Earticle

초록

영어

Recently due to natural disasters, the world is facing huge ecological, social, economic, and loss of precious lives. Traditionally during natural disasters, emergency response teams are physically visiting different areas to inspect and stop their further damages. Therefore, the existing monitoring system is facing issues such as human accessibility and unable to analyze disaster in real-time. To address these issues, we propose a machine learning inspired framework for automatically recognized disaster scenes that contains three main steps. In the first step preprocessing is applied for condense and normalize the image dimension. Next, histogram of oriented gradient (HOG) descriptor is utilize to extract discriminative features and extracted features are classified through SVM. Finally in testing step, in case of disaster scenes our system trigger notification to nearby disaster management centers to take an appropriate action. We provide comprehensive experiments on various machine learning approaches among them we obtain 64% accuracy on HOG with SVM.

목차

Abstract
1. Introduction
2. Proposed Framework
2.1. Preprocessing phase
2.2. Feature extraction phase
2.3. Classification phase
3. Experimental Results and Discussion
3.1. Dataset Description
3.2. Result and Discussion
4. Conclusion and Possible Future Work
Acknowledgement
References

키워드

저자

• Altaf Hussain [ Sejong University ]
• Samee Ullah Khan [ Sejong University ]
• Fath U Min Ullah [ Sejong University ]
• Mi Young Lee [ Sejong University ]
• Sung Wook Baik [ Sejong University ] Corresponding Author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  한국차세대컴퓨팅학회 학술대회

• 간기

  반년간

• 수록기간

  2021~2025

• 십진분류

  KDC 566 DDC 004