Integration of Plant Electrophysiology and Time-Lapse Video Analysis via Artificial Intelligence for the Advancement of Precision Agriculture

Abstract

Biological research and agriculture are increasingly benefiting from the use of artificial intelligence algorithms, which are becoming integral to various areas of human activity. Fundamental knowledge of the mechanisms of plant germination, growth/development, and reproduction is the basis for plant cultivation. Plants provide food and valuable biochemicals and are an important element of a sustainable natural environment. An interdisciplinary approach involving basic science (biology and informatics), technology (artificial intelligence), and farming practice can contribute to the development of precision agriculture, which in turn increases crop and food production. Nowadays, a progressive elucidation of the mechanisms of plant growth/development involves studies of interrelations between electrical phenomena occurring inside plants and movements of plant organs. Recently, there have been increasing numbers of reports on methods for classifying plant electrograms using statistical and artificial intelligence algorithms. Artificial intelligence procedures can identify diverse electrical signals—signatures associated with specific environmental abiotic and biotic factors or stresses. At the same time, a growing body of research shows methods of precise and fast analysis of time-lapse videos via automated image analysis and artificial intelligence to study the movement and growth/development of plants. In both research fields, scientists introduce modern and promising methods of studying plant growth/development. Such basic research along with technological innovations will contribute to the development of precision agriculture and an increase in yields and production of healthier food in future.

1. Introduction

Plants are part of the natural environment, from where they receive and process nutrient elements and stimuli and as a result show induced and spontaneous activity that changes the environment. Physiological processes occurring in plants interact with each other. The subject of my research and the following review are processes related to plant growth/development related to movements of organs and electrical changes (Figure 1). These processes exhibit a mutual relationship, which is currently being studied in greater detail. Plant movements also inspire modern architectural designs that support environmental sustainability [1]. Studying the influence of environmental factors on plant growth and development is essential for considerations of a sustainable natural and agricultural environment [2,3,4,5].

1.1. Plant Electricity

Plant electrophysiology explores the role of electricity in plant physiology, including organ movements and growth/development processes. Electrical phenomena occurring in living organisms are generated by the presence of ions and electrons and are related to the properties of the cell membrane and photosynthetic and respiratory electron carrier chains. Changes in the trans-membrane electrical potential are essential in triggering the motor activity in animals, i.e., they exhibit stimulus-induced and spontaneous electrical activity. Electro-motive phenomena in animals rely on essential structural elements: the nervous system and the muscular system. Currently, comparative analyses of animal and plant behaviour are being carried out, and some processes that are important in animal and human biology are being studied in plants [6,7,8,9,10,11,12].

1.2. Plant Organ Movements

Organ movements, ubiquitous in growing plants, are studied at the level of cells, tissues, organs, and the whole plant. The whole plant exhibits rapid and slow endogenously and exogenously driven movements of organs. Endogenous ultradian nutations and circadian-regulated movements are a background for various exogenously induced tropisms and nastic movements. The time scale of plant movements ranges from milliseconds to hours. Plants can twine, bend towards and against stimuli, pull leaves down and up, and track the sun. The common classification of plant movement distinguishes tropisms, nastic movements, and nutations [13]. The motor behaviour is a collection of all types of movements performed by the plant. In a single plant, movements of various organs appear simultaneously and their combination results in its complex motor behaviour. Charles Darwin tried to explain the origin of organ movement in terms of the evolution of behaviour. He explained how plants achieved the ability to climb in his famous work “The Power of Movement in Plants ” [14]. Nowadays, plant movements are an inspiration for biomimetic and biomechanic investigations, search for actuators, and robotic studies [15,16]. In animal studies, the connections between electrical and motor activity are obvious. In this work, we present both these activities in plants. A broad collection of videos of plant movements, especially nutations, and examples of electrical activity are available on my website: http://circumnutation.umcs.lublin.pl, accessed on 6 April 2025.

1.3. Method

This review was written based on literature searched in Google Scholar using a combination of such keywords as artificial intelligence, electrome, electrical signals, electrodes, electrophysiology, machine learning, phenomics, plant electrograms, plant movements, precision agriculture, and time-lapse video. Logical operators (and, or) were used to narrow down the results. The work focuses on articles published between 2000 and 2025, with an emphasis on the latest research from the last 5 years (2020–2025) to reflect the current state of knowledge and technological advances. Older primary publications by Charles Darwin (1880) [14] and Burdon-Sanderson and John Scott (1873) [17] and two reviews from 1975 and 1976 are also included. This review includes 68 literature items. Papers on image analysis of only single biological images were excluded. The focus was placed on time-lapse video sequences of plant movements analysed using special software and AI. Papers on general changes in electric potential and electrical changes related to microelectrode measurements were also excluded. The latest works from extracellular research with additional AI analysis in plants were the main focus. In this review, the possibility of systematic errors, which may affect the presented results, should be taken into account. The systematic errors that may have occurred in this work include publication bias, selection bias, and methodological heterogeneity of the studies, which makes direct comparisons difficult. The publication bias may result from the greater frequency of publication of studies with positive results. The selection bias is related to the limitation of searching for publications in English and using Google Scholar. Additionally, interpretation errors related to the assessment of the quality of the studies and their results cannot be excluded.

