PKLot

该数据集名为PKLot，包含了在不同天气条件下从多个停车位拍摄的图像，主要关注于停车位检测。此外，数据集中还手动标注了车辆和额外的停车位，并为停车位创建了新的标注多边形。具体规模上，数据集分别包含了来自43、46和134个停车位的3791、4152和4117张图像。该数据集的任务是进行停车位检测以及车辆实例分割。

This dataset is named PKLot. It comprises images captured from multiple parking spaces under diverse weather conditions, with the primary objective of parking space detection. Furthermore, vehicles and supplementary parking spaces are manually annotated within the dataset, and dedicated annotation polygons are created for each parking space. In terms of scale, the dataset contains 3791, 4152, and 4117 images originating from 43, 46, and 134 parking spaces respectively. The target tasks of this dataset are parking space detection and vehicle instance segmentation.

搜集汇总

背景与挑战

背景概述

PKLot数据集包含12,417张停车场图像，涵盖多种天气条件，附带XML文件标注停车位状态，适用于停车场分类研究。