Studies in Computational Intelligence, pages 175-183

Roof Defect Segmentation on Aerial Images Using Neural Networks

Publication typeBook Chapter
Publication date2020-10-02
Quartile SCImago
Q4
Quartile WOS
Impact factor
ISSN1860949X, 18609503
Abstract
The paper describes usage of deep neural networks for flat roof defect segmentation on aerial images. Such architectures as U-Net, DeepLabV3+ and HRNet+ OCR are studied for recognition five categories of roof defects: “hollows”, “swelling”, “folds”, “patches” and “breaks”. Paper introduces RoofD dataset containing 6400 image pairs: aerial photos and corresponding ground truth masks. Based on this dataset different approaches to neural networks training are analyzed. New SDice coefficient with categorical cross-entropy is studied for precise training of U-Net and proposed light U-NetMCT architecture. Weighted categorical cross-entropy is studied for DeepLabV3+ and HRNet+ OCR training. It is shown that these training methods allow correctly recognize rare categories of defects. The state-of-the-art model multi-scale HRNet+ OCR achieves the best quality metric of 0.44 mean IoU. In sense of inference time the fastest model is U-NetMCT and DeeplabV3+ with worse quality of 0.33–0.37 mean IoU. The most difficult category for segmentation is “patches” because of small amount of images with this category in the dataset. Paper also demonstrates the possibility of implementation of the obtained models in the special software for automation of the roof state examination in industry, housing and communal services.

Citations by journals

1
Journal of Physics: Conference Series
Journal of Physics: Conference Series, 1, 33.33%
Journal of Physics: Conference Series
1 publication, 33.33%
IEEE Transactions on Intelligent Transportation Systems
IEEE Transactions on Intelligent Transportation Systems, 1, 33.33%
IEEE Transactions on Intelligent Transportation Systems
1 publication, 33.33%
Bulletin of Belgorod State Technological University named after V G Shukhov
Bulletin of Belgorod State Technological University named after V G Shukhov, 1, 33.33%
Bulletin of Belgorod State Technological University named after V G Shukhov
1 publication, 33.33%
1

Citations by publishers

1
IOP Publishing
IOP Publishing, 1, 33.33%
IOP Publishing
1 publication, 33.33%
IEEE
IEEE, 1, 33.33%
IEEE
1 publication, 33.33%
Belgorod State Technological University named after V.G. Shoukhov
Belgorod State Technological University named after V.G. Shoukhov, 1, 33.33%
Belgorod State Technological University named after V.G. Shoukhov
1 publication, 33.33%
1
  • We do not take into account publications that without a DOI.
  • Statistics recalculated only for publications connected to researchers, organizations and labs registered on the platform.
  • Statistics recalculated weekly.
Metrics
Share
Cite this
GOST |
Cite this
GOST Copy
Yudin D. et al. Roof Defect Segmentation on Aerial Images Using Neural Networks // Studies in Computational Intelligence. 2020. pp. 175-183.
GOST all authors (up to 50) Copy
Yudin D., Adeshkin V., Dolzhenko A. V., Polyakov A., Naumov A. Roof Defect Segmentation on Aerial Images Using Neural Networks // Studies in Computational Intelligence. 2020. pp. 175-183.
RIS |
Cite this
RIS Copy
TY - GENERIC
DO - 10.1007/978-3-030-60577-3_20
UR - https://doi.org/10.1007%2F978-3-030-60577-3_20
TI - Roof Defect Segmentation on Aerial Images Using Neural Networks
T2 - Studies in Computational Intelligence
AU - Yudin, D.
AU - Adeshkin, Vasily
AU - Dolzhenko, Alexandr V
AU - Polyakov, Alexandr
AU - Naumov, Andrey
PY - 2020
DA - 2020/10/02 00:00:00
PB - Springer Nature
SP - 175-183
SN - 1860-949X
SN - 1860-9503
ER -
BibTex
Cite this
BibTex Copy
@incollection{2020_Yudin
author = {D. Yudin and Vasily Adeshkin and Alexandr V Dolzhenko and Alexandr Polyakov and Andrey Naumov},
title = {Roof Defect Segmentation on Aerial Images Using Neural Networks},
publisher = {Springer Nature},
year = {2020},
pages = {175--183},
month = {oct}
}
Found error?