phenotyping has garnered growing research interest in recent years.
However, ground vehicles are typically constrained by crop types and locations, which can also destroy
the crops and fields, and handheld-based phenotyping does not have a high throughput. Because of
their inefficiency and time requirements, these methods cannot be used widely. Imaging from satellites
is dependent on the weather. Additionally, satellites’ relatively low image resolution prevents them from
being used in situations where high resolution is required (Jannoura et al. 2005).
UAV has been utilised in agriculture and general plant sciences over the past few years. To
achieve high-throughput phenotyping, more and more researchers are fusing airborne remote sensing
methods with plant science. The use of unmanned aerial systems (UAS) has introduced a fresh
method for crop management and observation. The UAS-based phenotyping is a good replacement for
the earlier techniques, which could get around the drawbacks.
High-throughput phenotyping has sped up plant breeding efforts by allowing for the rapid screening
of numerous plants at various phenological stages. As a result, it is unnecessary to wait for plant
maturity in the field and can quickly screen for desired traits at the beginning of the process. It can be
applied in controlled and natural settings in both the lab and the field. Plant performance can be quickly
evaluated in the field, which encourages a thorough evaluation of the entire life cycle using less
destructive methods. In addition, data recording is enhanced and environment-controlled high-
throughput facilities require fewer replications. Phenotypic data are unquestionably the best predictors
of biological outcomes, such as plant health, disease-related traits, and mortality, in the majority of
cases. They are also particularly helpful for obtaining precise measurements of plant characteristics,
enhancing the abilities of plant breeders and agronomists (Mir et al. 2019; Rebetzke et al. 2019; Zhao
et al. 2019; Ku et al. 2023).

3.1. Crop Modeling and Simulation interconnected with HTP

AI algorithms can be used to develop crop models that simulate plant growth and predict crop
performance under different environmental conditions. These models can aid in optimizing cultivation
practices, predicting yield outcomes, and adapting cropping systems to changing climatic conditions
(Hassan et al. 2022). High-throughput phenotyping and crop modeling are closely linked. The data
obtained from high-throughput phenotyping, such as plant trait measurements, can be used as inputs
for crop models to improve their accuracy and predictive capabilities. In turn, crop models can provide
a framework for interpreting and analyzing the large volumes of data generated by high-throughput
phenotyping, allowing for a better understanding of crop performance and optimizing agricultural
practices. By combining high-throughput phenotyping and crop modeling, researchers and farmers can
gain deeper insights into crop physiology, identify key traits for breeding programs, optimize resource
allocation, enhance crop productivity, and develop sustainable and resilient agricultural systems
(Bustos-Korts et al. 2019; York 2019; Gill et al. 2022).
3.2. Phenotypic platforms
 Journal of Integrative Agriculture

Numerous high-throughput phenotyping (HTP) platforms are currently in use to assess diverse biotic
and abiotic stress-related characteristics across various crops. These platforms are pivotal in
advancing our understanding of plant responses to stressors. A selection of such platforms is outlined
below:
i. “PHENOPSIS”: An automated platform developed for analyzing plant responses to soil water
stress in Arabidopsis.
ii. “GROWSCREEN FLUORO": Designed to phenotype leaf growth and chlorophyll fluorescence,
enabling the identification of abiotic stress tolerance in Arabidopsis.
iii. "LemnaTec 3D Scanalyzer System": Employed for non-invasive screening of salinity tolerance
traits in rice, facilitating comprehensive assessments.
iv. "HyperART": Enables non-destructive quantification of leaf traits, including chlorophyll content
and disease severity, across various crop
v. "PhenoBox": Utilized for identifying diseases such as head smut and corn smut in Brachypodium
and maize, respectively, along with assessing salt stress response in tobacco.
vi. "PHENOVI-SION": Developed for detecting drought stress and recovery patterns in maize plants,
contributing to stress response analyses.
vii. "PhénoField": Geared towards characterizing various abiotic stresses in wheat, advancing
insights into stress-related plant traits.
viii. "PlantScreen™ Robotic XYZ System": Used to analyze diverse traits associated with drought
tolerance in rice, enhancing our understanding of resilience mechanisms.These platforms
collectively contribute to the systematic assessment of stress-associated traits in a range of crops,
fostering insights into plant responses and stress tolerance mechanisms (Gill et al. 2022).

Fig. 4 Different aspects of AI for utilization in high throughput phenotyping inplant breeding.

4. Advances in imaging technologies and AI for crop improvement

Advanced imaging technologies, high-throughput phenotyping, and AI have emerged as powerful tools
for crop improvement. They enable researchers and breeders to rapidly and accurately assess the
traits and performance of large populations of plants, leading to more efficient and targeted breeding
strategies. High-resolution imaging technologies such as hyperspectral imaging, thermal imaging, 3D
imaging, and fluorescence imaging allow researchers to capture detailed information about various
plant characteristics, including leaf morphology, canopy structure, physiological processes, and stress
responses (Fahlgren et al. 2015; Ampatzidis and Partel 2019; Nabwire et al. 2021). Advanced imaging
technologies, such as hyperspectral imaging, thermal imaging, LiDAR (Light Detection and Ranging),
and drones equipped with high-resolution cameras, provide detailed and real-time data on crops.
 Journal of Integrative Agriculture

