ShanghaitechA

ShanghaiTechA人群计数数据集
包含482张图片和对应的密度图、透视图，Train：Test=300：182，
主要用来训练PAACNN
ShanghaiTech [44]是一个应用广泛的大规模的人群计数数据集之一，由1198幅人群图像和330165个被注释的头部中心组成。根据人群密度分布的不同，将其分为两部分，分别是ShanghaiTech Part A（SHTA）和ShanghaiTech Part B（SHTB）。SHTA是从互联网上随机选择的人群图像，SHTB是从上海大都市的一条繁忙街道拍摄的人群图像。SHTA中人群的密度比SHTB中的密度大得多。

ShanghaiTechA Crowd Counting Dataset
Consists of 482 images paired with their corresponding density maps and perspective maps, with a train-test split of 300:182 (300 for training and 182 for testing). It is primarily utilized for training PAACNN.
ShanghaiTech [44] is one of the widely used large-scale crowd counting datasets, which contains 1,198 crowd images and 330,165 annotated head centers. It is divided into two subsets based on differences in crowd density distribution: ShanghaiTech Part A (SHTA) and ShanghaiTech Part B (SHTB). SHTA comprises crowd images randomly selected from the Internet, while SHTB was captured from a busy street in the Shanghai metropolitan area. The crowd density in SHTA is significantly higher than that in SHTB.

搜集汇总

背景与挑战

背景概述

ShanghaitechA是一个大规模的人群计数数据集，包含482张图片和对应的密度图、透视图，训练集和测试集比例为300:182。数据集分为A、B两部分，A部分为互联网随机图像，人群密度较大，B部分为上海街道图像，密度较小，两部分共同构成更具挑战性的数据集。