The aim of this paper is to show electrical and motor phenomena in plants, which are currently being analysed anew by modern statistics and artificial intelligence methods. New methods of recording and analysing electrical and motor changes in plants will be presented on the basis of the literature review. The potential application of these methods in the development of precision agriculture will be considered.

The paper presents an interdisciplinary approach, highlighting the contribution of research on plant electrophysiology and movement to fundamental plant sciences. These basic sciences, combined with technological and applied research, support the development of precision agriculture (Figure 2). Figure 2 illustrates converging interdisciplinary pathways that advance precision farming and promote environmental sustainability. The role of studies on plant electricity and movement in enhancing our understanding of plant physiology and biology is particularly noteworthy. This foundational knowledge may, in turn, find future applications in precision agriculture.

2. Artificial Intelligence Procedures in Plant Phenomics

Artificial intelligence procedures increasingly support the plant biology research [18,19]. Studies at the level of the whole organism, i.e., the phenome, are currently much less developed than studies on the genome, proteome, or transcriptome. The studies described below mainly concern research on whole plants. Both the extracellular method of measuring electrical voltage and the time-lapse video technique are methods describing the plant phenome, contributing to the development of methods of plant phenotyping.

Subtle patterns of plant movements and electrical signals/signatures that are not visible in traditional statistical analyses can be detected by machine learning algorithms. This approach opens up new perspectives in the study of plant movements and behaviour [19]. To study the kinematic behaviour of plants, the following algorithms can be used: Unsupervised Learning (Anomaly Detection, Isolation Forest, One-Class Support Vector Machine, k-nearest neighbours), Supervised Learning (Decision Trees, random forest, support vector machines, Logistic regression), Ensemble Learning (Bagging, Boosting), and deep learning [19]. Similarly, various artificial intelligence algorithms are used in analyses of plant electrograms [20].

3. Plant Electrograms and Electrome Concept

Plant electrical activity is a set of all types of spontaneous and induced transmembrane potential changes, having different kinetics and occurring in different plant organs and tissues. The types of electrical changes include resting potential, action potential (AP), stimulus-induced long-lasting depolarisation and hyperpolarisation, and ultradian and circadian oscillations. All these changes in electrical activity, usually measured separately, create a system of changes in the electrical potential in the plant. It is an integral part of metabolic processes, in which flows of electric charges measured by electrodes take place. This system of diverse and dynamic distribution of electrical potential in the plant may perform signalling and coordination functions, similarly to animal systems. Recently, the concept of the electrome has appeared in reference to such concepts in biology as the proteome, transcriptome, and genome. The name electrome was proposed by De Loof (2016), and the plant electrome concept is currently being developed [21]. The set of ions, ion channels, and pumps located in the cell membrane determines the ionic flows taking place both in the cells and at the level of the entire organism, creating different levels of the electrome [21]. The electrome refers to the total electrical activity in plants, which plays a key role in their physiology, signalling, and adaptive capacity [20]. The dynamic development of the electrome concept is evidenced by the work addressed to young people [22], which is intended to emphasise its importance in plant life.

Mathematical modelling is also used in analyses of some electrophysiological data. An example is the mathematical model derived from the Hodgkin–Huxley equation for predicting the extracellular potential based on intracellular activity [23]. Such modelling allows predicting the value of extracellular signals. It was determined that extracellular signals are proportional to the transmembrane current and depend on the distance between cells and the resistance of the environment [23]. Mathematical modelling is also used to investigate changes in the AP response to stimuli. This work has been performed in Nitellopsis cells using salinity stimuli and modelling using the Thiel–Beilby model, which has previously been applied to Chara australis [24,25,26]. Most researchers have used the extracellular measurement method. Importantly, extracellular measurements reflect the total electrical activity of many cells rather than the exact intracellular potential [23]. Recent work using metal electrodes and extracellular measurements applies AI analyses to extracellular recordings, i.e., plant electrograms.

4. Effects of Environmental Factors and Stresses on Plant Electricity

Abiotic and biotic environmental factors affect the electrical state of the plant by inducing characteristic electrical signals (Figure 3, Table 1). Below are the types of environmental factors, discussed in the recent literature, that affect the electrical state of the plant and are detected by statistical and AI methods.

