Deep Learning in Arecanut Research: A Comprehensive Survey

Abstract

Arecanut (betel nut) is an important tropical crop grown mainly in India, which ranks second worldwide in its production and consumption. However, Arecanut plants face numerous diseases affecting their leaves, trunks, roots, and fruits, some of which are hard to detect visually. However, its production is severely threatened by various leaf and nut diseases, particularly Yellow Leaf Disease (YLD), which causes substantial economic losses. Early and accurate disease detection is therefore crucial for ensuring sustainable cultivation. Recent advancements in machine learning and deep learning have enabled automated approaches for identifying, classifying, and predicting Arecanut diseases through image-based analysis. This survey paper provides a comprehensive review of existing works ranging from conventional image processing techniques to modern (CNNs) and transfer learning models such as MobileNetV2, ResNet, and VGG-16. Studies have demonstrated the effectiveness of hybrid methods combining CNN with SVM, as well as region-specific assessments of disease severity. Additionally, research on causal agents of YLD and spatial disease pattern analysis highlights the growing role of AI in precision agriculture. By comparing methodologies, datasets, and performance metrics across multiple studies, this paper identifies current challenges such as limited annotated datasets, variations in environmental conditions, and the need for real-time mobile applications. Finally, it outlines potential research directions including multimodal disease prediction, lightweight deep learning models for deployment on edge devices, and integration with Internet of Things (IoT)-based monitoring systems to achieve scalable and farmer-friendly solutions.

Introduction

1. History & Importance of Arecanut

• Arecanut has been cultivated for centuries in South and Southeast Asia, especially in India, where states like Karnataka and Kerala rely on it economically.

• The crop holds both economic and cultural significance, supporting millions of farmers.

2. Major Arecanut Diseases

• Yellow Leaf Spot: Causes yellow-ringed dark spots, leading to leaf drop if untreated.

• Yellow Leaf Disease (YLD): Turns leaves pale yellow, weakens the plant, and reduces yield.

• Mahali/Koleroga (Fruit Rot): Fungal infection that causes nut rot during rainy seasons.

3. Traditional Detection Challenges

• Relies on manual inspection, which is labor-intensive, slow, and error-prone.

• Difficult to scale and unsuitable for early detection, leading to economic losses.

4. Deep Learning Approach for Disease Detection

Workflow Overview

1. Data Collection: Images from fields, satellites, or mobile devices.

2. Preprocessing: Resize, segment, augment images for quality.

3. Feature Extraction: Use texture features (e.g., GLCM, LBP) or deep CNN layers.

4. Model Training: CNNs (ResNet, VGG16, MobileNetV2), trained using accuracy, F1, etc.

5. Deployment: Models integrated into apps/web platforms to assist farmers.

Advantages

• High accuracy (up to 98.9% with ResNet).

• Real-time support for farmers.

• Works well with remote sensing and field-level monitoring.

5. Background and Literature

• Arecanut diseases include YLD, fruit rot, stem bleeding, nut split, etc.

• Traditional methods lack scalability and robustness.

• CNNs outperform classical methods (SVM, RF, AdaBoost) in disease detection.

• Studies use datasets ranging from 180 to 11,000+ images, including satellite data.

Research Metrics Used

• Accuracy, precision, recall, F1-score, confusion matrix, ROC curves.

Tools & Frameworks

• Python, MATLAB for prototyping.

• TensorFlow/Keras, OpenCV, and scikit-learn for model development.

• Pretrained models (ResNet, VGG16, Inception) improve performance on small datasets.

6. Key Findings from Literature Survey

• CNNs dominate in usage (49.3%), followed by SVM (19.3%).

• Accuracy has steadily improved from 2020 (~90%) to 2025 (~99%) due to model advancements and dataset expansion.

7. Challenges Identified

1. Dataset Limitations: Small, region-specific, or non-standardized datasets.

2. Image Quality Issues: Noise, variability in lighting, and background clutter.

3. Model Challenges: Overfitting, generalization, and explainability issues.

4. Evaluation Gaps: Lack of consistent benchmarks or validation methods.

5. Tooling Constraints: Integration complexity across platforms.

6. Deployment Difficulties: Real-time performance and user adoption barriers.

7. Biological Validation: Uncertainty in causative agents (e.g., for YLD).

8. Scalability: Models often do not generalize well across regions or climates.

Conclusion

The reviewed studies show that both machine learning and deep learning methods have strong potential in detecting and classifying Arecanut diseases. Convolutional Neural Networks (CNNs) and their variants, such as ResNet, VGG16, MobileNetV2, and EfficientNet, consistently outperformed traditional techniques like SVM, Random Forest, and Backpropagation Neural Networks. Reported accuracies usually ranged between 85% and 95%, with some models achieving almost perfect performance under controlled conditions. Transfer learning and data augmentation proved especially useful in overcoming the challenges of small datasets. Feature-based methods like GLCM, Gabor, and LBP provided valuable baselines, while the use of satellite imagery extended detection to plantation-level monitoring. Pathology-focused studies also highlighted the biological complexity of Yellow Leaf Disease (YLD), showing the need for accurate labelling and interdisciplinary approaches. Despite these advances, several challenges remain. Most datasets are still small, region-specific, and restrict the generalization of models. Many studies emphasized accuracy, while paying less attention to precision, recall, or real-world validation. Overfitting also remains a recurring problem due to limited data diversity. In addition, practical deployment aspects—such as mobile applications, offline usability, and decision-support tools for farmers—are still at an early stage. Addressing these gaps will require larger and more diverse datasets, along with hybrid models that integrate biological, climatic, and imaging data.

