ShanghaiTech

该数据集名为ShanghaiTech，包含了1198张标注图像，共标注了330,165个人。数据集分为两部分：Part A（SHA）包含高密度人群的图像，而Part B（SHB）则包含相对稀疏的人群场景。具体来说，Part A包含了482张图像（其中300张用于训练，182张用于测试），而Part B则包含了716张图像（其中400张用于训练，316张用于测试）。整个数据集的规模为1198张图像，330,165个标注人物，其任务是对人群进行计数。The dataset named ShanghaiTech consists of 1,198 annotated images with a total of 330,165 annotated human instances. It is divided into two subsets: Part A (SHA) which contains images of high-density crowd scenes, and Part B (SHB) which covers relatively sparse crowd scenarios. Specifically, Part A includes 482 images, among which 300 are used for training and 182 for testing, while Part B comprises 716 images with 400 for training and 316 for testing. The entire dataset has 1,198 images and 330,165 annotated individuals, and its targeted task is crowd counting.

搜集汇总

背景与挑战

背景概述

ShanghaiTech Dataset是CVPR 2016论文中提出的人群计数数据集，用于单图像人群计数研究，与多列卷积神经网络(MCNN)相关。数据集可通过百度云盘和Dropbox下载。