RESEARCH Research Contents

Manipulating Various Data to Create Cutting-Edge Technologies for the Future

In the Laboratory of Media Dynamics, we continue to create “knowledge”, which is the source of innovation, to solve social issues by analyzing data from various aspects. We are attempting to embody a super-smart society (Society 5.0) in which AI and humans collaborate, not only by proposing original new technologies that open up frontiers but also by aiming for their social implementation.

Research Contents

Advanced Image Video Recognition

  • Research on domain adaptation based on art style transformation
  • Research on representation learning for obtaining highly interpretable feature expressions

Vision Language Integration

  • Research on image/video retrieval from text based on image generation methods
  • Research on image/video retrieval based on user interaction
  • Research on image/video editing based on text

Web/SNS analysis

  • Research on cross-domain recommendation for e-commerce services
  • Research on generating sports highlight videos from SNS and broadcast video
  • Research on music recommendation systems based on reinforcement learning for music distribution services

Tacit Knowledge Understanding

  • Research on estimating the gazing area of images and videos
  • Research on interest estimation using skeletal information during video viewing