Dense Fog 目标检测数据集

Dense Fog 是一个用于目标检测的大型多模式数据集。该数据集包括真实世界驾驶场景中的 12,000 个样本和雾气室中受控天气条件下的 1,500 个样本。该数据集包括不同的恶劣天气条件，如雾、雪和雨，采集于北欧 10,000 多公里的驾驶里程中。该数据集总共有 10 万个目标被标注了精确的 2D 和 3D 边界框。

Dense Fog is a large-scale multimodal dataset for object detection. It consists of 12,000 samples from real-world driving scenarios and 1,500 samples collected under controlled weather conditions in a fog chamber. Covering diverse severe weather conditions such as fog, snow and rain, the dataset was gathered across more than 10,000 kilometers of driving mileage in Northern Europe. In total, 100,000 target objects in this dataset have been annotated with precise 2D and 3D bounding boxes.

搜集汇总

背景与挑战

背景概述

Dense Fog 是一个专注于目标检测的大型多模式数据集，包含真实驾驶场景和受控雾气室条件下的样本，总计13,500个样本，覆盖雾、雪、雨等多种恶劣天气条件。该数据集采集自北欧10,000多公里的驾驶里程，并提供了10万个目标的精确2D和3D边界框标注，适用于恶劣天气环境下的目标识别研究。