Controlling water resources in the current context of global warming is a worrying difficulty for the authorities. This work aims to control the amount of water in inland water bodies, including ponds, lakes, rivers, and streams, to provide appropriate solutions to meet the needs of agricultural use, fisheries, and daily life. Analysis and evaluation studies on remote sensing images contribute significantly to controlling these water areas. The image segmentation method is one of the techniques that strongly supports researchers when conducting experiments on remote-sensing high-resolution remote sensing images. In this study, we propose using UNet 3+ architecture to improve the efficiency of information extraction on remote sensing images for inland water areas, thereby improving model accuracy and resulting in high segmentation efficiency. Experimental results show that the proposed method with UNet 3+ helps improve image segmentation efficiency in satellite images. This is a basis for monitoring and analyzing remote sensing images over time to gain greater control over inland water flows.