4.1. Abiotic Factors and Stresses

Deep learning will play an increasingly important role in development of methods of precision fertilisation in agriculture. Work is already underway to use AI to analyse soil and plant nutrients to optimise plant growth [27].

Table 1. Environmental factors, electrogram parameters, and statistics/AI analysis used in investigation of plant electricity.

Plant	Environmental Factors	Electrogram Parameters	Statistics/ AI Analysis	Literature (Year)
Abiotic factors
tomato	calcium (Ca), nitrogen (N), manganese (Mn), and iron (Fe) deficiencies	silver-coated copper wire electrodes continuous 3 weeks, 500 Hz	signal decomposition sample space reduction feature extraction	[28] (2022)
peppers	low and high urea fertilization	three stainless steel needle electrodes stem electrical resistance many days greenhouse and field experiment	average value of triplicates one-way ANOVA Duncan’s multi-range test principal component analysis (PCA)	[29] (2023)
tomato cabbage	exposure to chemicals such as sulfuric acid (H2SO4), sodium chloride (NaCl) and ozone (O3)	two stainless steel needle electrodes laboratory, Faraday cage 10 Hz	fifteen statistical features eight different classification algorithms, PCA	[30] (2023)
Hedera helix	ozone exposure	laboratory, Faraday cage 300 Hz	generic automatic toolchain machine learning models	[31] (2024)
Zamioculcas zamiifolia Solanum lycopersicum (tomato)	wind, heat, red light blue light	electrical potential and impedance many minutes lasting measures laboratory, 0.58 Hz	discriminant analysis deep-learning methods	[32] (2023)
grapevine	water status	two silver-plated needle electrodes many days climate chambers, 256 Hz.	two machine learning approaches based on classification and regression the prediction models	[33] (2024)
bean	water status	needle electrodes two hours measurements Faraday’s cage 62.5 Hz.	arithmetic average of voltage variation, skewness, kurtosis, probability density function (PDF), autocorrelation, power spectral density (PSD), approximate entropy (ApEn), fast Fourier transform (FFT), and multiscale approximate entropy (ApEn(s), machine learning (ML)	[34] (2024)
Clivia	water gradient	patch electrodes Faraday’s cage in the thermostatic and humidified incubator 60 min measurements 30 sec samples, 30 Hz	lightweight convolutional neural network (CNN) model (PlantNet)	[35] (2024)
Biotic factors
barley (Hordeum vulgare)	fungal infection (Blumeria graminis, Bipolaris sorokiniana)	pair of needle electrodes Faraday cage, 48 h measurements	descriptive statistics, Bayesian change point (BCP) analysis, ApEn, autocorrelation, ML (cluster analysis)	[36] (2023)
tobacco	viral diseases (alfalfa mosaic virus)	two fine needle electrodes field conditions, 8 s at a sampling rate of 250 Hz	median, autoregressive coefficients, autocorrelation—ML models (support vector machine, k-nearest neighbours, random forest)	[37] (2024)
tomato	parasitic nematodes	many days measurements	ML model	[38] (2024)

Table 1. Environmental factors, electrogram parameters, and statistics/AI analysis used in investigation of plant electricity.

Plant	Environmental Factors	Electrogram Parameters	Statistics/ AI Analysis	Literature (Year)
Abiotic factors
tomato	calcium (Ca), nitrogen (N), manganese (Mn), and iron (Fe) deficiencies	silver-coated copper wire electrodes continuous 3 weeks, 500 Hz	signal decomposition sample space reduction feature extraction	[28] (2022)
peppers	low and high urea fertilization	three stainless steel needle electrodes stem electrical resistance many days greenhouse and field experiment	average value of triplicates one-way ANOVA Duncan’s multi-range test principal component analysis (PCA)	[29] (2023)
tomato cabbage	exposure to chemicals such as sulfuric acid (H2SO4), sodium chloride (NaCl) and ozone (O3)	two stainless steel needle electrodes laboratory, Faraday cage 10 Hz	fifteen statistical features eight different classification algorithms, PCA	[30] (2023)
Hedera helix	ozone exposure	laboratory, Faraday cage 300 Hz	generic automatic toolchain machine learning models	[31] (2024)
Zamioculcas zamiifolia Solanum lycopersicum (tomato)	wind, heat, red light blue light	electrical potential and impedance many minutes lasting measures laboratory, 0.58 Hz	discriminant analysis deep-learning methods	[32] (2023)
grapevine	water status	two silver-plated needle electrodes many days climate chambers, 256 Hz.	two machine learning approaches based on classification and regression the prediction models	[33] (2024)
bean	water status	needle electrodes two hours measurements Faraday’s cage 62.5 Hz.	arithmetic average of voltage variation, skewness, kurtosis, probability density function (PDF), autocorrelation, power spectral density (PSD), approximate entropy (ApEn), fast Fourier transform (FFT), and multiscale approximate entropy (ApEn(s), machine learning (ML)	[34] (2024)
Clivia	water gradient	patch electrodes Faraday’s cage in the thermostatic and humidified incubator 60 min measurements 30 sec samples, 30 Hz	lightweight convolutional neural network (CNN) model (PlantNet)	[35] (2024)
Biotic factors
barley (Hordeum vulgare)	fungal infection (Blumeria graminis, Bipolaris sorokiniana)	pair of needle electrodes Faraday cage, 48 h measurements	descriptive statistics, Bayesian change point (BCP) analysis, ApEn, autocorrelation, ML (cluster analysis)	[36] (2023)
tobacco	viral diseases (alfalfa mosaic virus)	two fine needle electrodes field conditions, 8 s at a sampling rate of 250 Hz	median, autoregressive coefficients, autocorrelation—ML models (support vector machine, k-nearest neighbours, random forest)	[37] (2024)
tomato	parasitic nematodes	many days measurements	ML model	[38] (2024)

