Institute of Post-LED Photonics, Tokushima University

SHISHIBORI Masami​ SHISHIBORI Masami​
Development of multimedia systems using AI technologySHISHIBORI Masami​[Professor​]

People

Development of multimedia systems using AI technology

SHISHIBORI Masami​[Professor​]

Division of Next-generation Photonics​ (Joint Faculty)

1997. Doctor degree from the Faculty of Engineering, Tokushima University.​
2014. Professor at the Graduate School of Technology, Industrial and Social Sciences, the University of Tokushima.​
Concurrently served to our institute in March 2019.

  • Medical Photonics
  • Visible
  • Infrared
  • Terahertz
  • Deep ultraviolet
  • Information Technology
  • Medical
  • Inspection
  • Light source / Sensing
  • etc.
  • Deep learning
  • Image classification
  • Object detection
  • ​ Data mining
  • Virtual Reality (VR)
Research Interests

With the recent development of AI technology, AI systems which are equivalent to human cognitive abilities have been developed. In particular, CNN (Convolutional Neural Network), which is one of deep learning models for image data, has a high accuracy classification ability of over 90% for images with large differences in their features​
Our research group has been developed a AI cytology system which can detect cancer cells from Whole Slide Images (WSI) of the urothelial epithelium using the AI technologies. This system can visualize the basis for CNN decisions. It is possible to reduce the screening work by using this system.​
    We have also developed a system which can detect insects in the food by analyzing images of fresh food taken with the special equipment. By applying special preprocessing, this system can detect minute shadows of insects which are difficult for human to see. It is possible to improve the safety of food by using this system.​
    We have also developed a musical instrument performance support system and a medical equipment operation support system using cognitive interaction technologies such as VR/AR.​
We are researching technologies that enable users to more easily access desired information from large amounts of data.​