Skip to main content

Misc.

FixRes: Fixing the train-test resolution discrepancy​

  • It is desirable to use a lower training resolution and fine-tune the network at the larger resolution
  • This speeds up the full training and improves the accuracy under prevailing data augmentation schemes

Train-Test Resolution Discrepancy​

While the crops extracted at training and test time have the same size, they arise from different RoCs, which skews the distribution of data seen by the CNN

Train​

  • The training practice is to extract a rectangle with random coordinates from the image, which artificially increases the amount of training data.
  • This region is called the Region of Classification (RoC)

Test​

  • The RoC is instead set to a square covering the central part of the image, which results in the extraction of a so called "center crop"
  • This reflects the bias of photographers who tend to center important visual content

Strategy​

fixres

Determining the optimal sample size​

  • TODO