DLR-ACD

DLR-ACD 数据集是用于人群计数和密度估计以及大规模活动中的人员定位的航空图像集合。它包含 33 张大型航拍图像，这些图像是通过 16 次不同的飞行活动在各种群众活动和涉及人群的城市场景中获得的，例如体育赛事、市中心、露天集市和节日。这些图像是使用安装在直升机上的标准数码单反相机拍摄的，其空间分辨率（或地面采样距离 - GSD）范围为 4.5 至 15 厘米/像素。该数据集是用个人的点注释手动标记的，总共包含 226,291 个人注释，每张图像的注释范围从 285 到 24,368。DLR-ACD Dataset is a collection of aerial images designed for crowd counting, density estimation, and person localization in large-scale events. It contains 33 large aerial images, which were acquired across 16 distinct flight missions in various crowd activities and urban scenarios involving large populations, such as sporting events, downtown areas, open-air markets, and festivals. These images were captured using a standard digital single-lens reflex (DSLR) camera mounted on a helicopter, with their spatial resolution (or Ground Sampling Distance - GSD) ranging from 4.5 to 15 centimeters per pixel. This dataset was manually labeled with individual point annotations, totaling 226,291 person annotations, with the count of annotations per image ranging from 285 to 24,368.

搜集汇总

背景与挑战

背景概述

DLR-ACD是一个用于人群计数和密度估计的航空图像数据集，包含33张高分辨率航拍图像和226,291个人工标注点注释，适用于大规模活动的人员定位研究。