This repository contains code for our baselines, namely HCRN, ClipBERT, and All-in-one, which is migrated from their original implementations to fit our data structure.

see dataset for details.

see the HCRN, ClipBERT, and All-in-one folders for details.

The above baseline models are trained on the train set and evaluated on the val,test-dev and test sets, respectively.

This project is licensed under the Apache License 2.0.

If you use ANetQA in your research, we appreciate it if you cite our paper in the following.

@inproceedings{yu2023anetqa,
title={ANetQA: A Large-scale Benchmark for Fine-grained Compositional Reasoning over Untrimmed Videos},
   author={Yu, Zhou and Zheng, Lixiang and Zhao, Zhou and Wu, Fei and Fan, Jianping and Ren, Kui and Yu, Jun},
   booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
   year={2023}
}