Electric Resistivity Tomography for Quantifying Heartwood from Sandalwood Trees Using Machine Learning

Agriculture forms the core foundation of India's economy. Farmers in the Karnataka region seek information on the growth of sandalwood trees in advance. For this purpose, the early detection and calculation of heartwood in sandalwood trees would benefit these farmers and help them predict harvest times. This paper is a proposal and contribution to the Institute of Wood Science and Technology, highlighting the potential of utilizing machine learning techniques to quantify heartwood in sandalwood trees. The electric resistivity tomography (ERT) generated images were used to calculate the internal structure and resistivity of the trees. The Institute of Wood Science and Technology (IWST) provided a sample dataset of 27 sandalwood trees in the Kolar region. Due to the unavailability of an extensive dataset from the IWST, a synthetic dataset was created by studying the repeated patterns of the sample. This synthetic dataset was populated by applying random number generation, mathematical simulation equations, and generative adversarial networks (GANs). Machine learning techniques, including decision tree, random forest, logistic regression, and MLP methods, are compared using the synthetically prepared 'Kolar' dataset. Performance metrics such as accuracy, F1 score, recall, and precision were computed to assess model effectiveness. The MLP classifier emerged as the top-performing model, exhibiting an accuracy of 83.33%. These results were obtained from synthetic data to extend the application of machine learning techniques to real-time data. This work seeks to make a significant social impact by empowering farmers and agriculturists to plan their harvests and avert premature tree cutting, thus ensuring optimal profitability.