International Journal of Scientific Research in Science and Technology

Available online at : www.ijsrst.com

Print ISSN: 2395-6011 | Online ISSN: 2395-602X doi : https://doi.org/10.32628/IJSRST24114312

Enhancing Agricultural Sustainability through AI-Powered

Image Processing: Review Study on Plant Disease Detection
Khushwant Kaur, Meena Jindal*
Assistant Professor, Sri Guru Gobind Singh College, Chandigarh, India

ARTICLEINFO ABSTRACT
The agricultural field is encountering multiple problems with the climate
Article History:
changing, population booming and overusing chemical pesticide which all
Accepted : 18 Nov 2024 lead to unsustainable agriculture. Affecting quality and yield, plant
Published : 10 Dec 2024 diseases account for a heavy loss from the final production. Conventional
plant disease detection is definitely the aforementioned matter as well,
profound education analyzing with labor-intensive and time-consuming
Publication Issue : procedure yet not so accurate. By merging artificial intelligence (AI) with
Volume 11, Issue 6 image processing, plant disease diagnosis can be automated quickly and
November-December-2024 efficiently. It uses machine learning algorithms, combined with high-
resolution imagery to detect disease symptoms in the early stage of
Page Number : infestation thereby making the treatment process largely dependent on
490-496 chemical control. In this paper, we reviewed state-of-the-art methods
which have experience significant improvement and development in terms
of image processing approaches using AI for plant disease recognition. We
made a lot of progress however there are still many gaps to fill like other
data types, real-time processing and generalizability models that need to be
incorporated with farming practices as well accessibility considering all the
factors is important for economic viability. Overcoming these gaps requires
a holistic approach by combining AI innovations with perspectives from
the fields of agronomy and agricultural economics. Future research could
potentially concentrate in improving the real-time process, increasing
model interpretability and integration with current agricultural systems.
Overcoming these challenges, AI-powered image processing can be the
backbone of precision agriculture that could secure our food supply and
make farming more sustainable.
Keywords: AI, image processing, plant disease detection, sustainable
agriculture, precision agriculture, machine learning, early intervention,
food security.

Copyright © 2024 The Author(s): This is an open-access article distributed under the terms of the Creative 490
Commons Attribution 4.0 International License (CC BY-NC 4.0)
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

I. INTRODUCTION who does not (only) contributes on productivity

but also encourages environmentally friendly
The world is at a tipping point where agriculture, business trends to reduce the overuse of chemicals
one of the cornerstones of human civilization, [3].
confronts unprecedented challenge. Worldwide In this extensive review, the focus is to cover
increases in both temperature and population various recent approaches being undertaken in
place this sustainability at risk, with climate the domain of Machine learning for image
change threatening to disrupt food production processing which are further helping towards
through increasing the intensity of droughts or plant disease detection. This review will also
torrential downpours while introducing new pests detail the wide range of methodologies and
that previously could not survive in prohibitively algorithms used in that domain, assess where
cold regions. Among these challenges are plant current research stands, propose areas for
diseases, which can significantly impact crop improvements through unaddressed critical
productivity by reducing both the quality and questions within existing literature itself or by
quantity of agricultural yield. Currently, just outlining few grey spots to be further clarified
monitoring plant diseases in an accurate and before asserting them as limitations as well give
timely manner is essential to reduce the negative prospects which lays ahead for it. It will involve a
impacts on crops. Common traditional plant comprehensive survey of laboratory techniques,
disease detection methods require human experts the success rates that can be expected from these
and are time-consuming, labor-intensive, and methods and how they may support new
easily subjective.[2] philosophies in sustainable plant health [4].
Over the recent years, plant disease detection has The AI integration in plant disease detection is
been revolutionized through advancements in seen as an arm of precision agriculture, a broader
artificial intelligence (AI) and image processing revolution where data-fueled decision-making
technologies. Image processing using AI in plants optimizes the sustainability and profitability of
can solve this problem by making automated farming. In the case of agriculture, precision
faster and accurate diagnosis of plant diseases. monitoring and management-meaning high
Using machine learning algorithms and high- accuracy in all operations-are achieved by using a
resolution imaging, this technology is able to series of technologies such as remote sensing (e.g.,
identify disease symptoms that the human eye satellite or drone imagery), geographic
may not catch. Using AI for agriculture in the information systems (GIS) for mapping every
field of plant disease detection will be a mile stone little aspect on agricultural lands, alongside
to increase sustainability, and reduce dependence artificial intelligence. Among the aforementioned,
on chemical treatment due to early prediction one of the most important tools for disease
through timely intervention [2]. management is AI-powered image processing
The applications of AI and image processing can with a high scalability over numerous crop types
immensely change the game in agriculture. and farming conditions [5].
Farmers can do appropriate treatment to In essence, this study is valuable because it
eliminate the pathogens and deficit reduction at suggests a complete understanding on how AI-
early stages of disease detection, this would motivated image processing can be used to
regulate crop losses causes by diseases and maybe mitigate one of the most challenging problems
gain a good yield quality. Extra virgin olive oil currently applying in agriculture. Through

