Vulnerability Analysis and Quality Improvement of Early Wildfire Detection Datasets for Machine-Learning Applications

Atef Shalan, Khaled Tarmissi, Nafeeul Alam Walee

Research output: Contribution to book or proceedingConference articlepeer-review

2 Scopus citations

Abstract

Wildfires are recognized as highly devastating natural disasters, capable of causing irreversible harm to the environment, structures, and human lives and the aftermath of this dreadful occurrence often involves exorbitant expenses for repairs. Detecting wildfires poses a significant challenge, but identifying scenarios for early detection could empower cities and countries to proactive prepare strategies for wildfire management. The primary objective of this research paper is to provide analysis and improvement of the major vulnerabilities of existing public datasets for early wildfire detection using machine learning. We discuss the current weaknesses in these datasets and illustrate their impacts on machine learning solutions. We then describe the required dataset quality aspects and discuss their advantages for early wildfire detection using machine learning. A sample quality dataset and a sample deep learning model using TensorFlow are made publicly available through GitHub.

Original languageEnglish
Title of host publicationProceedings - 2024 13th IEEE International Conference on Communication Systems and Network Technologies, CSNT 2024
EditorsG.S. Tomar
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages754-760
Number of pages7
ISBN (Electronic)9798350305463
DOIs
StatePublished - 2024
Event13th IEEE International Conference on Communication Systems and Network Technologies, CSNT 2024 - Hybrid, Jabalpur, India
Duration: Apr 6 2024Apr 7 2024

Publication series

NameProceedings - 2024 13th IEEE International Conference on Communication Systems and Network Technologies, CSNT 2024

Conference

Conference13th IEEE International Conference on Communication Systems and Network Technologies, CSNT 2024
Country/TerritoryIndia
CityHybrid, Jabalpur
Period04/6/2404/7/24

Scopus Subject Areas

  • Computer Networks and Communications
  • Hardware and Architecture
  • Signal Processing
  • Safety, Risk, Reliability and Quality
  • Instrumentation

Keywords

  • Dataset
  • Deep Learning
  • Detection
  • Wildfire

Fingerprint

Dive into the research topics of 'Vulnerability Analysis and Quality Improvement of Early Wildfire Detection Datasets for Machine-Learning Applications'. Together they form a unique fingerprint.

Cite this