GithubHelp home page GithubHelp logo

sonwyang / building-extraction Goto Github PK

View Code? Open in Web Editor NEW
13.0 5.0 1.0 7.28 MB

Python 28.47% Shell 0.01% C 0.17% Cuda 0.22% C++ 0.12% Jupyter Notebook 71.00%
building-extraction deep-learning pytorch ndsm high-resolution-rs-image attention-mechanism hnn weight-mapping instance-segmentation watershed-algorithm

building-extraction's Introduction

building-extraction-based-on-deep-learning

This repo is the official pytorch implementation of paper "HA U-Net: Improved Model for Building Extraction From High Resolution Remote Sensing Imagery".

前言

语义分割在基于高分影像的建筑提取中不能区分不同的建筑个体,因为在建筑密集区其通常将同一个类的对象紧密地打包成一个连接的组件,即使是优秀的语义分割网络模型也不可避免地会产生这样的结果。而实例分割可以较好地解决这个问题,精准到建筑个体的分割结果在绘制地图、城市规划以及人口估计中将会大有可为。

image-20200514161921651

要解决的问题

  1. 目前,在基于深度学习技术对高分遥感影像进行建筑提取的研究中,通常是将其归类为二分类的任务,即将所有的像素视为建筑和非建筑这两类,而对网络模型区分建筑物个体的能力关注较少;
  2. 在建筑密集区进行遥感影像分类后,易出现“房屋粘连”现象,即预测后的边界通常不够友好,而且同一个类的对象被紧密地打包成一个连接的组件,因此需要研究相应的影像后处理方法来优化分割结果。

技术路线

image-20200514162328700

HA U-Net模型结构

image-20200514162902586

基于权重映射的影像分类流程

image-20200514163432391

基于分水岭算法的影像后处理

对所获得的概率分布图进行双阈值操作,分别获取内部标记和外部标记,其中高阈值对应内部标记,低阈值对应外部标记,然后用分水岭算法进行处理(对于无房屋粘连区域不进行处理),保留分水线,最后将分水线叠加到预测结果上。

image-20200514163531289

提取效果概览

image-20200514162815984

相关指标

Methods\Metrics IOU Kappa 实例化F1
U-Net 73.19 82.16 85.76
HA U-Net+IWM 75.32 83.71 89.36
HA U-Net+IWM+ Watershed (0.5, 0.9) 75.28 83.69 89.90

Reference

基于分水岭算法的影像后处理

weight mapping

attention U-Net

HNN

实例化F1

PS

如果你不熟悉深度学习在提取建筑物的使用,你可以尝试运行下这个kaggle上的notebook

Citation

If you find our work useful in your research, please cite

building-extraction's People

Contributors

sonwyang avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

Forkers

yangpanhzau

building-extraction's Issues

kaggle上的notebook的问题。

Hi bro,

您的kaggle上的notebook在kaggle上运行不了,总是无法下载模型,我把您kaggle上的notebook复制到谷歌的colab上跑,可以正常跑,但是跑一百个epoch之后得到的模型,把用于训练的图像弄进去进行预测,总是得不到建筑类型,请问是什么问题?

期待您的回复,

谢谢!
zhen

后处理

您好,我想问一下您文章中提到的基于分水岭的后处理方法是怎么做的呢?看您的代码中似乎并没有用到这个方法?可以提供相关代码参考嘛?非常感谢

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.