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.

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

Object Detection in Optical Remote Sensing Images Based on Weakly Supervised Learning and High-Level Feature Learning #6

Open
KenichiSasaki opened this issue Oct 11, 2019 · 0 comments
Labels
Feature Engineering Feature Extraction Machine Learning Method Employ machine learning method Object Detection Satellite Imagery Paper about satellite image classification

Comments

@KenichiSasaki
Copy link
Collaborator

KenichiSasaki commented Oct 11, 2019

概要

  • アルゴリズム論文(2015)
  • モチベーション:データアノテーションを簡易化
  • WSL(Weakly Supervised Learning)
  • 画像に正解ラベル(飛行機,車,空港)が存在するかどうかのみを教師とする

アルゴリズム

  • 各Image patchクロップ
  • Sift計算,Kmeansで各特徴量の距離算出
  • 距離が近いもののHistogram取得しImage patch同士で連結(LLC: Locally-restricted Linear Coding)
  • DBM (Deep Boltzman Machine)で高次特徴量取得
  • 各画像から特徴量自動生成
  • 生成されたImage patchのクラス間およびクラス内の分布を繰り返しGMMで計算し,Mislabelingを修正
  • Bayesianで識別
    2019-10-10_18h18_03

実装

  • Dataset (Resolution): Google Earth (0.5m), ISPRS (8-15cm), Landsat (30m)
    • それぞれ飛行機,車,空港の検出に使用
  • 既存のWSLより精度向上(精度0.5くらい)だが,教師あり学習(SVM)の精度に少し劣る

所感

  • クラシックな特徴量抽出を行ってアノテーションを自動化する発想は面白い
  • 職人芸の域を脱していない
@KenichiSasaki KenichiSasaki added the Feature Engineering Feature Extraction label Oct 11, 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
Labels
Feature Engineering Feature Extraction Machine Learning Method Employ machine learning method Object Detection Satellite Imagery Paper about satellite image classification
Projects
None yet
Development

No branches or pull requests

2 participants