International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 491
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

synthesis of existing knowledge and possible 2. Less Chemical Usage: Accurate detection also
directions to be followed in the research, this means fewer chemicals are used in treating
communication will help endeavor sustainable the field - which would help avoid
agricultural practices. Plant diseases contribution unnecessary blanket applications of pesticides
will negatively affect the enhancement of food and promote more targeted, environmentally
supplies and ecological equilibrium in future friendly pest management.
given that demand for plant-derived products 3. It will also help in Enriching Food Security:
increases with population increase around the There is an increasing consumption
world [6,7]. worldwide, therefore securing the same
The motivation behind the work is described, standard of agricultural growth year after
followed by a table summarizing three traditional another to produce a uniform high-quality
methods of detection and five AI approaches for output. AI-based detection facilitates this by
plant disease identification (sect. 4), current ensuring that the losses arising out of disease
knowledge issues are identified from there after are brought to a minimum.
relevant research questions highlighted in section 4. Fansler labeled this system as eco-unfriendly,
four subsection six with challenges encountered promoting sustainable practices of reducing
while implementing this idea. crops chemical inputs and improving plant
disease management rather than damaging
II. MOTIVATION the environment over a long period which
gets in line with goals of sustainable
The primary goal of investigating AI-driven agriculture - an ecosystem approach to create
image processing for plant disease detection as an conditions for ecological balance where
agronomical approach is the prompt application diverse forms of fully integrated agro-
to promote agricultural sustainability. Current ecosystems decreased sharply any losses that
methods of disease detection are often slow and threaten its long-term productivity.
inefficient resulting high crop losses and overuse This study aims to comprehensively review the
of chemical treatments. AI has been proposed to advancements and applications of AI in plant
revolutionise this process, providing possible disease detection, providing insights that can
quick and accurate diagnosis that is based on drive further research and practical
automated detection of endotypes [8]. implementations in the field.
Key motivations include:
1. Increasing Crop Yields: This will help III.LITERATURE REVIEW
farmers catch the disease early and be able to
treat it before it spreads, which is vital in
order to maintain crop yields [9].
Study Methods Used Key Findings Strengths Limitations
Convolutional High accuracy in
Sladojevic et High accuracy, robust Requires large
Neural Networks detecting 13 different
al. (2016) model datasets
(CNN) plant diseases
Accurate classification of Effective use of
Mohanty et Deep Learning, Computationally
26 diseases across 14 crop transfer learning,
al. (2016) Transfer Learning intensive
species extensive dataset
International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 492
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

Study Methods Used Key Findings Strengths Limitations

Enhanced model
Amara et al. CNN, Data Improved detection of Limited to specific
performance with data
(2017) Augmentation tomato leaf diseases crops
augmentation
High performance in
Ferentinos Deep Learning, High real-world Requires high-quality
identifying plant diseases
(2018) Image Segmentation applicability images
in real-field conditions
Comparison of traditional
Comprehensive
Barbedo Image Processing, and machine learning Variability in dataset
comparison, practical
(2018) Machine Learning methods for disease quality
insights
detection
Efficient disease High efficiency, Trade-off between
Too et al. CNN, MobileNet,
detection using potential for mobile model size and
(2019) VGG16
lightweight models applications accuracy
Zhang et al. Deep Learning, Simultaneous detection Enhanced detection Complexity in model
(2020) Multi-task Learning of multiple diseases capability training
Generative
Improved disease Effective data
Li et al. Adversarial High computational
detection through data generation, enhanced
(2021) Networks (GAN), cost
augmentation detection
CNN
Ensemble Learning,
Hasan et al. High accuracy in multi- Robust performance, Complexity in
Random Forest,
(2021) class disease classification enhanced accuracy ensemble methods
SVM
Effective use of
Kaur et al. CNN, Transfer High accuracy with Dependence on pre-
transfer learning,
(2021) Learning limited training data trained models
reduced training time
Hyperspectral Enhanced disease
Picon et al. High precision, useful Requires specialized
Imaging, Machine detection using spectral
(2019) for specific crops equipment
Learning data
Limited by
Ramcharan Smartphone-based Practical application for High usability, mobile
smartphone camera
et al. (2019) Detection, CNN field use integration
quality
Liu et al. Deep Learning, IoT Real-time disease Real-time capability, Complexity in system
(2020) Integration detection and monitoring IoT integration integration
Chen et al. Attention Improved accuracy with Enhanced focus on Increased model
(2021) Mechanisms, CNN attention-based models relevant features complexity
Integrating various data
Saleem et al. Deep Learning, Improved accuracy Complexity in data
sources for better
(2020) Data Fusion through data fusion integration
detection

