ASOD60K

ASOD60K是由法国国立应用科学学院雷恩分校的Yi Zhang创建的大型全景视频音频诱导显著对象检测数据集。该数据集包含62,455个4K分辨率视频帧，通过音频诱导的眼动标记显著对象，具有六级层次结构，以丰富性、多样性和质量著称。每个序列都标有超类/子类，每个子类对象进一步标注了人眼注视、边界框、对象/实例级掩码和相关属性。ASOD60K旨在推动全景视频显著对象检测的研究，特别是在增强现实和虚拟现实应用中。

ASOD60K is a large-scale panoramic video-audio guided salient object detection dataset created by Yi Zhang from Institut National des Sciences Appliquées de Rennes (INSA Rennes), France. This dataset comprises 62,455 4K-resolution video frames, where salient objects are annotated using audio-guided eye movement data. It features a six-level hierarchical structure and is renowned for its richness, diversity and high annotation quality. Each video sequence is labeled with superclass and subclass annotations, and each subclass object is further annotated with human eye fixations, bounding boxes, object- and instance-level masks, as well as relevant attributes. ASOD60K aims to advance research on panoramic video salient object detection, particularly in augmented reality (AR) and virtual reality (VR) applications.