TSV360
收藏资源简介:
TSV360数据集由来自160个具有多样化视觉内容的360度视频的约16,000个ERP帧、文本描述和地面实况突出图组成。该数据集旨在支持基于文本的360度视频突出检测方法的训练和客观评估。数据集的内容包括约16,000个ERP帧、文本描述和地面实况突出图。数据集的创建过程涉及到从160个具有多样化视觉内容的360度视频中提取ERP帧、文本描述和地面实况突出图。数据集的应用领域是文本驱动的360度视频突出检测,旨在解决如何根据用户的需求进行定制化突出检测的问题。
The TSV360 dataset consists of approximately 16,000 ERP frames, textual descriptions, and ground-truth saliency maps extracted from 160 360-degree videos with diverse visual content. This dataset is intended to support the training and objective evaluation of text-based 360-degree video saliency detection methods. The process of constructing the TSV360 dataset involves extracting the aforementioned data elements from the 160 360-degree videos with diverse visual content. Its application domain is text-driven 360-degree video saliency detection, which aims to address the challenge of performing customized saliency detection based on user requirements.

- 1通过塞萨洛尼基希腊CERTH ITI · 2025年



