This repository is the PyTorch implementation for the paper Out-of-Distribution Detection for Generalized Zero-Shot Action Recognition, published at CVPR 2019. This repo is based on the codebase of f-CLSWGAN, which was originally proposed for zero-shot object classification.
- PyTorch 0.3.1
- Python 3.5
The I3D features (concatenated RGB and Flow) for HMDB51 and UCF101 are provided here. Place the unzipped sub-folders under data_action folder before running the experiments.
Shell scripts for running the baseline ZSL/GZSL/GZSL-OD experiments on HMDB51 and UCF101 datasets are provided.
Please cite the following work if you use this package.
@InProceedings{Mandal_2019_CVPR,
author = {Mandal, Devraj and Narayan, Sanath and Dwivedi, Sai Kumar and Gupta, Vikram and Ahmed, Shuaib and Khan, Fahad Shahbaz and Shao, Ling},
title = {Out-Of-Distribution Detection for Generalized Zero-Shot Action Recognition},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}