ML based sustainable precision agriculture: A future generation perspective

doi:10.1016/j.suscom.2020.100439

Sustainable Computing: Informatics and Systems

Volume 28, December 2020, 100439

https://doi.org/10.1016/j.suscom.2020.100439 Get rights and content

Highlights

•
An architectural representation of ML models for addressing the agricultural issues is presented.
•
The study focusses on the various issues in the agricultural field.
•
Various ML algorithms are discussed to handling different type of data is discussed.
•
The advanced approaches to overcome is presented in a precise manner.
•
The gap in research and application is presented.

Abstract

Agriculture is an essential source of survival and also accounts for the economic growth of any country. Rapid advancement in precision agriculture has helped agriculture and its ancillary enter into the era of machine learning and big data technologies. The advanced technology would help in the improvement in various dimensions of the agro-sector. The agro science community needs to create its architectural model to enjoy the advantages of parallel computing and storage management of large datasets, which would help to discover novel analytical structure to extract useful information from the data patterns. These patterns would help understand the field and various issues and also help in identifying the solutions to eradicate multiple problems. Machine learning offers favorable computational as well as analytical solutions for the integrated study of different types of datasets from various sources. This study introduces the core concept of machine learning and systematic processes to comprehend its application in agriculture. It also discusses various machine learning algorithms that can be utilized to build models to address various agricultural issues.

Introduction

Agriculture is an essential source of survival and is also responsible for the economic growth of any country. To address the various problems related to agriculture, complex agricultural ecosystems need to be understood [1]. It can be done through modern digital technologies that continuously monitor the physical environment, producing massive data at unprecedented speeds. Analyzing this data will help farmers and agricultural companies to increase crop production by extracting valuable information [2]. Although big data technology has seen significant progress in various fields, it is still in its growing phase in agriculture, and many obstacles need to be overcome [3].

Precision agriculture is a practice where the agricultural activities are carried out accurately and controlled so that the crop yield can be maximized without exploiting the available resources. To maximize crop yield, fields are deployed with sensor nodes, which are used to collect data related to the farm to perform analytical processing of the collected data. These nodes are connected to the base station, which sends the collected data to the final server from where the data is moved to various data storage and data processing platforms. Further, data processing is done, and machine learning algorithms are applied to gain insight within the data. The analysis helps develop farmer-friendly and interactive GUI, through which useful farm-related tips and instruction are provided. Various research and business analytics are also carried out with the help of machine learning.

This paper focuses on the overall architecture of the precision agriculture model as shown in Fig. 1, and also discusses the various machine learning algorithms which can be utilized to address multiple agricultural problems. This study will help identify common issues encountered in developing any model, the problem that is already addressed, and the area that needs more attention.

People have collected a large volume of environmental data through the years of observation and learning and have helped in the in-depth investigation, fluctuation-estimation, and unusual pattern recognition of the climatic conditions. Despite endless efforts and analysis, there is a limitation to human intelligence and capabilities; hence people developed artificial intelligence so that the complex and the conglomerate behaviour of the dataset could be studied precisely. With the advancement in computing technologies, the development of artificial intelligence also started [4]. It can be thought of as the large universe of which analytics, statistics, reasoning, and probabilistic calculations form the popular subparts. Machine learning is also one of the subsections of artificial intelligence, which can be considered as a superset of statistical calculations [5]. The major difference between machine learning and statistics is that the statistical models find the solutions based mainly on the likelihood. In contrast, in machine learning, no prior assumptions of the results are made, and the results are calculated based on a set of rules. Apart from this, other step and approaches in both cases are similar. The strategy for finding the results can be commonly termed as a process of estimation.

According to many analysts and scientists, machine learning works on the fundamental principles of estimation of all the data characteristics and values. For evaluation of the results, it does not take into account the number of occurrences of the terms or the past results since the basic principle of estimation in machine learning is based on the structures and patterns in the dataset. Hence, for developing any model, a new approach is considered, and it does not depend on any previously developed model. Therefore it only takes into account a set of examples from which it infers the information and structure of data and learning following the set of input and corresponding output provided to the model. Fig. 3illustrates some machine learning algorithms; like artificial intelligence, machine learning can be divided into three subcategories depending on the data processing approaches.