Plant yield and health are largely dependent on the availability of nutrients in the nutrient solution [28]. To date, the assessment of the plant’s nutrient supply has been based on the visual observation of the plant. Recently, studies have been conducted to assess whether electrical signals from plants can indicate specific nutrient deficiencies. An attempt was made to use electrophysiological signals to detect calcium (Ca), nitrogen (N), manganese (Mn), and iron (Fe) deficiencies in tomato plants [28]. The analysis was based on signal decomposition using empirical mode decomposition and statistical feature extraction. The analysis of electrical signal patterns showed their specificity in relation to a specific deficiency. The development of this method may support real-time diagnosis of deficiencies as well as effective and precise intervention in the future agriculture [28].

Another paper presents studies in which the growth of peppers was monitored with different urea applications in the soil (low and high urea fertilisation) in relation to induced electrical signals [29]. A relationship was demonstrated between the electrical activity of the pepper plants and the degree of fertilisation and pepper growth. The advantages of the method were indicated as non-invasive, and the fertilisation status was presented in real time [29].

Electrical signals were also studied in tomato and cabbage plants after exposure to such chemicals as sulphuric acid (H₂SO₄), sodium chloride (NaCl), and ozone (O₃) [30]. An analysis of fifteen statistical features of the electrical signal was performed using eight types of algorithms in statistics and AI. The research aimed to develop more precise agriculture and check the level of environmental pollution based on electrical signalling in plants [30].

Ozone exposure was studied in Hedera helix plants, where the electrophysiological response was measured and analysed using machine learning methods [31]. The plant was used here as a sensor of ozone in the air. The tool chain presented in this work automates the development of algorithms for monitoring plants as air quality sensors [31]. The authors believe that a network of such sensor plants could monitor air quality in sustainable cities in the future [31].

Further studies were performed on Zamioculcas zamiifolia and Solanum lycopersicum (tomato) plants exposed to four different stimuli: wind, heat, red light, and blue light. The electric potential and impedance signals of the tissues were measured and analysed with a wide range of methods from discriminant statistical analysis to deep learning [32].

Another important problem for farmers worldwide is the real-time assessment of water status in grapevine plants [33]. This study proposed the use of plant electrophysiology as a new approach to assess water status. Experiments were conducted in a climatic chamber with potted grapevines under different irrigation regimes. Various morphological and physiological assessments were performed in parallel with electrophysiological measurements to correlate classical water status assessment methods with plant electrophysiological signals [33]. A binary classification model was used, and a regression model that effectively distinguishes between weak and strong irrigation signals was developed. These results therefore represent a promising new tool for future real-time monitoring and remote irrigation of grapevines. In the future, the development of this method will improve irrigation management and agronomic strategies [33].

In modern sustainable agriculture, saving water in crop planning is extremely important. Optimisation models balancing irrigation needs equipped with soil moisture and plant hydration sensors are currently being developed [27].

The study of water availability and accompanying electrical signals was the subject of experiments on common beans [34]. Extensive statistical methods and machine learning techniques were also used to classify electrical signals. This work indicated the possibility of using the electrome as a physiological indicator of water status in bean plants [34].

Another way to monitor water use by a plant is through the use of Clivia to determine the water gradient in the environment [35]. In this study, specific electrical signals in Clivia were documented under different soil moisture gradients. A lightweight convolutional neural network model was used for the analysis. Such studies demonstrate the potential of plants to be used as environmental sensors [35].

