数据堂—760,607张监控场景下车辆检测数据

760,607张监控场景下车辆检测数据采集场景为地下停车场、地面停车场、出入口闸口、室外道路（高速公路、城市道路等）。数据多样性包括不同监控场景、不同时间段、不同摄像头、车辆分布多样（密集、稀疏）。在标注方面，标注车身矩形框、车辆类型属性。数据可用于监控场景下的车辆检测等任务

This dataset consists of 760,607 vehicle detection surveillance data collected from multiple scenarios: underground parking lots, ground parking lots, entrance and exit gates, and outdoor roads (including expressways, urban roads, etc.). The data exhibits rich diversity across multiple dimensions, including different surveillance scenarios, time periods, camera devices, and vehicle distribution densities (both dense and sparse scenarios). In terms of annotation, both the bounding boxes of vehicle bodies and vehicle type attributes are labeled. This dataset can be applied to tasks such as vehicle detection in surveillance scenarios.

搜集汇总

背景与挑战

背景概述

该数据集包含760,607张监控场景图像和5,796,265个车辆边界框，覆盖地下停车场、地面停车场、出入口及户外道路等多种环境，具有场景、时间段和车辆分布的多样性。标注内容包括车辆矩形边界框和车型属性，准确率高，偏移不超过3像素且合格率至少97%，适用于监控场景下的车辆检测任务。