特任研究員
国立情報学研究所 (NII)

: 〒101-8430 東京都千代田区一ツ橋 2-1-2 国立情報学研究所 佐藤真一研究室

: matsui(at)nii.ac.jp / : Resume / : Google scholar / : Github / : kaggle

私はNIIの特任研究員で,佐藤真一研究室にて研究を進めています. 私は2016年に東京大学にて相澤清晴教授のもとで博士号を取得しました. 私の研究分野はコンピュータビジョン,マルチメディア処理です.特に,大規模データ処理(画像検索,近似最近傍探索,クラスタリング)を専門としています.

News

Projects


Reconfigurable Inverted Index
Yusuke Matsui, Ryota Hinami, and Shin'ichi Satoh
ACMMM 2018 (oral).
[Paper] [arXiv] [Project] [Code]

直積量子化を用いた近似最近傍探索に関するサーベイ
松井勇佑, 内田祐介, Hervé Jégou, 佐藤真一
ITE Transactions on Media Technology and Applications 2018.
[Paper] [Project]

PQk-means: Billion-scale Clustering for Product-quantized Codes
Yusuke Matsui* , Keisuke Ogaki* , Toshihiko Yamasaki, and Kiyoharu Aizawa
*Joint first authors.
ACMMM 2017.
[Paper] [Project] [Code] [News release]

古典籍画像検索
松井勇佑, Chairath Sirirattanapol, 佐藤真一, 松田訓典, 山本和明.
[Project] [古典籍dataset] [News release]

PQTable: Nonexhaustive Fast Search for Product-Quantized Codes Using Hash Tables
Yusuke Matsui , Toshihiko Yamasaki, and Kiyoharu Aizawa
ICCV 2015, TMM 2018.
[Paper] [Project] [Code]

1時間で画像検索エンジンを作る
Yusuke Matsui
招待講演:1時間で画像検索エンジンを作る, CGVI研究会, 2017.
[Project] [Slide] [Code] [Demo]

DrawFromDrawings: 2D Drawing Assistance via Stroke Interpolation with a Sketch Database
Yusuke Matsui , Takaaki Shiratori, and Kiyoharu Aizawa
IEEE Transactions on Visualization and Computer Graphics (TVCG), 2017.
[Paper] [Project] [Video (YouTube)]

Sketch-based Manga Retrieval using Manga109 Dataset
Yusuke Matsui , Kota Ito, Yuji Aramaki, Azuma Fujimoto, Toru Ogawa, Toshihiko Yamasaki, and Kiyoharu Aizawa
Multimedia Tools and Applications (MTAP), Springer (in press).
[Paper] [Project] [Springer]

Illustration2Vec: A Semantic Vector Representation of Illustrations
Masaki Saito and Yusuke Matsui
ACM SIGGRAPH Asia, Technical Brief, 2015.
[Paper] [Project] [Pre-trained model] [ACM]

Separation of Manga Line Drawing and Screentones
Kota Ito, Yusuke Matsui , Toshihiko Yamasaki, and Kiyoharu Aizawa
Eurographics, Short paper, 2015.
[Code]

Reference-based Manga Colorization by Graph Correspondence Using Quadratic Programming
Kazuhiro Sato, Yusuke Matsui , Toshihiko Yamasaki, and Kiyoharu Aizawa
ACM SIGGRAPH Asia, Technical Briefs, 2014.
[Paper] [ACM]

Codes

  • manga109api: Simple python API to read annotation data of Manga109
  • Rii: Fast and memory-efficient ANN with a subset-search functionality
  • nanopq: Pure python implementation of product quantization for nearest neighbor search
  • PQk-means: Fast and memory-efficient clustering
  • faiss_tips: Some useful tips for faiss
  • PQTable: Fast search algorithm for product-quantized codes via hash-tables
  • Sis: Simple image search engine

Slides

billion-scaleの近似最近傍探索
松井勇佑
招待講演資料, サイバーエージェントおよびDeNA, 2018.
  • 【ツリー系(FLANN, Annoy)】について
  • 【LSH(FALCONN)】について
  • 【グラフ探索系(HNSW)】について
  • 【直積量子化(Faiss)】の概略について
  • コード:nanopq
  • コード:faiss_tips

ショードコードによる大規模近似最近傍探索
松井勇佑
講義資料, 大阪大学, 2016.
  • 【バイナリハッシング】について
  • 【直積量子化】について
  • コード:nanopq

【招待ショートサーベイ】直積量子化を用いた近似最近傍探索
松井勇佑
パターン認識・メディア理解研究会(PRMU), 2016.
  • 【直積量子化】について
  • コード:nanopq

