Uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models
This paper shows uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models. Model ensembles and Monte Carlo Dropout (MCD) were used to evaluate uncertainty. Three methodologies were examined as performance improvement...
Saved in:
| Published in | Journal of the Japan society of photogrammetry and remote sensing Vol. 61; no. 6; pp. 368 - 386 |
|---|---|
| Main Authors | , , , , , |
| Format | Journal Article |
| Language | English Japanese |
| Published |
Tokyo
Japan Society of Photogrammetry and Remote Sensing
2022
Japan Science and Technology Agency |
| Subjects | |
| Online Access | Get full text |
| ISSN | 0285-5844 1883-9061 |
| DOI | 10.4287/jsprs.61.368 |
Cover
| Abstract | This paper shows uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models. Model ensembles and Monte Carlo Dropout (MCD) were used to evaluate uncertainty. Three methodologies were examined as performance improvement methodologies. The methodologies were slide processing, recall/precision emphasized models, and transfer learning with an inherent factor of landslide. The recall/precision emphasized models were developed by the improved loss function. The result showed that MCD could not be an alternative to model ensembles. In performance improvement methodologies, the transfer learning with geology distribution scored at 80% of precision. The recall/precision emphasized models inferred the distribution of landslide mass adequately. The effectiveness of the slide processing was found to be dependent on the performance of the trained model. |
|---|---|
| AbstractList | This paper shows uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models. Model ensembles and Monte Carlo Dropout (MCD) were used to evaluate uncertainty. Three methodologies were examined as performance improvement methodologies. The methodologies were slide processing, recall/precision emphasized models, and transfer learning with an inherent factor of landslide. The recall/precision emphasized models were developed by the improved loss function. The result showed that MCD could not be an alternative to model ensembles. In performance improvement methodologies, the transfer learning with geology distribution scored at 80% of precision. The recall/precision emphasized models inferred the distribution of landslide mass adequately. The effectiveness of the slide processing was found to be dependent on the performance of the trained model. |
| Author | TAKEUCHI, Yutaro FURUKI, Hirokazu YOSHIDA, Kazuya NAKAMURA, Yoshio YAMAMOTO, Yoshiyuki UTSUKI, Shinji |
| Author_xml | – sequence: 1 fullname: NAKAMURA, Yoshio – sequence: 1 fullname: TAKEUCHI, Yutaro – sequence: 1 fullname: FURUKI, Hirokazu – sequence: 1 fullname: YOSHIDA, Kazuya – sequence: 1 fullname: UTSUKI, Shinji – sequence: 1 fullname: YAMAMOTO, Yoshiyuki |
