Machine Learning for a Better Agriculture Calendar

Pascal Francois Faye

1 a

, Jeanne Ana Awa Faye

and Mariane Senghor

Universit

e du Sine Saloum El Hadj Ibrahima NIASS (USSEIN), Sing Sing, Kaolack, Senegal

Universit

e Aix-Marseille, Institut de Management Public et Gouvernance Territoriale, Aix-Marseille, France

Keywords:

Agriculture Calendar, Internet of Things, Data Analysis, Artiﬁcial Intelligence, Crop Yield.

Abstract:

In Senegal, agriculture is subsistence, low-input, and signiﬁcantly less mechanized than many other nations in

Africa, and is also highly dependent on soil, climate, and water. Food crops take up to 46% of the total land

and make up 15% of the Gross Domestic Product (GDP), ensuring between 70% and 75% employment. In this

work, we provide a set of mechanisms that uses a set of trust database of agro-climatic parameters and a set

of artiﬁcial intelligence algorithm in order to assess agricultural calendar for a good distribution of the farm’s

activities over time and ﬁnd the relationship between crops. Our results show the effectiveness of our solution

to overcome the abandonment of agricultural perimeters or an agriculture depending on the raining season.

That means, taking these data into account makes possible to understand crops dependencies and anticipate

the agroecological phenomena, the crop diseases and pests that impact the planning of production facilities

and variations in agricultural yields.

1 INTRODUCTION

With the arduousness of agriculture works and the

gradual abandonment of land due to the demographic

pressure of cities, climate change and soil deteriora-

tion, several research questions are done in order to

ﬁnd solutions or advice with the reduction of agricul-

tural perimeters. In order to continuously feed the fu-

ture, in computer sciences we have several advances

on sensor network, robot automation, computer vision

and artiﬁcial intelligence for pest detection, predic-

tion, decision making, etc.

Computer vision recognition has been increas-

ingly applied to numerous ﬁeld of agricultural with

the advancement of computer graphics and image

processing technology. The development of sensors

technologies for smart agricultural, such as soil tem-

perature and humidity sensors, air temperature and

humidity sensors, etc., enhance data collection and

processing for decision making in the agricultural en-

vironment. Usually, the data is wirelessly transferred

from the sensor to the sink node for data collection

and the server for processing and decision making.

Between them, the gateway changes the protocol into

one that can be communicated over the Internet when

it receives data from the sink node. In addition, the

use of robots in agriculture that uses a variety of sen-

https://orcid.org/0000-0002-2078-5891

sors to sense the dynamic of the agricultural environ-

ment and then picks the target using this knowledge

and a decision-making algorithm based on artiﬁcial

intelligence help farms in order to manage (plough-

ing, sowing, harvesting, attack detection, etc.) their

ﬁeld.

In this work we tackle the planning of the agri-

cultural calendar for a good distribution of activities

over time. This in order to take into account climate

changes, soil degradation, etc. for improving the farm

crop plan and to overcome the seasonality of the agri-

culture.

In the rest of this paper, section 3 presents our re-

search problem, section 4 highlights our objectives

and section 5 discusses related works. Section 6

presents some preliminaries and methodologies. Sec-

tion 7 highlights the results that lead to the proposed

cropping calendar before the conclusion.

2 RESEARCH PROBLEM

An agriculture that deals with the environment and

the climate change has become an imperative if we

aim to feed the future. All ﬁelds of agriculture are

affected and need to limit the disadvantages of cli-

mate change and soil degradation. A better under-

standing of the changes of resources (water, energy,

Faye, P., Faye, J. and Senghor, M.

Machine Learning for a Better Agriculture Calendar.

DOI: 10.5220/0012691800003753

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 19th International Conference on Software Technologies (ICSOFT 2024), pages 307-314

ISBN: 978-989-758-706-1; ISSN: 2184-2833

307

etc.) in the farm’s environment is needed to make

the right varietal choices and crop options. This is

not to mention the rise in temperatures, which leads

to increased evaporation or evapotranspiration, which

inﬂuences yields and the seasonality of the agricul-

ture. That is why, in this work, we propose to process

historical agro-ecological data using Machine Learn-

ing (ML) algorithms and probability laws in order to

make fair decisions about cropping calendars and un-

derstand changes in cropping practices. In addition,

we aim to ﬁnd the relation between a set of crop in

order to give the farmers the possibility to test other

crop type in the peanut basin of Senegal.

3 OUTLINE OF OBJECTIVES

Knowledge of climate change and its effects on the

various sectors of the national economy is a major

challenge for the country’s developers. Various ini-

tiatives are therefore being developed to better iden-

tify the implications of climate variability in the agri-

cultural sector. However, while a causal relationship

has clearly been established between the vulnerabil-

ity of the agricultural sectors and a set of parame-

ters like: the climate, the challenge of accurate infor-

mation, the sharing and pooling efforts (Faye et al.,

2022). This calls for a review of existing frameworks,

but it should also help to ﬁnd ways of collaborating,

sharing and other one in order to provide better sup-

port for decision-making, particularly for grassroots

users such as producers. Supporting farmers to bet-

ter manage the risks associated with climate variabil-

ity is now a major necessity. All economic activities

which promote food security and suitable agriculture

must incorporate the risks of climate change into their

planning. The aims of this work are proposing :

1. a crop calendar that is adapted to variations in

endogenous resources and meteorological factors

during the annual seasons.

2. to ﬁnd similarities between different crops in or-

der to propose an annual soil occupation manage-

ment strategy.

This by using historical agro-meteorological and

agroecological database to ensure sustainable crop

yields. In this way, farmers can make decisions about

the technical itineraries for their crops and the dates

set for the cropping calendars (ploughing dates, sow-

ing dates, fertiliser application dates, irrigation hours

and other inputs), which can enable precision farm-

ing. This work combines:

1. laws of probability to model the dynamics and un-

predictable events;

2. Machine learning algorithms (ML) to ﬁnd the bet-

ter decision making;

This combination delivers a solution that addresses

well the dynamism and uncertainty challenges tar-

geted in this work.

4 STATE OF THE ART

In (Sellam and Poovammal, 2010), the authors per-

sist to research the environmental parameters that af-

fect the crop yield and related parameters. Here a

multivariate Regression Analysis is applied for the

same. A sample of environmental factors considers

a period of 10 years. The System is applied to ﬁnd

the relationship between explanatory variables like

annual rainfull,area under cultivation, food price in-

dex and hence the crop yield as a response variable

and R

value clearly shows that, the yield is espe-

cially hooked into annual rainfall,area under cultiva-

tion and food price index are the opposite two factors

that are inﬂuencing the crop yield. This research is

often enhanced by considering other factors like min-

imum support price, cost price index, wholesale price

index, etc. and their relationship with crop yield. In

paper (Paswan and Begum, 2013), the authors have

compared feed forward neural networks with tradi-

tional statistical methods through linear regression.

This work presents the capability of neural networks

and their statistical counterparts used in the world of

crop yield prediction. In (Zhang et al., 2010), the

authors have done a comparison between the linear

regression model based on the ordinary least square

(OLS) and special autoregressive model for crop yield

prediction in Iowa. The special autoregressive model

has shown enormous enhancement in the model per-

formance over the OLS model. The model can pro-

vide better prediction than the OLS model and has

capability of adjust with the special autocorrelation,

which is not considered by the OLS model. This

work has shown that NDVI (Normalized Difference

Vegetation Index) and precipitation are the most im-

portant predictors for corn yield in Iowa. In (Zingade

et al., 2018), the authors have presented an android

based application and an internet site that uses Ma-

chine learning methods to predict the foremost prof-

itable crop in the current weather and soil conditions

and with current environmental conditions. This sys-

tem helps the former with a sort of option for the crops

that will be cultivated, which will be helping them

over the long run. In (Sun et al., 2022b) they im-

proved a density peak cluster segmentation algorithm

for RGB (Red Green Blue) images with the help of a

gradient ﬁeld of depth images to locate and recognize

ICSOFT 2024 - 19th International Conference on Software Technologies

308

target fruit during the process of green apple harvest-

ing or yield estimation. Speciﬁcally, the image depth

information is adopted to analyse the gradient ﬁeld of

the target image. In (Feng et al., 2022a), the automatic

separation between two diseases was examined using

image processing technologies. The acquired disease

images were preprocessed using morphological open-

ing and closing reconstruction, color image contrast

stretching, and image scaling. Then, two crop leaf

lesion segmentation algorithms based on circle ﬁt-

ting were suggested and applied. Support vector ma-

chine (SVM) models and random forest models were

used based on individual LBP histogram features and

various LBP (Local Binary Pattern) histogram fea-

ture combinations. (Fu et al., 2022) created rape-

seed dataset (RSDS) using eight categories of data

gathered. The target-dependent neural architecture

search (TD-NAS) was proposed. Usually, smart agri-

cultural produces enormous quantities of multidimen-

sional time series data. However, due to the techno-

logical’s limitations, data loss and misrepresentation

are frequent problems with the smart agricultural’s

IOT devices. In order to solve the issues (Cheng

et al., 2022) proposes a anomaly detection model that

can handle these multidimensional time series data.

Meanwhile, a multi-objective strategy based on super-

vised machine learning was utilized in (Uyeh et al.,

2022) to identify the ideal number of sensors and in-

stallation locations in a protected cultivation system.

A machine learning tree-based model in the form of a

gradient boosting technique was speciﬁcally adapted

to observed (temperature and humidity) and derived

circumstances (dew point temperature, humidity ra-

tio, enthalpy, and speciﬁc volume). Time series fore-

casting was used for feature variables. In (Maia et al.,

2022), sensor data analysis over two irrigation sea-

sons in three cotton ﬁelds from two cotton-growing

regions of Australia revealed a connection between

soil matric potential and cumulative crop evapotran-

spiration (ETcn) derived from satellite measurements

between irrigation events. In (Ma et al., 2022) ex-

plore the distributed averaging issues of agriculture

picking multi-robot systems under directed commu-

nication topologies by utilizing the sampled data. A

distributed protocol based on nearest-neighbor infor-

mation is presented using the principles of algebraic

graph theory and matrix theory. The brown planthop-

per (BPH), Nilaparvata lugens (Stl; Hemiptera: Del-

phacidae), is a piercing-sucking insect that seriously

harms rice plants by sucking out their phloem sap and

spreading viruses. For reducing mating rates, a phys-

ical control mechanism based on BPH courting dis-

ruption is a viable approach to reducing environmen-

tal pollution. To gather effective courtship disrupting

signals. (Feng et al., 2022b) created a vibration sig-

nal recording, monitoring, and playback system for

BPHs. This technology was used to gather and eval-

uate male competitiveness and BPH courting signals

in order to determine their frequency spectra. Accord-

ing to the ﬁndings, the mean main vibration frequency

and mean pulse rate of female courtship signals are

234 Hz and 23 Hz, respectively. Male courting sig-

nals had mean main vibration and pulse frequencies

of 255 Hz and 82 Hz, respectively. Furthermore,

Cnaphalocrocis medinalis, Sogatella furcifera, and

Nilaparvata lugens are three kinds of migratory pests

that severely reduce rice yield and result in economic

losses each year. (Sun et al., 2022a) create an intel-

ligent monitoring system of migrating pests based on

searchlight trap and computer vision to replace man-

ual identiﬁcation of migratory pests in. The system

consists of a cloud server, a Web client, a migratory

pest automatic identiﬁcation model, and a searchlight

trap based on computer vision. The searchlight trap

uses lights at night to draw in high-altitude migrat-

ing insects. All captured insects are distributed using

rotary brushes and multi-layer insect conveyor belts.

