Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Object-Based Convolutional Neural Network for High-Resolution Imagery Classification #11

Open
KenichiSasaki opened this issue Oct 25, 2019 · 0 comments
Open

Object-Based Convolutional Neural Network for High-Resolution Imagery Classification #11

KenichiSasaki opened this issue Oct 25, 2019 · 0 comments
Labels
CNN Feature Engineering Feature Extraction Hyperspectral Image Image Analysis with hyperspectral image data handling Satellite Imagery Paper about satellite image classification Segmentation Statistical Method Employ classic statistical method

Comments

@KenichiSasaki
Copy link
Collaborator

KenichiSasaki commented Oct 25, 2019 •
edited

概要

  • アルゴリズム系論文(2017)
  • CNNによる特徴量抽出とObject-based の分類を繰り合わせ9割以上の精度建物の識別に成功
    • CNNが高度の特徴量抽出を可能とするがPixelレベルの形状を識別するのが困難である欠点をObejct-basedのアルゴリズムを統合することで補完可能

アルゴリズム

  • Train dataからCNNで特徴量計算
  • 同時に統計特徴量用いて似た要素を持つ領域をセグメンテーション
  • セグメンテーション結果からShape constraintを先ほどのCNNの特徴量に付与
  • 各領域に対して2層の全結合層で識別
    2019-10-25_16h02_24

実装

  • Beijing, Pavia, Vailingen (China, Italy, Germany)の画像を利用
    • それぞれWorld-view2, ROSIS sensor data, German Association of Photogrammetry and Remote sensing
  • 畳み込み層は1-5層で実装，
    • 層が増すにつれ精度が向上
  • Soilの精度が最も低い(7割くらい)が既存のアルゴリズムより大幅に精度向上
    • SVM, EMAP, Pixel based CNN, SSFC

所感

  • CNNによるアップサンプリングの弱点である正確なセグメンテーションを統計量を利用することで向上したが特徴量エンジニアリングの要素が多い，
  • Obejct-basedの具体的なアルゴリズムの言及はないため要調査
  • CNNの弱点や傾向の勉強になった
@KenichiSasaki KenichiSasaki added CNN Feature Engineering Feature Extraction Hyperspectral Image Image Analysis with hyperspectral image data handling Object Detection Segmentation Statistical Method Employ classic statistical method and removed Object Detection labels Oct 25, 2019
@Shaw0202 Shaw0202 added the Satellite Imagery Paper about satellite image classification label Oct 31, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
CNN Feature Engineering Feature Extraction Hyperspectral Image Image Analysis with hyperspectral image data handling Satellite Imagery Paper about satellite image classification Segmentation Statistical Method Employ classic statistical method
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@KenichiSasaki @Shaw0202