CHUG
收藏资源简介:
CHUG数据集是一个大规模的用户生成内容(UGC)高动态范围(HDR)视频质量数据集,由德克萨斯大学奥斯汀分校的研究人员创建。该数据集包含856个UGC-HDR源视频,通过多个分辨率和比特率进行转码,总计5,992个视频。通过Amazon Mechanical Turk进行大规模研究,收集了211,848个主观评分。CHUG旨在为分析UGC特定失真在HDR视频中的影响提供一个基准,并推动无参考(NR)HDR视频质量评估(VQA)研究。该数据集适用于研究UGC-HDR视频的失真、压缩和编码对视频质量的影响,并用于开发无参考VQA模型。
The CHUG Dataset is a large-scale User-Generated Content (UGC) High Dynamic Range (HDR) video quality dataset created by researchers at The University of Texas at Austin. This dataset contains 856 UGC-HDR source videos, which are transcoded into multiple resolutions and bitrates, totaling 5,992 videos. A large-scale subjective study was conducted via Amazon Mechanical Turk, and 211,848 subjective ratings were collected. The CHUG Dataset aims to provide a benchmark for analyzing the impact of UGC-specific distortions on HDR video quality and advance research on No-Reference (NR) HDR Video Quality Assessment (VQA). This dataset is suitable for studying the effects of distortions, compression and encoding on the quality of UGC-HDR videos, and for developing no-reference VQA models.

- 1通过德克萨斯大学奥斯汀分校 · 2025年



