Grenze International Journal of Engineering and Technology, June Issue
Machine Learning Techniques for Banana Disease
Detection: A Comprehensive Review and Analysis
Dr. Pankaj Dashore1 and Dr. Rachana Dashore2
1
Professor SOCSE, Sandip University. Nashik
Email: dashorepankaj@[Link]
2
Professor, MBA dept, SITRC, Sandip Foundation, Nashik
Email: dashorerahna@[Link]
Abstract— This research paper investigates the utilization of machine learning methods in the
domain of banana disease detection, with the objective of improving the accuracy and
effectiveness of disease diagnosis. Many diseases pose serious risks to banana crops, which are
essential for maintaining global food security and can result in large yield losses. This study offers
a comprehensive analysis of the situation of both conventional and contemporary methods for
detecting banana disease. The study focuses on current developments that have improved the
precision and effectiveness of disease diagnosis, including the use of cutting-edge technology like
machine learning and remote sensing.
This research could completely change the way banana diseases are managed by giving farmers
a reliable, accurate, and timely tool for early disease detection. For banana growers, this can
result in higher-quality produce, higher yields, and more profitability. The created system can
also be modified to identify illnesses in different crops, which will greatly enhance agricultural
output and food security overall.
Index Terms— Global food security Banana diseases, image processing, machine learning, disease
detection, classification.
I. INTRODUCTION
Bananas, a staple food for millions, are vulnerable to diseases like Panama disease, Black Sigatoka, and Bunchy
Top. These diseases can have devastating consequences, threatening not only farmers' livelihoods but also global
food security.
Historically, visual inspection and symptom-based identification have been the main methods used for disease
detection in banana crops. Although these techniques have proven useful, they are time-consuming, labor-
intensive, and prone to human error. Modern technologies are being investigated as a result of the demand for
methods that are more automated and efficient.
Numerous diseases and pests have an impact on banana production. Novel and rapid methods for detecting pests
and diseases will allow for more efficient surveillance and development of control measures. Deep learning, neural
networks and transfer learning have been successfully applied in a variety of fields for crop disease detection.
One of the mainstays of global agriculture, banana farming is essential to tackling issues related to food security.
However, a number of illnesses that can negatively affect yield and quality pose a threat to the sustainable
Grenze ID: [Link].10.2.501
© Grenze Scientific Society, 2024
production of bananas. The application of deep learning techniques to agricultural practices has demonstrated
encouraging potential in recent years for revolutionizing disease detection and providing a more accurate and
efficient substitute for conventional methods. This work aims to provide insights into the current state-of-the-art,
challenges, and future directions in this important field by presenting a thorough review and analysis of the
application of deep learning techniques for the detection of banana disease.
Fig1:- Banana Disease Detection Fig 2 Banana Disease Detection
It was observed that diseases affect all banana crops. Leaf, stem, flower, fruit, roots, and suckers all exhibit
symptoms. The most common leaf diseases are Xanthomonas wilt, Fusarium wilt, and black and yellow Sigatoka.
Banana Leaf Diseases
Banana Xanthomonas Wilt (BXW) is a bacterial infection caused by Xanthomonas campestris.
Wilt Fusarium It is commonly known as Panama infection, a deadly fungal infection caused by the soil-borne fungus
Fusarium oxysporum.
The Bunchy Top Virus is a plant pathogenic virus ancestor nanoviridae that contaminates banana plants.
Black Sigatoka is called a black leaf streak caused by mycosphaerella. The multifaceted Sigatoka infection is a
collection of intimately related fungi.
1. Methodologies
(i) Image Acquisition: Using smartphones or specialized cameras, high-resolution pictures of banana leaves with
symptoms of different diseases will be taken.
(ii) Image Processing : To improve the characteristics important for illness identification, acquired images will be
enhanced. Techniques like background removal, noise reduction, and color space conversion may be used in this.
(iii) Feature extraction: From the previously processed photos, pertinent features will be taken out. Color, texture, shape,
and the spatial relationships between the various affected areas are examples of these features.
(iv) Disease Identification and Classification: To analyze the retrieved features and identify the existence and kind of
disease, machine learning algorithms such as Support Vector Machines (SVM), Convolutional Neural Networks
(CNNs), or Artificial Neural Networks (ANNs) will be trained.
(v) Evaluation and Optimization: Using a sizable dataset of labeled photos of banana leaves, the developed system will
be assessed. Performance will be assessed using metrics such as F1-score, recall, accuracy, and precision. The
system will undergo additional optimization to enhance its performance, taking into account the evaluation results.
Fig 3. Shows the methodologies for Banana Disease Detection and the Expected outcomes.,
Fig 3.:- Methodologies for Banana Disease Detection
2
Root diseases can be avoided by keeping trees healthy and avoiding wounds in the root system and near the tree's
base. It is difficult to control a root disease once it has infected a banana tree.
II LEAST SQUARES REGRESSION:
The least squares method is a mathematical approach for fitting a model to data points that is widely used. While
it is not typically used directly for image-based classification tasks such as banana disease detection, it can be
useful in scenarios involving quantitative measurements or predictions. Alternative machine learning methods,
particularly those suited to image analysis, are more commonly used in the context of banana disease detection.
Convolutional neural networks (CNNs) have demonstrated significant success in image-based tasks such as
disease detection in bananas. CNNs excel at capturing spatial patterns and features within images, making them
useful for disease classification based on visual symptoms. Transfer learning, which involves leveraging pre-
trained models on large datasets, can also improve CNN performance when dealing with limited datasets.
