New Ordinal Regression For Severity Grading In Fundus Images
Takafumi Sakura, Yusuke Kondo, Hidenori Takahashi
-
Members: FreeSPS
IEEE Members: $11.00
Non-members: $15.00Length: 00:04:15
28 Mar 2022
Ordinal Regression (OR) aims to solve classification problems in which the categories follow an ordering relation. This study empirically investigates the effects of state-of-the-art (SOTA) OR methods on severity grading. Besides, to address the lack of data of ordinal annotation, we combined OR and weakly supervised learning to reduce the mean absolute error (MAE).
