Publication

Abstract : The coconut tree, known as the 'Kalpavriksha,’ holds a significant relationship with nature and people in India. For many, it serves as a source of livelihood, either directly or indirectly. The health status of coconut trees is crucial for agricultural productivity and ecosystem conservation. Tree parameters play a vital role in identifying the health conditions and growth of coconut trees, aiding researchers and farmers in cultivation and health prediction. This paper explores various clustering and classification methods to mitigate time consumption, over fitting, and complexity in data classification. This paper proposes a machine learning algorithm for classifying coconut trees based on tree parameters such as Height, Inclination, and Diameter at breast height (DBH). The proposed approach commences with the extraction of tree parameters using an unsupervised learning algorithm. Parameters such as tree height obtained through a mobile application, inclination measured by an inclinometer, and DBH are used as a database for classifying the coconut tree into different classes. These pre-processed parameters serve as inputs to supervised learning models, including Support vector Machine, Random Forest, and K-Nearest Neighbor, to classify trees into different categories. This paper endorses K-means clustering followed by the support vector machine (SVM) approach for classification. Evaluation of each algorithm is conducted by calculating accuracy, precision, and Fl score. The dataset comprising 60 sensor data points is utilized for clustering, parameter training, and testing. By implementing machine learning algorithms with real-time morphological data, greater accuracy is achieved for the classification of coconut tree health.