| BookMark | eNpFkMtOwzAQRS1UJErpjg-wxJaUOPErK4QQL6kSG7q2nHjSukrsYKdI_Qc-GkOrsprFPXNHcy7RxHkHCF2TfEELKe62cQhxwcmi5PIMTYmUZVblnEzQNC8ky5ik9ALNY7R1XuRUFLIqp-h75RoIo7Zu3GP40t1Oj9Y7rJ3BQ_CDj7qLuPUBDxDS6HXise1T9gU9uBH3MG688Z1fW4jYuhTqNeA1OAiHLt_iLvXFzhrAvY4R13tsAAbcgQ7OujXuvYEuXqHzNp2D-XHO0Or56ePxNVu-v7w9PiyzhrBKZlJw1nBgdcsNk0CAlYaLWgpSCVM3XEgJJjd10QpaCdJSY4QkZVUnDZQTXs7QzaE3ffG5gziqrd8Fl06qomIiF5RSlqjbA9UEH2OAVg0h_Rb2iuTqV7n6U644UUl5wu8P-DaOScAJ1mG0TQf_MD9unJJmo4MCV_4ATS6Ruw |
| Cites_doi | 10.2208/jscejhe.77.2_I_319 10.1109/CVPR.2017.632 10.4287/jsprs.61.14 10.2208/jscejipm.77.4_400 |
| ContentType | Journal Article |
| Copyright | 2022 Japan Society of Photogrammetry and Remote Sensing Copyright Japan Science and Technology Agency 2022 |
| Copyright_xml | – notice: 2022 Japan Society of Photogrammetry and Remote Sensing – notice: Copyright Japan Science and Technology Agency 2022 |
| DBID | AAYXX CITATION 7SP 8FD FR3 H8D KR7 L7M |
| DOI | 10.4287/jsprs.61.368 |
| DatabaseName | CrossRef Electronics & Communications Abstracts Technology Research Database Engineering Research Database Aerospace Database Civil Engineering Abstracts Advanced Technologies Database with Aerospace |
| DatabaseTitle | CrossRef Aerospace Database Civil Engineering Abstracts Engineering Research Database Technology Research Database Advanced Technologies Database with Aerospace Electronics & Communications Abstracts |
| DatabaseTitleList | Aerospace Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Visual Arts |
| EISSN | 1883-9061 |
| EndPage | 386 |
| ExternalDocumentID | 10_4287_jsprs_61_368 article_jsprs_61_6_61_368_article_char_en |
| GroupedDBID | ABJNI ACGFS ALMA_UNASSIGNED_HOLDINGS JSF KQ8 RJT ~02 AAYXX CITATION 7SP 8FD FR3 H8D KR7 L7M |
| ID | FETCH-LOGICAL-c1598-8765c6e5bf6d58e1e53d67b87197dbc6788ed0db2f74971f4dd78139b28546163 |
| ISSN | 0285-5844 |
| IngestDate | Sun Jun 29 12:41:51 EDT 2025 Tue Jul 01 04:16:10 EDT 2025 Wed Sep 03 06:30:28 EDT 2025 |
| IsDoiOpenAccess | true |
| IsOpenAccess | true |
| IsPeerReviewed | false |
| IsScholarly | true |
| Issue | 6 |
| Language | English Japanese |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-c1598-8765c6e5bf6d58e1e53d67b87197dbc6788ed0db2f74971f4dd78139b28546163 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
| OpenAccessLink | https://www.jstage.jst.go.jp/article/jsprs/61/6/61_368/_article/-char/en |
| PQID | 2957074445 |
| PQPubID | 2048407 |
| PageCount | 19 |
| ParticipantIDs | proquest_journals_2957074445 crossref_primary_10_4287_jsprs_61_368 jstage_primary_article_jsprs_61_6_61_368_article_char_en |
| ProviderPackageCode | CITATION AAYXX |
| PublicationCentury | 2000 |
| PublicationDate | 2022 2022-00-00 20220101 |
| PublicationDateYYYYMMDD | 2022-01-01 |
| PublicationDate_xml | – year: 2022 text: 2022 |
| PublicationDecade | 2020 |
| PublicationPlace | Tokyo |
| PublicationPlace_xml | – name: Tokyo |
| PublicationTitle | Journal of the Japan society of photogrammetry and remote sensing |