These technologies allow researchers and farmers to monitor crop growth, detect stress factors, and
identify diseases or pests at an early stage (Ampatzidis and Partel 2019; Hassan et al. 2022; Nabwire
et al. 2021. By capturing information about plant health, nutrient status, water availability, and other
vital parameters, imaging technologies enable precise and targeted interventions for improved crop
management. For example, hyperspectral imaging can analyze the reflected light spectrum from plants
and identify specific biochemical markers related to stress or disease . This enables farmers to address
these issues promptly and prevent yield losses. Thermal imaging helps detect variations in plant
temperature, indicating stress caused by factors such as water shortage or disease. LiDAR technology
assists in creating 3D models of crop canopies, which can aid in estimating biomass, canopy structure,
and crop density (Guo et al. 2018; Kim et al. 2023). Drones equipped with cameras or other sensors
can capture high-resolution images of vast agricultural areas, providing comprehensive information for
monitoring and decision-making.
High-throughput phenotyping involves automated and non-destructive measurement of plant traits on
a large scale. It enables researchers to collect phenotypic data from thousands of plants quickly and
accurately. This data can include traits related to growth, yield, disease resistance, nutrient status, and
stress tolerance. High-throughput phenotyping involves capturing large amounts of data on plant traits,
such as leaf area, plant height, biomass, flowering time, and disease resistance, on a large scale and in
a time-efficient manner (Kim 2020; Yang et al. 2020). Traditional manual phenotyping methods are
labor-intensive, time-consuming, and may not capture accurate data due to human error or subjectivity.
High-throughput phenotyping addresses these challenges by employing automated systems, robotic
platforms, and non-destructive imaging techniques. By using imaging technologies and machine vision
algorithms, high-throughput phenotyping platforms can rapidly measure and analyze multiple traits
across thousands of plants. This enables researchers and breeders to conduct large-scale screening
experiments, assess the performance of diverse germplasm, identify desirable traits, and accelerate the
breeding process. It also facilitates the collection of data in diverse environmental conditions, helping to
understand genotype-environment interactions and develop more resilient and adaptable crop varieties
(Parry et al. 2011; Yang et al. 2017; Xu and Li 2022). AI plays a crucial role in analyzing the vast
amount of data generated by advanced imaging technologies and high-throughput phenotyping. AI
algorithms, including machine learning and deep learning techniques, can process and interpret
complex datasets, identify patterns, and make predictions or recommendations (Yang et al. 2021;
Hassan et al. 2022). This enables AI to assist in various aspects of crop improvement

4.1. High-throughput phenotyping and AI in Disease Detection

Using the integrated phenotypic data and AI algorithms, disease detection models are built. These
models can identify subtle relationships and associations between phenotypic patterns and specific
diseases. They can be trained on large, labeled datasets to learn the patterns and characteristics
associated with different diseases. HTP and AI algorithms together can analyze images or sensor data
to identify signs of diseases or pests, allowing for early detection and targeted interventions. For
example, machine learning models trained on large datasets of diseased plant images can accurately
classify and diagnose plant diseases (Nabwire et al. 2021; Hassan et al. 2022; Kim et al. 2023).

4.2. Trait discovery and quantification

Advanced imaging technologies combined with AI can facilitate the discovery and quantification of
novel or complex traits that are otherwise challenging to measure. For example, AI algorithms can
identify subtle differences in leaf coloration, texture, or disease symptoms that may be indicative of
specific genetic traits. By analyzing historical data, weather patterns, and crop traits using HTP, AI
models can predict crop yields and help optimize management practices. This information allows
farmers to plan harvesting, storage, and marketing strategies effectively (Shakoor et al. 2017;
Harfouche et al. 2019; Karunathilake et al. 2023; Kim et al. 2023).

4.3. Crop management and precision agriculture

Advanced imaging technologies combined with AI can provide real-time recommendations for irrigation
scheduling, fertilizer application, and pest control based on sensor data, weather conditions, and plant
responses. This optimization of crop management practices contributes to resource conservation and
increased productivity. Advanced imaging technologies and AI can also be applied to monitor and
optimize crop growth in real time. By continuously analyzing plant health, stress levels, and nutrient
requirements, farmers can make informed decisions regarding irrigation, fertilization, and pest
 Journal of Integrative Agriculture

management, leading to improved resource efficiency and higher crop yields (Chawade et al. 2019;
Maes and Steppe 2019; Kim et al. 2020; Karunathilake et al. 2023; Kim et al. 2023).

