lzrobots / LearningToCompare_ZSL

PyTorch code for CVPR 2018 paper: Learning to Compare: Relation Network for Few-Shot Learning (Zero-Shot Learning part)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

about test part

wang1104014663 opened this issue · comments

Thank you for sharing the code. However, in the part of calculating the accuracy rate, the accuracy of each class should be used, and then the category average should be obtained instead of the accuracy of all images.This compares to the accuracy data you quoted.

@wang1104014663 Thanks for pointing out. It seems we released the old evaluation protocol version -- "top1 over all samples", as used in table 3. We revise it to "averaged per class accuracy" now. Thanks again.