4.2. Biotic Stresses

In barley (Hordeum vulgare), bioelectric signals are produced as a response to fungal infection before visible disease symptoms appear [36]. Two fungal pathogens were studied: Blumeria graminis—a biotrophic fungus causing powdery mildew—and Bipolaris sorokiniana—a hemibiotrophic fungus causing brown leaf spot. The signals obtained after infection with both pathogenic fungi differed, giving a specific electrome signature. Blumeria graminis caused a decrease in entropy (more orderly signals), while B. sorokiniana increased entropy (more chaotic signals) [36]. In the study of barley signals, machine learning was applied to detect and classify bioelectrical changes in this plant. The combination of time series analysis (Bayesian Change Point, approximate entropy, autocorrelation), and AI clustering (K-means, K-modes, Balanced Iterative Reducing, and Clustering using Hierarchies) helped to create algorithms for early detection of fungal infection [36]. Interestingly, these algorithms can detect signals of infection within minutes of contact with the pathogen.

A very important problem in plant productivity today is posed by viral plant diseases [37]. It seems necessary to develop fast and non-invasive methods for early detection of virus infection to mitigate losses and limit the spread of viruses. Alfalfa mosaic virus infection of tobacco plants and accompanying electrical signatures were the subject of the latest research [37]. Electrical signals in uninfected and infected plants were examined. Precise techniques for feature selection (median, autoregressive coefficients, and autocorrelation) and machine learning models, including support vector machine, k-nearest neighbours, and random forest, were selected [37]. Computational techniques facilitated extraction of electrical signals indicating virus infection even before visual symptoms appeared. Moreover, an easy-to-use tool was built which, after improvement, will contribute to fast and inexpensive diagnosis of virus infection in the future, enabling timely interventions in the crop [37].

Another problem in agriculture is caused by parasitic nematodes that invisibly parasitize in the soil [38]. Analysis of electrical signals generated by uninfected, infected, and timely treated tomato plants was carried out. It was shown that the analysis of electrical signals effectively indicates the condition of plants in real time and facilitates therapeutic intervention [38].

As shown above, studying the electrical signature allows detection of early-stage infections by fungi [36], viruses [37], and parasitic nematodes [38]. In the future, this may help to reduce the use of chemicals, such as fungicides and nematicides, for pest control. Therefore, understanding the plant electrome may aid in making adequate decisions about the management of chemicals and support the production of healthier food in future.

5. Software for Time-Lapse Video Analysis for Investigation of Organ Movements in Plants

The studies on plant electricity presented above show that the phenomenon is related to plant responses to various environmental stimuli. In animals, electrical activity regulates many physiological processes and is inextricably linked to the induction and modulation of motor activity. Comparative analyses between the electrical activity of plants and their ability to move have been conducted for many years [7,8]. The best studied plants in this area are model plants Dionaea muscipula [17,39,40,41] and Mimosa pudica [42,43], which exhibit easily observable rapid movements of the trap and complex leaf. Recently, thanks to the rapidly developing techniques of recording and analysing digital video images, new possibilities have emerged in studying the biology of such slow processes as plant movements and growth/development (Table 2).

The brilliantly developing time-lapse video recording technique is currently at the forefront of research into the movements of plant organs and video imaging of growth and development. Digital video image processing techniques contribute to the objectivity, repeatability, and acceleration of scientific research [44]. The basic steps of the bioimage analysis workflow include preprocessing, segmentation, feature extraction, object tracking, and classification—these algorithms use machine learning and deep learning to streamline the operations [44].

Investigations of plant movement and growth/development start with the study of cell biology and video imaging [45,46,47,48]. Then, increasingly complex physiological processes can be imaged including organ (leaves, flower stem, roots) movement and growth/development [19]. The time-lapse video method, which records and visualises plant slow movements and growth, is invaluable in these studies. Specifically, there is a close link between elongation growth and slow, endogenous, rhythmic movements called nutations [49,50].

The slow nutational movements of various plant organs are being studied increasingly often. One of the first software applications for analysing nutational movements was the Circumnutation Tracker built by our team [51]. The software was built for semi-automatic analysis of standardised parameters of circumnutation movements. It was tested mainly on sunflower plants (Helianthus annuus) but has applications for analysis of the movements of various organs in different plant species [51]. In previous studies, rhythmic leaf movement and growth in Arabidopsis thaliana, Petunia hybrida, and Solanum lycopersicum plants were analysed by an inexpensive infrared vision system called OSCILLATOR [52]. Characteristic patterns of movement and growth were determined for day/night conditions [52].

Table 2. Software for time-lapse video analysis for investigation of organ movements in plants.

