云梯系统电梯拥堵分析数据

依据告警数据，评估电梯的运行状态和安全性能，及时在轿厢内提醒乘客安全乘梯，设备抓图便于在发生突发状况时，快速了解现场状况，定位电梯及乘客所在位置，对梯内乘客的安全起到实际保障作用；定期对告警数据设备抓图所反映的现场情况进行分析，方便物业制定对电梯的维护保养计划；根据一段时间内告警的数据规模、现场图片、时间等，帮助电梯生产销售企业了解该地的乘梯情况，选择销售合适的电梯类型；公开报道一段时间内电梯拥堵告警数据，可以提高公众对乘梯安全的重视程度。1.通过基于传感器的实时电梯门开关状态检测算法检出处于运行状态的电梯； 2.开启电梯内摄像头抓图服务(抓图频率1FPS),并将图片上传到华为云obs。云端算法服务器从华为云obs拉取到图片后，通过基于深度学习神经网络训练的目标检测算法进行图片检测，输出图片相关属性；3.输出的图片属性包含：目标类型category、目标坐标框box、目标结果置信度score、目标匹配groupid； 4.对输出的目标类型进行识别，category所代表的含义分别为：0-人体，1-人头，groupid为人头和其他类别的匹配数量，当算法检出人数超过当前电梯配置拥堵人数后触发电梯拥堵告警，告警时间为时间戳，更加精准。

Based on alarm data, this dataset evaluates the operating status and safety performance of elevators: it timely sends safety reminders to passengers in elevator cabs; the device-captured images enable rapid acquisition of on-site conditions and location information of elevators and passengers during emergencies, effectively safeguarding the safety of in-car passengers.
Regular analysis of on-site scenarios reflected by alarm data and captured images helps property management formulate elevator maintenance plans.
By analyzing alarm data volume, on-site images, timestamps and other metrics over a period of time, elevator manufacturing and sales enterprises can gain insights into local elevator riding demands and select appropriate elevator models for sales.
Publicly disclosing elevator congestion alarm data over a specified period can also enhance public awareness of elevator safety.
1. Detect operating elevators using a real-time elevator door switch state detection algorithm based on sensors;
2. Enable the image capture service for in-car cameras (with a capture frequency of 1 FPS) and upload the images to Huawei Cloud OBS. After the cloud-based algorithm server pulls the images from Huawei Cloud OBS, it conducts image detection via a target detection model trained on deep learning neural networks, and outputs relevant image attributes;
3. The output image attributes include: target category, target bounding box, target confidence score, and target matching groupid;
4. Identify the output target categories, where the values of category correspond to: 0 for human body, 1 for human head. The groupid indicates the number of matched pairs between human heads and other categories. When the algorithm detects a passenger count exceeding the configured congestion limit of the elevator, an elevator congestion alarm will be triggered, with the alarm time recorded as a timestamp for higher accuracy.

搜集汇总

数据集介绍

特点

云梯系统电梯拥堵分析数据是一个包含2251条记录的数据集，用于分析电梯拥堵情况，通过传感器和深度学习算法采集和处理数据，旨在提升电梯运行安全和乘客乘梯体验。