International Journal of Scientific Research in Engineering and Management (IJSREM)

Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

Crop Recommendation System using Satellite Images

1
Sushmitha K Jadav, 2 Sindhu S L
[1]. Student, Department of MCA, BIET, Davangere
[2].Assistant professor ,Department of MCA ,BIET ,Davangere

ABSTRACT

The "Crop Recommendation System Using Multi- methods, the system encourages crop management
Temporal Satellite SAR Data" is a creative project that is both efficient and sustainable.By bridging the
that combines cutting edge technologies to modernize gap between conventional wisdom and cutting-edge
agriculture. The study introduces a data-driven technology, the project's holistic approach offers a
approach to crop selection by utilizing multi-temporal revolutionary solution for improving productivity and
Synthetic Aperture Radar (SAR) data, deep learning, resource usage. The technology seeks to transform
and machine learning. The method provides farmers agricultural decision-making processes by providing
with precise advice based on their specific agricultural farmers with useful insights obtained from satellite
needs by evaluating temporal fluctuations in SAR data, promoting resilience and adaptation in the face
imagery. Comprehensive data analysis has yielded of shifting environmental conditions.
insights into crop dynamics, moisture levels, and soil Keyword:Crop Detection, Satellite Imagery, Image
health that guide these suggestions. By fusing Processing, Agriculture, Machine Learning
advanced analytics with conventional agricultural

1. INTRODUCTION utilizing the power of multi-temporal Synthetic

A new initiative at the intersection of modern Aperture Radar (SAR) data in conjunction with
technology and agriculture is the "Crop advanced deep learning and machine learning
Recommendation System Using Multi-Temporal algorithms. With the help of this innovative method,
Satellite SAR Data" study. This work presents an which examines temporal fluctuations in SAR
innovative data-centric method to crop selection
imaging, farmers can receive advice that are With the help of satellite data, this endeavor seeks to
specifically customized to meet their needs. revolutionize agricultural decision-making by
These customized crop recommendations are based on providing farmers with useful information. It
important discoveries about crop dynamics, moisture ultimately transforms farming's future by fostering
content, and soil health that were made possible by the resilience and adaptation in the face of shifting
extensive data analysis carried out for this project. environmental conditions. The system gives a
The system facilitates efficient and sustainable crop continuous, detailed image of the agricultural
management tactics by fusing advanced analytics with environment by using multi-temporal SAR data,
conventional agricultural methods. This innovative which helps to capture important data on crop growth
method effectively combines the abundance of stages, variations in soil moisture, and general field
conventional agricultural knowledge with state-of-the- health. By using these information, farmers may
art technology to provide a comprehensive solution maximize yields and reduce their impact on the
intended to maximize resource use and productivity in environment by choosing crops wisely, planning
agriculture. planting dates, and allocating resources.Furthermore,