In supervised learning, a set of examples of input and corresponding outputs are feed into the algorithm or the machine such that the model gets trained from the example sets. The goal of the algorithm is to find the deterministic characteristic that maps each input to the related output, focusing on minimizing the error so that accurate prediction or the forecast can be made. Classification and regression techniques are prime examples of supervised learning [6]. In contrast, unsupervised learning is an approach in which informational measures are used for setting up the learning procedures that do not include predefined labels. Also, the training datasets contain just the information and a cost function, which decreases throughout the training phase. Clustering, forecasting future inputs, decision making, and dimensionality reduction are a few examples of unsupervised learning. In addition to this, the inter-machine and interaction of the machine with the environment is done using some action. In such a situation, the machine is termed as agent, and the surroundings in which it works are termed as the environment. The machine tries to use actions whereby the maximum interaction is stabilized with the environment, which helps reduce errors in the final output. These support calculations or the actions are considered as a critical group of estimations that could be utilized for improvements of the ideas related to dynamic programming.

The paper organization in the following sections is as follows. Section 2 discusses the various factors influencing agriculture, and the general architecture is also presented. Section 3 discusses the evaluation of the factors affecting crop yield. The working of various machine learning techniques are discussed in Section 4. Section 5 discusses the future challenge and also presents suggestions. Finally, the present study is concluded in Section 6.

Section snippets

Factor influencing agriculture

As reported, the world's surface temperature has expanded to 0.89 $°$ C in the last century [7]. There is an estimate of the excessive climate crisis arising due to global warming. The changing environment and climatic fluctuations can cause tremendous difficulties in agriculture. The changing climate is among the primary deciding factor for affecting crop production and food protection for the future generation. Moreover, the agriculture in India is weather dependent and susceptible since a large

Evaluation of various parameters affecting crop yield

It is a known fact that the economic conditions and the living standards of any farmer mainly depend on the harvest obtained by the crop. Also, it is a basic fact and can be inferred to develop and strengthen any decision support system. There is a need to study the relationship between the environmental factors and other yield influencing characteristics [14], [15]. Most of the time, the relationship between the various elements is complicated and has higher unpredictability in general. For

Machine learning tools and applications in agriculture

A prominent subpart of artificial intelligence is machine learning, which can adapt and improve itself by training without significant modifications. Although machine learning is a subpart of AI, it is very much different in terms of working. In the machine learning approach, the model is developed by training the algorithm based on the available data, such that accurate forecasts can be made. Despite the fact the machine learning is a sub-domain of artificial intelligence, this area itself

Future challenges

Since the Indian economy is dependent on agriculture and has varied climate in different parts of the country, it is essential to incorporate and use advanced methodologies and techniques to explore the various resources for increasing the productivity. Various government policies have been made for the development of the agriculture sector. New machinery for plowing fields, watering crops, and harvest are being used in various parts of the country. The various innovations have the capability

Conclusion

Machine learning models provide a great way to analyze the data collected on a large scale and help in acquiring the information gathered to generate a progressive and important process. These technologies are an ideal way to develop different profound models to introspect the relation between various factors and their influence. Hence they can be used for multiple forecasts and predictions depending on the particular conditions. The focus is to present the overview, and the potentials of

Declaration of Competing Interest

The authors report no declarations of interest.

References (61)

D. Elavarasan et al.
Forecasting yield by integrating agrarian factors and machine learning models: a survey
Comput. Electron. Agric.
(2018)
J.H. Lee et al.
Machine learning: overview of the recent progresses and implications for the process systems engineering field
Comput. Chem. Eng.
(2018)
E.P. Xing et al.
Strategies and principles of distributed machine learning on big data
Engineering
(2016)
P. Ramos et al.
Automatic fruit count on coffee branches using computer vision
Comput. Electron. Agric.
(2017)
S. Amatya et al.
Detection of cherry tree branches with full foliage in planar architecture for automated sweet-cherry harvesting
Biosyst. Eng.
(2016)
S. Sawant et al.
Interoperable agro-meteorological observation and analysis platform for precision agriculture: a case study in citrus crop water requirement estimation
Comput. Electron. Agric.
(2017)
S.K. Dubey et al.
Assessment of climate change impact on yield of major crops in the Banas river basin, India
Sci. Total Environ.
(2018)
J.A. Mathieu et al.
Assessment of the agro-climatic indices to improve crop yield forecasting
Agric. For. Meteorol.
(2018)
A. Chlingaryan et al.
Machine learning approaches for crop yield prediction and nitrogen status estimation in precision agriculture: a review
Comput. Electron. Agric.
(2018)
G.E. Ogutu et al.
Probabilistic maize yield prediction over East Africa using dynamic ensemble seasonal climate forecasts
Agric. For. Meteorol.
(2018)