The intelligent monitoring system can automatically

monitor the three migratory pests in time.

In contrast to our work, these works do not pro-

pose a cropping calender in order to minimize the risk

depending to climate change, pest migration and other

agrometeorological information and soils parameters

like ours. In addition, we aim to enhance productivity

and sustainability in the peanut basin of Senegal by

taking into account the socio-economic impact. This,

because in Senegal, there are different types of farms

and various levels of complexities in terms of organi-

zation. We have on the one hand, family farms with

limited levels of organizations, ﬁnancial capabilities

and standard procedures. On the other hand, there are

some farms working on fruits and vegetables exporta-

tion with better organizations and procedures.

5 METHODOLOGY

Our study is motivated by the environmental and cli-

mate challenges that make difﬁcult the prediction on

crop yield, crop diseases and pests in the peanut basin

of Senegal (Fatick, Kaolack and Kaffrine) (cf. Fig-

ure 1). The peanut basin of Senegal has a set of

different climatic characteristics (Faye et al., 2023).

Many agroclimatic challenges are observable due to

changes in land degradation, soil salinization, temper-

ature, rainfall, etc. We conduced a two years study

that has shown the correlation relationship (Figure 3)

between a set of climatic parameters and the NDVI of

Machine Learning for a Better Agriculture Calendar

309

Figure 1: Peanut basin of Senegal.

Figure 2: Data correlation of the agents data.

the peanut basin by using our dataset from our sensor

network (ﬁgure 6).

To do this, we use an agent concept. An agent is a

device or an application which can sense the environ-

ment compute some processes and provide results or

acts on its environment.

The ﬁgure 2 is a two-dimensional representation of

data which highlights the dependencies between our

set of variables. Each square shows the correlation (a

measure of dependencies) ranges from -1 to +1. Val-

ues closer to zero means there is no linear trend be-

tween the variables. Close to 1 the variables are more

positively correlated, and stronger is the relationship.

This means, as one increases so does the other. A cor-

relation closer to -1 means one variable will decrease

as the other increases. The legend on the right side

help to interpret this heatmap.

In (Faye et al., 2023), a set of results in this peanut

basin shown the crop yield relative to the rainfall (cf.

Figure 4) and to the temperature (cf. Figure 5).

To study the set of real-time interactions between at-

mospheric phenomena and all parameters of agrom-

eteorology (set of scientiﬁc and technical tools that

take into account meteorological and agronomic data

to help farm management and agricultural forecast-

ing), we have to deal with the real needs of farmers.

There are three different types of agrometeorological

information: short term (from day to day), medium

Figure 3: Data clustering between 2022 and 2023 using De-

cisionTreeClassiﬁer.

Figure 4: Crop yield relative to the rainfall.

ICSOFT 2024 - 19th International Conference on Software Technologies

310

Figure 5: Crop yield relative to the temperature.

term (from ﬁfteen days to two months), and long

term (from one year to more year). Such information

must meet the agricultural requirements for planning

work, forecasting the development risks of certain

climate-related diseases, monitoring the water bal-

ance of soils, monitoring of temperatures in connec-

tion with the plant development schedule. The func-

tioning of the crop, soil and water system depends

mainly on ﬁve meteorological variables, namely:

• The air temperature measured under cover at 2

meters above the ground.

• The partial pressure of water vapour in the air

measured under cover at 2 meters above the

ground.

• Wind speed measured at 10 meters above the

ground.

• The overall solar radiation or the daily insolation

time.

• The rainfall.

The ﬁrst three physical variables are intensive be-

cause they describe the state of a system at a given

time, while the other two are extensive variables that

quantify an exchange of energy or mass between the

atmosphere and the ground. Derived variables are

also used, such as relative air humidity, which de-

pends on the temperature and partial pressure of wa-

ter vapour. An agent which extract, organize, aggre-

gate and interpret sensor data based on our machine

learning algorithm is constrained by the parameters:

Figure 6: Sensors network for local climate and soil data.

{Rs, Hs, ϑ

, U, L

Net

}. Rs is its resource(s) and Hs is

its history set which consists of a set of previous deci-

sions. A view ϑ

is the set of sensors in its neighbor-

hood with whom it can directly communicate at time

t. U is its private utility function. L

Net

deﬁnes the de-

pendence level between the received data in a given

sensors network (Net). The utility function U of the

agent is the score used in order to help to improve the

learning rate U =

errors rate

good decision

To provide decision the agent may compute the

following steps (cf. Figure 7 and Figure 8) by taking

into account its constraints and the agrometeorology

parameters. This help us to reﬁne the crop calender.

6 CROPPING CALENDAR

PROPOSAL

To obtain the agrometeorology parameters we used a

set of trusted open source database like powerlarc.

Dates (month/day/year): From 01/01/2000 to

12/31/2023

Location: Latitude 14.1635 Longitude -16.1268

The parameter(s)collected are:

• PS = Surface Pressure (kPa)

• QV2M = Speciﬁc Humidity at 2 Meters (gkg)

• T2M MAX = Temperature at 2 Meters Maximum (C)

• T2M MIN = Temperature at 2 Meters Minimum (C)

• WS2M MAX = Wind Speed at 2 Meters Maximum

(ms)

• WS2M MIN = Wind Speed at 2 Meters Minimum (ms)

• PREC = Precipitation Corrected (mmday)

Machine Learning for a Better Agriculture Calendar

311

Figure 7: Agent’s main steps.

Figure 8: Agent’s Machine Learning processing.

• UVA = All Sky Surface UVA Irradiance (Wm

)

• UVB = All Sky Surface UVB Irradiance (Wm

)

Temperatures in Senegal, range from very warm

to hot, with an annual average temperature of 35 Cel-

sius. At least 4 months of the year are tropical and

frequently sultry with temperatures above 35 Celsius.

The distribution of crop types grown in Senegal cor-

relates with the timing of seasonal rainfall (ﬁgure

9 and ﬁgure 14). Moreover, some of the practices

of the Green Revolution, especially the use of mod-

ern crop varieties and the addition of synthetic fer-

tilizers and pesticides/herbicides are not sustainable

practices, especially under climate change conditions.

NPK(nitrogen, phosphorus, potassium) input are the

most important parameters in maximizing yields and

economic returns to farmers. However, in the peanut

Figure 9: The study of the mean variations of the parameters

from 2000 to 2023.

Figure 10: Comparative study of the needs of speculation.

basin it is required to take into account to the soil pH

(potential of hydrogen) and salinity ((Electro Conduc-

tivity)) for agriculture calendar. In Figure 10, the se-

quence of bars represents the following speculations

list :

1-Carrot, 2-Sweet potato, 3-Eggplant, 4-Lettuce, 5-

Cabbage, 6-Okra, 7-Tomato, 8-Turnip, 9-Melon, 10-

Zucchini, 11-Cucumber, 12-Bell pepper, 13-Chilli,

14-Onion, 15-Cassava, 16-Potato, 17-Hibiscus sab-

dariffa, 18-Parsley. For each speculation we consider:

• NPK

hec(Kg): nitrogen, phosphorus, potassium

for each hectare.

• Water Day(L): stream-day water requirements

(litre).

• Temperature(C): Temperature (Celsius).

• Light(UV): Means of Ultraviolet (UVA and

UVB).

• Soil type: Soil type (e.g. tropical ferruginous

soils, hydromorphic soils, loamy soils, clay soils,

etc.).

ICSOFT 2024 - 19th International Conference on Software Technologies

312

Figure 11: Elbow Method to evaluate the optimal cluster

number.

Figure 12: Output of the clustering.

• Ripening time(day): Number of days before har-

vest.

• Salinity tolerance: Electro conductivity (EC).

After this comparison of needs, we used an AI cluster-

ing algorithm (KMeans) in order to ﬁnd the similari-

ties in the following crops. In order to be sure about

the better number of clusters with these data, we used

also the Elbow method (Umargono et al., 2020). Fig-

ure 11 shows that, the optimal number of clusters with

our data is four.

Figure 12 shows the similarity between a set of

speculations regarding the parameters that are deter-

mining the crop adaptation.

Depending to the analyse of the ﬁgure 9, ﬁgure 13

and ﬁgure 14, we propose the following crop calender

(ﬁgure 15) by taking into account the possibility of an

out-of-season cultivation. This in order to overcome

the abandonment of agricultural perimeters or an agri-

culture depending on the raining season. This ﬁgure

15 in combination with the ﬁgure 12 permit to ﬁnd a

Table 1: Cluster assignment by similarity.

Cluster 1

Cluster

3 Cluster 4

Carrot,

Eggplant,

Zucchini,

Cucumber,

Onion,

Potato

Sweet

potato,

Okra Lettuce

Cabbage, Tomato,

Turnip, Melon, Bell

pepper, Chilli,

Cassava, Hibiscus

sabdariffa, Parsley

Figure 13: The study of the mean variations of the parame-

ters from 2000 to 2023 except the surface pressure (PS).

Figure 14: Comparative variations of the humidity and the

temperature from 2000 to 2023.

Figure 15: Agricultural calendar for a good distribution of

farm activities over seasons.

set of other speculations adaptable in the peanut basin

when the salinity (Electro Conductivity) is between

0.1 and 1.2 and Ph (Potential of Hydrogen) between

5 and 8.

Machine Learning for a Better Agriculture Calendar

313

7 CONCLUSION

In Senegal, particularly in his peanut basin, agricul-

ture is subsistence, low-input, and signiﬁcantly less

mechanized than many other parts of the country, and

is also highly dependent on soil, climate, soil salin-

ity and water. In addition, due to the lack of the use

of new ﬁeld in agriculture like data-sciences, artiﬁcial

intelligence, etc. the distribution of crop types grown

correlates with the timing of seasonal rainfall. In this

work, we provide a set of mechanisms that uses a set

of trust database of agro-climatic parameters and a set

of artiﬁcial intelligence algorithm in order to assess

agricultural calendar for a good distribution of agri-

culture activities over time and ﬁnd the relationship

between crops. Our results show the effectiveness

of our solution. That means, taking these data into

account makes possible to understand crops depen-

dencies and anticipate the agroecological phenomena,

the crop diseases and pests that impact the planning

of production facilities and variations in agricultural

yields.

In the future we aim to disseminate this technique

in the other agroecological area of the Senegal. In-

corporate an analysis of the socio-economic impact

of our agricultural calendar on local communities by

selecting performances metrics and comparison with

traditional methods. As we have already done the

tests on the peanut basin of Senegal, it would be valu-

able to discuss the scalability of the approach to other

regions and crops. And, the work will be expanded

to potential collaborations to further develop and im-

plement. In addition, to reﬁne our predictions we aim

to compare our methods with the tools provided by

FAO (CROPWAT and CLIMWAT) (Food and of the

United Nations, 2023) to measure positive or negative

deviations from the predictions.

REFERENCES

Cheng, W., andXiaoting Wang, T. M., and Wang,

G. (2022). Anomaly detection for internet of

things time series data using generative adversar-

ial networks with attention mechanism in smart

agriculture. Frontiers in Plant Science, page

https://doi.org/10.3389/fpls.2022.890563.

Faye, P. F., Sall, M. M., Dembele, B., and Faye, J. A. A.

(2023). Plateforme num

erique de guides culturaux

dans un contexte de changement climatique et de

baisse des rendements agricoles. RAMReS Sciences

des Structures et de la Mati

ere, pages vol. 7, no. 1, pp.

85–115, ISSN : 2630–1180.