© 2024, IJSREM | www.ijsrem.com | Page 1

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

the incorporation of deep learning techniques The primary goal of the study ""A Novel Strategy for
improves the precision and applicability of the Field Kind Identification in Landsat Images Utilizing
recommendations, guaranteeing that they are both Mixed Deep Capsules an Autoencoder" is to use a
flexible and accurate in a range of agricultural hybrid deep capsule autoencoder to increase crop type
scenarios. This initiative demonstrates the mapping accuracy. By combining the advantages of
revolutionary potential of fusing cutting-edge autoencoders and capsule networks, this novel
technology with agriculture, opening the door for a approach seeks to efficiently extract and represent
more resilient, productive, and sustainable agricultural spatial data from satellite photos for improved pattern
industry. The "Crop Recommendation System Using identification. The project aims to enable more
Multi-Temporal Satellite SAR Data" represents the informed agricultural decision-making and sustainable
future of smart agriculture by bridging the gap land management practices by gaining higher
between conventional farming methods and precision in crop type identification[4].The goal of the
contemporary technological breakthroughs. study "An Intelligent Crop Recommendation System
using Deep Learning" is to create a complex system
2. LITERATURE REVIEW that can provide accurate crop recommendations by
Constructing and validating a comprehensive crop utilizing deep learning techniques. It seeks to increase
recommendation system that employs ensembling crop selection processes' accuracy in comparison to
strategies to increase crop output is the primary traditional approaches by incorporating deep neural
objective of the study "Boosting Plant Yield With A networks. Through the analysis of various agriculture
Plant Advice System Using Assembly Technique" information, including past crop performance, climate
,This method combines many machine learning variables, and soil conditions, the system places an
models with multi-temporal satellite data to deliver emphasis on enabling precision agriculture. It also
accurate, data-driven recommendations for crop seeks to promote environmentally friendly farming
choices. The following are the specific goals[1].The methods by recommending crops that maximize
study aims to increase crop recognition accuracy by resource efficiency and reduce negative
using convolutional neural networks (CNNs) and environmental effects. The goal of the research is to
synthetic photos that integrate different timings and confirm its efficacy through real-world application,
spectral data. Improving agricultural monitoring is the giving farmers access to cutting-edge instruments for
goal since it will improve crop management, resource increased production and sustainable agriculture
use, and overall productivity. To illustrate this management[5].This article's primary goal is to
method's potential advantages for farmers and other discuss machine learning (ML) applications in
agricultural stakeholders, its efficacy will be agriculture, with a particular emphasis on disease
compared to currently used procedures[2]. detection, crop production prediction, and soil
By using innovative image processing techniques, the parameter prediction. It also discusses intelligent
study "Crop Detection Using Satellite Image irrigation methods and the application of ML to
Processing" seeks to create an accurate approach for animal production. The intention is to demonstrate
identifying crops from satellite imagery. Its goals are how ML may improve the quality and production of
to increase the precision of detection, facilitate well- sustainable agriculture[9].This work proposes a
informed agricultural decision-making, and encourage recommendation system to address the problem of
sustainable farming methods by making effective use Indian farmers selecting the incorrect crop based on
of available resources. The study aims to enhance soil requirements. It makes highly accurate and
crop management, productivity, and sustainability in efficient crop recommendations by using an ensemble
agriculture by verifying the suggested method's model involving Randomized The Tree, CHAID, the
efficacy against benchmark datasets and current k-n Others, and naive Bayes models with a majority
methodologies[3]. vote. Increasing productivity through precision
farming is the aim[10].The paper seeks to emphasize

© 2024, IJSREM | www.ijsrem.com | Page 2

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

the use of soil analysis in contemporary agriculture as fluctuations in Synthetic Aperture Radar imaging and
a means of maximizing crop yield and reducing derive pertinent data concerning crop dynamics, soil
environmental impact. It attempts to show how health, and moisture content. To extract complex
resource efficiency and sustainable agricultural features from the SAR data, methods including
methods can be improved by data-driven crop polarimetric decomposition, texture analysis, and
suggestions based on soil attributes. Enhancing yields temporal change detection are used. The suggestion
and quality is the main goal, along with halting soil generation module uses these features that have been
erosion and promoting food security[11].The primary extracted as input.
goal of the study “Machine Learning Techniques for Model Training:
Remote Sensing-Based Crop Identification Data in The model training component complements the
Fresno County, California” is to use machine learning feature extraction module by imparting interpretative
and remote sensing data to reliably identify and machine learning models that utilize the features that
classify various crop types in Fresno County. This have been extracted. Models are trained on historical
seeks to improve agricultural management and data employing supervised methods for learning such
monitoring through the application of cutting-edge as support vector machines (SVM), random forests,
technology[12]. and neural networks. By optimizing their predicting
skills, the models discover patterns and connections
3. METHODOLOGY between the crop results and the retrieved variables.
Data Acquisition: Recommendation Generation:
The recommendation generating module creates
First, multi-temporal SAR data is obtained from customized crop recommendations for farmers based
satellite sources and fed into the system. By capturing on the collected information and trained models. To
microwave signals reflected from the Earth's surface, determine the best crop selections for a given set of
satellites equipped with SAR sensors are able to agricultural conditions, the models evaluate historical
produce high-resolution photographs of agricultural trends, soil properties, and current SAR data. These
regions. These pictures are sent to ground stations so suggestions are made with the intention of increasing
that they can be processed. The SAR imagery is agricultural productivity, reducing resource
gathered and arranged by the data gathering module, consumption, and advancing sustainable farming
guaranteeing a steady and dependable data stream for methods.
analysis.
Preprocessing: 4. RESULT and ANALYSIS
The collected SAR data is preprocessed to improve its Choice of Spectral Features.
quality and usefulness before analysis. This covers
operations like calibration, radiometric correction, Multispectral satellite images with 3 m × 3 m pixels
geometric correction, and noise reduction. In order to are included in the study. Every pixel has the ability to
align SAR images with other spatial datasets, they reflect four spectrum values, which are all Infrared
might also require georeferencing or resampling. radiation that is invisible and visible in RGB lights.
Accuracy and consistency are ensured during the The cadastral data is used to determine the values of
preprocessing module's preparation of the data for the required pixels. The association between sixteen
feature extraction and testing. attributes—the average, standard deviation, maximum
and lowest values of each pixel on each cadastre, and
Feature Extraction the number of days that cabbage develops—is then
The system's primary function is to extract significant ascertained by analyzing the data (Figure 4). To
characteristics from multi-temporal SAR data. DL demonstrate the values of the spectrum parameters,
algorithms are utilized to examine temporal such as the mean, max, and min of the infrared
radiation, a growth period of 20 to 60 days for

