Shin'ichi Satoh's Lab

Welcome to Shin'ichi Satoh's Lab homepage at NII!

Research in our lab focuses on multimedia understanding and knowledge discovery. Especially, we aim to create an intelligent computer system which can see and understand the visual world.

We accept graduate students from Department of Information and Communication Engineering, Graduate School of Information Science and Technology, the University of Tokyo. Our lab is in National Institute of Informatics, Japan.


News

  • [2017.7.18] One paper was accepted to ICCV 2017!
  • [2017.7.10] We had a TMM paper invited for presentation, and another paper accepted at ICME 2017!
  • [2017.7.3] Three full papers were accepted to ACM Multimedia 2017!
  • [2017.7.3] We will present two invited talks, five poster presentations, and a demo at MIRU 2017.
  • [2017.6.16] Dr. Uchida (our former PhD student) and Prof. Satoh received the best paper award at ICMR 2017! [paper]

Research projects

Large-scale fast object detection

We extended R-CNN to larger scale, which enables immediate and accurate object category detection from a large image databas. R. Hinami and S. Satoh, "Large-scale R-CNN with Classifier Adaptive Quantization", ECCV 2016

Multimedia Analytics

Explore, analyze, and visualize archives of multimedia content by bringing together data science and computer vision for the support of real world applications such as social sciences, media studies, and even marketing.

Temporal Matching Kernel with Explicit Feature Maps for Video Event Retrieval

We propose a new video representation for video event retrieval. Given a video query, the method is able to efficiently retrieve similar video events or near-duplicates along with a precise temporal alignment. ``Temporal matching kernel with explicit feature maps,'' ACM Multimedia 2015.

Video Event Detection by Exploiting Word Dependencies

We exploited word dependencies as a new semantic video representation for recognizing complex events S. Phan, Y. Miyao, D-D Le and S Satoh, "Video Event Detection by Exploiting Word Dependencies from Image Captions", COLING 2016