On Aliased Resizing and Surprising Subtleties in GAN Evaluation. (arXiv:2104.11222v2 [cs.CV] UPDATED)

Metrics for evaluating generative models aim to measure the discrepancy
between real and generated images. The often-used Frechet Inception Distance
(FID) metric, for example, extracts "high-level" features using a deep network
from the two sets. However, we find that the differences in "low-level"
preprocessing, specifically image resizing and compression, can induce large
variations and have unforeseen consequences. For instance, when resizing an
image, e.g., with a bilinear or bicubic kernel, signal processing principles
mandate adjusting prefilter width depending on the …

arxiv cv gan