• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Deep State Space Modeling Methods for Video Understanding

Research Project

Project/Area Number 19K12039
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 61010:Perceptual information processing-related
Research InstitutionChiba University

Principal Investigator

Kawamoto Kazuhiko  千葉大学, 大学院工学研究院, 教授 (30345376)

Project Period (FY) 2019-04-01 – 2023-03-31
Project Status Completed (Fiscal Year 2022)
Budget Amount *help
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2021: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2020: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2019: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Keywords状態空間モデル / 深層学習 / 解きほぐし / 状態空間表現 / 動画生成 / 行動認識 / 動画像理解 / 深層マルコフモデル / 一人称行動認識
Outline of Research at the Start

本研究では,深層学習モデルと状態空間モデルを統合し,行動認識,複数人物追跡,あるいは動画生成といったコンピュータビジョン分野における動画像理解タスクへ応用展開する.深層学習(ディープラーニング)は,画像認識等の知的情報処理を実現するための強力な方法である.一方,状態空間モデルは,時系列解析やシステム制御・同定に広く利用されている.これら2つのモデルを統合することにより,知的情報処理を含む時系列解析法を発展させることができる.

Outline of Final Research Achievements

This study tackled video understanding based on integrating deep learning and state-space models. First, we introduced a deep Markov model for predicting chaotic dynamics. Next, we extend the deep Markov model to a 2D convolutional neural Markov model that handles both time series and spatial data. Furthermore, we developed deep models for video generation and action recognition. Then, we worked on building a deep model that enables control of video generation and developed zero-shot image generation. Furthermore, we developed a sequential variational autoencoder that separates static and dynamic features in video images. These studies demonstrated the effectiveness of our approach.

Academic Significance and Societal Importance of the Research Achievements

深層学習モデルと状態空間モデルの統合により、コンピュータビジョンにおける動画像理解タスクを適切にモデル化でき、行動認識、人物追跡、動画生成といったタスクがより精度高く、効率的に行えるようになる。これは、監視システム、自動運転車、ロボティクスなどの分野に貢献できる。また、動画生成技術は、エンターテイメントや広告への応用も期待できる。

Report

(5 results)
  • 2022 Annual Research Report   Final Research Report ( PDF )
  • 2021 Research-status Report
  • 2020 Research-status Report
  • 2019 Research-status Report
  • Research Products

    (24 results)

All 2023 2022 2021 2020 2019

All Journal Article (9 results) (of which Peer Reviewed: 9 results,  Open Access: 4 results) Presentation (15 results) (of which Int'l Joint Research: 2 results)

  • [Journal Article] Sequential Variational Autoencoder with Adversarial Classifier for Video Disentanglement2023

    • Author(s)
      Haga Takeshi、Kera Hiroshi、Kawamoto Kazuhiko
    • Journal Title

      Sensors

      Volume: 23 Issue: 5 Pages: 2515-2515

    • DOI

      10.3390/s23052515

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Zero-Shot Action Recognition with Three-Stream Graph Convolutional Networks2021

    • Author(s)
      Wu Nan、Kawamoto Kazuhiko
    • Journal Title

      Sensors

      Volume: 21 Issue: 11 Pages: 3793-3793

    • DOI

      10.3390/s21113793

    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Audio-Visual Model for Generating Eating Sounds Using Food ASMR Videos2021

    • Author(s)
      Uchiyama Kodai and Kawamoto Kazuhiko
    • Journal Title

      IEEE Access

      Volume: 9 Pages: 50106-50111

    • DOI

      10.1109/access.2021.3069267

    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Animating Cloud Images With Flow Style Transfer2021

    • Author(s)
      Kurisaki Kazuma and Kawamoto Kazuhiko
    • Journal Title

      IEEE Access

      Volume: 9 Pages: 3269-3277

    • DOI

      10.1109/access.2020.3048160

    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Three-Stream Graph Convolutional Networks for Zero-Shot Action Recognition2020

    • Author(s)
      Wu Nan and Kawamoto Kazuhiko
    • Journal Title

      Proc. of SCIS & ISIS

      Volume: 1 Pages: 1-5

    • DOI

      10.1109/scisisis50064.2020.9322783

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Reinforcement Learning with Randomized Physical Parameters for Fault-Tolerant Robots2020

    • Author(s)
      Okamoto Wataru and Kawamoto Kazuhiko
    • Journal Title

      Proc. of SCIS & ISIS

      Volume: 1 Pages: 1-5

    • DOI

      10.1109/scisisis50064.2020.9322775

    • NAID

      130007857117

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Object-Action Interaction Region Detection in Egocentric Videos2020

