故宫博物院冰窖餐厅区域排队监测数据集

本数据集基于故宫冰窖餐厅的监控视频，利用YOLOv8深度学习模型进行排队人数与进出流量的实时统计与分析。数据采集时间范围为2025年6月20日至2025年7月13日，每天10:30至14:30时段内，且不包括周一。数据采集通过两台摄像头，监控排队区域和门口区域，捕捉的视频流通过YOLOv8的目标检测与跟踪算法实时计算进出人数与排队人数。YOLOv8结合深度学习的目标检测算法，通过设定感兴趣区域（ROI）对视频中的行人进行识别，并统计每一帧的人员流动数据。每分钟内的人员流动数据通过统计每帧目标的框选与分类，生成包括进出人数和当前排队人数的统计信息。为了确保数据的准确性和连续性，我们对数据集中出现的缺失值和异常值进行了补充与改正。数据集的时间精度为每分钟一次。该数据集主要应用于排队时间估算模型的构建及未来排队人数预测，为餐厅运营优化提供数据支持。此外，数据集能够为排队管理提供实时人员流动分析，也为类似场景中的人员流动预测提供技术支撑，具有较高的实际应用价值和研究意义,数据及大小为166KB。

This dataset is constructed using surveillance footage from the Ice Cellar Restaurant of the Palace Museum, where the YOLOv8 deep learning model is employed for real-time statistics and analysis of queue size and passenger flow. The data collection period spans from June 20 to July 13, 2025, during the time slot of 10:30 to 14:30 each day, excluding Mondays. Two cameras were deployed to monitor the queuing zone and entrance/exit area for data collection. The captured video streams are processed using YOLOv8's object detection and tracking algorithms to compute the in-flow and out-flow of people as well as the current queue size in real time. As a deep learning-based object detection algorithm, YOLOv8 identifies pedestrians in videos by setting Regions of Interest (ROIs) and counts passenger flow data for each frame. Per-minute passenger flow data is generated by counting and classifying the detected objects in each frame, producing statistical information including the number of people entering/exiting and the current queue size. To ensure the accuracy and continuity of the dataset, missing and outlier values in the collected data were supplemented and corrected. The dataset has a temporal precision of one record per minute. This dataset is primarily applied to the construction of queue time estimation models and future queue size prediction, providing data support for restaurant operation optimization. In addition, the dataset can provide real-time passenger flow analysis for queuing management, as well as technical support for passenger flow prediction in similar scenarios, holding high practical application value and research significance. The total size of this dataset is 166 KB.

搜集汇总

背景与挑战

背景概述

该数据集基于故宫博物院冰窖餐厅监控视频，利用YOLOv8深度学习模型实时统计排队人数与进出流量，采集时间为2025年6月20日至7月13日（每日10:30-14:30，排除周一）。数据集主要用于构建排队时间估算模型和预测未来排队人数，支持餐厅运营优化和人员流动分析，具有实际应用价值。