🌟 Official Access to the THQA database in "THQA: A PERCEPTUAL QUALITY ASSESSMENT DATABASE FOR TALKING HEADS" 🌟
| Research Gate | Arxiv |
The THQA database is a quality assessment database designed specifically for the person independent driving methods. It compares 8 popular speech-driven methods and contains a total of 800 talking head videos. If there are any questions or suggestions for improvement against the database, please feel free to contact us by zyj2000@sjtu.edu.cn.
- Open Platform to validate your talking head driving methods subjectively.
- Open Image and Audio (code = thqa) Resources are provided which can be employed to driving heads.
- More objective methods will be provided soon!!!
- More Function is going to be explored.
- If you have more upadated talking head results, welcome to share them to us by email or PR.
- We will carry out more subjective experiments on your results and update work.
- Thank for your support. Hope this work can promote each other!!!
[10.12.2024] The poster for ICIP2024 THQA has been released.
[9.25.2024] A derivative work ReLI-QA from the THQA dataset is accepted by VCIP2024.
[7.17.2024] A further research THQA-3D is accepted by ACM MM2024.
[6.14.2024] Release the original audios in Baidu YunPan.
[6.07.2024] Our paper is accepted by ICIP2024.
[4.22.2024] Poster our paper in arxiv.
[4.13.2024] Release the original images.
[3.01.2024] Poster our paper in Reseach Gate.
The database is nearly 732MB, and 2 links are provided here.
- Baidu Yun Pan (thqa)
- Onedrive
@article{zhou2024thqa,
title={THQA: A Perceptual Quality Assessment Database for Talking Heads},
author={Zhou, Yingjie and Zhang, Zicheng and Sun, Wei and Liu, Xiaohong and Min, Xiongkuo and Wang, Zhihua and Zhang, Xiao-Ping and Zhai, Guangtao},
journal={arXiv preprint arXiv:2404.09003},
year={2024}
}