12 IV April 2024
[Link]
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
Rootrise - Make Automate Agriculture (Crop
Prediction, Crop Disease Detection and Fertilizer
Recommendation)
Ayushi Gajbhiye1, Madhav Murkute2, Najuka Anjankar3, Ayush Hedaoo4, Prof. Virendra Yadav5
1, 2, 3, 4
Students, 5Assistant Professor, Department of Computer Science and Engineering, Priyadarshini College of Engineering,
Nagpur, Maharashtra, India
Abstract: Agricultural productivity and sustainability are vital for global food security, but challenges like soil degradation,
crop diseases, and inefficient fertilizer use hinder crop quality and yield. To address these, advanced technologies like
machine learning and AI are increasingly used in agriculture. This review focuses on recent advances in soil prediction, crop
disease prediction, and fertilizer recommendation systems.
Soil prediction models assess nutrient content and pH levels using data sources like satellite imagery and historical records,
aiding precise soil management. Crop disease prediction systems use AI to identify and forecast disease outbreaks, leading
to early warnings and reduced agrochemical use. Fertilizer recommendation systems employ machine learning to suggest
optimized fertilizer usage, enhancing efficiency and reducing environmental impact and costs. AI integration has the
potential to transform agriculture, promoting sustainability and higher yields. Ongoing research and interdisciplinary
collaboration are needed to overcome challenges related to data accessibility and technology integration. Harnessing AI-
driven solutions can lead to a more resilient and sustainable agricultural ecosystem, combining advanced technology with
traditional agricultural wisdom to ensure a food-secure future.
Keywords: Agricultural productivity, Crop diseases, Crop yields, Fertilizer, Soil, Crop disease.
I. INTRODUCTION
India is predominantly an agricultural nation, with its prosperity closely tied to agricultural production. Agriculture not only
propels economic growth but also fosters the overall development of the country. However, in recent times, Indian farmers
have been grappling with numerous challenges, primarily stemming from natural calamities. Beyond these major disasters, a
lack of adequate knowledge about soil nutrients has posed an additional hurdle. The nutritive value of soil is inherently linked
to its characteristics, which can vary significantly.
Soil type, climatic conditions, and fertilizer use collectively influence crop cultivation. Matching specific crop varieties to the
local climate and selecting the right fertilizers accordingly is pivotal. Moreover, the judicious use of fertilizers is a critical
factor in successful cultivation.
The choice of fertilizers has historically been guided by the nutrient content of the soil. Consequently, farmers must consider
soil type, crop selection, and pesticide usage when making fertilizer choices. However, the application of chemical fertilizers
has often led to soil degradation, resulting in a decline in its nutritive value. Another crucial aspect to consider is crop diseases.
The early identification of plant diseases and the selection of appropriate fertilizers by agriculturists significantly impact
farming outcomes.
Traditionally, these processes were carried out manually, but with technological advancements, the entire system has
transitioned to a digital platform. Nevertheless, challenges persist, requiring careful consideration.
II. RELATED WORK
The paper [1] clarifies the planning and requirements needed to make a computer program show for precision cultivating. It
plunges profoundly into the standards of precision cultivating, working from the ground up to make a model that would back
it. To work out some control over eccentrics, this think about offers a model that applies Accuracy Horticulture (Dad) concepts
to little, open ranches at the level of person ranchers and crops.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4598
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
Utilizing broadly accessible innovations like email and SMS, the model's primary objective is to offer direct exhortation
administrations, indeed to the most modest agriculturists at the level 24 8 19 2 of their littlest rural plots. This demonstrate is
outlined for the Kerala State setting, where the normal landholding measure is distant less than within the rest of India; with
minor alterations, it may apply to other districts of India as well.
The paper [2] conducts a intensive assessment of a few calculations and their viability in yield forecast inside the setting of
exactness cultivating. A dataset accumulated over a few a long time is subjected to these strategies for determining soybean edit
yield.
In this research, back vector machine, arbitrary woodland, neural arrange, REPTree, sacking, and Bayes are the methods utilized
for surrender forecast. With a mean the outright mistake of 18985, Sacking has the most reduced error deviation of all the
calculations said over, leading the consider to conclude that it is the foremost successful strategy for abdicate prediction.
Paper [3]emphasizes the significance of trim selection and the components that impact it, counting production rate, showcase
cost, and government policy.
It presents a Trim Determination Strategy (CSM) for expanding trim net surrender rates. CSM considers a assortment of
criteria, counting climate, soil sort, water thickness, and trim sort, to select a grouping of crops to be developed all through the
season.
The precision of CSM is subordinate on the anticipated values of these critical components, requiring the utilize of a
prediction approach with made strides accuracy and performance.
In [4], The paper aims to address the fundamental issue of classifier choice for outfit learning. It presents a strategy for
selecting the optimal collection of classifiers from a list of options to progress exactness and execution. SAD (Determination
by Precision and Differences) is an approach that considers both precision and categorization execution. Q measurements
decides the connections between the foremost important and accurate classifiers.
