YouTube-UGC 视频数据集

YouTube-UGC 是一个大规模 UGC 数据集，包括 YouTube 官网上 1500 个时长 20 秒的视频，这些内容均由用户遵循知识共享协议上传，可用于视频压缩和质量评估领域的研究。视频内容涵盖了游戏、体育等热门类别，以及高动态范围成像（HDR）等新特性。质量评估使用了三个参考指标：Noise, Banding, SLEEQ 。

YouTube-UGC is a large-scale user-generated content (UGC) dataset that includes 1500 20-second-long videos from the official YouTube platform. All these contents are uploaded by users under the Creative Commons license, and it can be applied to research in the fields of video compression and quality assessment. The video content covers popular categories such as gaming and sports, as well as new features like high dynamic range (HDR) imaging. Three reference metrics are used for quality assessment: Noise, Banding, and SLEEQ.

搜集汇总

背景与挑战

背景概述

YouTube-UGC 视频数据集是一个包含1500个时长20秒的YouTube用户生成内容（UGC）的大规模数据集，涵盖游戏、体育等热门类别及高动态范围成像（HDR）等新特性，主要用于视频压缩和质量评估研究。数据集的质量评估参考了Noise、Banding和SLEEQ三个指标。