5. Challenges in implementing high-throughput phenotyping(HTPP) and

Artificial Intelligence (AI) in crop breeding programs
Undoubtedly, employing high-throughput plant phenotyping (HTPP) techniques provide an assessment
of desired characteristics of ample genotypes indirectly via robust, cost-friendly, wholesome, and
concurrent techniques (Kim et al. 2023). But; there are certain aspects which restrict the
implementation of this technology in a massive way in plant breeding. In this context, impediments viz.,
traits of quantitative origin, phenotyping of roots, impact of environment, trials (multilocation and
replicated) at field plot levels are a bit challenging. The use of HTPP resolutions at the spatial and
temporal levels for tissues or cells, morphological aspects, phenotypes of small size, and the characters
which are below the soil isalso risky (Zhao et al. 2019). Besides, the data which is generated is a
difficult task to store, manage, and process along with the creation of worthwhile information at
biological levels (Ates and Bukowski 2021; Jo et al. 2021). Various aspects like phenotyping using crop
models involve occlusionwhich could not be avoided while the structure of the canopy is estimated,
especially during the later stages ofcrop growth, in which the architecture offoliage is a challenge
(Karunathilake et al. 2023). Also, during the later growing season ofsoybean crop total area of
leavesanda number of leaves using Unmanned Aerial Systems (UAS) imaging is inaccurate compared
to an early period of growth (Liu et al. 2021). The root phenotyping involves, the core-break method, for
obtaining the depth of roots (Wasson et al. 2017) and minirhizotrons having sensors for observing the
growth of roots (Svane et al. 2019). But in both methods only a few roots can be detected along with
measured roots subjected to the position of the sample. Non-destructive tools of detection, viz., ground-
penetrating radar (Delgado et al. 2017) and electrical impedance tomography (Corona-Lopez et al.
2019), can detect the biomass of roots and help to observe the development of roots. But, the
resolution at a spatial level for both is less (~ cm/pixel), with low detection of fine roots at the individual
level. The genetic architecture of abiotic tolerance, such as drought resistance (DR), is complex and
influenced by many alleles with small effects (Fukao and Xiong 2013). Thus, the search for generic
drought tolerance using single major-effect genes has almost always been disappointing (Passioura et
al. 2012). There are many constraints which need to be properly with respect to (HTPP) alongwith
high-throughput analysis of imagery and artificial intelligence (AI) which will createan exact list of 2-
Dimensional and 3-Dimensional traits of various phenotypes regarding the stress aspects withthe latent
knowledge of tolerability with time. ollecting data under field conditions via (HTTP) and (AI) is quite in
an advancedphasefora few years. But, management and mining of the data collected in vast amounts
is quite complicated. As firstly, it needs clear-cut aims and objectives at biological levels. Secondly,
vibrant and friendly methods for the processing and analysis aspects at later levels in order to interpret
the rawdata need to be improvised at a larger scale (Poorter et al. 2012).
Meanwhile, every year, a lot of phenotypic data through experimentations of phenotypes under a
controlledenvironment or in the fieldgenerates huge. But, to replicate the output by the same
researcher or to reproduce resultsviavarious labs in different research experimentations is non-
satisfactory in many casesdueuncontrolledvariabilityin environmental conditions (Kim et al. 2023).
Therefore,parameters of environmenthave quite great impact andmustobtainsimilar attention from
researchers asthat for traitswhich need to be measured, leading to anotherquestion as through what a
researcher have to measure theeffect of environment. Envirotyping, which is a next-generation
technology forenvironmental aspects that helps in addressing this issue (Xu 2016). Besides, through
integration and multiplication of the information, the impact of genotype x environment x management
(GxExM) might be calculated, and phenomic prediction might be possibly done (Xu 2016; Araus et al.
2018). Since many years, yield of crop growth i.e., genetic gain is quite slow influenced by various
aspects, viz., population, genotypic data, heritable traits, geospatial (GS) data, and breeding strategy
(Xu et al. 2020). Theformat of imagery datachangesbroadly depends onvarious imagery sensors viz.,
RGB, thermal aspects, hyperspectral, computed tomography (CT) making a general view to analyze
image data achallenge (Yang et al. 2020). Also, it’s quite rare to re-use data of phenotypic origin as
compared to data of omics level. Data of phenotypic level is in addition to comply to the FAIR criteria is
not accessible easily (Goff et al. 2011). Elaborative descriptive metadata and the ideologies which can
be agreed upon are required for constructing a database system for storing multiomicsdata. But, huge
data which is collected is either scattered pieces in various laboratories as pieces or they differ in
format and quality aspects (Li et al. 2018). Sharing and standardization of data among various
communities is still an impediment. Other lacuna is deficient funds and infrastructure of data to
 Journal of Integrative Agriculture

handling (Bolger et al. 2019). Challenges of implementing high-throughput phenotyping(HTPP) and

artificial intelligence (AI) in crop breeding programs are depicted in Table 2.

Table 2 Applications of High-Throughput Phenotyping Platforms in various crops with limitations

Crop Trait Technique Challenges Reference
 Common bean Phenotyping of X-ray Cost intensive;time- Metzner et al. (2015)
(Phaseolus vulgaris L. roots mediatedComputed taking; specialized
cv. ‘Shiny Fardenlosa’) Tomography (CT) and prototype not available
Magnetic Resonance tostudyacrop
Imaging (MRI)
 Rapeseed seeds Automaticindivi PhenoSeeder Less measurement Jahnke et al. (2016)
(Brassica napus) dual seed ofspeed;
genotypesviz., (Wotan, handling and Threshing needed
Expert, and Pirola) phenotyping
 Barley seeds (Hordeum
vulgare) genotypes viz.,
(Barke, HOR13719, and
HOR9707)
 Arabidopsis seeds
(Arabidopsis thaliana)
genotypes viz., ( Co1-0,
lag2-2 and Agu-1)
 Maize (Zea mays) Canopyanalysis PocketPlant3D Fewerfeatures and Confalonieri et al.
hybrids viz., (Pioneer structurewith a single function; (2017)
P1223, smartphone reliable models needed
Syngenta GSS1477) for
complicatedfield
situations
 Spring wheat (Triticum Morphometric X-ray mediatedmicro Cost-intensive; time- Hughes et al. (2017)
aestivum cv. Paragon) traits of spike computed tomography taking;
and grains bespoke image analysis
needed for
pipelining novel species
type
 Wheat (Triticum Heading and AutomaticFieldScanalyze Cost-intensive; less Yoosefzadeh-
aestivum L.) cvs. Avalon, Flowering r, a rail-based gantry imaging area; varied Najafabadi et al.
Cadenza, Crusoe, phenotyping system ambient light (2023)
Gatsby, Soissons and
Maris Widgeon
 Zea mays genotypes Phenotyping of PhenoBox Requires labor for Sadeghi-Tehran et
EGB seeds viz., (Olds shoots screeningat large level al. (2017)
Seeds, Madison, WI,
USA)
 Nicotiana
benthamiana,Brachypodi
umdistachyon ecotype
ABR4 seeds
 237 grains of Triticum Cost-intensive; time- Hughes et al. (2019)
monococcum subsp. taking;
aegilopoides (wild) and morphometric μ Computed bespoke image analysis
 513 for variations Tomography needed topipeline new
Triticiummonococcum (CT )analysis of traits species
subsp. monococcum
(domesticated),grain
traits of wild (Hordeum
spontaneum) and
domesticated (Hordeum.
vulgare) 2‐row barley.
 Cultivated rice (Oryza Concentration Hyperspectral analysis Cost-intensive; bespoke Sun et al. (2019)
sativa L.) of protein image analysis required
and updationofmodel for
novel species,
recentindicators for
physiologyor
biochemical based
features
 Journal of Integrative Agriculture

