Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition
Accurate fire load (combustible objects) information is crucial for safety design and resilience assessment of buildings. Traditional fire load acquisition methods, such as fire load survey, which are time-consuming, tedious, and error-prone, failed to adapt to dynamic changed indoor scenes. As a st...
Guardado en:
Autores principales: | , , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
IEEE
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/dd9e5e1f07ec4de3900d6d304d3d0d35 |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:dd9e5e1f07ec4de3900d6d304d3d0d35 |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:dd9e5e1f07ec4de3900d6d304d3d0d352021-11-18T00:05:10ZDeep Learning-Based Instance Segmentation for Indoor Fire Load Recognition2169-353610.1109/ACCESS.2021.3124831https://doaj.org/article/dd9e5e1f07ec4de3900d6d304d3d0d352021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9598928/https://doaj.org/toc/2169-3536Accurate fire load (combustible objects) information is crucial for safety design and resilience assessment of buildings. Traditional fire load acquisition methods, such as fire load survey, which are time-consuming, tedious, and error-prone, failed to adapt to dynamic changed indoor scenes. As a starting point of automatic fire load estimation, fast recognition and detection of indoor fire load are important. Thus, this research proposes a computer vision-based method to automatically detect indoor fire loads using deep learning-based instance segmentation. First, indoor elements are classified into different categories according to their material composition. Next, an image dataset of indoor scenes with instance annotations is developed. Finally, a deep learning model, based on Mask R-CNN, is developed and trained using transfer learning to detect fire loads in images. Experimental results show that our model achieves promising accuracy, as measured by an average precision (AP) of 40.5% and AP<sub>50</sub> of 59.2%, for instance segmentation on the dataset. A comparison with manual detection demonstrates the method’s high efficiency as it can detect fire load 1200 times faster than humans. This research contributes to the body of knowledge 1) a novel method of high accuracy and efficiency for automated fire load recognition in indoor environments based on instance segmentation; 2) training techniques for a deep learning model in a relatively small dataset of indoor images which includes complex scenes and a variety of instances; and 3) an image dataset with annotations of indoor fire loads. Although instance segmentation has been applied for several years, this is a pioneering research on using it for automated indoor fire load recognition, which paves the foundation for automatic fire load estimation and resilience assessment for the built environment.Yu-Cheng ZhouZhen-Zhong HuKe-Xiao YanJia-Rui LinIEEEarticleBuilding resiliencedeep learningfire load recognitionfire safetyindoor sceneinstance segmentationElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 148771-148782 (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
Building resilience deep learning fire load recognition fire safety indoor scene instance segmentation Electrical engineering. Electronics. Nuclear engineering TK1-9971 |
spellingShingle |
Building resilience deep learning fire load recognition fire safety indoor scene instance segmentation Electrical engineering. Electronics. Nuclear engineering TK1-9971 Yu-Cheng Zhou Zhen-Zhong Hu Ke-Xiao Yan Jia-Rui Lin Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
description |
Accurate fire load (combustible objects) information is crucial for safety design and resilience assessment of buildings. Traditional fire load acquisition methods, such as fire load survey, which are time-consuming, tedious, and error-prone, failed to adapt to dynamic changed indoor scenes. As a starting point of automatic fire load estimation, fast recognition and detection of indoor fire load are important. Thus, this research proposes a computer vision-based method to automatically detect indoor fire loads using deep learning-based instance segmentation. First, indoor elements are classified into different categories according to their material composition. Next, an image dataset of indoor scenes with instance annotations is developed. Finally, a deep learning model, based on Mask R-CNN, is developed and trained using transfer learning to detect fire loads in images. Experimental results show that our model achieves promising accuracy, as measured by an average precision (AP) of 40.5% and AP<sub>50</sub> of 59.2%, for instance segmentation on the dataset. A comparison with manual detection demonstrates the method’s high efficiency as it can detect fire load 1200 times faster than humans. This research contributes to the body of knowledge 1) a novel method of high accuracy and efficiency for automated fire load recognition in indoor environments based on instance segmentation; 2) training techniques for a deep learning model in a relatively small dataset of indoor images which includes complex scenes and a variety of instances; and 3) an image dataset with annotations of indoor fire loads. Although instance segmentation has been applied for several years, this is a pioneering research on using it for automated indoor fire load recognition, which paves the foundation for automatic fire load estimation and resilience assessment for the built environment. |
format |
article |
author |
Yu-Cheng Zhou Zhen-Zhong Hu Ke-Xiao Yan Jia-Rui Lin |
author_facet |
Yu-Cheng Zhou Zhen-Zhong Hu Ke-Xiao Yan Jia-Rui Lin |
author_sort |
Yu-Cheng Zhou |
title |
Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
title_short |
Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
title_full |
Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
title_fullStr |
Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
title_full_unstemmed |
Deep Learning-Based Instance Segmentation for Indoor Fire Load Recognition |
title_sort |
deep learning-based instance segmentation for indoor fire load recognition |
publisher |
IEEE |
publishDate |
2021 |
url |
https://doaj.org/article/dd9e5e1f07ec4de3900d6d304d3d0d35 |
work_keys_str_mv |
AT yuchengzhou deeplearningbasedinstancesegmentationforindoorfireloadrecognition AT zhenzhonghu deeplearningbasedinstancesegmentationforindoorfireloadrecognition AT kexiaoyan deeplearningbasedinstancesegmentationforindoorfireloadrecognition AT jiaruilin deeplearningbasedinstancesegmentationforindoorfireloadrecognition |
_version_ |
1718425254851772416 |