Last Update: 2017/04/05 14:06:49

Self Introduction
Keisuke DOMAN
Ph.D. (Information Science)
Lecturer at Department of Information Engineering, School of Engineering, Chukyo University
Visiting Scholar at Graduate School of Info. Science, Nagoya University

Table of Contents

Contact

Department of Information Engineering, School of Engineering, Chukyo University

Top

Biography

Academic Background

Employment History

Top

Research

Object detection and recognition with an in-vehicle camera

In this research, as an element technique in DAS (Driver Assistance System) for automobiles, we propose a method for constructing a traffic sign detector using a generative learning method. Our generative learning method defines color change models (discoloration and lighting change), shape change models (rotation, shifting, and stretching), and texture change models (optical blurring, background change). By generating a lot of artificial traffic sign images from a few original images, we can efficiently construct an rapid and accurate appearance-based classifier, such as a classifier cascade proposed by Viola and Jones.
Examples of sign images generated by the proposed method

Visibility estimation of visual objects with an in-vehicle camera

In this research, as an element technique in DAS (Driver Assistance System) for automobiles, we propose a method for quantifying traffic sign visibility. The proposed method extract several kinds of image features that are contrasts of luminance, color, complexity between the target and its surrounding region, and appearances in luminance and color of the target. By integrating these image features based on the knowledge on cognitive science, the proposed method accurately estimates traffic sign visibility.
Examples of visibility values estimated by the proposed method

Analysis of cooking videos

In this research, as an element technique for cooking activities, we propose a method for multimediatization of cooking recipes. Multimediatization of cooking recipse enables visual explanations of cooking motions which are not provided with conventional cooking recipes. The proposed method automatically constructs a cooking video database by applying a text and image processing to a recipe text, closed-captions, and a cooking video in a cook show. Once a cooking video database is constructed, we can multimediatize an arbitrary cooking recipe on the Web.
Multimedia cooking recipe

Research Achievements

See here
Top

Academic & Social Activities

Academic Society

Society Committee

Editorial Committee

Conference Committee

Social Contribution

Other Activities

Top

Educational Activities

Classes

Department of Information System Technology, School of Information Science & Techinology, Chukyo University

Information Science Major, Graduate School of Sociology, Chukyo University

Department of Information Engineering, School of Engineering, Chukyo University

Computer Science Major, Graduate School of Engineering, Chukyo University

Extracurricular Activities

Top

Links

Top