© 2024, IJSREM | www.ijsrem.com | Page 3

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

cabbage was chosen (Table 1). NDVI, and the average the brightness index (ABI)
were among the eight distinct kinds of vegetation
indices that were computed using the spectral values
of the pixels. The study examined the relationship
between these markers and the cabbage's development
phases, or growth days.employed sixteen features in
total, included the mean as well as the standard
deviation of each unique vegetation index. The results
are presented in Figures 5 and 6.
There is a strong positive link with the vegetation
indicators during the 20–60day timeframe.
Specifically, Figure 6 demonstrates that the vegetation
indicators have a stronger correlation compared to the
spectrum data, with the cabbages' growth days.
However, the bulk of vegetation indices are computed
using values of red light and near-infrared light, which
have a high association. As a result, Table 2 shows
that only one of the three vegetation indices that are
Figure 4.:Relationship between spectral properties fully associated (correlation coefficients equal to 1 or
and the stages of growth of cabbage -1)—the BR (SQBR), the V, or the NDVI. (IPVI,
MSAVI, CMFI)—is was chosen.
Table 1 shows the coefficients of connection between
the spectral properties and the developmental phases
of cabbage.

Association
Spectral Features
Ratio

average_nir NIR's
0.486128
standard deviation

std_nir The NIR's standard

0.393943
deviation

max_nir: NIR Maximum 0.478906

Figure 5: connections between the cabbage
growth stages and the vegetation index's
Feature Selection for the Vegetation Index characteristics.

This showing leadership factor the index (CMFI),

infrared proportion index of vegetation (IPVI), the
changed dirt organisms the index in changes
(MSAVI), the width to height ratio (BR), the square
root of the arrive relation (SQBR), the vegetable index
(VI), the mean of the brightness the index (ABI),

© 2024, IJSREM | www.ijsrem.com | Page 4

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

resolution. As a result, the model training did not

incorporate these properties.

Texture Coefficients of
Specification Correlation

haralick1 −0.012057

haralick2 −0.011231

haralick3 0.004898

haralick4 −0.029392

haralick5 0.065032

Figure 6. Strong or even perfect relationships haralick6 −0.123112

exist between vegetation index featurepairs.
haralick7 −0.029654

Index of Vegetation Coefficients of Correlation haralick8 −0.054594

NDVI 0.47121 haralick9 −0.016616

IPVI 0.47121 haralick10 0.016184

CMFI −0.47121 haralick11 −0.009838

BR 0.443567 haralick12 0.060944

SQBR 0.458123 haralick13 −0.014681

VI 0.508448 gxy_mean −0.004639

MSAVI 0.468535 gxy_std 0.022363

Table 2 presents an analysis of the relationship gx_mean −0.002482

between cabbage growth stages and vegetation
gx_std 0.005792
indexes.
gy_mean 0.000629
Selection of Texture Features:
Thirteen characteristics and six image gradient gy_std 0.045106
attributes (gx, gy, and gxy mean and standard
deviations) (a total of 19 attributes) determined using Table 3.relationships between growth phases of
the GLCM technique represent the various textural cabbage and textural characteristics.
qualities of the satellite pictures under examination.
Nevertheless, the correlation analysis reveals no
relationship between any of the textural characteristics
and the cabbages' growth days (Table 3). It's likely
that the satellite imagery used had insufficient

© 2024, IJSREM | www.ijsrem.com | Page 5

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

5. CONCLUSION
All things considered, the "Crop Recommendation
System Using Multi-Temporal Satellite SAR Data" is
a noteworthy development in agricultural technology.
The system offers farmers accurate crop
recommendations based on crop dynamics, soil health,
and moisture levels by utilizing deep learning and
SAR data processing. This fusion of cutting-edge
technology and conventional agricultural methods
encourages crop management efficiency and
sustainability. Enhancing decision-making and
optimizing resource utilization are the goals of the
system, which provides farmers with useful insights
Figure 7:Backstage interpretations of the field survey
from satellite data. Advances in satellite technology
images by an agricultural professional.
and continuous algorithmic refining will greatly
increase the system's accuracy and usability. In the
end, this approach aims to foster resilience in farming
methods and environmental stewardship while
boosting agricultural productivity.