 Rice (Oryza sativa) Chalkiness Convolutional neural coloured rice cannot be Wang et al. (2022)
networks (CNNs) and screened , needs extra
Gradient-weighted Class trainings and fine-tuning
Activation Mapping for quantifying
(Grad-CAM) chalkiness under varied
proportions of grain
overlapping

The next decade has to addressabove mentioned challenges, which will emphasize below ground
innovatory tools, advanced technology for dissecting stress like abiotic aspects oranother complicated
features, robust and feasible phenotyping at field levels, standardized data, and dissecting data for
multiomicsapproach. Despite, creating phenomic technology advances talent or potential is a
considering aspect, before constructing phenotyping facilities.But, the experienced and trained
manpower for phenotyping is still a lacuna to be tackled alongwith research personnel (e.g., data
imagery analysts) which is a challenge. These challenges can be overcome by innovative
technologytoimprovisethe conventional agricultural knowledgefor attracting industry potential personnel,
multidisciplinary phenomic trainings at multidisciplinary level especially agricultural universities or
colleges, sharing ofrightdata and questionnaire attracting more computer experts forproblem solving at
cost-free levels, even ifaccurate solutions are not there (Tsaftaris and Scharr 2019). Phenotyping of
plants is entering big-data levels with the advanced high throughput platforms. Information of
phenotypes at individual stage does not suffice associational analysis. Therefore, exact and accurate
phenomic data, multi-scale interactions viz., physiological, structural, omics approach, and environment
form the basis of research in coming years. To overcome vagaries of environment studies on
phenotyping of plants should create novel cost-friendly techniques on the basis of artificial intelligence
(AI)/remote sensing (RS) to advance the imagery-based phenotyping approaches. Deep or machine
learning models along with simulation platforms is the need of hour to develop latest applications for
(HTP) and (AI). Besides, an automatic model ofphenotyping should be identified forprecisetasksamong
the novel species of plant kingdomwithrobust and vibrant outputson the basis of large sampling
statistical analysiswithrelationshipsto traits of agronomic origin (Tayade et al. 2022). With the
availability ofmulti-disciplinary information at phenotypic levels,there is a need to
employrecentapplications of (AI) for in depth learning, fusion of data, hybrid intelligence and swarm
intelligencefor generating big-data managing productions to support integrated data, coordination,
reasonability, divisibility and universality.The prominent method to influence
(genotype×environment×management) interaction is to find desired traits for a particular environment.
Although, at scientific and technological levels many problems are there which need remedies viz.,
validation and practical processes of modelingalong with their interactions which require verified
procedures along with inter-connection and reaction of multi-scale phenotypic aspects among modeling
tools which require answers to gear up and for organizing the twisted genotype-vs-phenotype path for
sustainable agriculture (Zhao et al. 2019). The path of (GxPxE) requires elaborated analytical tools.
Therefore, Coppens et al. (2017) described that “phenotyping of plant has a future by synergising the
national and globalscenarios.”Since, HTP and AI is still an inception, the pioneers and the stalwarts of
research are combing the unmanned aerial systems based High throughput phenotyping platforms with
spaceborne (RS), Artificial Intelligence, and simulation modelling for crops to create agricultural
features for larger area. Also, there is a needto build multidomain groupswhich can tackle different
challengesthroughout the fields of biology, environment sciences, and computer sciences. There is an
urge to put efforts for developing standardized collection of, processing, and interpretation of data.
Meanwhile, as the core of digital data-based agriculture, the need of having quality raw-data cannotbe
neglected (Jung et al. 2021).

6. Conclusion and future perspective

Despite rapid advancements in robust and high-quality genetic and genomic technologies, the quality
and speed of high-throughput phenotyping are currently limiting crop genome functional analysis.
Continuous advances in genomics and HTP generate multiple layers of valuable information that can
be used to advance crop breeding and disease monitoring at a rapid pace. Significant contributions
have been made in recent years by government and private organizations to the development and use
of HTP tools to accelerate the development and deployment of phenotyping and breeding technologies
to benefit researchers and farmers. Integrating heterogeneous data from reliable, automated,
 Journal of Integrative Agriculture

multifunctional, and high-throughput phenotyping platforms will necessitate the continued development
of novel technologies, with a greater emphasis on developing low-cost, high-performance HTP
technologies. With multifunctional phenotyping platforms generating massive amounts of sensor data
and images, crop HTP will face new challenges in data storage, management, and analysis. The data
volume is determined by the sensor/imager resolution and the number of acquired readings. To further
promote the use of HTP in crop improvement programs, data analysis infrastructure that is less
expensive and more accessible will need to be developed.
Statements
Conflict of interest: Authors have no conflict of interest to declare
Funding: The Basic Science Research Program supported this research through the National
Research Foundation of Korea (NRF), funded by the Ministry of Education (2019R1A6A1A11052070).

