Glaucoma data has been a difficult task due to the presence of hard samples and data imbalance. They proposed a curriculum-learning EGDCL to learn it through a network that generates sample difficulty, label rarity, and feature map importance. Results beyond previous research.
TL;DR
Glaucoma data has been a difficult task due to the presence of hard samples and data imbalance. They proposed a curriculum-learning EGDCL to learn it through a network that generates sample difficulty, label rarity, and feature map importance. Results beyond previous research.
Why it matters:
Paper URL
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3685_ECCV_2020_paper.php
Submission Dates(yyyy/mm/dd)
Authors and institutions
Rongchang Zhao, Xuanlin Chen, Zailiang Chen, Shuo Li
Methods
Results
Comments