PA-100K (PA-100K Dataset)

PA-100K是最近提出的大型行人属性数据集，总共从室外监控摄像机收集了100,000个图像。它分为训练集的80,000图像和验证集的10,000和测试集的10,000。此数据集由26个二进制属性标记。两个所选数据集中都存在的共同特征是，由于分辨率相对较低，图像模糊，并且每个二进制属性的正比率较低。

PA-100K is a recently proposed large-scale pedestrian attribute dataset. A total of 100,000 images were collected from outdoor surveillance cameras. The dataset is split into three subsets: 80,000 training images, 10,000 validation images, and 10,000 test images. It is annotated with 26 binary attributes. A common feature shared by the validation and test subsets is that the images are blurry due to their relatively low resolution, and the positive ratio of each binary attribute is relatively low.

搜集汇总

背景与挑战

背景概述

PA-100K是一个大型行人属性识别数据集，包含10万张室外监控图像，分为8万训练集和1万验证集、1万测试集，标注了26个二进制属性。特点是图像分辨率低、模糊且属性正比率低。