This code is for our CVPR 19 paper: Attentive Region Embedding Network for Zero-shot Learning
-
We provide a trained model (https://drive.google.com/file/d/1EIuOPi0MjdKUD7mtKlFkbaMF7tCsYedV/view?usp=sharing) of ARE on CUB:PS setting for ZSL. The default parameters in main.py will achieve 72.7% ZSL result on CUB.
-
Contact: gsxiehm@gmail.com