Super-Resolution for Video Compression benchmark
收藏资源简介:
本数据集名为‘Super-Resolution for Video Compression benchmark’,由莫斯科国立大学和MSU人工智能研究所创建。该数据集包含基于五种压缩标准(H.264, H.265, H.266, AV1, AVS3)的视频编码器压缩的视频,用于分析超分辨率模型对压缩视频的上采样能力。数据集的创建涉及从多个来源收集视频,包括Vimeo、相机拍摄和游戏视频,并使用K-Means聚类进行视频选择。该数据集主要应用于视频处理领域,旨在评估和改进超分辨率技术在视频压缩环境中的性能,解决视频质量与带宽消耗之间的平衡问题。
This dataset, named 'Super-Resolution for Video Compression benchmark', was developed by Lomonosov Moscow State University and the MSU AI Research Institute. It contains videos compressed by video encoders adhering to five compression standards: H.264, H.265, H.266, AV1, and AVS3, which is used to analyze the upsampling performance of super-resolution models on compressed videos. The dataset construction involved collecting videos from multiple sources, including Vimeo, camera-captured footage, and gaming videos, and selecting suitable videos via K-Means clustering. This dataset is primarily applied in the field of video processing, aiming to evaluate and improve the performance of super-resolution technologies in video compression scenarios, and address the trade-off between video quality and bandwidth consumption.




