View of A Survey on Crop Yield Prediction using Machine Learning

A Survey on Crop Yield Prediction using Machine Learning

Article History: Received: 11 January 2021; Revised: 12 February 2021; Accepted: 27 March 2021; Published

online: 4 June 2021

Abstract: . In a country within the economy of the state, the agricultural sector plays a vital role. several advances within the

field of agriculture have conjointly been enforced in recent years. Serious analysis is happening everywhere the planet with the event of IoT, massive knowledge technology and machine learnings regarding however farmers and even the govt. square measure learning a way to do this. whereas the fortification of agriculture may be classified and analyzed in several areas, like crop management, ground management, weed identification, water management, yield management, eutherian management, we'll think about crop yield management as a result of we have a tendency to believe it to be the surroundings within which an excellent deal of technical progress goes to permit the farmer to urge his crop to his right level. during this article, we'll study specially machine learning and profound learning ways of varied researchers likewise as their impact within the agriculture business on crop yield management.

Keywords: Precision Agriculture, Crop yield, machine learning, deep learning,

1. Introduction

Although farming work are often classified and analyzed in varied areas, like crop management, soil management, weed identification, water management, yield management, farm farming management, we'll specialize in crop yield management, as we believe that this is often the setting which will alter farmers to require advantage of their crops. This paper focuses on laptop and in-depth learning ways for various scientists and also the impact on crop yield management within the farming trade.[1]. Although farming is distinctive and analyzable in numerous areas like crop management, soil management, weed identification, water management, crop yield management, we'll target crop farming as we have a tendency to believe this is often the climate which will enable farmers to learn from their crops. This paper is regarding the machine and in- depth learning for varied researchers and the way crop yield management within the agriculture sector is influenced. The new technical methodologies within the agriculture sector should be incorporated to feed India's growing population. Moreover, farmers want prompt recommendation on crop productivity to develop favorable methods to reinforce crop yields. exactness agricul-2 true could be a technological methodology to confirm that soil and crops square measure supplied with what they need to deliver the goods most production and health. Precious agriculture uses sensors to capture period of time field and weather knowledge and to forecast that farmers can build the proper decision. The farm deploys little sensors that capture and transmit the info to the datastore node in question [2]. the info gathered square measure terribly massive, thus huge knowledge analysis will be accustomed method it. huge knowledge offers correct options like knowledge assortment, data transmission and knowledge analysis. this can conjointly favor farmers and eco- nomic development within the field of agriculture. [3]. The knowledge collected square measure terribly high and might be processed by huge data analyses. Exact functions like knowledge storage, knowledge transport and processing square measure provided by huge data. this can conjointly promote agricultural farmers and economic process. Years once year ML spreads to additional and additional totally different fields of agricultural activity, like plant management, crop production, soil protection, etc. within the past decade, cubic centimeters and IoT have demonstrated, together with huge knowledge Analytics, that this technological assistance improves improved management of the agricultural sector as a whole [6]. This paper presents an in-depth assessment of the employment of cubic centimeter in agricultural prediction management, usually of crop yield. many connected papers that highlight the most options of common ML models are given. This divides the composition of the rest of the document. In Section a pair of, we have a tendency to shall examine the agricultural industry's crop yield management as a literature study. Section three describes the Machine Learning half and connected work of Machine Learning in Agriculture Crop Yield prediction. Section four focuses on our proposed methodologies and in section five we have a tendency to had our conclusion.

2. LiteratureReview

One of the foremost vital predictions for preciseness agriculture, the yield forecast is of high significance to maximize production within the mapping of returns, yield estimates, reconciliation crop provide with demand and crop management. samples of metric capacity unit applications embrace the work of [7]; an efficient, cheap, and non-destructive method that counts occasional fruits mechanically on branches. The system measures the fruit of the occasional in 3 categories: harvestable, not harvestable and unsoundness. The system conjointly calculable the

that will optimize their economic gains and schedule their farm work. The developers of [8] during which they created a laptop vision technique for automation throughout harvest, are used for yield prediction. even though these don't seem to be evident, the segments of the system and notice occluded cherry boughs with an entire perform. The system's primary goal was to cut back manual gather and process necessities. Authors created associate early yield mapping technique in an- alternative study [9] to spot untimely inexperienced citrus in an outside citrus grove. like all alternative relative analysis, the report aimed to produce farmers with yield-specific data to assist them optimize the yield and profit growth. In another study [10], the authors elaborate on ANNs and multi-temporary remote sensing information to gauge parcel biomass (kg of dry matter /ha / day). In another study [11], another study specializing in yield prediction and particularly on the prediction of wheat yield was introduced. the tactic developed used satellite imaging and was consolidated with soil information for additional precise predictions. [12] the authors planned a method for detective work tomatoes supported EM images that were taken by associate aerial remote-controlled vehicle and remotely detected red inexperienced blue (RGB) (UAV). Also, the authors developed a rice stage statement system supported SVM and basic geographical details from China's earth science stations in the [13] work. Another analysis has planned a generalized technique for predicting agricultural yield [14]. The approach relies on a long agronomical information ENN application (1997– 2014). The analysis focuses on regional pre- dictions of farmers facilitate to stop worth offers and demand imbalance triggered or hastened by crop quality. in Taiwan particularly.