B. Krawczyk et al.

Ensemble learning for data stream analysis: a survey

Inf. Fusion

(2017)

C. Voyant et al.

Machine learning methods for solar radiation forecasting: a review

Renew. Energy

(2017)

A. Singh et al.

Machine learning for high-throughput stress phenotyping in plants

Trends Plant Sci.

(2016)

I. Portugal et al.

The use of machine learning algorithms in recommender systems: a systematic review

Expert Syst. Appl.

(2018)

R.H. Ip et al.

Big data and machine learning for crop protection

Comput. Electron. Agric.

(2018)

M. Sirsat et al.

Classification of agricultural soil parameters in India

Comput. Electron. Agric.

(2017)

D. Li et al.

Prediction analysis of a wastewater treatment system using a Bayesian network

Environ. Model. Softw.

(2013)

G. Kou et al.

Evaluation of clustering algorithms for financial risk analysis using MCDM methods

Inf. Sci.

(2014)

W. Xing et al.

A novel digital analysis method for measuring and identifying of wool and cashmere fibers

Measurement

(2019)

R. Rupnik et al.

Agrodss: a decision support system for agriculture and farming

Comput. Electron. Agric.

(2019)

S. Wang et al.

Crop type mapping without field-level labels: random forest transfer and unsupervised clustering techniques

Remote Sens. Environ.

(2019)

C.E. Ndehedehe et al.

Is terrestrial water storage a useful indicator in assessing the impacts of climate variability on crop yield in semi-arid ecosystems?

Ecol. Indic.

(2018)

S. Park et al.

Drought assessment and monitoring through blending of multi-sensor indices using machine learning approaches for different climate regions

Agric. For. Meteorol.

(2016)

K.L. Roldán-Serrato et al.

Automatic pest detection on bean and potato crops by applying neural classifiers

Eng. Agric. Environ. Food

(2018)

M.T. Harrison et al.

Modelling the sensitivity of agricultural systems to climate change and extreme climatic events

Agric. Syst.

(2016)

Y. Dash et al.

Rainfall prediction for the Kerala state of India using artificial intelligence approaches

Comput. Electr. Eng.

(2018)

M.E. Holzman et al.

Early assessment of crop yield from remotely sensed water stress and solar radiation data

ISPRS J. Photogramm. Remote Sens.

(2018)

J. Rhee et al.

Meteorological drought forecasting for ungauged areas based on machine learning: using long-range climate forecast and remote sensing data

Agric. For. Meteorol.

(2017)

R. Whetton et al.

Nonlinear parametric modelling to study how soil properties affect crop yields and NDVI

Comput. Electron. Agric.

(2017)

W. Wieder et al.

Comparison of soil strength measurements of agricultural soils in Nebraska

J. Terramech.

(2018)

Cited by (29)