Publications

Journal

  • A Survey of Product Quantization
    Yusuke Matsui, Yusuke Uchida, Hervé Jégou, Shin'ichi Satoh
    ITE Transactions on Media Technology and Applications (ITE), 2018. [pdf]
  • PQTable: Nonexhaustive Fast Search for Product-Quantized Codes Using Hash Tables
    Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    IEEE Transactions on Multimedia (TMM), 2018. [pdf] [project]
  • Sketch-based Manga Retrieval using Manga109 Dataset
    Yusuke Matsui, Kota Ito, Yuji Aramaki, Azuma Fujimoto, Toru Ogawa, Toshihiko Yamasaki, Kiyoharu Aizawa
    Multimedia Tools and Applications (MTAP), 2017. [pdf] [project]
  • DrawFromDrawings: 2D Drawing Assistance via Stroke Interpolation with a Sketch Database
    Yusuke Matsui, Takaaki Shiratori, Kiyoharu Aizawa
    IEEE Transactions on Visualization and Computer Graphics (TVCG), 2017. [pdf] [project]

Conference

  • Revisiting Column-Wise Vector Quantization for Memory-Efficient Matrix Multiplication
    Yusuke Matsui, Shin'ichi Satoh
    IEEE International Conference on Image Processing (ICIP), 2018.
  • Reconfigurable Inverted Index
    Yusuke Matsui, Ryota Hinami, Shin'ichi Satoh
    ACM International Conference on Multimedia (ACMMM), 2018. [project]
  • Deep Image Retrieval Applied on Kotenseki Ancient Japanese Literature
    Chairath Sirirattanapol, Yusuke Matsui, Shin'ichi Satoh, Kuninori Matsuda, Kazuaki Yamamoto
    IEEE International Workshop on Machine Learning and Computing for Visual Semantic Analysis (MLCSA), 2017. [project]
  • Energy based Fast Event Retrieval in Video with Temporal Match Kernel
    Junfu Pu, Yusuke Matsui, Fan Yang, Shin'ichi Satoh
    IEEE International Conference on Image Processing (ICIP), 2017.
  • PQk-means: Billion-scale Clustering for Product-quantized Codes
    Yusuke Matsui, Keisuke Ogaki, Toshihiko Yamasaki, Kiyoharu Aizawa
    ACM International Conference on Multimedia (ACMMM), 2017. [pdf] [project]
  • Region-based Image Retrieval Revisited
    Ryota Hinami, Yusuke Matsui, Shin'ichi Satoh
    ACM International Conference on Multimedia (ACMMM), 2017. [pdf]
  • Manga109 Dataset and Creation of Metadata
    Azuma Fujimoto, Toru Ogawa, Kazuyoshi Yamamoto, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    International Workshop on Comics Analysis, Processing and Understanding (MANPU), 2016.
  • Interactive Region Segmentation for Manga
    Kota Ito, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    International Conference on Pattern Recognition (ICPR), 2016.
  • Sketch Simplification by Classifying Strokes
    Toru Ogawa, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    International Conference on Pattern Recognition (ICPR), 2016.
  • Text Detection in Manga by Combining Connected-omponent-based and Region-based Classifications
    Yuji Aramaki, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    IEEE International Conference on Image Processing (ICIP), 2016.
  • PQTable: Fast Exact Asymmetric Distance Neighbor Search for Product Quantization using Hash Tables
    Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    IEEE International Conference on Computer Vision (ICCV), 2015. [pdf] [project]
  • Separation of Manga Line Drawings and Screentones
    Kota Ito, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    Eurographics, Short Paper, 2015. [project]
  • Interactive Segmentation for Manga using Lossless Thinning and Coarse Labeling
    Yuji Aramaki, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    APSIPA ASC, 2015.
  • Illustration2vec: A Semantic Vector Representation of Illustrations
    Masaki Saito, Yusuke Matsui
    ACM SIGGRAPH Asia, Technical Brief, 2015. [project]
  • Challenge for Manga Processing: Sketch-based Manga Retrieval
    Yusuke Matsui
    ACM International Conference on Multimedia (ACMMM), Doctoral Symposium, 2015. [project]
  • Sketch2Manga: Sketch-based Manga Retrieval
    Yusuke Matsui, Kiyoharu Aizawa, Yushi Jing
    IEEE International Conference on Image Processing (ICIP), 2014. [pdf] [project]
  • Diffusion : Change an Ambience of the Space with a Small Amount of Ink
    Saemi Choi, Yusuke Matsui, Kiyoharu Aizawa
    ACM SIGGRAPH Asia, Poster, 2014.
  • Reference based Manga Colorization by Graph Correspondence Using Quadratic Programming
    Kazuhiro Sato, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    ACM SIGGRAPH Asia, Technical Brief, 2014. [pdf]
  • IllustStyleMap: Visualization of Illustrations based on Similarity of Drawing Style of Authors
    Daiki Matsumoto, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa, Takanori Katagiri
    ACM SIGGRAPH, Poster, 2014.
  • Interactive Segmentation for Manga
    Yuji Aramaki, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    ACM SIGGRAPH, Poster, 2014.
  • Interactive Manga Retargeting
    Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    ACM SIGGRAPH, Poster, 2011. [pdf]

