M2の岸田俊文君と森 健太郎君が、The International Conference of Machine Learning and Cybernetics (ICMLC) 2017 にて、発表を行いました。

M2の岸田俊文君と森 健太郎君が、2017年7月9日(月) – 12(水) に中国の寧波市で開催されたThe International Conference of Machine Learning and Cybernetics (ICMLC) 2017 にて、以下の口頭発表を行いました。

 発表者 Toshifumi Kishida, Mizuki Higuchi, Tadahito Egawa, Kazuhiko Taniguchi, and Yutaka Hata
 題 Electric Wire Inspection by Discriminant Analysis Method
 概 This paper described a method of automatic extraction of abnormal points from electric cable. The cable changes the color depending on the weather, environment and used period. We define low brightness area and classify size of it. We divide them into abnormal and normal with discriminant analysis method. As the result, we can detect all arc-marks for no rust cables. On the other hand, we can detect almost all arc-mark and much rust for rust cables. We examined that discriminant analysis method successfully detected abnormal points automatically.
 発表者 Kentaro Mori, Atsushi Yukawa, Atsushi Kono, and Yutaka Hata
 題 Heart Failure Diagnosis for Tagged Magnetic Resonance Images
 概 This paper proposes an automated method to track the tag intersections in the Tagged Magnetic Resonance Image (MRI) to diagnose Heart failure. Heart Failure brings about decline of a function to supply blood by contraction of myocardium. Physician always analyzes myocardial motion using the images. The proposed method employs Mean shift belief propagation (MSBP) that integrates all information about image to track tag intersections. This method enables us to extract tag intersections with disappearing tags over time. The experimental results show a clinical ability of tracking of tag intersections.