講演名 2019-02-28
[招待講演]Visual Question Generation for Class Acquisition of Unknown Objects (ECCV2018)
上原 康平(東大), Antonio Tejero-De-Pablos(東大), 牛久 祥孝(東大), 原田 達也(東大/理研),
PDFダウンロードページ PDFダウンロードページへ
抄録(和) Traditional image recognition methods only consider objects belonging to already learned classes. However, since training a recognition model with every object class in the world is unfeasible, a way of getting information on unknown objects (i.e., objects whose class has not been learned) is necessary. A way for an image recognition system to learn new classes could be asking a human about objects that are unknown. In this paper, we propose a method for generating questions about unknown objects in an image, as means to get information about classes that have not been learned. Our method consists of a module for proposing objects, a module for identifying unknown objects, and a module for generating questions about unknown objects. The experimental results via human evaluation show that our method can successfully get information about unknown objects in an image dataset. Our code and dataset are available at https://github.com/mil-tokyo/vqg-unknown
抄録(英) Traditional image recognition methods only consider objects belonging to already learned classes. However, since training a recognition model with every object class in the world is unfeasible, a way of getting information on unknown objects (i.e., objects whose class has not been learned) is necessary. A way for an image recognition system to learn new classes could be asking a human about objects that are unknown. In this paper, we propose a method for generating questions about unknown objects in an image, as means to get information about classes that have not been learned. Our method consists of a module for proposing objects, a module for identifying unknown objects, and a module for generating questions about unknown objects. The experimental results via human evaluation show that our method can successfully get information about unknown objects in an image dataset. Our code and dataset are available at https://github.com/mil-tokyo/vqg-unknown
キーワード(和) Visual question generation / Unknown object recognition / Unknown object class acquisition / Real world recognition
キーワード(英) Visual question generation / Unknown object recognition / Unknown object class acquisition / Real world recognition
資料番号 PRMU2018-114,CNR2018-37
発行日 2019-02-21 (PRMU, CNR)

研究会情報
研究会 PRMU / CNR
開催期間 2019/2/28(から2日開催)
開催地(和) 徳島大学
開催地(英)
テーマ(和) ロホ?ティクスとそれを支えるヒ?シ?ョン技術
テーマ(英)
委員長氏名(和) 佐藤 真一(NII) / 小野 哲雄(北大)
委員長氏名(英) Shinichi Sato(NII) / Tetsuo Ono(Hokkaido Univ.)
副委員長氏名(和) 井尻 善久(オムロン) / 玉木 徹(広島大) / 神原 誠之(奈良先端大) / 高汐 一紀(慶大)
副委員長氏名(英) Yoshihisa Ijiri(Omron) / Toru Tamaki(Hiroshima Univ.) / Masayuki Kanbara(NAIST) / Kazunori Takashio(Keio Univ.)
幹事氏名(和) 石井 雅人(NEC) / 菅野 裕介(阪大) / 坂本 大介(北大) / 吉岡 康介(パナソニック)
幹事氏名(英) Masato Ishii(NEC) / Yusuke Sugano(Osaka Univ.) / Daisuke Sakamoto(Hokkaido Univ.) / Kosuke Yoshioka(Panasonic)
幹事補佐氏名(和) 入江 豪(NTT) / 牛久 祥孝(東大) / 水戸 和(セコム) / 小林 優佳(東芝) / 石原 達也(NTT)
幹事補佐氏名(英) Go Irie(NTT) / Yoshitaka Ushiku(Univ. of Tokyo) / Wataru Mito(SECOM) / Yuka Kobayashi(Toshiba) / Tatsuya Ishihara(NTT)

講演論文情報詳細
申込み研究会 Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Cloud Network Robotics
本文の言語 JPN-ONLY
タイトル(和) [招待講演]Visual Question Generation for Class Acquisition of Unknown Objects (ECCV2018)
サブタイトル(和)
タイトル(英)
サブタイトル(和)
キーワード(1)(和/英) Visual question generation / Visual question generation
キーワード(2)(和/英) Unknown object recognition / Unknown object recognition
キーワード(3)(和/英) Unknown object class acquisition / Unknown object class acquisition
キーワード(4)(和/英) Real world recognition / Real world recognition
第 1 著者 氏名(和/英) 上原 康平
第 1 著者 所属(和/英) 東京大学(略称:東大)
第 2 著者 氏名(和/英) Antonio Tejero-De-Pablos
第 2 著者 所属(和/英) 東京大学(略称:東大)
第 3 著者 氏名(和/英) 牛久 祥孝
第 3 著者 所属(和/英) 東京大学(略称:東大)
第 4 著者 氏名(和/英) 原田 達也
第 4 著者 所属(和/英) 東京大学/理化学研究所(略称:東大/理研)
発表年月日 2019-02-28
資料番号 PRMU2018-114,CNR2018-37
巻番号(vol) vol.118
号番号(no) PRMU-459,CNR-460
ページ範囲 pp.5-5(PRMU), pp.5-5(CNR),
ページ数 1
発行日 2019-02-21 (PRMU, CNR)