Comparing U-Net Convolutional Network with Mask R-CNN in Agricultural Area Segmentation on Satellite Images
Thinh Tran Pham Quoc, Tam Tran Linh, Thu Nguyen
Abstract
Deep learning is the fastest-growing trend in statistical analysis of remote sensing data. Deep learning models are used for information processing of spectral steps, identification statistics, segmentation and classification of the objects in satellite images, etc. Image segmentation could help to make the object statistics more accurate by separating the objects from the background. In this paper, we propose knowledge of Mask R-CNN and U-Net in satellite imagery segmentation, and we also make an experiment for these models to show the appropriateness in this field. Experimental result of the mean average precision (mAP) on dataset of Vietnam satellite images is 95.21% for Mask R-CNN and 92.69% for U-Net.