System disruptions
We are currently experiencing disruptions on the search portals due to high traffic. We are working to resolve the issue, you may temporarily encounter an error message.
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • apa.csl
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Fire Detection using Deep Learning Approach
Karlstad University, Faculty of Health, Science and Technology (starting 2013), Department of Engineering and Physics (from 2013).
2021 (English)Independent thesis Basic level (degree of Bachelor), 15 credits / 22,5 HE creditsStudent thesis
Abstract [en]

Fire in forests and urban areas has been and is still a serious issue all over the world. Owing to the different colors, intensity, textures, and shapes of the fires, the development of a robust and accurate fire detection system is a challenging task. Fire detection in images through computer vision and image processing techniques such as convolutional neural networks (CNNs) has gained significant attention due to its real-time detection capability. However, fire detection using CNNs in the previous studies has limitations such as evaluation on balanced datasets that may mislead information on a real scenario. This study proposed a new approach using modern technology to detect and control fire. An Unmanned Aerial Vehicle (drone) integrated with two state-of-the-art detection models Yolo-v5 and EfficinetDet is used. The deep learning-enabled drone has the capability of patrolling over the potentially threatened fire areas to detect smoke and fire accurately, based on static images and video clips taken through drone cameras. The two detection models Yolo-v5 and EfficinetDet were individually tested using a huge fire dataset in a real wildfire environment and their performances were comparatively evaluated in terms of precision, recall, and accuracy. The experimental results showed that the new approach based on the strong detection models (Yolo-v5 and EfficientDet) has high fire detection accuracy. Both models have satisfactory detection accuracy but observable false-positive rates, however, in the in-depth comparative analysis, EfficientDet outperforms the Yolo model in terms of detection accuracy and as well as in other evaluation parameters.

Place, publisher, year, edition, pages
2021. , p. 42
Keywords [en]
deep learning, Yolov5; EfficientDet
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kau:diva-86241OAI: oai:DiVA.org:kau-86241DiVA, id: diva2:1603636
Supervisors
Examiners
Available from: 2021-10-29 Created: 2021-10-16 Last updated: 2021-10-29Bibliographically approved

Open Access in DiVA

No full text in DiVA

By organisation
Department of Engineering and Physics (from 2013)
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 472 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • apa.csl
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf