Vol. 7, Issue 5 (2019)
Machine learning classifiers on sentinel-2 satellite image for the classification of banana (Musa Sp.) plantations of Theni district, Tamil Nadu, India
Author(s): Manjunath CB, Gnanappazham L, Vijayakumar RM, Kavino M, Jagadeeswaran R and Ramsubramoniam S
Abstract: Remote sensing offers to visualize, map and monitor the agricultural systems at varied resolution. Classification methods have been established to identify crop, patterns and intensity. This paper analyses the potentials of machine learning classification techniques namely RF (Object-based), RF, CART and SVM technique in identification of Banana crop of Theni District, Tamil Nadu through Sentinel 2 satellite image. Initial Principal Component Analysis of data depicted that majority of the information could be derived from Near Infrared (NIR) and Short wave Infra-red (SWIR). This information in original bands is maximized to the lowest number of primary bands. The overall classification accuracy using algorithms was 91.1%, 92.1%, 89.6% & 87.6% with the Kappa co-efficient of 0.89, 0.90, 0.88 and 0.85 for RF (OB), RF, CART and SVM respectively. However, RF was found to be robust against RF (OB), CART and SVM in classification of banana plantations while visually inspecting the outputs. The outputs illustrate that machine learning algorithms on Sentinel2 image are suitable for classifying banana plantations from other agricultural crops.
Pages: 1419-1425 | 350 Views 91 Downloads
download (11263KB)
How to cite this article:
Manjunath CB, Gnanappazham L, Vijayakumar RM, Kavino M, Jagadeeswaran R, Ramsubramoniam S. Machine learning classifiers on sentinel-2 satellite image for the classification of banana (Musa Sp.) plantations of Theni district, Tamil Nadu, India. Int J Chem Stud 2019;7(5):1419-1425.