Table one summarizes the on top of papers for the case of yield prediction sub-category

Table 1. Outline of Crop Yield Prediction literature Article Crop Observed Features Functionality Algorithms Results

[4] Coffee Forty-two (42) colour options in digital picture illustrating coffee fruits

Automatic count of low fruits on a low branch SVM Harvestable: (1) Ripe/overripe: 82.54– 87.83% visibility percentage (2) Semi-ripe: 68.25–85.36% visibility percentage Not harvestable:

(1) Unripe: 76.91–81.39% visibility percentage

[5] Cherry Coloured digital

picture depicting leaves, branches, cherry

fruits, and the back- ground

Detection of cherry branches with full foliage

BM/GNB 89.6% accuracy

[6] Green

citrus

Image optiond

(form twenty nine twe nty pixels

digital pictures of unripe inexperienced c itrus

fruits) like coarseness, contrast,

directionality- ty,line-likeness, regu- larity, roughness, graininess, irregular- ty, brightness, smoothness, and fineness Identification of the amount of immature inexperience d citrus beneath natura l outside conditions

SVM 80.4% accuracy

[7] Grass Vegetation indices, spectral bands of red and NIR

Estimation of grass- land biomass (kg dry matter/ha/day) for 2 managed grass- land farms inIreland; Moorepark &Grange ANN/ANFIS R2 = 0.85 RMSE = 11.07 Grange: R2 = 0.76 RMSE = 15.35

[8] Wheat Normalized values of on-line forseen soil parameters and also the satellite NDVI

Wheat yield predic- tion within-field varia- tion

ANN/SNKs 81.65% accuracy

[9] Tomato High spatial resolu- tion RGB images Detection of toma- toes via RGB images cap- tured by UAV Cluster- ing/EM Recall: 0.6066 Precision: 0.9191 F-Measure: 0.7308

3. Machine Learning inAgriculture

To learn from the "experience" (training data), associate cubic centimeter resolution generallyinvolves a learning project. asetof in-stances consists ofcubic centimeter data. Usually, a collection offeatures or variables defines one specific example. The operator is also halfway, conditional (e.g., 0 or 1), ordinal (e.g., A+ or B+), or numerical (integer, real, etc.). associative improvement method, that is improved over time with observation, evaluates the consistencyofthemodelsforagiventask.Severalmany laptopsandnumericalmodel’sarea unit used to live the cubic centimeter models and algorithms' accuracy. The coaching knowledge are going to beused once’s the lecture, victimization the information nonheritable throughoutthe course to spot, forecast or cluster new examples (test data). Typical cubic centimeter as seen in figureone.

Fig. 1. Machine Learning Process

MLtasksare historically divided into several broad classes reckoning on the kindof education (supervised/unsupervised), models of learning (classification, regression, clustering, and dimensional reduction), and learning models used for the chosentask.

4. Related work victimizationMachine Learning inAgriculture

A review on the Crop Yield Prediction VictimizationGroundnut and Naive mathematiciantechnique within the field of information mining has been completed by Siju, H.L., & Patel, P. J. et al. [6]. Totally different data mining applications in horticulture is depicted right from the bat and and the study work on the groundnut yield expectation was examined at thisstageinconjunctionwithdNaiveBayesprocessupdatedforvariousapplications lastly. Creators also proposed that different knowledge mining techniques willbe used to build the more accurate model of groundnut crop yield.

The audits conducted by Chlingaryan, A., Sukkarieh, S., & Whelan, B. et al. [10] centered on AI processes, ROEs and Board matter, were mentioned. The analysis reveals the technique of the worth of context proliferation and its consistency in harvest expectations for various vegetation lists. above all, they show that gaussian ways square measure helpful for the prediction and determination of various qualities of plant leaves. The importance of M5-Prime Regression trees is usually monitored by creators as a technique of characteristics various perceptions of yield. Finally, the survey conjointly shows Fuzzy psychological feature Map (FCM), that is to be used for the model and illustration of master information for crop yield expectations. In comparison to type of data processing techniques like K suggests that and Support Vector Machinery (SVM), and multiple regression toward the mean (MLR) to administer high accuracy, K. Samundeeswari, K. Srinivasan et al.[11] was researched within the Krishna- giri District for the dirt data survey. The essential parameter wont to increase the event of crops is named the Sun during this article. Creators conjointly enclosed extension and improvement of jobs, victimization climate conditions and anticipation of harvests for the long run. The various connected properties just like the location, from that the pH scale of soil is resolved, square measure counteracted by Bhanumathi, S., Vineeeth, M., & Rohit, N. et al. [12]. additionally, the number of chemical element (N), element (P) and atomic number 19 supplements is additionally thought of (K). They used outside applications like climatical and temperature arthropod genus, soil type, further dirt assessment and precipitation calculation in this sector to deal with soil generation. Nevavuori, P.,

in Nursing Adadelta coaching algorithmic rule, SGD-force Associate in Nursingd RMSprop and updated the utilization of CNNs – an in-depth learning method, through the gathering of knowledge victimization unmanned Air Vehicles (UAVs), satellite and cam- era bundles, and space-filled lightweight. Creators assumed that the RMSprop was deceptive and thus prevented from testing. Adadelta outflanked the SGD power be- tween the 2 residual calculations and was taken to calculate the readiness for extra exams and This model are going to be developed in future so as to develop the mod- el for accuracy in conjunction with a broader set of highlights like climate and soil.

