Abstract:
In the modern day, remote sensing presents the greatest problems. Numerous studies in the field of remote sensing have been conducted. Classifying land use and cover in remote sensing involves both supervised and unsupervised procedures. Previous research on unsupervised learning has employed many classifiers. On the supervised categorization, no parallel work has been completed. This study examines the parameter-based performance of supervised classifiers, such as neural networks and support vector machines. Two situations are used in this study. A remotely sensed image is captured using ENVI 5.0 for assessment. Three distinct classes-forest, buildup, and aquatic bodies-make up the selected image. The first instance involved applying support vector machines, of supervised classification, to an image. A neural network was applied directly to a picture in the second situation, by employing a confusion matrix, the most accurate classifier is determined. The SVM provides an overall accuracy of 93.6805% in scenario 1. While the neural network's total accuracy for the image in scenario 2 was 50.4237%.The best results are obtained from Support Vector Machine for the chosen image; however, when a neural network is applied directly to the chosen image, the outcome is not precise. The goal is to arrive at the optimal classification pattern.
Page(s):
1-1
DOI:
DOI not available
Published:
Journal: Second International Conference on Computing Technologies, Tools and Applications (ICTAPP-24), June 4-6,2024 (Abstract Book), Volume: 0, Issue: 0, Year: 2024