{"created":"2024-04-19T07:05:23.736764+00:00","id":6586,"links":{},"metadata":{"_buckets":{"deposit":"53184cf3-d672-4575-82d5-a3325781293b"},"_deposit":{"created_by":10,"id":"6586","owner":"10","owners":[10],"owners_ext":{"displayname":"","username":""},"pid":{"revision_id":0,"type":"depid","value":"6586"},"status":"published"},"_oai":{"id":"oai:nied-repo.bosai.go.jp:00006586","sets":["631:1:1713510153521"]},"author_link":["293","149"],"item_10001_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"2024-04","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"89","bibliographicPageEnd":"10","bibliographicPageStart":"1","bibliographic_titles":[{"bibliographic_title":"防災科学技術研究所 研究報告","bibliographic_titleLang":"ja"},{"bibliographic_title":"Report of the National Research Institute for Earth Science and Disaster Resilience","bibliographic_titleLang":"en"}]}]},"item_10001_description_5":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"航空撮影画像からの斜面崩壊領域検出に対する画像認識技術の適用可能性の検証を行った．画像認識手法は，画像内の検出対象領域を画素単位で抽出するimage segmentation 技術の1 つであるMaskR-CNN を採用した．また学習・検証には平成30 年7 月豪雨直後の広島地域の航空撮影画像（オルソ画像）および専門家による斜面崩壊領域検出結果を用いた．斜面崩壊領域は撮影画像中の限られた領域であるため必然的に画像に含まれる斜面崩壊領域の面積はそれ以外の領域に較べ非常に少なくなる．精度向上策として，この教師データの不均衡への対処法についても検証を行った．検証の結果，教師データの不均衡に応じた重みを学習時の損失計算に導入することの効果が確認でき，最終的な斜面崩壊領域検出の精度はF 値0.68 であった．\n今後の課題は，データ増加も含めた検出精度の向上や地域や災害種別が異なる場合の精度確認・対応である．また災害対応として本手法を活用する方法や必要機能についての検討も必要である．","subitem_description_language":"ja","subitem_description_type":"Abstract"},{"subitem_description":"We studied the utility of image recognition technology for detecting slope failure areas from aerial imagery by using Mask R-CNN (Region-based Convolutional Neural Network), which is a kind of image segmentation. In addition, aerial imagery (orthoimagery) and slope failure area data from a torrential rain event in 2018 in western Japan was used. To improve accuracy, treating imbalanced training data was considered because of its high probability in the case of data for slope failure areas detection. We confirmed that the loss calculation weighting according to the area ratio of the slope failure areas was effective in improving accuracy. The F-value for the detection accuracy of Mask R-CNN with weighted loss was 0.68.","subitem_description_language":"en","subitem_description_type":"Abstract"}]},"item_10001_description_6":{"attribute_name":"内容記述","attribute_value_mlt":[{"subitem_description":"(ONLINE先行公開につき、ページ数および発行年情報は仮のものになります。引用にはDOIをご利用ください)","subitem_description_language":"ja","subitem_description_type":"Other"}]},"item_10001_identifier_registration":{"attribute_name":"ID登録","attribute_value_mlt":[{"subitem_identifier_reg_text":"10.24732/NIED.00006586","subitem_identifier_reg_type":"JaLC"}]},"item_10001_publisher_8":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"防災科学技術研究所","subitem_publisher_language":"ja"}]},"item_10001_source_id_9":{"attribute_name":"収録物識別子","attribute_value_mlt":[{"subitem_source_identifier":"1347-7471","subitem_source_identifier_type":"EISSN"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"内山, 庄一郎","creatorNameLang":"ja"},{"creatorName":"ウチヤマ, ショウイチロウ","creatorNameLang":"ja-Kana"},{"creatorName":"UCHIYAMA, Shoichiro","creatorNameLang":"en"}],"familyNames":[{"familyName":"内山","familyNameLang":"ja"},{"familyName":"ウチヤマ","familyNameLang":"ja-Kana"},{"familyName":"UCHIYAMA","familyNameLang":"en"}],"givenNames":[{"givenName":"庄一郎","givenNameLang":"ja"},{"givenName":"ショウイチロウ","givenNameLang":"ja-Kana"},{"givenName":"Shoichiro","givenNameLang":"en"}],"nameIdentifiers":[{"nameIdentifier":"293","nameIdentifierScheme":"WEKO"}]},{"creatorNames":[{"creatorName":"中西, 良成","creatorNameLang":"ja"}],"familyNames":[{"familyName":"中西","familyNameLang":"ja"}],"givenNames":[{"givenName":"良成","givenNameLang":"ja"}],"nameIdentifiers":[{"nameIdentifier":"149","nameIdentifierScheme":"WEKO"}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_access","date":[{"dateType":"Available","dateValue":"2024-04-30"}],"filename":"R89-1.pdf","filesize":[{"value":"1.7 MB"}],"format":"application/pdf","mimetype":"application/pdf","url":{"url":"https://nied-repo.bosai.go.jp/record/6586/files/R89-1.pdf"},"version_id":"9310f953-a564-4810-9d88-e316fe0f9072"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"Slope failure, Image segmentation, Mask R-CNN (Region-based Convolutional Neural Network)","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"斜面崩壊","subitem_subject_language":"ja","subitem_subject_scheme":"Other"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"journal article","resourceuri":"http://purl.org/coar/resource_type/c_6501"}]},"item_title":"航空撮影画像からの画像認識技術による斜面崩壊領域検出の検討","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"航空撮影画像からの画像認識技術による斜面崩壊領域検出の検討","subitem_title_language":"ja"},{"subitem_title":"Using Aerial Imagery to Detect Slope Failure Areas","subitem_title_language":"en"}]},"item_type_id":"10001","owner":"10","path":["1713510153521"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2024-04-27"},"publish_date":"2024-04-27","publish_status":"0","recid":"6586","relation_version_is_last":true,"title":["航空撮影画像からの画像認識技術による斜面崩壊領域検出の検討"],"weko_creator_id":"10","weko_shared_id":-1},"updated":"2024-04-22T06:49:35.255760+00:00"}