References

Ampatzidis Y, Partel V. 2019. UAV-based high throughput phenotyping in citrus utilizing multispectral
imaging and artificial intelligence. Remote Sensing, 11, 410.
Araus J L, Kefauver S C, Zaman-Allah M, Olsen M S, Cairns J E. 2018. Translating high-throughput
phenotyping into genetic gain. Trends in Plant Science, 23, 451-466.
Arend D, Beier S, König P, Lange M, Memon J A, Oppermann M, Scholz U, Weise S. 2022. From
genotypes to phenotypes: A plant perspective on current developments in data management and data
publication. In: Integrative Bioinformatics: History and Future. Springer Singapore, Singapore. pp. 11-
43.
Ataş M, Yardimci Y, Temizel A. 2012. A new approach to aflatoxin detection in chili pepper by machine
vision. Computers and Electronics in Agriculture, 87, 129-141.
Ates A M, Bukowski M. 2021. 22 Global soybean stocks fall in tandem with South American production
economic research service| situation and outlook report domestic outlook International Outlook. 2022.
Benos L, Tagarakis A C, Dolias G, Berruto R, Kateris D, Bochtis D. 2021. Machine learning in agriculture:
A comprehensive updated review. Sensors, 21, 3758.
Bolger A M, Poorter H, Dumschott K, Bolger M E, Arend D, Osorio S, Gundlach H, Mayer K F, Lange M,
Scholz U, Usadel B. 2019. Computational aspects underlying genome to phenome analysis in plants.
The Plant Journal, 97, 182-198.
Busby P E, Soman C, Wagner M R, Friesen M L, Kremer J, Bennett A, Morsy M, Eisen J A, Leach J E,
Dangl J L. 2017. Research priorities for harnessing plant microbiomes in sustainable agriculture. Plos
Biology, 15, e2001793.
Bustos-Korts D, Boer M P, Malosetti M, Chapman S, Chenu K, Zheng B, Van Eeuwijk F A. 2019.
Combining crop growth modeling and statistical genetic modeling to evaluate phenotyping strategies.
Frontiers in Plant Science, 10, 1491.
Chawade A, van Ham J, Blomquist H, Bagge O, Alexandersson E, Ortiz R. 2019. High-throughput field-
phenotyping tools for plant breeding and precision agriculture. Agronomy, 9, 258
Cobb J N, DeClerck G, Greenberg A, Clark R, McCouch S. 2013. Next-generation phenotyping:
Requirements and strategies for enhancing our understanding of genotype–phenotype relationships
and its relevance to crop improvement. Theoretical and Applied Genetics, 126, 867-887.
 Journal of Integrative Agriculture

Confalonieri R, Paleari L, Foi M, Movedi E, Vesely F M, Thoelke W, Agape C, Borlini G, Ferri I, Massara
F, Motta R. 2017. PocketPlant3D: Analysing canopy structure using a smartphone. Biosystems
Engineering, 164, 1-12.
Coppens F, Wuyts N, Inzé D, Dhondt S. 2017. Unlocking the potential of plant phenotyping data through
integration and data-driven approaches. Current Opinion in Systems Biology, 4, 58-63.
Corona-Lopez D D, Sommer S, Rolfe S A, Podd F, Grieve B D. 2019. Electrical impedance tomography
as a tool for phenotyping plant roots. Plant Methods, 15, 1-15.
Crossa J, Pérez-Rodríguez P, Cuevas J, Montesinos-López O, Jarquín D, De Los Campos G, Burgueño
J, González-Camacho J M, Pérez-Elizalde S, Beyene Y, Dreisigacker S. 2017. Genomic selection in
plant breeding: methods, models, and perspectives. Trends in Plant Science, 22, 961-975.
Delgado A, Hays D B, Bruton R K, Ceballos H, Novo A, Boi E, Selvaraj M G. 2017. Ground penetrating
radar: A case study for estimating root bulking rate in cassava (Manihot esculenta Crantz). Plant
Methods, 13, 1-11.
van Dijk A D J, Kootstra G, Kruijer W, de Ridder D. 2021. Machine learning in plant science and plant
breeding. Iscience, 24, 101890.
Fahlgren N, Gehan M A, Baxter I. 2015. Lights, camera, action: High-throughput plant phenotyping is
ready for a close-up. Current Opinion in Plant Biology, 24, 93-99.
Fiorani F, Schurr U. 2013. Future scenarios for plant phenotyping. Annual Review of Plant Biology, 64,
267-291.
Fukao T, Xiong L. 2013. Genetic mechanisms conferring adaptation to submergence and drought in rice:
simple or complex. Current Opinion in Plant Biology, 16, 196-204.
Furbank R T, Tester M. 2011. Phenomics–technologies to relieve the phenotyping bottleneck. Trends in
Plant Science, 16, 635-644.
Ganesan M, Andavar S, Raj R S P. 2021. Prediction of land suitability for crop cultivation using
classification techniques. Brazilian Archives of Biology and Technology, 64, e21200483.
Ghimire A, Seong-Hoon K, Areum C, Naeun J, Seonhwa A, Mohammad S I, Sheikh M, Yong S C,
Yoonha K. 2023. Automatic evaluation of soybean seed traits using RGB image data and a python
algorithm. Plants, 12, 3078.
Gill T, Gill S K, Saini D K, Chopra Y, de Koff J P, Sandhu K S. 2022. A comprehensive review of high
throughput phenotyping and machine learning for plant stress phenotyping. Phenomics, 2, 156-183.
Goff S A, Vaughn M, McKay S, Lyons E, Stapleton A E, Gessler D, Matasci N, Wang L, Hanlon M,
Lenards A, Muir A. 2011. The iPlant collaborative: Cyberinfrastructure for plant biology. Frontiers in
Plant Science, 2, 34.
Gowda V D, Prabhu M S, Ramesha M, Kudari J M, Samal A. 2021. Smart agriculture and smart farming
using IoT technology. Journal of Physics: Conference Series, India. 2089, 012038.
Guo Q, Wu F, Pang S, Zhao X, Chen L, Liu J, Xue B, Xu G, Li L, Jing H, Chu C. 2018. Crop 3D-a LiDAR
based platform for 3D high-throughput crop phenotyping. Science China Life Sciences, 61, 328-339.
Hamadani A, Ganai N A. 2022. Development of a multi-use decision support system for scientific
management and breeding of sheep. Scientific Reports, 12, 19360.
 Journal of Integrative Agriculture