Advancements in variable rate spraying for precise spray requirements in precision agriculture using Unmanned aerial spraying Systems: A review
2024, Computers and Electronics in Agriculture
Pesticides suppress pest populations and maintain agricultural yield; however, their overuse causes ecological damage. To mitigate the ecological damage caused by the overuse of pesticides, this paper explores the application of a precise and adaptable technique known as unmanned aerial spraying system (UASS)-based variable rate spraying (VRS). Herein, the current state of precision agriculture is examined, and the application of UASS in variable rate spraying is discussed. Then, the role of advanced sensors, including multispectral and hyperspectral technologies, in optimizing UASS-based VRS missions is studied, followed by the challenges in UASS-based spraying, ranging from technical intricacies to regulatory considerations, and related solutions. In addition to delving into pesticides, the paper explores alternative solutions such as herbicides, encompassing an integrated approach that aligns with sustainable farming practices for more effective pest management. VRS has advantages, including increased yields, reduced resource usage, and environmental sustainability. This paper concludes by delineating current challenges and envisioning future innovations, spotlighting ongoing studies on sensor technology and AI-driven flow rate optimization. This review provides insights into UASS-based VRS for precision agriculture.
Harvesting a sustainable future: An overview of smart agriculture's role in social, economic, and environmental sustainability
2024, Journal of Cleaner Production
As climate change and population growth intensify, the agricultural sector's need for sustainable solutions is paramount. This paper presents an overview of smart agriculture, a strategy leveraging technology and data analytics, and its potential to advance sustainability socially, economically, and environmentally. Drawing on recent research, we examine the integration of smart technologies in agriculture, their impact on sustainability, and the architectural design and key factors influencing their adoption. We critically review the literature on the effectiveness of smart agriculture, illuminating how smart farming strategies can curb environmental harm, stimulate economic growth, and promote social inclusivity. We identify existing research gaps and propose directions for future investigation, emphasising the need for rigorous studies on implementation strategies and long-term impacts of smart agriculture on rural landscapes. Our comprehensive overview aims to engage and guide diverse stakeholders, enhancing understanding of sustainable agriculture practices and encouraging more sustainable farming implementations. This paper contributes to global efforts towards harvesting a sustainable future.
Data-efficient crop pest recognition based on KNN distance entropy
2023, Sustainable Computing: Informatics and Systems
With the development of deep learning technique, the intelligent pattern recognition studies mainly focus on the optimization of models but ignore the impact of data quality for a long time. From our viewpoint, data quality is very important to the sustainability of intelligent applications, in terms of task performance, power consumption and training time. In this paper, we proposed a novel data quality assessment method, called k-nearest neighbors distance entropy (KNN-DE), to evaluate the crop pest images and screen high informative data to complete the data-efficient pest recognition task. Many comparative experiments were carried out, and the results show that the high informative data selected by the proposed KNN-DE method outperform other related works. In specific, under the data budget of 100 samples per category, the performance of the model trained with high informative data is 17 % higher than that trained with low informative data. In addition, by referring to the model accuracy trained with 100 low informative samples per class, the alternative selection of high informative data can realize the improvement of data efficiency and training time efficiency by 2.5 times. Thus, this paper provides a data-centric research perspective, lays a foundation for data quality assessment, and makes a positive attempt to explore the data-efficient sustainable learning.
IoFT-FIS: Internet of farm things based prediction for crop pest infestation using optimized fuzzy inference system
2023, Internet of Things (Netherlands)
Citation Excerpt :
Zhang et al. (2017) proposed a comparative study of various machine learning models to find pest occurrence regions [61]. Machine learning is useful for data analytics to generate automated predictive models [5,62]. The field of computer science deals with pattern recognition using the computer learning theory of artificial intelligence.
Advanced farming techniques help to know the appropriate environmental conditions, soil quality, water and fertilizer needs, and crop monitoring during each plant’s growth phase, resulting in higher yields. Also, IoT-compatible crop monitoring and data collection systems can help identify crop diseases and the breeding status of pathogenic pests. Extracting data from samples collected from the complete crop growth cycle acknowledged a plausible relationship between weather parameters, crop yield, and insect reproduction. Data analysis done on the data collected through an IoT monitoring system helped determine the environmental factors supporting the higher pest breeding conditions. The proposed fuzzy inference system’s knowledge base is designed using these weather parameters. The multi-objective evolutionary algorithm uses fuzzy rules to find suitable cropping window and low pest breeding conditions. This proposal identifies crop-sowing windows based on fuzzy logic with maximum crop yield and minimum pest growth by deploying IEEE 802.15.4 wireless IoT-enabled sensor network monitoring infrastructure in medium grass vegetation. Experiments are being done on rice and Sugarcane crops. The experiments were conducted in the agriculture field of Gwalior, Madhya Pradesh, India. The soil moisture, rainfall, temperature, etc., data were collected using the wireless sensor network deployed in the field. Fuzzy logic-based identification of appropriate planting seasons by IoT application development services helps farmers prevent the development of pests and proactively take precautions to achieve maximum yields.
An intelligent IOT sensor coupled precision irrigation model for agriculture
2023, Measurement: Sensors
Citation Excerpt :
The three basic subcategories of ML approaches are reinforcement learning, unsupervised learning, and supervised learning. Dimensionality reduction, Anomaly detection, and cluster are examples of unsupervised learning applications, while regression and classification fall under the category of supervised learning [15,16]. Traditional irrigation methods [17] lead to field runoff and contamination of freshwater resources because they apply too much water.
Sustainable agricultural practices are required, especially for irrigation, to provide for a growing population. About 85% of the freshwater resources in the world are used for irrigation. "Us," current irrigation procedures must be either updated or replaced with cutting-edge, intelligent systems that utilize the ML, IoT (Internet of Things), and sensor networks which are wireless. In the proposed paper, a brainy system for tracking and scheduling accuracy irrigation using IoT, a LoRa-based machine learning (ML) is being introduced. We developed an independent irrigation technique that would provide the tomato and eggplant with the precise amount of water they required was set up using the information obtained from the soil moisture sensors. In comparison to traditional watering, which required 7541 mL for the banana plant and 8755 mL for the rice plant, the irrigation system irrigated the plants with an overall quantity of 5010.745 mL & 4421.09 mL in one month. Overall, this led to a 46% reduction in water usage, and the plants looked better than they would have with conventional watering. The simulation results clearly show that the suggested approach uses water more sensibly than cutting-edge models in the experiment farming region.
Environment: role of precision agriculture technologies
2023, Precision Agriculture: Evolution, Insights and Emerging Trends
Agriculture is one of the main contributors to the environmental pollution in the past few decades. Overapplication of agrochemicals on various components of the environment such as surface water, groundwater, and air has forced the researchers and farmers to adapt precise agriculture technologies to ensure that along with environment, crop yields are also not being negatively affected. The combination of advancements in precision agriculture technologies development with application of artificial intelligence algorithms in agriculture decision making will help scientists and farmers to develop and implement decision support systems for early warning applications for weather and disease/pests attacks. This could lead to reducing the negative impacts of agriculture industry on environment through enforcing and adopting mitigative strategies to minimize effects of climate change on crop production. Finally, how future climate changes could impact agricultural practices and what are the roles of precision agriculture technologies in mitigating the environmental pollution risks are discussed.

