Skip to main content

Table 1 Analysis on hierarchical features

From: Sparse self-attention aggregation networks for neural sequence slice interpolation

Extractor cremi_triplet A cremi_triplet B cremi_triplet C mouse_triplet
  PSNR SSIM PSNR SSIM PSNR SSIM IE PSNR SSIM IE
U-Net [40] 18.10 0.4295 16.75 0.3549 16.26 0.3542 28.82 14.59 0.2219 36.55
SU-Net 18.12 0.4352 16.71 0.3417 16.18 0.3280 29.71 14.75 0.2089 35.99
RDN [32] 17.73 0.4448 16.59 0.3521 16.44 0.3451 28.75 14.85 0.2195 35.53
SRDN(ours) 18.26 0.4374 16.79 0.3712 16.46 0.3575 28.38 15.04 0.2156 34.81
  1. We compared the actual effects of different feature extractors on the cremi_triplet datasets and mouse_triplet dataset. Lower IEs indicates better performance
\