Faye, P. F., Senghor, M., and Ndione, G. A. (2022). Stable

trade coordinations in smart agriculture. International

Joint Conference on Web Intelligence and Intelligent

Agent Technology (WI-IAT), pages 215–222.

Feng, Q., Wang, S., Wang, H., Qin, Z., and Wang,

H. (2022a). Circle ﬁtting based image segmen-

tation and multi-scale block local binary pattern

based distinction of ring rot and anthracnose on

apple fruits. Frontiers in Plant Science, 13:doi:

10.3389/fpls.2022.884891.

Feng, Z., Wei, Q., Ye, Z., Yang, B., Gao, Y., Lv, J.,

Dai, Y., Bao, J., and Yao, Q. (2022b). Vibrational

courtship disruption of nilaparvata lugens using arti-

ﬁcial disruptive signals. Frontiers in Plant Science,

13:https://doi.org/10.3389/fpls.2022.897475.

Food and of the United Nations, A. O. (2023). Land and

water databases and software.

Fu, L., Li, S., Rao, Y., Liang, J., Teng, J., and He, Q. (2022).

A novel heuristic target-dependent neural architecture

search method with small samples. Frontiers in Plant

Science, page 10.3389/fpls.2022.897883.

Ma, F., Yao, H., Du, M., Ji, P., and S, X. (2022). Distributed

averaging problems of agriculture picking multi-robot

systems via sampled control. Frontiers in Plant Sci-

ence, page https://doi.org/10.3389/fpls.2022.898183.

Maia, R. F., Lurbe, C. B., and Hornbuckle, J. (2022).

Machine learning approach to estimate soil matric

potential in the plant root zone based on remote

sensing data. Frontiers in Plant Science, page

https://doi.org/10.3389/fpls.2022.931491.

Paswan, R. P. and Begum, S. A. (2013). Regression and

neural networks models for prediction of crop produc-

tion. International Journal of Scientiﬁc and Engineer-

ing Research, pages Volume 4, Issue 9,ISSN 2229–

5518.

Sellam, V. and Poovammal, E. (2010). Prediction of crop

yield using regression analysis. IEEE Trans. Knowl.

Data Eng., pages vol.23, no. 10, pp. 1498–1512.

Sun, G., Liu, S., Luo, H., Feng, Z., Yang, B., Luo, J., Tang,

J., Yao, Q., and Xu, J. (2022a). Intelligent monitor-

ing system of migratory pests based on searchlight

trap and machine vision. Frontiers in Plant Science,

13:https://doi.org/10.3389/fpls.2022.897739.

Sun, M., Xu1, L., Luo, R., Lu, Y., and Jia, W. (2022b).

Fast location and recognition of green apple based

on rgb-d image. Frontiers in Plant Science, 13:doi:

10.3389/fpls.2022.864458.

Umargono, E., Suseno, J. E., and K., V. G. S. (2020).

K-means clustering optimization using the elbow

method and early centroid determination based-on

mean and median. International Conferences on In-

formation System and Technology (CONRIST 2019,

DOI: 10.5220/0009908402340240:pages 234–240.

Uyeh, D. D., Iyiola, O., Mallipeddi, R., Asem-Hiablie,

S., Amaizu, M., Ha, Y., and Park, T. (2022). Grid

search for lowest root mean squared error in pre-

dicting optimal sensor location in protected culti-

vation systems. Frontiers in Plant Science, page

https://doi.org/10.3389/fpls.2022.920284.

Zhang, L., Lei, L., and Yan, D. (2010). Comparison of

two regression models for predicting crop yield. IEEE

International Symposium on Geoscience and Remote

Sensing (IGARSS).

Zingade, P. D., Buchade, O., Mehta, N., Ghodekar, S., and

Mehta, C. (2018). Machine learning-based crop pre-

diction system using multi-linear regression. Interna-

tional Journal of Emerging Technology and Computer

Science(IJETCS), pages Vol 3, Issue 2.

ICSOFT 2024 - 19th International Conference on Software Technologies

314