Software Name	Plant	Types of Movement	Literature (Year)
OSCILLATOR	Arabidopsis thaliana Petunia hybrida Solanum lycopersicum	rhythmic leaf movements and growth	[52] (2012)
Circumnutation Tracker, https://circumnutation.umcs.lublin.pl/ct, accessed on 6 April 2025	Helianthus annuus various plant species	analysis of the movements of various organs	[51] (2014)
3D stereovision machine system	beans	nutation movements of climbing plants	[53] (2024)
PALMA (Plant Leaf Movement Analyzer), https://sourceforge.net/projects/palma-leafmov/ accessed on 6 April 2025	Arabidopsis thaliana	periodic movements of leaves	[54] (2017)
TRiP (Tracking Rhythms in Plants), http://github.com/KTgreenham/TRiP, accessed on 6 April 2025	Arabidopsis thaliana Brassica rapa Glycine max Cleome violacea Solanum lycopersicum Mimulus guttatus	whole-plant images and periodic movements of cotyledons and leaves	[55] (2015)
Oskam et al. system, https://github.com/Pierik-Lab, accessed on 6 April 2025	Arabidopsis thaliana	nastic movements and growth of leaf as a part of the shade avoidance response	[56] (2024)
Rehman et al. system	Arabidopsis thaliana mutants	movements of plant leaves	[57] (2020)
Mao et al. tracker	Arabidopsis thaliana	circumnutation of flowering shoot apex	[58] (2023)
SLEAP (Social LEAP Estimates Animal Poses), https://zenodo.org/record/5764169#.YbCK0_FBxqt https://doi.org/10.5281/zenodo.5764169, accessed on 6 April 2025	Arabidopsis thaliana sunflower bean	circumnutations, tropisms, twining	[59] (2022)
Gibbs et al. system	Triticum aestivum	wind-induced plant movement in field-grown wheat	[60] (2019)

Table 2. Software for time-lapse video analysis for investigation of organ movements in plants.

Software Name	Plant	Types of Movement	Literature (Year)
OSCILLATOR	Arabidopsis thaliana Petunia hybrida Solanum lycopersicum	rhythmic leaf movements and growth	[52] (2012)
Circumnutation Tracker, https://circumnutation.umcs.lublin.pl/ct, accessed on 6 April 2025	Helianthus annuus various plant species	analysis of the movements of various organs	[51] (2014)
3D stereovision machine system	beans	nutation movements of climbing plants	[53] (2024)
PALMA (Plant Leaf Movement Analyzer), https://sourceforge.net/projects/palma-leafmov/ accessed on 6 April 2025	Arabidopsis thaliana	periodic movements of leaves	[54] (2017)
TRiP (Tracking Rhythms in Plants), http://github.com/KTgreenham/TRiP, accessed on 6 April 2025	Arabidopsis thaliana Brassica rapa Glycine max Cleome violacea Solanum lycopersicum Mimulus guttatus	whole-plant images and periodic movements of cotyledons and leaves	[55] (2015)
Oskam et al. system, https://github.com/Pierik-Lab, accessed on 6 April 2025	Arabidopsis thaliana	nastic movements and growth of leaf as a part of the shade avoidance response	[56] (2024)
Rehman et al. system	Arabidopsis thaliana mutants	movements of plant leaves	[57] (2020)
Mao et al. tracker	Arabidopsis thaliana	circumnutation of flowering shoot apex	[58] (2023)
SLEAP (Social LEAP Estimates Animal Poses), https://zenodo.org/record/5764169#.YbCK0_FBxqt https://doi.org/10.5281/zenodo.5764169, accessed on 6 April 2025	Arabidopsis thaliana sunflower bean	circumnutations, tropisms, twining	[59] (2022)
Gibbs et al. system	Triticum aestivum	wind-induced plant movement in field-grown wheat	[60] (2019)

There are also studies of circumnutational movements of beans using stereovision [53]. The movement of the shoot tip was tracked in a 3D space, and reference was made to the support structures and their influence on the dynamics of the plant movement [53].

Studies of the rate of growth, development, and periodic endogenous movements were undertaken in the model plant Arabidopsis thaliana. To study the periodic movements of Arabidopsis thaliana leaves, the video system developed works on the basis of time-lapse photography and using fast Fourier transformation and nonlinear least squares fitting [54]. This system (PALMA, Plant Leaf Movement Analyzer) can effectively and automatically capture changes in the environment, such as iron deficiency, by detecting the extension of the period of leaf movement rhythm [54].

Also for Arabidopsis thaliana leaf movements, a new method for estimating the biological period was developed using a motion estimation algorithm that can be applied to whole-plant images (TRiP, Tracking Rhythms in Plants) [55]. The new system efficiently tracks the movement of cotyledons and leaves without having to select individual leaves for analysis [55]. The new method was also used to estimate the period of movement for five different plant species, underlining its broad applicability [55].