For Banana Disease numbers of bacteria and the corresponding scatter graph, the line of best fit is the straight
line that can be drawn through the scatter points to best represent the relationship between them. We have the
straight line equation:
Y= mX + Ci
Where Y stands for Dependent Variable.
M stands for incline and Ci for constant values of Banana disease
X stands for Independent Variable.
The goal here is to calculate the slope and y-intercept values and then plug them into the equation along with the values
of the independent variable X that indicates the number of becteria to determine the values of the dependent variable Y.
Assuming we have 'n' data points, we can calculate slope using the frightening-looking formula below:
Then substitute these values to equation Y = m X + c.
The Fig. 4 Shows the Regression Line for Banana Disease.
Fig. 4- Regression Line for Banana Disease
2. Expected Results
(i) An automated and reliable system for classifying and detecting banana disease.
(ii) Faster and more accurate disease detection than with conventional techniques.
(iii) Decreased losses for banana growers and improved disease control efficiency.
(iv) Possibility of use with additional crop types.
3
III. CONCLUSIONS
The importance of this research is that it highlights the immense potential of machine learning techniques in
transforming the detection of banana diseases, providing a quicker and more precise alternative to conventional
approaches. These findings significantly contribute to the continuous endeavors in precision agriculture, ultimately
fostering sustainable banana cultivation and ensuring food security.
Furthermore, the developed system can be adapted to detect diseases in other crops, significantly contributing to
the overall improvement of agricultural production and food security . Best fit Least square method is used to .
The analysis of current methodologies and the synthesis of existing literature reveal the transformative potential
of deep learning in improving the efficiency and accuracy of banana disease detection.
Although the least squares method is not frequently employed for directly detecting banana diseases, it can be
utilized in associated fields like regression analysis to forecast quantitative outcomes such as disease severity or
crop yield using input variables. Nevertheless, when it comes to image classification tasks, contemporary machine
learning techniques, especially CNNs, are more widely used and efficient.
REFERENCES
[1] [Link], aki Bhuiyan , Hasan Muhammad Abdullah , Shifat E. Arman , Sayed Saminur Rahman d, Kaies Al
Mahmud .”BananaSqueezeNet: A very fast, lightweight convolutional neural network for the diagnosis of three prominent
banana leaf diseases, Smart Agricultural Technology,Volume 4, August 2023, 100214.
[2] Selvaraj, M.G., Vergara, A., Ruiz, H. et al. AI-powered banana diseases and pest detection. Plant Methods 15, 92 (2019).
[Link]
[3] S. L. Sanga, D. Machuve, and K. Jomanga, “Mobile-based Deep Learning Models for Banana Disease Detection”, Eng.
Technol. Appl. Sci. Res., vol. 10, no. 3, pp. 5674–5677, Jun. 2020..
[4] Akshara Gupta, Nidhi Kamath, Pallavi Mali, A Deep Learning Based Approach for Banana Plant Leaf Diseases
Classification and Analysis. IJCRT Volume 6, Issue 2 April 2018 ISSN: 2320-2882.
[5] K. Lakshmi Narayanan, R. Santhana Krishnan, Y. Harold Robinson, E. Golden Julie, S. Vimal, V. Saravanan, M.
Kaliappan, "Banana Plant Disease Classification Using Hybrid Convolutional Neural Network", Computational
Intelligence and Neuroscience, vol. 2022, Article ID 9153699, 13 pages, 2022. [Link]
[6] V. N. P and P. R, "Detection and Classification of Banana Leaf diseases using Machine Learning and Deep Learning
Algorithms," 2022 IEEE 19th India Council International Conference (INDICON), Kochi, India, 2022, pp. 1-6, doi:
10.1109/INDICON56171.2022.10039912.
[7] N. Bharathi Raja & P. Selvi Rajendran (2023) An efficient banana plant leaf disease classification using optimal ensemble
deep transfer network, Journal of Experimental & Theoretical Artificial Intelligence, DOI:
10.1080/0952813X.2023.2241867 .
[8] M. U. Kothalawala, M. G. Gaveshith K, A. H. D. H. Tharaka, I. A. Punchihewa and D. Sriyaratna, "Banana Disease
Identification Using Machine Learning Based Technologies and Weather-Based Dispersion Analysis," 2022 4th
International Conference on Advancements in Computing (ICAC), Colombo, Sri Lanka, 2022, pp. 234-239, doi:
10.1109/ICAC57685.2022.10025325.
[9] Harshit Bhardwaj; Pankaj Dashore, A novel genetic programming approach to control bloat using crossover and mutation
with intelligence technique, IEEE International Conference. 10-12 September2015, 10.1109/IC4.2015.7375619.
[10] Narayanan KL, Krishnan RS, Robinson YH, Julie EG, Vimal S, Saravanan V, Kaliappan M. Banana Plant Disease
Classification Using Hybrid Convolutional Neural Network. Comput Intell Neurosci. 2022 Feb 23;2022:9153699. doi:
10.1155/2022/9153699. PMID: 35251158; PMCID: PMC8890843.
[11] Selvaraj, M.G., Vergara, A., Ruiz, H. et al. AI-powered banana diseases and pest detection. Plant Methods 15, 92 (2019).
[Link]