Hamadani A, Ganai N A, Alam S, Mudasir S, Raja T A, Hussain I, Ali H, Ahmad S M. 2022. Artificial
intelligence techniques for the prediction of body weights in sheep. Indian Journal of Animal Research,
B-4831, 1-6.
Hamadani H, Khan A. 2015. Automation in livestock farming–A technological revolution. International
Journal of Advance Research, 3, 1335-1344.
Haq M A. 2022. CNN based automated weed detection system using UAV imagery. Computer Systems
Science & Engineering, 42, 2.
Harfouche A L, Jacobson D A, Kainer D, Romero J C, Harfouche A H, Mugnozza G S, Moshelion M,
Tuskan G A, Keurentjes J J, Altman A. 2019. Accelerating climate resilient plant breeding by applying
next-generation artificial intelligence. Trends in Biotechnology, 37, 1217-1235.
Hartmann A, Czauderna T, Hoffmann R, Stein N, Schreiber F. 2011. HTPheno: An image analysis
pipeline for high-throughput plant phenotyping. BMC Bioinformatics, 12, 1-9. (not found in text?)
Hassan A R, Olanrewaju R O, Chukwudum Q C, Olanrewaju S A, Fadugba S E. 2022. Comparison
study of generative and discriminative models for classification of classifiers. International Journal of
Mathematics and Computers in Simulation, 16, 76-87.
Hughes A, Askew K, Scotson C P, Williams K, Sauze C, Corke F, Doonan J H, Nibau C. 2017. Non-
destructive, high-content analysis of wheat grain traits using X-ray micro computed tomography. Plant
Methods, 13, 1-16.
Hughes A, Oliveira H R, Fradgley N, Corke F M, Cockram J, Doonan J H, Nibau C. 2019. μCT trait
analysis reveals morphometric differences between domesticated temperate small grain cereals and
their wild relatives. The Plant Journal, 99, 98-111.
Islam M, Shehzad F. 2022. A prediction model optimization critiques through centroid clustering by
reducing the sample size, integrating statistical and machine learning techniques for wheat
productivity. Scientifica, 2022, pp?.
Jafari M, Shahsavar A. 2020. The application of artificial neural networks in modeling and predicting the
effects of melatonin on morphological responses of citrus to drought stress. PLoS ONE, 15, e0240427.
Jahnke S, Roussel J, Hombach T, Kochs J, Fischbach A, Huber G, Scharr H. 2016. Pheno seeder-a
robot system for automated handling and phenotyping of individual seeds. Plant Physiology, 172,
1358-1370.
Jannoura R, Brinkmann K, Uteau D, Bruns C, Joergensen R G. 2015. Monitoring of crop biomass using
true colour aerial photographs taken from a remote controlled hexacopter. Biosystems Engineering,
129, 341-351.
Jo H, Lee J Y, Cho H, Choi H J, Son C K, Bae J S, Bilyeu K, Song J T, Lee J D. 2021. Genetic diversity
of soybeans (Glycine max (L.) merr.) with black seed coats and green cotyledons in Korean
germplasm. Agronomy, 11, 581.
Jung D H, Kim C Y, Lee T S, Park S H. 2022 Depth image conversion model based on CycleGAN for
growing tomato truss identification. Plant Methods, 18, 83.
Jung J, Maeda M, Chang A, Bhandari M, Ashapure A, Landivar-Bowles J. 2021. The potential of remote
sensing and artificial intelligence as tools to improve the resilience of agriculture production systems.
Current Opinion in Biotechnology, 70, 15-22.
 Journal of Integrative Agriculture

Karunathilake E M B M, Tuan Le A, Seong H, Yong S C, Sheikh M. 2023. The path to smart farming:
Innovations and opportunities in precision agriculture. Agriculture, 13, 1593.
Kim J Y. 2020. Roadmap to high throughput phenotyping for plant breeding. Journal of Biosystems
Engineering, 45, 43-55.
Kim J, Lee C, Park J E, Mansoor S, Chung Y S, Kim K. 2023. Drought stress restoration frequencies of
phenotypic indicators in early vegetative stages of soybean (Glycine max L.). Sustainability, 15, 4852.
Koh J C, Spangenberg G, Kant S. 2021. Automated machine learning for high-throughput image-based
plant phenotyping. Remote Sensing, 13, 858.
Ku K B, Mansoor S, Han G D, Chung Y S, Tuan T T. 2023 Identification of new cold tolerant Zoysia
grass species using high-resolution RGB and multi-spectral imaging. Scientific Reports, 13, 13209.
Latif N A, Nain F N M , Malim N H A H, Abdullah R, Rahim M F A, Mohamad M N, Fauzi N S M. 2021.
Predicting heritability of oil palm breeding using phenotypic traits and machine learning. Sustainability,
13, 12613.
Li L, Zhang Q, Huang D. 2014. A review of imaging techniques for plant phenotyping. Sensors, 14,
20078-20111.
Li Y, Xiao J, Chen L, Huang X, Cheng Z, Han B, Zhang Q, Wu C. 2018. Rice functional genomics
research: past decade and future. Molecular Plant, 11, 359-380.
Liu F, Hu P, Zheng B, Duan T, Zhu B, Guo Y. 2021. A field-based high-throughput method for acquiring
canopy architecture using unmanned aerial vehicle images. Agricultural and Forest Meteorology, 296,
108231.
Liu H, Bruning B, Garnett T, Berger B. 2020. Hyperspectral imaging and 3D technologies for plant
phenotyping: From satellite to close-range sensing. Computers and Electronics in Agriculture, 175,
105621.
Maes W H, Steppe K. 2019. Perspectives for remote sensing with unmanned aerial vehicles in precision
agriculture. Trends in Plant Science, 24, 152-164.
McMullen M D, Kresovic S, Villeda H S, Bradbury P, Li H, Sun Q, Flint-Garcia S, Thornsberry J, Acharya
C, Bottoms C, Brown P. 2009. Genetic properties of the maize nested association mapping population.
Science, 325, 737-740.
Mentsiev A U, Amirova E F. 2020. IoT and mechanization in agriculture: problems, solutions, and
prospects. In: IOP Conference Series: Earth and Environmental Science. 548, 032035.
Merrick L F, Herr A W, Sandhu K S, Lozada D N, Carter A H. 2022. Optimizing plant breeding programs
for genomic selection. Agronomy, 12, 714.
Metzner R, Eggert A, van Dusschoten D, Pflugfelder D, Gerth S, Schurr U, Uhlmann N, Jahnke S. 2015.
Direct comparison of MRI and X-ray CT technologies for 3D imaging of root systems in soil: potential
and challenges for root trait quantification. Plant Methods, 11, 1-11.
Minervini M, Scharr H, Tsaftaris S A. 2015. Image analysis: The new bottleneck in plant phenotyping
[applications corner]. IEEE Signal Processing Magazine, 32, 126-131.
Mir R R, Reynolds M, Pinto F, Khan M A, Bhat M A. 2019. High-throughput phenotyping for crop
improvement in the genomics era. Plant Science, 282, 60-72.
 Journal of Integrative Agriculture