View all citing articles on Scopus

^☆: This work was supported by the Science and Research Board (SERB), Govt. of India with the grant number ECR/2017/001273. The authors also wish to express their gratitude and heartiest thanks to the Department of Computer Science & Engineering, Indian Institute of Technology (ISM), Dhanbad, India for providing their research support.

View full text

ML based sustainable precision agriculture: A future generation perspective☆

Highlights

Abstract

Introduction

Section snippets

Factor influencing agriculture

Evaluation of various parameters affecting crop yield

Machine learning tools and applications in agriculture

Future challenges

Conclusion

Declaration of Competing Interest

Comput. Electron. Agric.

Comput. Chem. Eng.

Engineering

Comput. Electron. Agric.

Biosyst. Eng.

Comput. Electron. Agric.

Sci. Total Environ.

Agric. For. Meteorol.

Comput. Electron. Agric.

Agric. For. Meteorol.

Inf. Fusion

Renew. Energy

Trends Plant Sci.

Expert Syst. Appl.

Comput. Electron. Agric.

Comput. Electron. Agric.

Environ. Model. Softw.

Inf. Sci.

Measurement

Comput. Electron. Agric.

Remote Sens. Environ.

Ecol. Indic.

Agric. For. Meteorol.

Eng. Agric. Environ. Food

Agric. Syst.

Comput. Electr. Eng.

ISPRS J. Photogramm. Remote Sens.

Agric. For. Meteorol.

Comput. Electron. Agric.

J. Terramech.