Land Cover Analysis with Fully Convolutional Network

Ihwan, Abib Raifmuaffah
Lapatta, Nouval Trezandy
Joefrie, Yuri Yudhaswana
Anshori, Yusuf
Syahrullah, Syahrullah

Jurnal Sistem Cerdas
Vol. 8 No. 1 (2025)

This study analyses land cover in Morowali Regency using Sentinel-2 satellite imagery and the Fully Convolutional Network (FCN) algorithm. Land cover analysis in this area is crucial for monitoring rapid industrialization, especially in the mining sector. The methodology includes retrieving image data from Google Earth Engine, image processing to eliminate cloud influences, and model training using the European Space Agency (ESA) datasets. The results of the analysis show that 50% of the Morowali Regency area has the potential to be planted with trees, followed by 20% for water areas, and the rest for bushes, development land, and empty land. This study proves that FCN can be relied on to predict land potential with high accuracy with a loss value of 1.3001.