International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 493
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

IV. GAPS Addressing these gaps will require a multidisciplinary

approach, combining advances in AI with insights
While progress has been made in the development of from agronomy, plant pathology, and agricultural
AI-based image processing for plant disease economics.
recognition, there are still a number of deficits that
must be addressed to ensure this technology will V. RESEARCH QUESTIONS
reach its full potential.
1. The focus on individual data sets may not be To address the gaps identified, the following research
scalable to the plethora of crops and disease questions are proposed:
conditions encountered worldwide. Additional 1. How to adapt AI-powered image processing
datasets, including those that span diverse plant techniques for disease detection in a variety of
species, growth stages and environmental real time environments on the farm?
backgrounds are required [11]. Develop lightweight models to run on field
2. Some methods show promise for real-time deployable, portable hardware.
processing, but many are computationally 2. Which are the methods that could be
intensive and not optimized to be used in near- implemented to ensure more general
real time out of the lab. More precisely, we need transferability of AI models between crops and
to research for building light and portable regions?
challengeable image capable models that can run Understand Technologies and Open Problems in
on phones, drones,[12] etc. Transfer Learning, Domain Adaptation and
3. Generalizability: The existing models failed to Diverse Dataset Creation.
generalize across regions and crops. There is a 3. How can we make AI models for plant disease
need for developing models that can transcend detection work smoothly with the existing
different agricultural contexts without requiring principal methods of farming and bring them
extensive retraining [13]. into common use?
4. Capability to integrate with Farming Practices - Develop intuitive interfaces and systems for use
The AI systems should be able to work closely or by all farmers irrelevant to their IT skills o SDG /
in alignment with existing farming practices and Milestones: ~ Commence implementation in year
technologies. User-friendly and scalable solutions 3 of grant.
are required for accepting it by all type of farmers 4. In Plant Disease Detection: Which Method
from the novice to technology savvy [14]. would Enhance the Explainability and
5. Explainability and Trust: Most AI models Transparency in AI Models?
(especially deep learning ones) are black boxes, Use explainable AI techniques to provide insights
i.e it is hard to understand how decisions were on the decision-making behind an artificial
made. This is critical in improving the agent.
trustability of models amongst farmers and 5. How do you ensure an AI based plant disease
agricultural experts [15]. detection system is economically viable for small
6. Cost Factor - It can be too expensive for a small- farmers?
scale farmer to actually employ these AI-powered Develop affordable response strategies and
solutions. Future research should evaluate cost- scalable technologies that can be applied to
effective approaches that can be transferred to multiple economic settings.
any farm scale and economic situation.

International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 494
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

VI. CHALLENGES VII. CONCLUSION

Implementing AI-powered image processing for plant The developing AI-enabled image processing for plant
disease detection involves several challenges: disease detection is an important innovation in
1. Data Collection and Annotation - To train a agriculture that can improve sustainability, increase
robust AI model, we need high-quality annotated crop yields and reduce chemical treatments. Though
datasets. The process of collecting and annotating it is a very promising technology, there are many
large datasets is time-consuming, labor-intensive challenges and gaps that need to be addressed for its
and domain-specific in nature (plant pathology) ubiquity. In the future, more research should be
[17]. carried out on real-time performance improvement
2. Computational Resources: Training and modelling generalizability integration with farm
deploying AI models, in particular deep learning practices model interpretability economic feasibility
based ones requires large computational of this technology for all farmers.
resources which may be unfeasible for resource Addressing these issues will enable AI-powered image
constrained settings [18]. processing to play a central role in precision
3. Complex Algorithm: To develop the most agriculture, thereby helping secure food for the world
effective models, complex algorithms are and avoiding unnecessary land harm. This review
sometimes required which can be difficult to reveals the existing literature and identifies emerging
design and support for non-experts. trends based on various study designs with
4. Adapting AI applications to work alongside recommendations for strengthening both applied in
current agricultural systems: When incorporating vivo research, empirical modeling of distal
an AI solution, it is essential that custom mechanisms as a transparent basis for theory-based
solutions are developed and implemented for hypothesis testing. Through further innovation and
different types of farming operations. interdisciplinary collaboration, AI-based plant disease
5. Lack of Model Interpretability: Many AI models detection can have an unprecedented impact on the
are functioning as a black box, which simply future of agriculture with sustainability in mind.
means that most people cannot truly trust the
output or even understand why model behaving VIII. REFERENCES
in certain way. Explainability matters more than
ever for the adoption of these models. [1]. Sladojevic, S., Arsenovic, M., Anderla, A.,
6. Another is economic and social barriers, where Culibrk, D., & Stefanovic, D. (2016). Deep
the cost of AI systems can be too high for small- Neural Networks Based Recognition of Plant
scale farmers to implement. It is critical that Diseases by Leaf Image Classification.
these solutions are affordable and available so Computational Intelligence and Neuroscience,
they finally get adopted. 2016.
7. Ethical and Privacy Issues: Agricultural data, [2]. Mohanty, S. P., Hughes, D. P., & Salathe, M.
especially if it gets integrated with IoT devices (2016). Using Deep Learning for Image-Based
alarms privacy and security of that particular Plant Disease Detection. Frontiers in Plant
region. It is critical to ensure ethical use of data. Science, 7, 1419.
[3]. Amara, J., Bouaziz, B., & Algergawy, A. (2017).
A Deep Learning-based Approach for Banana

