NDISPark（ Night and Day Instance Segmented Park）

该数据集包括大约250个图像，描述了几个停车位，描述了我们在真实场景中可以找到的大多数有问题的情况: 七个不同的摄像机在各种天气条件和视角下捕获图像。另一个具有挑战性的方面是在许多场景中存在部分遮挡模式，例如障碍物 (树木，灯柱，其他汽车) 和阴影汽车。从白天和晚上在不同天气条件下拍摄的视频镜头和用于车辆分割的摄像机角度拍摄的停车场图像。

This dataset contains approximately 250 images that depict multiple parking spaces and cover most problematic scenarios commonly encountered in real-world parking environments. All images were captured by seven distinct cameras under varying weather conditions and viewing perspectives. Another challenging aspect of this dataset is the widespread presence of partial occlusion patterns in numerous scenes, including obstacles such as trees, lampposts, other vehicles, as well as shadowed or partially obscured cars. Additionally, the dataset includes video clips shot during both daytime and nighttime under diverse weather conditions, alongside parking lot images acquired from camera angles optimized for vehicle segmentation tasks.

搜集汇总

背景与挑战

背景概述

NDISPark数据集包含约250张停车位图像，涵盖多种天气条件和视角，特别关注部分遮挡和阴影车辆等挑战性场景。该数据集由七个不同摄像机采集，包含白天和夜晚不同条件下的图像，适用于车辆分割和交通密度估计研究。