Plant Health Monitoring System Using Machine Learning

Abstract

Factory splint conditions are a significant trouble to crop yields and food security worldwide. Beforehand discovery and opinion of these conditions are pivotal for effective operation and control. Machine literacy ( ML ) ways have shown great pledge in factory splint complaint discovery, offering a rapid- fire and accurate volition to traditional styles. This paper provides an in- depth review of the different ML approaches used for factory splint complaint discovery, including image processing, point birth, and bracket ways. We also estimate the performance of colourful ML algorithms using criteria similar as delicacy, perfection, and recall. Our results show that ML- grounded styles can achieve high delicacy in detecting factory splint conditions, outperforming traditional styles. We bandy the counteraccusations of our findings for agrarian exploration and practice, pressing the eventuality of ML to ameliorate crop yields and reduce the profitable impact of factory splint conditions.

Introduction

The text discusses the need for automated plant disease detection to reduce global crop losses, especially in developing countries, due to the growing population and increasing food demand. Traditional manual inspection methods are costly, time-consuming, and impractical for large farms or rural areas. While hyperspectral imaging has been used for large-scale detection, it is expensive and unsuitable for machine learning due to high dimensionality and small sample sizes. Instead, RGB image analysis has become popular because of smartphone ubiquity and advancements in computer vision, allowing classical ML and deep learning (DL) approaches to identify plant diseases effectively. DL methods, particularly Convolutional Neural Networks (CNNs), are preferred for large datasets because they outperform classical ML algorithms without the need for manual feature extraction.

The literature survey highlights multiple studies proposing various techniques for plant disease detection, including advanced sensors, image processing methods, classical ML classifiers (e.g., KNN, logistic regression), and DL models like Deep Convolutional Neural Networks (DCNNs), all aimed at improving accuracy, reducing computational load, and handling noisy images.

The methodology involves:

1. Image Acquisition – Using the Plant Village dataset of 600 RGB images from apples, grapes, cherries, and maize, categorized into bacteria, fungi, virus, nematodes, and healthy leaves.
2. Image Processing – Noise removal, scaling, RGB-to-HSV conversion, median filtering, histogram equalization, and image enhancement.
3. Segmentation – Extracting potentially infected leaf areas by removing non-informative pixels.
4. Feature Extraction – Reducing dimensionality and extracting statistical and color features from images for classification.
5. Classification – Employing a CNN with 80/5-fold cross-validation for robust plant disease classification.

The study aims to compare DL approaches with classical ML algorithms, with CNNs expected to outperform traditional methods in accuracy and efficiency for plant disease identification.

In short, the work demonstrates how affordable, smartphone-accessible RGB image analysis combined with CNNs offers an effective, scalable solution for timely detection of plant diseases, reducing yield losses and supporting food security.

Conclusion

This project presents the dominance of the DL method over the classical ML algorithms. Both the simplicity of the approach and the achieved accuracy confirm that the DL is the way to follow for image classification problems with relatively large datasets. As the achieved accuracy of the DL method is already very high, trying to improve its results on the same dataset would be of little benefit. Further work with the DL model could be done by expanding the dataset with more diverse images, collected from multiple sources, in order to allow it to generalize better. The considered ML algorithms achieved relatively high accuracy, but with error rates still an order of magnitude higher than the DL model. Further work in improving accuracy of the classical approach can be done by experimenting with other algorithms and by improving the features, as most likely they are the limiting factor of this approach.

References

[1] Chittabarni Sarkar, Deepak Gupta, Umesh Gupta, Barenya Bikash Haz- arika Applied Soft Computing, 110534, 2023 [2] Monika Lamba, Yogita Gigras, Anuradha Dhull Open Computer Science 11 (1), 491 508, 2021 [3] Majji V Applalanaidu, G Kumaravelan 2021 Third International Con- ference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 716 724, 2021 [4] Vibhor Kumar Vishnoi, Krishan Kumar, Brajesh Kumar Journal of Plant Diseases and Protection 128, 19-53, 2021 [5] Saifi, Suhail, et al. ”A Review on Plant Leaf Disease Detection using Deep Learning.”2022 International Conference on Computational Intel- ligence and Sustainable Engineering Solutions (CISES). IEEE, 2022 [6] plant detection and classification techniques, Wubetu Barud Demilie [7] Reddy, Y. Harshavardhan, et al. ”Plant Leaf Disease Detection using IoT, DL and ML.”International Journal of Advanced Research in Science, Communication and Technology (2023): 368-379 [8] Plant Disease Detection and Classification Using Machine Learning and Deep Learning Techniques: Current Trends and Challenges Yasmin M.Alsakar, Nehal A. Sakr , and Mohammed Elmogy [9] Geetha, G., et al. ”Plant leaf disease classification and detection system using machine learning.” Journal of Physics: Conference Series. Vol. 1712. No. 1. IOP Publishing, 2020 [10] Plant Leaf Disease Detection using Machine Learning Amrita S. Tul- shan; Nataasha Raul [11] Devi, P. Rama. ”Leaf Disease Detection Using Deep Learning.” 2021 Second International Conference on Electronics and Sustainable Com- munication Systems (ICESC). IEEE, 2021 [12] Anandhakrishnan, T., and S. M. Jaisakthi. ”Deep Convolutional Neural Networks for image based tomato leaf disease detection.” Sustainable Chemistry and Pharmacy 30 (2022): 100793 [13] Plant leaf disease classification and damage detection system using deep learning models B.Sai ReddyS. Neeraja [14] Vasavi, Pallepati, Arumugam Punitha, and T. Venkat Narayana Rao. ”Crop leaf disease detection and classification using machine learning and deep learning algorithms by visual symptoms: A review.” Interna- tional Journal of Electrical and Computer Engineering 12.2 (2022) [15] Sarkar, Chittabarni, et al. ”Leaf disease detection using machine learning and deep learning: Review and challenges.” Applied Soft Computing (2023) [16] Babu, Vishnu S., R. Satheesh Kumar, and R. Sunder. ”A comparative study on disease detection of plants using machine learning techniques.” 2021 7th International Conference on Advanced Computing and Com- munication Systems (ICACCS). Vol. 1. IEEE, [17] Cham, Marcelo Fernandes Xavier, Radius Tanone, and Hendra Alexan- der T. Riadi. ”Identification of Rice Leaf Disease Using Convolutional Neural Network Based on Android Mobile Platform.” 2021 2nd Interna- tional Conference on Innovative and Creative Information Technology (ICITech). IEEE, 2021 [18] Sathya, R., et al. ”Vision Based Plant Leaf Disease Detection and Recognition Model Using Machine Learning Techniques.” 2023 7th International Conference on Electronic ,Communication and Aerospace Technology (ICECA). IEEE, 2023 [19] Blessy, Y. M., et al. ”Plant disease detection using deep learning and image processing.” Int. J.Electron. Devices Netw 2.2 (2021): 13 16 [20] Meenakshi, T. ”Automatic detection of diseases in leaves of medicinal plants using modified logistic regression algorithm.” Wireless Personal Communications 131.4 (2023): 2573-2597 [21] Martinelli, Federico, et al. ”Advanced methods of plant disease detec- tion. A review.” Agronomy for Sustainable Development 35 (2015)

Copyright

Copyright © 2026 G. Sindhu, Ammakolla Sreeja, A. Harika, Ms. Bhagyashree . 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.