Presentation | 2021-03-03 A Study of Road Segmentation in Disaster Situations Using UAV Shinta Muto, Jun Ohya, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In this paper, we propose a system for segmenting road areas from aerial images using machine learning, assuming that fixed-wing UAVs are used to collect information in case of disasters. In the training phase, a deep learning network is trained using the training images to which preprocessing is applied. In the road segmentation phase, unknown images are inputted to the trained model so that road detection results are obtained. We created our own dataset from disaster images taken in Japan, and also created road-hidden images by overlaying disaster images (e.g. landslip) onto roads in non-disaster images. For the training, we used SegNet and DeepLab v3+ as the network structure and different loss functions for comparison. As a result, the highest accuracy was obtained by DeepLab v3+ and Dice Loss. In addition, as a result of applying the proposed method for the road-hidden images, promising results were obtained. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Machine Learning / Segmentation / Road Detection / SegNet / DeepLab / Loss Function |
Paper # | IMQ2020-30,IE2020-70,MVE2020-62 |
Date of Issue | 2021-02-22 (IMQ, IE, MVE) |
Conference Information | |
Committee | MVE / IMQ / IE / CQ |
---|---|
Conference Date | 2021/3/1(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Masayuki Ihara(NTT) / Toshiya Nakaguchi(Chiba Univ.) / Hideaki Kimata(NTT) / Hideyuki Shimonishi(NEC) |
Vice Chair | Kiyoshi Kiyokawa(NAIST) / Mitsuru Maeda(Canon) / Kenya Uomori(Osaka Univ.) / Kazuya Kodama(NII) / Keita Takahashi(Nagoya Univ.) / Jun Okamoto(NTT) / Takefumi Hiraguri(Nippon Inst. of Tech.) |
Secretary | Kiyoshi Kiyokawa(Oosaka Inst. of Tech.) / Mitsuru Maeda(NTT) / Kenya Uomori(Univ. of ToKyo) / Kazuya Kodama(Shizuoka Univ.) / Keita Takahashi(Sony Semiconductor Solutions) / Jun Okamoto(KDDI Research) / Takefumi Hiraguri(Nagoya Inst. of Tech.) |
Assistant | Naoya Isoyama(NAIST) / Takenori Hara(DNP) / Mitsuhiro Goto(NTT) / Hiroaki Kudo(Nagoya Univ.) / Masaru Tsuchida(NTT) / Keita Hirai(Chiba Univ.) / Shunsuke Iwamura(NHK) / Shinobu Kudo(NTT) / Yoshiaki Nishikawa(NEC) / Takuto Kimura(NTT) / Ryoichi Kataoka(KDDI Research) |
Paper Information | |
Registration To | Technical Committee on Media Experience and Virtual Environment / Technical Committee on Image Media Quality / Technical Committee on Image Engineering / Technical Committee on Communication Quality |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study of Road Segmentation in Disaster Situations Using UAV |
Sub Title (in English) | |
Keyword(1) | Machine Learning |
Keyword(2) | Segmentation |
Keyword(3) | Road Detection |
Keyword(4) | SegNet |
Keyword(5) | DeepLab |
Keyword(6) | Loss Function |
1st Author's Name | Shinta Muto |
1st Author's Affiliation | Waseda University(Waseda Univ.) |
2nd Author's Name | Jun Ohya |
2nd Author's Affiliation | Waseda University(Waseda Univ.) |
Date | 2021-03-03 |
Paper # | IMQ2020-30,IE2020-70,MVE2020-62 |
Volume (vol) | vol.120 |
Number (no) | IMQ-389,IE-390,MVE-391 |
Page | pp.pp.97-102(IMQ), pp.97-102(IE), pp.97-102(MVE), |
#Pages | 6 |
Date of Issue | 2021-02-22 (IMQ, IE, MVE) |