SmokeBench
收藏资源简介:
SmokeBench是一个为早期火灾场景监控图像去烟而设计的真实世界数据集。该数据集包含在多种场景设置和烟雾浓度下捕获的图像对,提供精确对齐的降质和清洁图像,以支持监督学习和严格的评估。数据集通过一个专门的烟雾采集系统(SAS)收集,该系统能够在真实条件下收集烟雾降质和无烟雾图像对。SmokeBench数据集的发布为在真实火灾场景中推进稳健和实用的图像去烟技术提供了一个宝贵的基础。
SmokeBench is a real-world dataset designed for image desmoking in early fire scene monitoring. This dataset contains image pairs captured under various scene settings and varying smoke concentrations, providing accurately aligned degraded and clean images to support supervised learning and rigorous evaluation. The dataset is collected via a specialized Smoke Acquisition System (SAS), which can gather smoke-degraded and smoke-free image pairs under real-world conditions. The release of the SmokeBench dataset offers a valuable foundation for advancing robust and practical image desmoking technologies in real fire scenarios.