Mishra B, Kumar N, Mukhtar M S. 2019. Systems biology and machine learning in plant–pathogen
interactions. Molecular Plant (Microbe Interactions), 32, 45-55.
Mochida K, Koda S, Inoue K, Hirayama T, Tanaka S, Nishii R, Melgani F. 2019. Computer vision-based
phenotyping for improvement of plant productivity: A machine learning perspective. GigaScience, 8,
p.giy153??.
Montes J M, Melchinger A E, Reif J C. 2007. Novel throughput phenotyping platforms in plant genetic
studies. Trends in Plant Science, 12, 433-436.
Morisse M, Wells D M, Millet E J, Lillemo M, Fahrner S, Cellini F, Lootens P, Muller O, Herrera J M,
Bentley A R Janni M. 2022. A European perspective on opportunities and demands for field-based
crop phenotyping. Field Crops Research, 276, 108371.
Nabwire S, Suh H K, Kim M S, Baek I, Cho B K. 2021. Application of artificial intelligence in phenomics.
Sensors, 21, 4363.
Nagasubramanian K, Singh A, Singh A, Sarkar S, Ganapathysubramanian B. 2022. Plant phenotyping
with limited annotation: Doing more with less. The Plant Phenome Journal, 5, e20051.
Pabuayon I L B, Sun Y, Guo W, Ritchie G L. 2019. High-throughput phenotyping in cotton: A review.
Journal of Cotton Research, 2, 1-9.
Parmley K A, Higgins R H, Ganapathysubramanian B, Sarkar S, Singh A K. 2019. Machine learning
approach for prescriptive plant breeding. Scientific Reports, 9, 7132.
Parry M A, Reynolds M, Salvucci M E, Raines C, Andralojc P J, Zhu X G, Price G D, Condon A G,
Furbank R T. 2011. Raising yield potential of wheat. II. Increasing photosynthetic capacity and
efficiency. Journal of Experimental Botany, 62, 453-467.
Passioura J B. 2012. Phenotyping for drought tolerance in grain crops: when is it useful to breeders.
Functional Plant Biology, 39, 851-859.
Pieruschka R, Schurr U. 2019. Plant phenotyping: past, present, and future. Plant Phenomics, 2019, pp?
Poorter H, Fiorani F, Stitt M, Schurr U, Finck A, Gibon Y, Usadel B, Munns R, Atkin O K, Tardieu F,
Pons T L. 2012. The art of growing plants for experimental purposes: A practical guide for the plant
biologist. Functional Plant Biology, 39, 821-838.
Rai K K. 2022. Integrating speed breeding with artificial intelligence for developing climate-smart crops.
Molecular Biology Reports, 49, 11385-11402.
Rebetzke G J, Jimenez-Berni J, Fischer R A, Deery D M, Smith D J. 2019. High-throughput phenotyping
to enhance the use of crop genetic resources. Plant Science, 282, 40-48.
Rouphael Y, Spíchal L, Panzarová K, Casa R, Colla G. 2018. High-throughput plant phenotyping for
developing novel biostimulants: From lab to field or from field to lab. Frontiers in Plant Science, 9,
1197.
Sadeghi-Tehran P, Sabermanesh K, Virlet N, Hawkesford M J. 2017. Automated method to determine
two critical growth stages of wheat: heading and flowering. Frontiers in Plant Science, 8, 252.
Sagan V, Maimaitijiang M, Paheding S, Bhadra S, Gosselin N, Burnette M, Demieville J, Hartling S,
LeBauer D, Newcomb M, Pauli D. 2021. Data-driven artificial intelligence for calibration of
hyperspectral big data. IEEE Transactions on Geoscience and Remote Sensing, 60, 1-20.
 Journal of Integrative Agriculture

Sarić R, NguyenV D, Burge T, Berkowitz O, Trtílek M, Whelan J, Lewsey M G, Čustović E. 2022.

