Mango Leaf Disease Detection using YOLO v1 to YOLO v10

Abstract

This study explores and compares the performance of various YOLO (You Only Look Once) object detection models—ranging from YOLOv1 to YOLOv10—for identifying diseases in mango leaves. A dataset of 4,000 mango leaf images was prepared, covering eight distinct classes, including common diseases such as Anthracnose, Bacterial Canker, Powdery Mildew, Sooty Mould, Leaf Spot, Dieback, Algal Leaf Spot, along with healthy leaves. Each image was manually annotated to highlight the affected areas using bounding boxes. To ensure fairness in evaluation, all YOLO versions were trained on the same dataset under consistent conditions. The models were assessed based on standard performance metrics such as mean Average Precision (mAP), precision. The comparative results offer valuable insights into how YOLO has progressed over its different versions, revealing the strengths and weaknesses of each in terms of detection accuracy and computational efficiency. This work aims to guide researchers and developers in choosing the most effective YOLO version for real-time disease detection in agricultural settings.

Introduction

Mango is an important fruit crop but is vulnerable to various leaf diseases caused by fungi, bacteria, and environmental factors. Early, accurate disease detection is crucial for minimizing damage and supporting sustainable agriculture. Traditional manual detection methods are slow and error-prone. Advances in computer vision and deep learning, especially YOLO (You Only Look Once) object detection models, offer fast, automated solutions suitable for real-time agricultural applications.

This research compares ten versions of YOLO (from YOLOv1 to YOLOv10) for detecting eight common mango leaf diseases using a curated dataset of 4,500 annotated images. Each model was trained under uniform conditions to fairly evaluate performance based on accuracy (mAP), precision, recall, and inference speed.

YOLO models excel in real-time, single-pass object detection, making them ideal for deployment on mobile and edge devices in precision farming. Despite many studies on individual YOLO versions, comprehensive comparisons across all versions on mango disease detection were lacking, which this study addresses.

Methodology:

• Dataset collected and annotated for diseases like Anthracnose, Powdery Mildew, Bacterial Canker, etc.

• Images were preprocessed and enhanced for better feature visibility.

• YOLO models were trained to detect and classify healthy vs diseased leaves with bounding boxes around affected regions.

Feature Extraction:
YOLO’s deep CNN architecture automatically learns hierarchical features, identifying disease symptoms such as discoloration, spots, and texture changes without manual feature engineering.

Results:
YOLOv9 achieved the best balance of accuracy and efficiency, detecting diseases reliably even under varying lighting and backgrounds. The system demonstrated practical real-time disease localization and classification suitable for field deployment.

Discussion:
The YOLO-based approach provides fast, accurate, and automated disease detection with minimal hardware requirements, outperforming traditional computer vision methods. Challenges include occasional misclassification with faint symptoms and visually similar diseases. Future work suggests expanding disease classes, improving dataset diversity, integrating with mobile platforms, and enhancing statistical evaluation metrics.

Conclusion

This study presents a structured approach for detecting and classifying mango leaf diseases using an object detection framework. The method utilizes a step-by-step pipeline, where annotated images of diseased and healthy mango leaves are fed into a YOLO-based model. The dataset undergoes preprocessing including image resizing, annotation formatting to enhance detection accuracy. Through qualitative evaluation of sample leaf images, it was observed that the YOLO model accurately identified diseased regions and classified them based on visual symptoms. While the model performed reliably across most test cases, occasional challenges such as poor lighting or overlapping leaves introduced some inconsistencies in detection. Nevertheless, the lightweight and efficient architecture of the YOLO framework makes it suitable for real-time applications in agricultural settings, particularly in areas with limited hardware resources. Additionally, the modular design allows for future enhancements, such as incorporating domain-specific feature extraction or integrating with mobile platforms. Overall, the work underscores the practical value of YOLO in agricultural disease detection, offering a replicable foundation for deploying smart, low-cost plant health monitoring systems.

References

[1] Rahman, M. M., Hasan, M. T., Akter, S., & Islam, M. N. (2023). Prediction of Mango Leaf Diseases using CNN. In Proceedings of the 15th International Conference on Computing, Communication and Networking Technologies (ICCCI). [2] Kumar, S., Rani, R., & Singh, P. (2018). Expert System for Diagnosis Mango Diseases. In Proceedings of the International Conference on Data Analytics and Management Technologies (ICDAMT) [3] Sharma, A., Verma, R., & Gupta, N. (2024). Optimizing Performance: Mango Plant Leaf Disease. STM Journals. [4] Patil, S., Joshi, A., & Kulkarni, R. (2022). Mango Leaf Diseases Detection Using Deep Learning. I-Scholar Journal. [5] Khan, F., Mehta, R., & Desai, S. (2024). Mango Plant Disease Detection System Using Hybrid BBHE & CNN. STM Journals [6] Shelar, N., Shinde, S., & Sawant, S. (2022). Plant Disease Detection Using CNN. [7] Lakshmi, K. R. C., Reddy, S. K., & Kumar, A. S. (2023). YOLO for Detecting Plant Diseases [8] Aldakheel, E. A., Zakariah, M., & Alabdalall, A. H. (2024). Detection and Identification of Plant Leaf Disease Using YOLOv4. [9] Shelar, N., Shinde, S., & Sawant, S. (2022). Plant Disease Detection Using CNN. [10] Deshmukh, R., Kale, P., & Thakur, S. (2024). Mango Plant Disease Detection System Using Hybrid BBHE & CNN. STM Journals. [11] Pratap, V. K., & Kumar, N. S. (2024). Deep Learning based Mango Leaf Disease Detection for Classifying and Evaluating Mango Leaf Diseases. [12] Sajitha, P., John, A., Devika, V. L., Gayathri, S. V., & Sakhir, N. (2023). Leaf Disease Detection & Correction using YOLO V7 with GPT-3 Integrated. [13] Ponnusamy, V., et al. (2020). Smart Glass: Real-Time Leaf Disease Detection using YOLO Transfer Learning. [14] Sun, Z., Feng, Z., & Chen, Z. (2024). Highly Accuracy and Lightweight Detection of Apple Leaf Diseases Based on YOLO. [15] Pratap, V. K., & Kumar, N. S. (2024). Deep Learning based Mango Leaf Disease Detection for Classifying and Evaluating Mango Leaf Diseases.

Copyright

Copyright © 2025 Chandrashekar M B, Dr. R J Prathibha. 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.