AVT-VQDB-UHD-2-HDR - An Open 8K HDR Source Dataset for Video Quality Research
This open 8K HDR source dataset provides 31 video sequences of 15 seconds each, specifically created for video quality assessment research. The dataset was developed as part of the ILMETA initiative and the PoQuMo8K project (ZIM No. KK5112706ER1) at TU Ilmenau, addressing the growing need for video quality models that support resolutions beyond 4K.
Description
This open 8K HDR source dataset provides 31 video sequences of 15 seconds each, specifically created for video quality assessment research. The dataset was developed as part of the ILMETA initiative and the PoQuMo8K project (ZIM No. KK5112706ER1) at TU Ilmenau, addressing the growing need for video quality models that support resolutions beyond 4K.
As more screens with ultra-high resolution and High Dynamic Range capabilities become available on the market, and video streaming providers like YouTube offer content in HDR with resolutions up to 8K (UHD-2), the demand for valid estimation of video quality for 8K content is increasing. Currently, the most used video quality models only support resolutions up to 4K (UHD-1), making this dataset particularly valuable for advancing video quality research.
The videos were created from scratch to accurately represent real-life footage while taking into account video coding and video quality testing challenges. The paper describes the methodological approach of planning, creating, and post-processing the video sources, evaluating their suitability for video coding research and quality testing applications. The repository includes SHA-512 and MD5 checksums for verification and hosts objective metrics for the source content.
This work was funded by the Deutsche Forschungsgemeinschaft (DFG, No. 438822823) and the ZIM project (No. KK5112706ER1). The dataset contributes to the advancement of 8K video quality assessment methods and provides researchers with high-quality source material for codec evaluation and quality model development.
Access
Source videos available via Google Form at https://forms.gle/Ptq9NvZe8tKpnJXx6, checksums and objective metrics available on GitHub
License
Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)