Alkmaar Car Detection from Satellite Image

This dataset contains parking spots in Alkmaar. In each parking spot the standard deviation, range, and variance is calculated for 3 color bands and one near-infrared band of a commercial Pleiades satellite image. Besides that, the NDVI mean and max of each parking spot is also included. Additionally a classification of the images is provided in the matrix column, where it was determined using Triangle thresholding if a car is parked in the parking spot. 1 - marks a parked car, 0 - marks no car, and -1 means that the NDVI value was too high to classify the parking spot. Finally the occupied column represents ground truth. It was determined based on video footage taken two minutes before and after the satellite image was taken, 1 - represents an occupied parking spot, 0 - an empty parking spot and NULL means no data was recorded for that parking spot.本数据集收录荷兰阿尔克马尔（Alkmaar）的停车场泊位相关数据。针对每一个泊位，本数据集计算了商用Pleiades（Pleiades）卫星影像的3个可见光波段与1个近红外波段的标准差、极差与方差；同时包含每个泊位的归一化差分植被指数（NDVI, Normalized Difference Vegetation Index）均值与最大值。此外，图像分类结果存储于matrix列中，该分类通过三角阈值法（Triangle thresholding）判断泊位内是否停放车辆：1代表泊位内停放有车辆，0代表无车辆停放，-1代表NDVI值过高，无法完成该泊位的分类。最后，occupied列为真实标注值（ground truth），其标注基于卫星影像拍摄前后两分钟内采集的视频数据生成：1代表该泊位已被占用，0代表泊位为空，NULL表示该泊位无有效记录数据。