the source code is coming soon before 2022.1.1.
DSTS-Net code for the following papers:
###English simplified version
Yinhao Liu, Xiaofei Zhou, Haibing Yin*,and so on.
Effificient Video Quality Assessment with Deeper Spatiotemporal Feature Extraction and Integration. Journal of Electronic imaging,SPIE,2021.
###Chinese extend version Yinhao Liu, Wei Zhang and so on.
基于高阶深层时空信息的自媒体视频质量评价 信号处理,2021.
| DSTS-Net | KoNViD-1k | CVD2014 | LIVE-Qualcomm | LIVE-VQC |
|---|---|---|---|---|
| SROCC | 0.812 | 0.881 | 0.790 | 0.758 |
| PLCC | 0.817 | 0.876 | 0.799 | 0.778 |
SOC.py is used to extract second order covariance
DSTS_345_9ff_mean_fc.py is used to build deep temporal modeling network
It's the first time for me to use the deep-learning and share it in github.
So if you find some error, please feel free to contact me:
Yinhao Liu, [email protected].
Code and data prepration largely benefits from VSFA by Dingquan Li.