Counterfactual Zero-Shot and Open-Set Visual Recognition

This project provides implementations for our CVPR 2021 paper Counterfactual Zero-Shot and Open-Set Visual Recognition, where we propose a counterfactual-based binary seen/unseen classifier (GCM-CF) for Zero-Shot Learning (ZSL) and Open-Set Recognition (OSR). This repo contains

  • ZSL: Strong binary seen/unseen classifier that is plug-and-play with any ZSL method
  • ZSL: Integrations with TF-VAEGAN, RelationNet, GDAN, CADA-VAE, LisGAN, AREN
  • OSR: Complete the OSR code base on MNIST, SVHN, CIFAR10, CIFAR+10, CIFAR+50 with 5 fixed random seed
  • OSR: Strong baseline (F1 score) of Softmax, OpenMax, CGDL
  • OSR: Implementation of our GCM-CF

For technical details, please refer to:

Counterfactual Zero-Shot and Open-Set Visual Recognition
Zhongqi Yue*, Tan Wang*, Hanwang Zhang, Qianru Sun, Xian-Sheng Hua
* Equal contribution
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021


Please refer to the in ZSL and OSR folder, respectively.



If you find our work or the code useful, please consider cite our paper using:

  title={Counterfactual Zero-Shot and Open-Set Visual Recognition},
  author={Yue, Zhongqi and Wang, Tan and Zhang, Hanwang and Sun, Qianru and Hua, Xian-Sheng},
  booktitle= {CVPR},