It should be emphasised that systems for video imaging of growth and plant organ movements are relatively low-cost in relation to the amount of information they can provide [61]. Such a system was proposed by Oskam, Snoek, et al. [56] to analyse leaf movements and elongation in relation to lighting conditions and especially the influence of infrared light. These movements, called nastic movements, are part of the shade avoidance response and are important in the processes of adaptation to changing light conditions and regulation of leaf temperature and hydration [56]. These factors are extremely important for plant cultivation, which is why this type of research is important for future crop cultivation.

The effects of external (e.g., abiotic stress) and/or internal (e.g., gene mutation) disturbances on plant growth can be approximated by analysing the movements of plant leaves [57]. A dense optical flow algorithm was used to measure movement directly rather than detecting the leaf tip or cotyledon in each image. The authors [57] tested wild-type and drought-tolerant Arabidopsis thaliana mutants and applied two water levels and two nitrogen levels. The video method effectively distinguished environmental and genotypic differences in the plant response. These studies attempt to compare video movement patterns with real environmental conditions and indicate that video monitoring is suitable for plant phenotyping [57].

Also, a system for studying Arabidopsis seedling circumnutation has been recently developed [58]. A deep learning-based model was proposed to track the circumnutation of the flowering shoot apex in Arabidopsis thaliana from time-lapse videos. U-Net was used for vertex segmentation and combined with a model updating mechanism and pre- and post-processing steps, which significantly improved the efficiency and correctness of the analysis system [58].

The rapidly developing procedures of convolutional neural networks used to estimate the position of a human or animal are being introduced to estimate the dynamics of plant growth and movement (SLEAP, Social LEAP Estimates Animal Poses) [59]. These procedures are particularly accurate in tracking lateral views of shoots and roots of plants. These algorithms open up new possibilities for fast research focusing on plant dynamics [59].

Recently, work has emerged demonstrating wind-induced plant movement as an important factor in most agricultural plant crops [60,62]. A strong method for characterising movement in field-grown wheat (Triticum aestivum) plants is presented based on time-ordered image sequences and training a convolutional neural network. The authors [60] present an automated data extraction that can be used to inform lodging models, breeding programs, and link movement characteristics to canopy light distribution and dynamic light fluctuations [60].

Currently, algorithms are also used to track the velocity of plant root growth using high-resolution microscopic image sequences [63]. An observation and analysis system called RTip is helpful in tracking the root tip with transient perturbations and is used for plant phenotyping [63]. The RTip tracker is an example of the possibility of video tracking of plant root growth in laboratory conditions. The topic of video tracking of root growth is not addressed in the present review.

6. Discussion

6.1. Electrical Signals–Signatures

6.1.1. Abiotic Factors

Studying the plant electrograms and thus electrical signature (Table 1) that accompanies water status changes [33,34,35] and nutrient deficiencies [28] may help in the future to save water resources and nutrient and mineral fertilisers that are necessary during plant growth. Research on electrical signals may also support the assessment of the perception of environmental stimuli such as wind, heat, light, and pollution [30,31,32]. Particularly promising is the role of whole plants as biosensors and, together with the electrical signatures they produce, whose specificity is classified using statistical and AI methods [31,35], they may support the maintenance of a sustainable environment in the future.

6.1.2. Biotic Factors

As shown above (Table 1), studying the electrical signature may contribute to detection of early-stage infections by fungi [36], viruses [37], and parasitic nematodes [38]. In the future, this may help to reduce the use of chemicals, such as fungicides and nematicides, for pest control. Therefore, understanding the plant electrome may aid in making adequate decisions about the management of chemicals and support the production of healthier food in future.

6.1.3. Statistic and AI Analysis for Studying Plant Electrograms

In this work, Table 1 presents the various methods currently used to analyse plant electrograms. Analysis of these methods allowed to distinguish the key steps currently used in research. They are summarized in

Table 3. Key steps in plant electrograms analysis (on the basis Table 1).

Key Steps	Statistics/AI Analysis
data preparation and preprocessing	average value sample space reduction data
feature extraction and signal processing	signal decomposition
	statistical features (arithmetic average, skewness, kurtosis, probability density function (PDF), autocorrelation, power spectral density (PSD))
	transforms and entropies (fast Fourier transform (FFT), approximate entropy (ApEn), multiscale approximate entropy (ApEn(s))
statistical analysis of variation and differences	descriptive statistics, median, one-way ANOVA, Duncan’s multi-range test, Bayesian change point (BCP) analysis
multivariate analysis	principal component analysis (PCA)
machine learning methods (ML)	classification and regression (support vector machine (SVM), k-nearest neighbours (KNN), random forest (RF), discriminant analysis, different classification algorithms)
	predictive models (construction of regression and classification models using selected features: median, autoregressive coefficients, autocorrelation)
	other ML techniques (cluster analysis)
deep learning (DL)	lightweight convolutional neural network (lightweight CNN, e.g., PlantNet), deep-learning methods: neural networks with various architectures
automation and toolchains	automatic toolchain, integration of ML models within automated tools

and can be taken into account in the design of further studies.