International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 495
 Meena Jindal et al Int J Sci Res Sci & Technol. November-December-2024, 11 (6) : 490-496

Leaf Diseases Classification. Datenbank- [13]. Liu, B., Zhang, Y., He, D., & Li, Y. (2020).
Spektrum, 17(3), 11–21. Identification of apple leaf diseases based on
[4]. Ferentinos, K. P. (2018). Deep learning models deep convolutional neural networks.
for plant disease detection and diagnosis. Symmetry, 12(7), 1065.
Computers and Electronics in Agriculture, 145, [14]. Chen, J., Zhang, D., Jiang, L., & Mao, Y. (2021).
311-318. Attention-based convolutional neural network
[5]. Barbedo, J. G. A. (2018). Factors influencing the for leaf disease recognition. Computers and
use of deep learning for plant disease Electronics in Agriculture, 187, 106318.
recognition. Biosystems Engineering, 172, 84- [15]. Saleem, M. H., Potgieter, J., & Arif, K. M.
91. (2020). Plant Disease Classification: A
[6]. Too, E. C., Yujian, L., Njuki, S., & Yingchun, L. Comparative Evaluation of Convolutional
(2019). A comparative study of fine-tuning deep Neural Networks and Deep Learning
learning models for plant disease identification. Optimizers. Plants, 9(10), 1319.
Computers and Electronics in Agriculture, 161, [16]. Patel, K. K., & Kar, A. (2016). Advances in
272-279. image processing for detection of plant diseases.
[7]. Zhang, K., Zhang, L., & Wang, L. (2020). Crop Springer.
Disease Recognition via Attention Mechanism [17]. Barbedo, J. G. A. (2019). Plant disease
and Multiple Convolutional Neural Networks. identification from individual lesions and spots
Neural Processing Letters, 51, 1485-1501. using deep learning. Biosystems Engineering,
[8]. Li, Y., Qian, P., & Wang, Q. (2021). Improved 180, 96-107.
Generative Adversarial Networks-Based Crop [18]. Wang, G., Sun, Y., & Wang, J. (2017).
Disease Recognition Model. IEEE Access, 9, Automatic image-based plant disease severity
109781-109789. estimation using deep learning. Computational
[9]. Hasan, M. M., Chopdar, M. S., & Sarkar, S. Intelligence and Neuroscience, 2017, 1-8.
(2021). Detection and recognition of rice plant
diseases using deep convolutional neural
networks. Computers and Electronics in
Agriculture, 183, 105950.
[10]. Kaur, P., Singh, A., & Pandey, S. (2021). Plant
Disease Identification and Classification
through Leaf Images: A Deep Learning
Perspective. Journal of Plant Pathology, 103, 1-
11.
[11]. Picon, A., Bulanon, D. M., Burks, T. F., &
Alchanatis, V. (2019). Hyperspectral imaging
for plant disease detection: A review.
Biosystems Engineering, 178, 24-35.
[12]. Ramcharan, A., Baranowski, K., McCloskey, P.,
Ahmed, B., Legg, J., & Hughes, D. P. (2019). A
mobile-based deep learning model for cassava
disease diagnosis. Frontiers in Plant Science, 10,
272.

International Journal of Scientific Research in Science and Technology (www.ijsrst.com) | Volume 11 | Issue 6 496