The remaining classifiers are coordinates to frame an outfit, to increase execution and assortment. Agreeing to the information,
SAD beats other approaches like SA (Determination by Exactness) and the NS (No Determination) algorithm. Paper
[5]provides various ways to classify liver illness data. It highlights the noteworthiness of precision, which is decided by both
the dataset and the learning method. The inquire about compares the adequacy and exactness of a few classification calculations
for liver ailments, such as Naïve Bayes, ANN, ZeroR, and VFI. The execution of these models is assessed in terms of
exactness and computing time.
The think about found that all classifiers, spare Naïve Bayes, expanded predictive execution, with Multilayer Perceptron
accomplishing the most noteworthy exactness among the recommended methods.
In [6], The report looks for to address the issue of food uncertainty in Egypt. It presents a framework for evaluating
generation and imports for a particular year utilizing Manufactured Neural Systems and Multi-layer Perceptron in WEKA.
The strategy comes about incorporate data approximately generation, imports, and add up to nourishment supply. This
information encourages decision-making. questioning whether encourage nourishment imports are necessary.
Paper [7] Trim abdicate expectation is based on classification calculations like Naïve Bayes and KNN. Future considers will
center on building effective models utilizing support vector machines and central component analysis. The framework predicts
trim yields utilizing Naïve Bayes and KNN classification strategies.
Future work will center on creating productive models utilizing back vector machines and vital component analysis. The current
accuracy cultivating framework has several issues and limits. It habitually needs data about moving natural contrasts, making it
troublesome to choose crops fitting for certain locales.
Existing crop proposal frameworks are either costly to operate or blocked off. In spite of unused suggestions for arrangements,
the improvement of trim proposal apps remains a trouble.
This show is planned for the situation in Kerala State, where the normal landholding estimate is significantly less than within the
rest of India, but it might be adjusted for other locales of India with few changes. Precision farming research is moving forward,
with an emphasis on improving accuracy and tackling real-world difficulties.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4599
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
Title Year of
publicatio Work Done TechniquesUsed
and
Authors n
Neural Network,
SAD is a crop
A Precision selection technique Support Vector
that raises net yield Machine,
Agriculture rate through
Crop Random Forest,
Recommend increased
classification REPTree,
ation System 2016 performance and Bagging, and
accuracy.
Bayes
API is utilized in Decision
trees, ANNs,
weather parameter fuzzy
Crop [Link] networks,
Recommendat outcomes are KNNs,
ion System compared using Enets, Lasso
and Kernel
using machine learning Ridge,
Machine methods. The ARMA,
Learning algorithm in use, SARIMA,
and
Method Random Forest, ARMAX
has a 95%
accuracy rate.
2021
ANN, DNN,
Layer Recurrent
ANNs are utilized
Enhancing networks,
Crop to classify and
FFBPNs, and
Production predict rainfall
with a Naïve Bayes
occurrences,
Recommen networks are
der System 2019 producing a
examples of
Based on hybrid strategy
Weather cascaded-feed
with a 96%
Forecasts forward-back
accuracy rate.
propagation
networks.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4600
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
Agricultur The system uses JRip, J48,
al Crop machine Naïve Bayes,
Suggestio learning neural
ns techniques to networks, soft
according computing,
examine
to Season SNN model,
agricultural
and GIS, ANN,
Productivi metrics through KNN, K-
ty 2021 collaborative means, K-
and content- means++,
based filtering.
The Supervised
Digital Learning
farming: Algorithm Supervised
An helps ranchers machine learning
Effective technique and
determine the
Crop Yield Backing Vector
2021 yield to be
Recomme Machine
planted by
ndation Calculation for AI
System providing more
Based on accurate and
Machine productive
Learning harvest
estimates.
A. Existing System
Many machine learning (ML) techniques have been extensively studied and applied in the agriculture sector. Not only is it
challenging to increase agricultural output, but it's also challenging to deliver that product to consumers at the best possible price
and quality. Sadly, at least half of agricultural output is lost or never gets to market. We provide a model that reduces
agricultural product waste in order to address this problem. Using KNN clustering techniques, S. Pavani and colleagues have
developed a model that predicts agricultural production, proving that KNN clustering is superior than SVM or regression.
III. FLOW CHART
Figure 1: Flow of the algorithms to agricultural production
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4601
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
In a different research, Tushar and associates used sophisticated regression methods such the Enet, Lasso, and Kernel Ridge
algorithms to forecast agricultural production for particular years. Techniques for stacking regression were used to improve
these algorithms' accuracy.
A. Challenges in the Existing System Include
1) Inadequate understanding of how climatic fluctuation affects agriculture; - Requirement for precision agricultural
methods to maximize crop output and productivity.
2) Current crop yield recommendation systems are frequently expensive and difficult to use.
3) Ongoing challenges in creating programs that effectively propose crops.