| PublicationTitleAlternate | Journal of the Japan society of photogrammetry and remote sensing |
| PublicationYear | 2022 |
| Publisher | Japan Society of Photogrammetry and Remote Sensing Japan Science and Technology Agency |
| Publisher_xml | – name: Japan Society of Photogrammetry and Remote Sensing – name: Japan Science and Technology Agency |
| References | 戸田堅一郎，2014．曲率と傾斜による立体図法（CS立体図）を用いた地形判読，森林立地，Vol. 56，No.２，75-79 佐藤 剛，土志田正二，2018．地すべり地形分布図を作成するうえで地形判読の専門家は地すべり地の何を見ているのか？，公益財団法人国土地理協会第18回学術研究助成，https://www.kokudo.or.jp/grant/pdf/2018/sato206.pdf. (2022/９/４確認) 竹内祐太朗，山本義幸，古木宏和，宇津木慎司，吉田一也，中村吉男，2022．空間的不均一性・連続性に関する正規化処理による地すべり移動体の深層生成，写真測量とリモートセンシング，vol. 61，no.１，14-31 叶井和樹，久保 栞，山根達郎，全 邦釘，2021．Mask R-CNNによる航空写真からの土砂崩壊地自動検出手法．AI・データサイエンス論文集，２巻，223-231 青島亘佐，山本拓海，中野 聡，中村秀明，2020．深層学習によるセグメンテーション手法を用いたコンクリート表面の変状領域の検出．AI・データサイエンス論文集，１巻，481-490 三浦奈都，宮本 崇，天方匡純，安野貴人，石井 明，2021．ベイズ深層学習を用いた予報雨量の不確実性を考慮したダム流入量の確率的予測，AI・データサイエンス論文集，２巻，933-943 防災科学技術研究所，2002．地すべり地形GISデータ：国立研究開発法人防災科学技術研究所地すべり地形分布図，https://dil-opac.bosai.go.jp/publication/nied_tech_note/landslidemap/gis.html. (2022/９/１確認) 久野 元，蘇 迪，長山智則，2020．LSTMとモンテカルロドロップアウトに基づく浮きまくらぎ検知手法の数値的検討，AI・データサイエンス論文集，１巻，536-544 及川 康，片田敏孝，2016．避難勧告等の見逃し・空振りが住民対応行動の意思決定に及ぼす影響．災害情報，14巻，93-104 TensorFlow，2022，pix2pix：条件付きGANによる画像から画像への変換，https://www.tensorflow.org/tutorials/generative/pix2pix (2022/９/２確認) 山口真一，1967．地すべりの素因と誘因について．地すべり，４巻，１号，４-11 平野英孝，相馬一義，宮本 崇，石平 博，馬籠 純，黒田 晴，倉上 健，2021．富士川周辺地域における深層学習を活用した土砂災害危険度現況推定手法の構築とその評価，水文・水資源学会研究発表会要旨集，34巻，水文・水資源学会/日本水文科学会 2021年度研究発表会，講演番号PP-B-26 古木宏和，2021．３次元地形情報を用いた深層学習による地すべり移動体抽出，日本地すべり学会誌，Vol. 58，No.２，65-72 内閣府，2014．避難勧告等の判断・伝達マニュアル作成ガイドライン，https://www.bousai.go.jp/oukyu/hinankankoku/guideline/guideline_2014.html (2022/９/２確認) Sam Maddrell-Mander, 2019, pix2pix GAN in TensorFlow 2.0, https://towardsdatascience.com/pix2pix-gan-in-tensorflow-2-0-fe0ab475c713 (2022/９/２確認) 木村延明，皆川裕樹，福重雄大，木村匡臣，馬場大地，2021．深層学習モデルに物理モデルを融合させた排水機場水位予測への転移学習の適用．土木学会論文集B1（水工学），77巻，２号，I_319-I_324 児島利治，Chantsal NARANTSETSEG，大橋慶介，2021．深層学習を用いた地形図の土地利用分類，土木学会論文集D3（土木計画学），77巻，４号，400-411 岡谷貴之，2022．深層学習改訂第２版，講談社，東京 産総研地質調査総合センター 地質図類データダウンロード，https://gbank.gsj.jp/datastore/download.php. (2022/９/１確認) Fort, S., Hu, H., Lakshminarayanan, B., 2019. Deep Ensembles : A Loss Landscape Perspective. arXiv, 1912.02757. Gal, Y., Ghahramani, Z., 2016. Dropout as a Bayesian Approximation : Representing Model Uncertainty in Deep Learning. Proceedings of the 33rd International Conference on Machine Learning, 48, 1050-1059. Isola, P., Zhu, J.-Y., Zhou, T., and Efros, A.A., 2016. Imageto-image translation with conditional adversarial networks, CoRR, Vol. abs/1611.07004. 11 22 12 13 14 15 16 17 18 19 1 2 3 4 5 6 7 8 9 20 10 21 |
