ANNOUNCEMENTS
Trees outside Forest (TOF) have much-needed social and environmental impacts from the local scale to the rest of the world. In the current context of climate change, their importance is increasingly important for human life and the national economy, as well as for various global processes facing global environmental and economic challenges such as carbon emissions, biodiversity loss, desertification, poverty eradication etc.
Tree outside forest (TOF) are widespread as they are recognized as an important body, in the management of sustainable natural resources, due to their role in providing a variety of goods, such as timber, fruit and fodder and as services such as water, carbon, and biodiversity. In this study two datasets and Machine learning and Deep learning Algorithm such as Random Forest classifier , Support Vector Machine classifier , U – Net model were used first high resolution Kompsat data (0.7 m spatial resolution) and second Medium resolution Sentinel 2 A data (10 m spatial resolution). All sentinel data bands have been resampled with a resolution of 10 m.
Training samples were then made to train ML and DL models. Basically this research focuses on drawing a Tree outside of the Forest and various other categories. The overall accuracy of the DL Net network is 96% with high resolution data and 88% with medium resolution data while the total SVM accuracy is 85% and the total RF accuracy is 74%. This study finds that the performance of the DL model is good for mapping Built-up, Water bodies, Wasteland, Fallow land, Agricultural field, Plantation and Forest.
KEY WORDS: Trees outside forest , Machine Learning , Deep Learning , Random Forest classifier , Support Vector Machine classifier , U – Net model.