Applications of hyperspectral imaging in plant phenotyping. Trends in Plant Science, 27, 301-315.
Shakoor N, Lee S, Mockler T C. 2017. High throughput phenotyping to accelerate crop breeding and
monitoring of diseases in the field. Current Opinion in Plant Biology, 38, 184-192.
Singh A, Ganapathysubramanian B, Singh A, Sarkar S. 2016. Machine learning for high-throughput
stress phenotyping in plants. Trends in Plant Science, 21, 110-124.
Singh P, Singh N, Singh K K, Singh A. 2021. Diagnosing of disease using machine learning. In: Machine
Learning and the Internet of Medical Things in Healthcare. Academic Press, (no country?), pp. 89-111.
(not found in text?)
Šulc M, Matas J. 2017. Fine-grained recognition of plants from images. Plant Methods, 13, 1-14.
Sun D, Cen H, Weng H, Wan L, Abdalla A, El-Manawy A I, Zhu Y, Zhao N, Fu H, Tang J, Li X. 2019.
Using hyperspectral analysis as a potential high throughput phenotyping tool in GWAS for protein
content of rice quality. Plant Methods, 15, 1-16.
Sundari V, Anusree M, Swetha U. 2022. Crop recommendation and yield prediction using machine
learning algorithms. World Journal of Advanced Research and Reviews, 14, 452-459.
Svane S F, Jensen C S, Thorup-Kristensen K. 2019. Construction of a large-scale semi-field facility to
study genotypic differences in deep root growth and resources acquisition. Plant Methods, 15, 1-16.
Taranto F, Nicolia A, Pavan S, De Vita P, D’Agostino N. 2018. Biotechnological and digital revolution for
climate-smart plant breeding. Agronomy, 8, 277.
Tayade R, Yoon J, Lay L, Khan A L, Yoon Y, Kim Y. 2022. Utilization of spectral indices for high-
throughput phenotyping. Plants, 11, 1712.
Teshome F T, Bayabil H K, Hoogenboom G, Schaffer B, Singh A, Ampatzidis Y. 2023. Unmanned aerial
vehicle (UAV) imaging and machine learning applications for plant phenotyping. Computers and
Electronics in Agriculture, 212, 108064.
Tian Z, Ma W, Yang Q, Duan F. 2022. Application status and challenges of machine vision in plant
factory—A r eview. Information Processing in Agriculture, 9, 195-211.
Tripodi P, Nicastro N, Pane C, Cammarano D. 2022. Digital applications and artificial intelligence in
agriculture toward next-generation plant phenotyping. Crop and Pasture Science, 74, 597-614
Tsaftaris S A, Minervini M, Scharr H. 2016. Machine learning for plant phenotyping needs image
processing. Trends in Plant Science, 21, 989-991.
Tsaftaris S A, Scharr H. 2019. Sharing the right data right: A symbiosis with machine learning. Trends in
Plant Science, 24, 99-102.
Varshney R K. 2021. The plant genome special issue: Advances in genomic selection and application of
machine learning in genomic prediction for crop improvement. The Plant Genome, 14, 3.
Veeragandham S, Santhi H. 2020. A review on the role of machine learning in agriculture. Scalable
Computing: Practice and Experience, 21, 583-589.
Wang C, Caragea D, Kodadinne Narayana N, Hein N T, Bheemanahalli R, Somayanda I M, Jagadish S
K. 2022. Deep learning based high-throughput phenotyping of chalkiness in rice exposed to high night
temperature. Plant Methods, 18, 9.
 Journal of Integrative Agriculture

Wasson A P, Chiu G S, Zwart A B, Binns T R. 2017. Differentiating wheat genotypes by Bayesian

hierarchical nonlinear mixed modeling of wheat root density. Frontiers in Plant Science, 8, 282.
Xu R, Li C. 2022. A review of high-throughput field phenotyping systems: Focusing on ground robots.
Plant Phenomics, 2022, 9760269.
Xu Y. 2016. Envirotyping for deciphering environmental impacts on crop plants. Theoretical and Applied
Genetics, 129, 653-673.
Xu Y, Liu X, Fu J, Wang H, Wang J, Huang C, Prasanna B M, Olsen M S, Wang G, Zhang A. 2020.
Enhancing genetic gain through genomic selection: From livestock to plants. Plant Communications, 1,
100005.
(no page?).
Xu Y, Zhang X, Li H, Zheng H, Zhang J, Olsen M S, Varshney R K, Prasanna B M, Qian Q. 2022. Smart
breeding driven by big data, artificial intelligence, and integrated genomic-enviromic prediction.
Molecular Plant, 15, 1664-1695.
Yang G, Liu J, Zhao C, Li Z, Huang Y, Yu H, Xu B, Yang X, Zhu D, Zhang X, Zhang R. 2017. Unmanned
aerial vehicle remote sensing for field-based crop phenotyping: current status and perspectives.
Frontiers in plant Science, 8, 1111.
Yang W, Feng H, Zhang X, Zhang J, Doonan J H, Batchelor W D, Xiong L, Yan J. 2020. Crop
phenomics and high-throughput phenotyping: past decades, current challenges, and future
perspectives. Molecular Plant, 13, 187-214.
Yang Y, Saand M A, Huang L, Abdelaal W B, Zhang J, Wu Y, Li J, Sirohi M H, Wang F. 2021.
Applications of multi-omics technologies for crop improvement. Frontiers in Plant Science, 12, 563953.
Yoosefzadeh-Najafabadi M, Hesami M, Eskandari M. 2023. Machine learning-assisted approaches in
modernized plant breeding programs. Genes, 14, 777.
York L M. 2019. Functional phenomics: An emerging field integrating high-throughput phenotyping,
physiology, and bioinformatics. Journal of Experimental Botany, 70, 379-386.
Younas M. 2019. Research challenges of big data. Service Oriented Computing and Applications, 13,
105-107.
Zhao C, Zhang Y, Du J, Guo X, Wen W, Gu S, Wang J, Fan J. 2019. Crop phenomics: Current status
and perspectives. Frontiers in Plant Science, 10, 714.