Technical report

  • Object Detection for Comics using Manga109 Annotations
    Toru Ogawa, Atsushi Otsubo, Rei Narita, Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    CoRR, 2018. [project]
  • Unsupervised Adversarial Learning of 3D Human Pose from 2D Joint Locations
    Yasunori Kudo, Keisuke Ogaki, Yusuke Matsui, Yuri Odagiri
    CoRR, 2018. [project]
  • PQk-means: Billion-scale Clustering for Product-quantized Codes
    Yusuke Matsui, Keisuke Ogaki, Kiyoharu Aizawa, Toshihiko Yamasaki
    CoRR, 2017. [project]
  • PQTable: Non-exhaustive Fast Search for Product-quantized Codes using Hash Tables
    Yusuke Matsui, Toshihiko Yamasaki, Kiyoharu Aizawa
    CoRR, 2017. [project]
  • Sketch-based Manga Retrieval using Manga109 Dataset
    Yusuke Matsui, Kota Ito, Yuji Aramaki, Toshihiko Yamasaki, Kiyoharu Aizawa
    CoRR, 2015. [project]

Awards

  • 船井研究奨励賞, 船井情報科学振興財団, 2018. [link]
  • Student Best Paper Award, 第10回 IEEE Signal Processing Society Japan, 2016.
  • 研究会推薦博士論文速報, IPSJ, 2016.
  • 研究科長賞, 東京大学大学院情報理工学系研究科, 2016.
  • 学生論文賞, IMPS, 2015.
  • 最優秀インタラクティブ発表賞, HCGシンポジウム, 2014.
  • 学府長賞, 東京大学大学院学際情報学府, 2013.
  • 原島博学術奨励賞, 電気・電子情報学術振興財団, 2012.
  • 工学部長賞, 東京大学工学部, 2011.
  • 7th place, Google Landmark Recognition Challenge, 2018. [link]
  • 7th place, Google Landmark Retrieval Challenge, 2018. [link]
  • Outstanding Reviewer Award, ACM ICMR, 2018. [link]
  • ACM Student Research Competition 3rd Place, ACM SIGGRAPH, 2011.

Funding sources

  • ACT-I 加速フェーズ, 科学技術振興機構. (2018/4 - 2020/3)
  • 共同研究, 株式会社ドワンゴ. (2016/12 - 2017/3)
  • ACT-I, 科学技術振興機構. (2016/12 - 2018/3)
  • 研究活動スタート支援, 日本学術振興会. (2016/10 - 2018/3)
  • 特別研究員(DC1), 日本学術振興会. (2013/4 - 2016/5)

Invited talks

  • billion-scaleの近似最近傍探索, DeNA, 2018.
  • billion-scaleの近似最近傍探索, サーバーエージェント, 2018.
  • 1時間で画像検索エンジンを作る, コンピュータグラフィックスとビジュアル情報学研究会, 2017.
  • ショートコードによる近似最近傍探索, 大阪大学 講義, 2016.
  • 漫画に対する画像処理, 第15回情報科学技術フォーラム(FIT) ,助教が吼える! 各界の若手研究者大集合, 2016.
  • PQTable: Fast Exact Asymmetric Distance Neighbor Search for Product Quantization using Hash Tables, 第19回画像の認識・理解シンポジウム(MIRU), 招待講演, 2016.
  • PQTable: ハッシュテーブルを用いたプロダクト量子化ベクトルの高速探索, 第20回 Image Processing Tokyo, 2016.
  • Sketch-based Manga Retrieval, Handai IDS-JFLI joint Workshop on Media and Graphics, 2018.
  • Sketch-based Image Retrieval And Data-Driven Drawing Assistance, Vietnam-Japan Scientific Exchange Meeting (VJSE), 2017.

Book

  • コンピュータビジョン―広がる要素技術と応用―
    米谷竜, 斎藤英雄, 池畑諭, 牛久祥孝, 内山英昭, 内海ゆづ子, 小野峻佑, 片岡裕雄, 金崎朝子, 川西康友, 齋藤真樹, 櫻田健, 高橋康輔, 松井勇佑
    共立出版, 2018. [link]

Link