The above analysis is based on a small number of studies conducted mainly in laboratory conditions. Terms of accuracy, computational cost, and practical applicability should be taken into account in future investigation.

6.2. Time-Lapse Video

As evidenced by the works presented above, considerable progress has been made in the methodology of studying movements and growth mainly in the model plant Arabidopsis thaliana [52,54,55,57,58]. Work is in progress on cultivated plants, such as sunflower [51], bean [53], wheat [60,64], and tomato [52]. Some video systems for studying the growth and movement of plant organs are universal for many species [51,52,55,59].

The current techniques used in modern basic science may find application in smart agriculture in the future. Non-invasive and relatively cheap imaging also deserves to be emphasised. There is increasing interest in vision systems involving time-lapse video imaging and deep learning algorithms that could be used in future agricultural activities [65,66]. The research presented above shows how the time-lapse video technique can support investigations of plant growth. The development of these methods in the future will provide better solutions to the problems of modern plant cultivation.

6.3. Advantages and Difficulties

It should also be added and emphasised that the presented methods of measuring electrical and motor activity are relatively cheap and not burdensome for the environment in relation to the increase in new information and knowledge about plant growth/development. Based on the analysed literature, it can be concluded that the research on electrical and motor activity of plants is associated with not only new application possibilities but also many difficulties to be solved (Figure 4).

6.3.1. Advantages and Possibilities

• Application of measurement methods in laboratories, greenhouses, and field conditions;
• Successful attempts to record and analyse some electrograms and time-lapse videos in field conditions;
• Optimisation of universal measurement and analytical methods aimed at crop plants (not only the model plant Arabidopsis);
• Relatively cheap and either not at all or minimally invasive methods;
• Detection of stress and impact on the accuracy of biomass estimation;
• Real-time monitoring of environmental factors and stresses;
• Support for the decision-making process in agriculture in real time.

Difficulties and Problems

• Calibration of measurement and analysis to a specific plant species and environmental factors (currently a large variety of measurement and analytical methods that are incomparable (Table 1 and Table 2));
• Determination of electrogram and time-lapse video parameters;
• Elimination of artifacts and noise;
• Researcher supervision during method introduction;
• Impact of field and atmospheric conditions;
• Electrodes resistant to field conditions;
• Wireless connections enabling remote sensing;
• Selection of appropriate and standardised AI analyses;
• Integrated simultaneous bioelectrical studies and organ movement estimation;
• Standardisation of recording methods, electrode placement, sampling frequency, and sampling time during the day, taking into account the plant growth phase.

7. Conclusions

Various types of methods for measuring changes in electrical voltage have potential in monitoring the physiological state of plants and diagnosing plant diseases. Specific electrical signatures occurring in association with a specific type of environmental stress can be used in the future in early detection of stress in plants even before visual symptoms appear. These studies are promising but require further work on the interpretation and standardisation of measurements. These relatively cheap environmental sensors—electrodes and digital video cameras—facilitate relatively cheap monitoring of the condition of a plant or a cultivated field. Their advantage is non-invasiveness or, in the case of electrodes, very low invasiveness. A huge advantage of such electrical and visual systems is the possibility of continuous real-time monitoring of the condition of plants or the environment. Such systems will facilitate decision-making related to irrigation or the use of nutrients or chemical plant protection substances. This application will be particularly precise, which will affect the health of the plants themselves and the soil environment. Equally important will be more economical calculations resulting from such precisely managed crops [67]. In the future, specific electrical signatures together video patterns of movement and growth associated with real environmental conditions could support decisions about plant cultivation.

8. Future Directions

Currently, precision agriculture is in its early stages of development. The ongoing cooperation with scientists dealing with plant biology will contribute to a proper transfer of modern solutions and technologies to farming practice in future. Most of the time-lapse analyses have been performed on Arabidopsis thaliana, which indicates the need to extend the research to other plants important in food production processes, as in the case of the recently undertaken research on wheat, sunflower, and bean. Similarly, the study of specific electrical signatures and the whole electrome in plants is promising but requires improvement of procedures and expansion of its scope. Threats to human health and the environment are related to the still insufficient knowledge of ecological, physiological, social, and economic processes, and therefore the use of inappropriate solutions in practice. Observing natural processes by basic science and applying technical solutions that are in accordance with the laws of nature will help to improve human daily lives in the future. Attempts to apply both artificial intelligence and precision agriculture are bringing new solutions towards creating healthier food and environments, and therefore healthier lives in the future.