6. REFERENCES
Figure 8 :Agricultural professionals interpret the
growth days of cabbages according to the field survey 1. N. H. Kulkarni, G. N. Srinivasan, B. M. Sagar and
images. N. K. Cauvery, "Improving Crop Productivity
Through A Crop Recommendation System Using
Ensembling Technique," 2018 3rd International
Conference on Computational Systems and
Information Technology for Sustainable Solutions
(CSITSS), Bengaluru, India, 2018, pp. 114-119,doi:
10.1109/CSITSS.2018.8768790.

2. Siesto, G.Fernández-Sellers, M. Lozano-Tello, A.

Crop Classification of Satellite Imagery Using
Synthetic Multitemporal and Multispectral Images in
Figure 9: Satellite images of cabbages (60 days).
Convolutional Neural Networks. Remote
Sens.2021,13,3378.
https://doi.org/10.3390/rs13173378

3. Pratik Jadhav, Vishwambhar Kachave, Aakash

Mane, Kavita Joshi, “Crop Detection using Satellite
Image Processing”April - June'2023. DOI
:https://doi.org/10.26634/jip.10.2.19800.

4. M. Venkatanaresh, Kullayamma “A new approach

for crop type mapping in satellite images using hybrid

© 2024, IJSREM | www.ijsrem.com | Page 6

 International Journal of Scientific Research in Engineering and Management (IJSREM)
Volume: 08 Issue: 07 | July - 2024 SJIF Rating: 8.448 ISSN: 2582-3930

deep capsule auto encoder” 28 November 8.Kulkarni,Nidhi,H,G.N.Srinivasan,B.M.Sagar,andN.

2022https://doi.org/10.1016/j.knosys.2022.109881. K. Cauvery. “Improving Crop Productivity through
aCrop Recommendation System using
5. Kumar, Y.Jeevan Nagendra,
Ensembletechnique” in 2018 3rd International
V.Spandana,V.S.Vaishnavi, K.Neha and
Conference onComputational Systems and
V.G.R.R.Devi. “Supervisedmachinelearning approach
Information Technology forSustainable Solution
for crop yield prediction inagriculture sector”. In 2020
(CSITSS), pp. 114-119. IEEE, 2018.
5th International Conferenceon
Communication and Electronics 9.Rajak, Rohit Kumar, Ankit Pawar,
systems(ICCES),pp.736-741.IEEE, 2020. MitleePendke,Pooja Shinde, Suresh Rathod and
Avinash Devare,“Croprecommendation system to
7. Reddy,D.Anantha, Bhagyashri Dadore and
maximise crop yield usingmachine learning
AartiWatekar “Crop recommendation system to
technique”.International ResearchJournal
maximisecrop yield in ramtek region using machine
learning”.International Journal of Scientific Research
in Scienceand Technology 6, no. 1 (2019): 485-489
of Engineering and Technology 4, no. 12(2017):950-
953.

6. A. Sharma, A. Jain, P. Gupta and VChowdary,

"Machine Learning Applications for Precision
Agriculture: A Comprehensive Review," in IEEE
Access, vol. 9, pp. 4843-4873, 2021, doi:
10.1109/ACCESS.2020.3048415.

10. S. Pudumalar,E. Ramanujam, R. Harine Rajashree,

C. Kavya “Crop recommendation system for precision
agriculture” January 2017Conference: 2016 Eighth
International Conference on Advanced Computing
(ICoAC) DOI:10.1109/ICoAC.2017.7951740.

11. Suyash Gholap, Harsh Singh, Rohan Rajesh,

Shweta Barshe, February 11, 2024, "Crop
Recommendation", IEEE Dataport, doi:
https://dx.doi.org/10.21227/3ays-5h79.

12.Sunanda Das, Aparna Menon, Ashima Malik, Jie

Hu “Crop Identification Based on Remote Sensing
Data using Machine Learning Approaches for Fresno
County, California”August 2021The 7th IEEE
International Conference on Big Data Service
Computing and Machine
LearningDOI:10.1109/BigDataService52369.2021.00
019.

© 2024, IJSREM | www.ijsrem.com | Page 7