Early Forest Fire Region Segmentation Based on Deep Learning

Guangyi Wang, Youmin Zhang, Yaohong Qu, Yanhong Chen, Hamid Maqsood

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

23 Scopus citations

Abstract

As the forest fire can bring about great property loss and ecological disaster, artificial intelligence-based forest fire monitoring system has gained popularity in recent years to enable the fire alarm quickly and accurately. In this paper, considering that the fire area is very small and hard to be detected using traditional method for detection early forest fire, we propose a novel forest fire monitoring framework based on convolutional neutral networks. In order to validate that the proposed framework can improve effectiveness and accuracy of detecting the early forest fires, many groups of fire detection experiments using a self-generated forest fire dataset and two real forest fire monitor videos are conducted. The experiment results demonstrate its capability to work in various challenging fire and illumination conditions presented in the study, and show that the framework can effectively detect the early forest fire.

Original languageEnglish
Title of host publicationProceedings of the 31st Chinese Control and Decision Conference, CCDC 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages6237-6241
Number of pages5
ISBN (Electronic)9781728101057
DOIs
StatePublished - Jun 2019
Event31st Chinese Control and Decision Conference, CCDC 2019 - Nanchang, China
Duration: 3 Jun 20195 Jun 2019

Publication series

NameProceedings of the 31st Chinese Control and Decision Conference, CCDC 2019

Conference

Conference31st Chinese Control and Decision Conference, CCDC 2019
Country/TerritoryChina
CityNanchang
Period3/06/195/06/19

Keywords

  • Artificial Intelligence
  • Deep Learning
  • Forest Fire
  • Semantic Segmentation

Fingerprint

Dive into the research topics of 'Early Forest Fire Region Segmentation Based on Deep Learning'. Together they form a unique fingerprint.

Cite this