4.0 Proposed Methodology

Most of the Machine Learning and deep learning algorithms used for crop yield pre- diction had used terribly less parameters like solely rain therein region and space on that the crop production is meant to be taken. The rain and space get defeminated because the district or state gets modified. though the accuracy of all higher than model’s area unit sensible and in vary of higher than ninety fifth planning to ninety-eight however being solely 2-3 parameters that we have a tendency to known as the options had been known for analysis and predicting the results for crop yield prediction, however there's still is scope for improvement altogether such models of computer science.

Hence, we have a tendency to area unit proposing the machine learning model within which we'll work on different options of like rain, ph level, temperature, variety of soil, precipitation, Humidity which is able to offer United States precise prediction of crop yield therein explicit district supported all higher than options. we'll implement the assorted machine learning and deep learning model 1st to check and so by analyzing the most effective model supported performance

Metrix, we'll ready to offer the most effective prediction results. Figure a pair of represents the pro- display system design.

Fig. 2. System Architecture of Proposed System

5. Conclusion

We had tried to point out that milliliter models are applied in multiple applications of agriculture management like crop yield management. This trend within the application distribution reflects the knowledge-intense applications inside the crop and high use of im- ages and different data values associated with crop yield like precipitation, temp etc. numerous milliliter rule like RF, Cubist, soil samples was tested and prediction is completed consequently by numerous researchers.

It is additionally evident from the analysis that the majority of the studies used ANNand SVM milliliter models. a lot of specifically, ANNs were used principally for implementations for crop yield management and prediction. This leads the long run work for US to require a specific space of the Indian landmass and use the offered dataset of a vicinity and apply Machine learning and Neural network to predict the crop yield by choosing multiple options in order that results are a lot of acceptable for crop yield prediction

References

1. R.Dharavath,“Cropyieldpredictionvictimization collectiverainfall-based standardartificial neural networks and support vector regression, Aug. 2020.

2. A McBratney, B. Whelan, T. Ancev, and J. Bouma, “Future directions of exactitudeagriculture,”inexactitudeAgriculture,Feb. 2005,

3. P. J. Ramos, F. A. Prieto, E. C. Montoya, and C. E. Oliveros, “Automatic fruitestimate low branches victimization laptopvision,” May 2017.

4. S. Sengupta and W. Lee, “Identification and Determination of the amount of inexperienced ediblefruit underneath completely different close light-weight Conditions,” 2012.

5. Ali, F. Cawkwell, E. Dwyer, and S. Green, “Modeling Managed Grassland BiomassEstimationbyvictimizationMultitemporalRemoteSensingData-AMachine

LearningApproach,”IEEEJul. 2017

6. X.E.Pantazi,D.Moshou,T.Alexandridis,R.L.Whetton,andA.M.Mouazen, “Wheat yield prediction using machine learning and advanced sensing techniques,” 2016.

7. J.Senthilnath,A.Dokania,M.Kandukuri,R.K.N.,G.Anand,andS.N.Omkar,

“Detectionoftomatoesvictimizationspectral-spatialmethodsinremotelysensedRGB images captured by UAV2016.

8. Chlingaryan, S. Sukkarieh, and B. Whelan, “Machine learning approaches forcropyieldpredictionandnitrogenstatusestimationinprecisionagriculture: Aug. 01, 2018,

9. K. Samundeeswari and K. Srinivasan, “CROP YIELD PREDICTION AND SOIL DATA ANALYSIS USING DATA PROCESSINGTECHNIQUES IN KRISHNAGIRIDISTRICT,”Oct. 2018,

10. “PredictingCropyield andEffectiveuseofFertilizersusingMachineLearning Techniques.

11. P. Nevavuori, N. Narra, and T. Lipping, “Crop yield prediction with deep convolutionalneuralnetworks,”Aug.2019.

12. H. H. Patel and P. Prajapati, “Study and Analysis of Decision Tree Based ClassificationAlgorithms,”Oct. 2018,

13. S. Kiranyaz, O. Avci, O. Abdeljaber, T. Ince, M. Gabbouj, and D. J. Inman, “1D convolutional neural networks and applications 2019.

14. L.Shi,Q.Duan,X.Ma,andM.Weng,“Theresearchofsupportvectormachine in agricultural data classification,” Oct. 2012.