Compressed Feature Quality Assessment Dataset
收藏资源简介:
该数据集由新加坡南洋理工大学的研究团队创建,旨在评估压缩特征的质量。数据集包含300个原始特征和12000个压缩特征,这些特征来自三个视觉任务(分类、分割、深度估计)和四种特征编解码器。数据集提供了跨任务、比特率和编解码器的语义退化定量分析。同时,每个压缩特征都提供了任务特定的语义退化标签,作为训练和评估质量指标的真实标签。该数据集有助于推动压缩特征质量评估(CFQA)领域的研究,并为社区提供了一个探索CFQA的基础资源。
This dataset was developed by a research team from Nanyang Technological University, Singapore, with the core objective of assessing the quality of compressed features. It contains 300 original features and 12,000 compressed features derived from three vision tasks (classification, segmentation, and depth estimation) and four feature codecs. The dataset enables quantitative analyses of semantic degradation across tasks, bit rates, and codecs. Moreover, each compressed feature is paired with task-specific semantic degradation labels, which act as ground truth for training and evaluating quality metrics. This dataset contributes to advancing research in the field of Compressed Feature Quality Assessment (CFQA) and serves as a foundational resource for the global research community to conduct CFQA-related explorations.