References

[1] A. Hegde, V. S. Sadanand, C. G. Hegde, K. M. Naik, and K. D. Shastri, “Identification and categorization of diseases in arecanut: a machine learning approach,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 31, no. 3, pp. 1803–1810, Sep. 2023, doi: 10.11591/ijeecs. v31.i3.pp1803-1810. [2] A. Karthik V., J. Shivaprakash, and R. D., “Disease detection in arecanut using convolutional neural network,” in Proc. Int. Conf. [SRM Institute of Science and Technology], Chennai, India, 2023. [3] Dhanuja K C and H. P. Mohan Kumar, “Areca nut disease detection using image processing technology,” in Proc. Conf. on Computer Applications, PES Engineering College, Mandya, India, 2023. [4] J. Guo, Y. Jin, H. Ye, W. Huang, J. Zhao, B. Cui, F. Liu, and J. Deng, “Recognition of Areca Leaf Yellow Disease Based on PlanetScope Satellite Imagery,” Agronomy, vol. 12, no. 1, p. 14, Dec. 2021, doi: 10.3390/agronomy12010014. [5] Latif Ullah Khan, R. Zhao, H. Wang, and X. Huang, “Recent advances of the causal agent of yellow leaf disease (YLD) on areca palm (Areca catechu L.),” Tropical Plants, vol. 2, no. 7, pp. 1–12, 2023, doi: 10.48130/TP-2023-0007. [6] Madhu B G., R. Kumar G., S. S. Rao, A. R. Shetty, and C. M., “Detection of diseases in arecanut using convolutional neural network,” in Proc. 2024 Second Int. Conf. on Advances in Information Technology (ICAIT), Chikmagaluru, India, 2024, pp. 1–6, doi: 10.1109/ICAIT.2024. [7] P. A. Chougale, A. Ammanagi, M. T. Hussain, A. K., and B. K. Byregowda, “Classification and detection of diseases of areca nut and leaf using machine learning,” Journal of Emerging Technologies and Innovative Research (JETIR), vol. 10, no. 5, pp. 376–382, May 2023. [8] K. Premalatha, G. Naik, M. K. Naik, V. Hegde, B. C. Dhananjaya, and S. K. M., “Exploring yellow leaf disease patterns in areca plantations in Chikkamagaluru District of Karnataka,” Saudi Journal of Pathology and Microbiology, vol. 9, no. 3, pp. 63–70, Mar. 2024, doi: 10.36348/sjpm.2024.v09i03.001. [9] S. N. Hegde and H. S. Vijaya Kumar, “Karnataka Malenadu region arecanut leafspot disease severity assessment using machine learning,” in Proc. 2025 3rd Int. Conf. on Smart Systems for Applications in Electrical Sciences (ICSSES), Tumkur, India, 2025, pp. 1–6, doi: 10.1109/ICSSES64899.2025.11009869. [10] M. Balipa, P. Shetty, A. Kumar, P. B. R., and A. Hebbar, “Arecanut disease detection using CNN and SVM algorithms,” in Proc. Int. Conf. on Artificial Intelligence and Data Engineering (AIDE), Nitte, Karnataka, India, 2022. [11] S. Chikkanayakanahalli, S. B. Srinivas, P. D. Ravindra, V. Singh, V. Gangadhar, and A. K. Dwivedi, “Disease Diagnosis and Sustainable Prevention in Arecanut Farming,” in 2025 Int. Conf. on Electronics and Renewable Systems (ICEARS), Bengaluru, India, 2025, pp. [insert page numbers], doi: 10.1109/ICEARS64219.2025.10940189. [12] L. M. Mohammed and Y. Yusoff, “Detection and classification of plant leaf diseases using digital image processing methods: A review,” ASEAN Engineering Journal, vol. 13, no. 1, pp. 1–9, Feb. 2023, doi: 10.11113/aej.v13.17460. [13] Beena K., Nidhi G. D., Sangeetha V., Varshini S., H. K. Lohar, and Y. K. N., “Arecanut disease prediction using MobileNet V2, ResNet and VGG-16 framework,” in Proc. 2024 4th Int. Conf. on Ubiquitous Computing and Intelligent Information Systems (ICUIS), Bangalore, India, 2024, pp. 1–6, doi: 10.1109/ICUIS64676.2024.10866077. [14] V. R. K., S. G. Prasad, S. Karegoudra, L. K. N. D., S. S. Shetty, S. S. Poojary, and H. Koten, “Deep learning-based classification of areca nut yellow leaf disease with ResNet-50 CNN,” in 2024 International Conference on Recent Advances in Science and Engineering Technology (ICRASET), 2024, pp. 1–6, doi: 10.1109/ICRASET63057.2024.10895610.

Copyright

Copyright © 2025 Ashwini S P, Kavyashree S J, Niranjan B R, Ritish M Tilavalli, Sunil V. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.