Introduction

This repo is the official implementation of our paper: Extraction of cropland field parcels with High-Resolution Remote Sensing Images using multi-task learning. The prediction process is shown in figure below.

Data and checkpoint

Google cloud: https://drive.google.com/drive/folders/1f1jZwUCS4892bkne7Ob1iWmcFhj8aDIZ?usp=sharing

Denmark datastet

link：https://pan.baidu.com/s/1l9W0ekFdvyj8FhKA44CNuQ pw：qazj

How to use

1. To obtain the cropland parcel(without using patch refinement)

bash python main.py --img ${image_path} --weights ${weights_path} --shp ${the_path_of_output_in_shapefile}

1. The next step is repair the break line

bash python PatchRefinement.py --lineDN ${lineDN_path} --img ${image_path} --weights ${weights_path} PS:The training code can be found here.

1. In final, merge the results of the two steps.

bash python postprocess.py --img ${image_path} --shp ${the_path_of_output_in_shapefile}

Training code

The code works a bit cluttered and without comments, download it if you need. link：https://pan.baidu.com/s/1AKwMW9-0t8MA6LoEzgt6VA?pwd=kqau pwd：kqau