    • Author(s)
      Shinobu Takahashi and Kazuhiko Kawamoto
    • Journal Title

      Proc. of ISCIIA

      Volume: 1 Pages: 1-5

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Deep Markov Models for Data Assimilation in Chaotic Dynamical Systems2020

    • Author(s)
      Halim Calvin Janitra、Kawamoto Kazuhiko
    • Journal Title

      Advances in Artificial Intelligence

      Volume: - Pages: 37-44

    • DOI

      10.1007/978-3-030-39878-1_4

    • NAID

      130007658483

    • ISBN
      9783030398774, 9783030398781
    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Journal Article] DCVGAN: Depth Conditional Video Generation2019

    • Author(s)
      Nakahira Yuki、Kawamoto Kazuhiko
    • Journal Title

      IEEE International Conference on Image Processing

      Volume: - Pages: 749-753

    • DOI

      10.1109/icip.2019.8803764

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Presentation] 動画像の解きほぐしに向けた敵対的補助分類器の効果検証2022

    • Author(s)
      芳賀壮、計良宥志、川本一彦
    • Organizer
      電子情報通信学会パターン認識・メディア理解研究会
    • Related Report
      2022 Annual Research Report
  • [Presentation] 動画像の解きほぐしに向けた補助分類器の効果検証2022

    • Author(s)
      芳賀壮、計良宥志、川本一彦
    • Organizer
      第25回画像の認識・理解シンポジウム
    • Related Report
      2022 Annual Research Report
  • [Presentation] Adversarial Bone Length Attack on Action Recognition2022

    • Author(s)
      Nariki Tanaka, Hiroshi Kera, and Kazuhiko Kawamoto
    • Organizer
      Association for the Advancement of Artificial Intelligence
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] Conditional Motion and Content Decomposed GAN for Zero-Short Video Generation2021

    • Author(s)
      Shun Kimura and Kazuhiko Kawamoto
    • Organizer
      the 7th International Workshop on Advanced Computational Intelligence and Intelligent Informatics
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 変分オートエンコーダを用いた動画像の解きほぐし2021

    • Author(s)
      芳賀壮, 計良宥志, 川本一彦
    • Organizer
      第24回画像の認識・理解シンポジウム
    • Related Report
      2021 Research-status Report
  • [Presentation] メイン適応を用いた一人称行動認識2021

    • Author(s)
      木内拓実, 計良宥志, 川本一彦
    • Organizer
      第24回画像の認識・理解シンポジウム
    • Related Report
      2021 Research-status Report
  • [Presentation] Food ASMR動画を用いたマルチモーダル深層学習による食感音の生成2020

    • Author(s)
      内山光大,川本一彦
    • Organizer
      第23回画像の認識・理解シンポジウム
    • Related Report
      2020 Research-status Report
  • [Presentation] 一人称視点での物体と動作のインタラクション領域検出2020

    • Author(s)
      高橋忍,川本一彦
    • Organizer
      第23回画像の認識・理解シンポジウム
    • Related Report
      2020 Research-status Report
  • [Presentation] ゼロショット動画生成のための条件付きMoCoGAN2020

    • Author(s)
      木村駿,川本一彦
    • Organizer
      第34回人工知能学会全国大会
    • Related Report
      2020 Research-status Report
  • [Presentation] 状態遷移差分の学習による耐故障ロボットのための強化学習2020

    • Author(s)
      大里虹平,川本一彦
    • Organizer
      第34回人工知能学会全国大会
    • Related Report
      2020 Research-status Report
  • [Presentation] Shared 2D-convolutions netを用いた深層マルチタスク学習による一人称行動認識2019

    • Author(s)
      小林 春平, 川本 一彦
    • Organizer
      第22回画像の認識・理解シンポジウム講演
    • Related Report
      2019 Research-status Report
  • [Presentation] デプスからカラーへのドメイン変換を用いたGANによる動画生成2019

    • Author(s)
      中平有樹,川本一彦
    • Organizer
      計測自動制御学会システム・情報部門学術講演会
    • Related Report
      2019 Research-status Report
  • [Presentation] Residual Networksに対する確率的正則化の提案: Shake-ResDropとShake-SENet2019

    • Author(s)
      白濱 淳也、川本 一彦
    • Organizer
      第33回人工知能学会全国大会
    • Related Report
      2019 Research-status Report
  • [Presentation] 模倣学習を用いた動画からの動作獲得2019

    • Author(s)
      大里虹平, 川本一彦
    • Organizer
      情報処理学会研究報告
    • Related Report
      2019 Research-status Report
  • [Presentation] ダイナミクススタイル変換を用いた自然風景画像の動画化2019

    • Author(s)
      栗崎一真,川本一彦
    • Organizer
      情報処理学会研究報告
    • Related Report
      2019 Research-status Report

URL: 

Published: 2019-04-18   Modified: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi