By replacing U-Net components with trainable versions of guided and Frangi filters, the authors achieve comparable performance on retinal vessel segmentation with 10x fewer parameters and direct mapping to signal processing tools.
IEEE Trans Med Imaging 23(4), 501–509 (2004)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Divide-and-Conquer Approach towards Understanding Deep Networks
By replacing U-Net components with trainable versions of guided and Frangi filters, the authors achieve comparable performance on retinal vessel segmentation with 10x fewer parameters and direct mapping to signal processing tools.