HOA-SSR数据库

HOA-SSR数据库是由技术力量SenseLab创建的高质量多模态数据集，包含12个沉浸式音视频场景，主要用于评估全方位媒体格式的多模态感知质量。该数据集包含8K分辨率的360度视频和4阶高阶Ambisonic音频，以及从音视频主观实验中收集的平均意见分数。数据集的创建旨在解决沉浸式音视频内容感知质量评估的研究需求，特别是在缺乏高质量多模态数据集的情况下。通过该数据集，研究者可以量化音频、视频及其组合的感知质量，并探索与各种客观质量指标的相关性。

The HOA-SSR database is a high-quality multimodal dataset developed by SenseLab. It contains 12 immersive audio-visual scenarios, and is primarily designed to evaluate the multimodal perceptual quality of omnidirectional media formats. This dataset includes 360-degree videos with 8K resolution, fourth-order higher-order Ambisonic audio, as well as mean opinion scores (MOS) collected from subjective audio-visual experiments. The database was constructed to address the research demands of perceptual quality assessment for immersive audio-visual content, particularly in contexts where high-quality multimodal datasets are lacking. With this dataset, researchers can quantify the perceptual quality of audio, video, and their combined content, and explore the correlations between these qualities and various objective quality metrics.

搜集汇总

背景与挑战

背景概述

HOA-SSR数据库是由SenseLab创建的高质量多模态数据集，包含12个沉浸式音视频场景，提供8K 360度视频和4阶高阶Ambisonic音频，并附带主观实验收集的平均意见分数。该数据集旨在评估全方位媒体格式的多模态感知质量，支持研究者量化音频、视频及其组合的感知质量，并探索与客观质量指标的相关性，以解决沉浸式音视频内容感知质量评估的研究需求。