An intelligent video fire detection approach based on object detection technology

Loading...
Thumbnail Image

Date

2018

Journal Title

Journal ISSN

Volume Title

Publisher

Mary Kay O'Connor Process Safety Center

Abstract

Fire that is one of the most serious accidents in chemical factories, may lead to considerable product losses, equipment damages and casualties. With the rapid development of computer vision technology, intelligent fire detection has been proposed and applied in various scenarios. This paper presents a new intelligent video fire detection approach based on object detection technology using convolutional neural networks (CNN). First, a CNN model is trained for the fire detection task which is framed as a regression problem to predict bounding boxes and associated probabilities. In the application phase, videos from surveillance cameras are detected frame by frame. Once fire appears in the current frame, the model will output the coordinates of the fire region. Simultaneously, the frame where the fire region is localized will be immediately sent to safety supervisors as a fire alarm. This will help detect fire at the early stage, prevent fire spreading and improve the emergency response.

Description

Presentation

Keywords

Fire Detection Approach

Citation