| References_xml | – reference: Sam Maddrell-Mander, 2019, pix2pix GAN in TensorFlow 2.0, https://towardsdatascience.com/pix2pix-gan-in-tensorflow-2-0-fe0ab475c713 (2022/９/２確認) – reference: 叶井和樹，久保 栞，山根達郎，全 邦釘，2021．Mask R-CNNによる航空写真からの土砂崩壊地自動検出手法．AI・データサイエンス論文集，２巻，223-231． – reference: 防災科学技術研究所，2002．地すべり地形GISデータ：国立研究開発法人防災科学技術研究所地すべり地形分布図，https://dil-opac.bosai.go.jp/publication/nied_tech_note/landslidemap/gis.html. (2022/９/１確認) – reference: 佐藤 剛，土志田正二，2018．地すべり地形分布図を作成するうえで地形判読の専門家は地すべり地の何を見ているのか？，公益財団法人国土地理協会第18回学術研究助成，https://www.kokudo.or.jp/grant/pdf/2018/sato206.pdf. (2022/９/４確認) – reference: 山口真一，1967．地すべりの素因と誘因について．地すべり，４巻，１号，４-11． – reference: 児島利治，Chantsal NARANTSETSEG，大橋慶介，2021．深層学習を用いた地形図の土地利用分類，土木学会論文集D3（土木計画学），77巻，４号，400-411． – reference: Fort, S., Hu, H., Lakshminarayanan, B., 2019. Deep Ensembles : A Loss Landscape Perspective. arXiv, 1912.02757. – reference: 青島亘佐，山本拓海，中野 聡，中村秀明，2020．深層学習によるセグメンテーション手法を用いたコンクリート表面の変状領域の検出．AI・データサイエンス論文集，１巻，481-490． – reference: Isola, P., Zhu, J.-Y., Zhou, T., and Efros, A.A., 2016. Imageto-image translation with conditional adversarial networks, CoRR, Vol. abs/1611.07004. – reference: 竹内祐太朗，山本義幸，古木宏和，宇津木慎司，吉田一也，中村吉男，2022．空間的不均一性・連続性に関する正規化処理による地すべり移動体の深層生成，写真測量とリモートセンシング，vol. 61，no.１，14-31． – reference: 岡谷貴之，2022．深層学習改訂第２版，講談社，東京． – reference: Gal, Y., Ghahramani, Z., 2016. Dropout as a Bayesian Approximation : Representing Model Uncertainty in Deep Learning. Proceedings of the 33rd International Conference on Machine Learning, 48, 1050-1059. – reference: 内閣府，2014．避難勧告等の判断・伝達マニュアル作成ガイドライン，https://www.bousai.go.jp/oukyu/hinankankoku/guideline/guideline_2014.html (2022/９/２確認) – reference: 木村延明，皆川裕樹，福重雄大，木村匡臣，馬場大地，2021．深層学習モデルに物理モデルを融合させた排水機場水位予測への転移学習の適用．土木学会論文集B1（水工学），77巻，２号，I_319-I_324． – reference: 平野英孝，相馬一義，宮本 崇，石平 博，馬籠 純，黒田 晴，倉上 健，2021．富士川周辺地域における深層学習を活用した土砂災害危険度現況推定手法の構築とその評価，水文・水資源学会研究発表会要旨集，34巻，水文・水資源学会/日本水文科学会 2021年度研究発表会，講演番号PP-B-26． – reference: 三浦奈都，宮本 崇，天方匡純，安野貴人，石井 明，2021．ベイズ深層学習を用いた予報雨量の不確実性を考慮したダム流入量の確率的予測，AI・データサイエンス論文集，２巻，933-943． – reference: 古木宏和，2021．３次元地形情報を用いた深層学習による地すべり移動体抽出，日本地すべり学会誌，Vol. 58，No.２，65-72． – reference: 及川 康，片田敏孝，2016．避難勧告等の見逃し・空振りが住民対応行動の意思決定に及ぼす影響．災害情報，14巻，93-104． – reference: 戸田堅一郎，2014．曲率と傾斜による立体図法（CS立体図）を用いた地形判読，森林立地，Vol. 56，No.２，75-79． – reference: 久野 元，蘇 迪，長山智則，2020．LSTMとモンテカルロドロップアウトに基づく浮きまくらぎ検知手法の数値的検討，AI・データサイエンス論文集，１巻，536-544． – reference: 産総研地質調査総合センター 地質図類データダウンロード，https://gbank.gsj.jp/datastore/download.php. (2022/９/１確認) – reference: TensorFlow，2022，pix2pix：条件付きGANによる画像から画像への変換，https://www.tensorflow.org/tutorials/generative/pix2pix (2022/９/２確認) – ident: 2 – ident: 17 – ident: 3 – ident: 18 – ident: 4 – ident: 5 doi: 10.2208/jscejhe.77.2_I_319 – ident: 1 – ident: 20 doi: 10.1109/CVPR.2017.632 – ident: 12 – ident: 11 doi: 10.4287/jsprs.61.14 – ident: 10 – ident: 19 – ident: 13 – ident: 16 – ident: 14 – ident: 15 – ident: 6 – ident: 9 – ident: 8 – ident: 21 – ident: 7 doi: 10.2208/jscejipm.77.4_400 – ident: 22 |
| SSID | ssib020472893 ssib036263174 ssib004305393 ssj0000400259 ssib031741132 |
| Score | 2.1720786 |
| Snippet | This paper shows uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models.... |
| SourceID | proquest crossref jstage |
| SourceType | Aggregation Database Index Database Publisher |
| StartPage | 368 |
| SubjectTerms | Deep learning Image processing Landslides Proposals Recall Uncertainty |
| Title | Uncertainty evaluation and proposals for performance improvement methodologies in image generation of landslide mass by deep learning models |
| URI | https://www.jstage.jst.go.jp/article/jsprs/61/6/61_368/_article/-char/en https://www.proquest.com/docview/2957074445 |
| Volume | 61 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| ispartofPNX | Journal of the Japan society of photogrammetry and remote sensing, 2022, Vol.61(6), pp.368-386 |
| journalDatabaseRights | – providerCode: PRVAFT databaseName: Open Access Digital Library customDbUrl: eissn: 1883-9061 dateEnd: 99991231 omitProxy: true ssIdentifier: ssj0000400259 issn: 0285-5844 databaseCode: KQ8 dateStart: 19750101 isFulltext: true titleUrlDefault: http://grweb.coalliance.org/oadl/oadl.html providerName: Colorado Alliance of Research Libraries |
| link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV1Lb9NAEF61hQMcEE81UNAe4IRcvM6uvT5apVXaqDySGvVm-bFO3aLYsuND8xv40czu-hUCUunFSryOs_HMN_vNZGYWofdEpDGsApFhmrFpUNAhg4fcNFhipSyJYIzJauTzL_bEp2eX7HJnd3-QtVSvosN4_de6kvtIFc6BXGWV7H9ItrspnIDXIF84goTheCcZ-yAx9Y8-MOm-bXdT-58XeSV7I8s8wmJQHpCpMIKKCjb7Ryv7pxKzYFDm8CxUL-qWS6pq4J-ZSnUFswh8NRGiaPebWOjNdKp_sFzJa89gQV6q6LzQ-R_FVb5SaWHw_aVuAFUK0BnxsZL59M1iKkPU3tQ792eeWijy6konjalIgzc99o8mp2qkXoVlN3Liz_ypOj_JyvwmXNedZfs6n5x-9nQSybq-DYchD6t3jfV85_18v23Pd6bnOx_MV5tTizMD6JYOXwht7jkfG66p28G360HzLtsy7mO9AVDDE8a6hfefS5B0QeUSVBVldWiTw-5DG029G5UJ1GWBTQJbHuDaoB2RRXeg47vogeUAh5LpBt-HjdnAePaW1ZI9PwddhIAUUkL6imHVc4g0RF9xFCrJrvIN28eiq0Lk9D8NJ7_B1x5eg8uy2OYtioxdPEVPGv3Cnv4Vz9DOdfgcPf6RVbU-W71AvwbgwD04MAgPd-DAgAo8AAcegANvgANnS6zAgXtw4DzFHTiwBAeObrEEB27BgTU4XiL_5PjiaGI0W48YMfB7LjkCi23BotROGBdEsHFiOxF3iOuABQOGx0ViJpGVOtR1SEqTxOHgTEWyINkGH-cV2lvmS7GPMIghJJawSEgJdRwRUirSyIoT13TH8HqEPrQPOCh0h5kAPHMpiF49QBAjxPXT7666sxKN0EErr6AxX1VgucwB_4FS9vr-d36DHkmA6pjkAdpblbV4Cyx9Fb1TCvsbiBnz2w |
| linkProvider | Colorado Alliance of Research Libraries |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Uncertainty+evaluation+and+proposals+for+performance+improvement+methodologies+in+image+generation+of+landslide+mass+by+deep+learning+models&rft.jtitle=Journal+of+the+Japan+society+of+photogrammetry+and+remote+sensing&rft.au=NAKAMURA%2C+Yoshio&rft.au=TAKEUCHI%2C+Yutaro&rft.au=FURUKI%2C+Hirokazu&rft.au=YOSHIDA%2C+Kazuya&rft.date=2022&rft.pub=Japan+Society+of+Photogrammetry+and+Remote+Sensing&rft.issn=0285-5844&rft.eissn=1883-9061&rft.volume=61&rft.issue=6&rft.spage=368&rft.epage=386&rft_id=info:doi/10.4287%2Fjsprs.61.368&rft.externalDocID=article_jsprs_61_6_61_368_article_char_en |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0285-5844&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0285-5844&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0285-5844&client=summon |