Waterline Extraction for Artificial Coast With Vision Transformers

Le Yang,Jingsheng Zhai,Xing Wang

doi:10.3389/fenvs.2022.799250

Abstract

Accurate acquisition for the positions of the waterlines plays a critical role in coastline extraction. However, waterline extraction from high-resolution images is a very challenging task because it is easily influenced by the complex background. To fulfill the task, two types of vision transformers, segmentation transformers (SETR) and semantic segmentation transformers (SegFormer), are introduced as an early exploration of the potential of transformers for waterline extraction. To estimate the effects of the two methods, we collect the high-resolution images from the web map services, and the annotations are created manually for training and test. Through extensive experiments, transformer-based approaches achieved state-of-the-art performances for waterline extraction in the artificial coast.

Highlights

Coastline extraction is a very challenging problem because it is obtained from a region not an instantaneous line
For the convolutional neural network (CNN) methods, the methods with the ResNet101 backbone are better than the methods with HRNet, and the methods with the UNet backbone achieve the cheapest results
The methods with ResNet101 and HRNet extract the small striped object in Image 6, but no methods can avoid the influence of the ship

Summary

Introduction

A coastline is the boundary between the dry and wet part in the coastal area when the high tide water is in the mean level Toure et al (2018). Coastline extraction is a very challenging problem because it is obtained from a region not an instantaneous line. The waterline extraction is the precondition for computing the natural coastline, so the waterline extraction is very important and meaningful. The waterline is the instantaneous boundary between the land and sea. It can be extracted from the highresolution images without other tools. The waterline can be considered as the coastline because the waterline is very slightly influenced by the tides

Methods

Results

Discussion

Conclusion