gsx0 / Attentive-Region-Embedding-Network-for-Zero-shot-Learning

Attentive Region Embedding Network for Zero-shot Learning (CVPR 2019)
58 stars 6 forks source link

your accuracy have some problem #2

Open ghost opened 4 years ago

ghost commented 4 years ago

Hi
image

This object(class) is not same as table2 in your paper. Your format in table2 is ts, ta and H, however your code only have mean class accuracy. How to modify your code to show ts, ta and H?

qq230091457 commented 4 years ago

This code only shows the recognition of unknown objects with known objects. This precision has no meaning.