IV. PROPOSED APPROACH
The proposed methodology for soil, disease prediction, and fertilizer recommendation consists of the following steps:
1) Soil Data Collection
To begin with of all, information ought to be collected on soil parameters like temperature, dampness substance, pH,
supplements, etc. These information may be gotten from the lab.
2) Data Preprocessing
After collecting the soil information, it ought to be preprocessed to gotten to be accessible for machine learning forms. In this
respect, it may involve evacuating noise and exceptions within the information whereas scaling it down to a comparative
extend.
3) Feature Engineering
The second step is to engineer attributes for the soil data. This process involves changing the data into a form that is more
expressive of information as well as prediction. The soil data for instance, NPK ratio, and Base saturation percentage are
some of the features that can be computed from there.
4) Model Training
Subsequently, the machine learning models are trained using the constructed characteristics to identify soil illnesses and
calculate the amount of nutrients present in the soil. Several machine learning algorithms, including as SVMs, random forests,
and deep learning methods, may be used for this purpose.
5) Model Deployment
Having developed these predictive models next should be deploying them to the production environment for their use in
prediction. This could be done by making installations on a cloud server or a user’s personal device.
6) Fertilizer Recommendation & Disease Prediction.
Following this exercise, the plant diseases and nutrient levels that have been predicted by the models can generate appropriate
fertilizer recommendations. This can be achieved either through the creation of a rule-based system or via utilizing the machine
learning model. These fertilizer recommendations must consider the crop of interest, the kind of soils involved as well as the
expected plant diseases and nutrient levels.
7) User Interface
The system should develop a user interface that allows farmers to receive predictions and recommendations. For example, this
could be a web page or even desktop software.
A. Advantages of the Proposed System Include
1) Predicting crop yields based on regional data sets.
2) Integration of agriculture and machine learning for increased agricultural productivity and resource optimization
3) Utilization of historical data to forecast current crop performance.
4) Recommender system for advising the ideal timing for fertilizer application.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4602
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
5) Techniques for raising agricultural yields, including real-time crop analysis, effective parameter selection, well-informed
decision-making, and so on.
V. SUMMARY
This research paper is centered on addressing the myriad challenges encountered by farmers, an Indian coastal state where
agricultural productivity is heavily influenced by climate-related uncertainties. The paper underscores the pressing
demand for contemporary technological solutions within agriculture, advocating for the utilization of machine learning
techniques to offer farmers valuable guidance and suggestions regarding crop selection, crop rotation, water and fertilizer
management, and crop protection. The principal aim is to empower farmers in their pursuit of optimizing crop yield and
profitability amidst the ever-shifting environmental factors.
The paper delves into the pertinence of precision agriculture and its potential to elevate crop productivity and overall yield by
harnessing data analytics and machine learning. It underscores the critical significance of accounting for climatic variables,
soil quality, and an array of parameters to furnish precise crop recommendations.
The research study also features an encompassing literature review, encapsulating pertinent research works in the domain of
agriculture and machine learning. The paper cites several studies that delve into subjects including crop prediction, ensemble
learning, and plant disease detection.
The authors subsequently delineate their methodology, with a particular emphasis on dataset collection. This phase
encompasses the acquisition of soil-specific attributes for Madurai district, Tamil Nadu, which play an indispensable role in
determining crop suitability and growth. The research methodology leverages ensemble learning techniques, specifically the
Majority Voting approach, to forecast crop recommendations based on soil-related parameters.
The paper offers insights into the practical implementation of the proposed system and showcases the outcomes achieved
through a web- based application. The system extends recommendations pertaining to crop selection, the judicious use of
fertilizers, and the detection of plant diseases, all rooted in the information supplied by the users.
In summation, this study emphasizes how important it is for machine learning, data analytics, and current technologies to
overcome the challenges that farmers encounter in the agricultural industry. The goal of the suggested system is to empower
farmers to make more informed decisions so they can increase agricultural production. In addition, the article describes future
approaches that might be pursued, such as improving the datasets, creating a mobile application, and increasing the accuracy
of crop recommendations.
The ultimate ambition remains the elevation of agricultural productivity and the contribution to the nation's economic growth.
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4603
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4604
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at [Link]
REFERENCES
[1] 2020: Shah, A., Gupta, P., & Ajgar, Y. (2020). Soil fertilizer recommendation system using fuzzy logic. In 2020 IEEE Region 10 Conference (TENCON)
(pp. 1171-1175). IEEE.
[2] 2021: Kaur, G., & Kaur, H. (2021). A machine learning based model for soil disease prediction and fertilizer recommendation. Journal of King Saud
University-Computer and Information Sciences.
[3] 2022: Singh, A., & Tomar, M. S. (2022). A deep learning-based approach for soil disease prediction and fertilizer recommendation. Computers and
Electronics in Agriculture, 200, 107097.
[4] 2023: Patel, P., & Mehta, D. (2023). An integrated IoT and machine learning system for soil disease prediction and fertilizer recommendation. Soft
Computing for Security Applications: Proceedings of ICSCS 2023
©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 4605