CN116403134A - Method and device for detecting abnormal events in carriage, electronic equipment and storage medium - Google Patents

Abstract

The embodiment of the application provides a carriage abnormal event detection method and device, electronic equipment and storage medium. The method comprises the following steps: acquiring a video of a scene of a train carriage shot by a camera; determining a detection target in a train carriage scene; obtaining world coordinates of the detection target through a calibration algorithm; encrypting the video privacy information of the detection target through the federal learning model; identifying an action of the detection target occurring in the video; and judging whether a carriage abnormal event exists according to the action. Based on the method, the embodiment of the application utilizes the federal learning algorithm model to protect the video information privacy of the target and detect the action of the target in the video, so that the abnormal event in the train carriage scene can be timely judged to ensure the running safety of the train. Therefore, the embodiment of the application detects the abnormal event in the carriage scene while reducing the risk of privacy disclosure of the user by adopting the federal learning algorithm model.

Description

Translated fromChinese

车厢异常事件检测方法和装置、电子设备及存储介质Method and device for detecting abnormal event in carriage, electronic equipment and storage medium

技术领域technical field

本申请涉及轨道交通列车安全管理技术领域，尤其涉及一种车厢异常事件检测方法和装置、电子设备及存储介质。The present application relates to the technical field of rail transit train safety management, and in particular to a method and device for detecting an abnormal event in a carriage, electronic equipment, and a storage medium.

背景技术Background technique

现阶段的轨道列车车厢异常事件检测系统，在异常外事件检测方面，还是以人力检测为主要手段，人力检测效率低下，检测的效果单一，只能检测某一特定场景下某一种特定的异常事件，人力检测范围有限，检测效果不具有实时性，而且人力成本高，需要克服昼夜变化带来的不便，不能全天检测列车车厢内部情况。目前主流的列车车厢内部异常事件检测系统，在面对不同种类的异常事件时，只能判断某一种单一的异常事件，不能兼顾不同的异常事件，同时由于采用传统的学习模型，不能保证用户数据的隐私安全。At present, the abnormal event detection system of rail train carriages still uses human detection as the main method in the detection of abnormal events. The efficiency of human detection is low, and the detection effect is single. It can only detect a specific abnormality in a specific scene. Incidents, human detection range is limited, the detection effect is not real-time, and the labor cost is high. It is necessary to overcome the inconvenience caused by day and night changes, and it is impossible to detect the internal conditions of the train compartment all day. At present, the mainstream abnormal event detection system inside the train compartment can only judge a single abnormal event when faced with different types of abnormal events, and cannot take into account different abnormal events. Data privacy and security.

发明内容Contents of the invention

本申请实施例的主要目的在于提出一种车厢异常事件检测方法和装置、电子设备及存储介质，能够达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。The main purpose of the embodiments of the present application is to provide a method and device for detecting abnormal events in the carriage, electronic equipment and storage media, which can reduce the risk of user privacy leakage and detect abnormal events in the carriage scene.

为实现上述目的，本申请实施例的第一方面提出了一种车厢异常事件检测方法，所述方法包括：In order to achieve the above purpose, the first aspect of the embodiment of the present application proposes a method for detecting an abnormal event in a carriage, the method including:

获取相机拍摄列车车厢场景的视频；Obtain the video of the camera shooting the scene of the train carriage;

确定列车车厢场景内的检测目标；Determine the detection target in the train compartment scene;

通过标定算法得到所述检测目标的世界坐标；Obtaining the world coordinates of the detection target through a calibration algorithm;

通过联邦学习模型对所述检测目标的视频隐私信息进行加密；Encrypting the video privacy information of the detection target through a federated learning model;

识别所述检测目标在视频中发生的动作；Identifying the action of the detected target in the video;

根据所述动作判断是否存在车厢异常事件。According to the action, it is judged whether there is an abnormal event in the carriage.

在一些实施例，在所述获取相机拍摄列车车厢场景的视频之前，还包括：In some embodiments, before the acquisition of the video of the scene of the train car taken by the camera, it also includes:

对不同的所述列车车厢场景进行定义；Define different scenarios of the train carriage;

确定不同的所述列车车厢场景对应车厢异常事件的种类。It is determined that the different train compartment scenarios correspond to the types of compartment abnormal events.

在一些实施例，所述通过标定算法得到所述检测目标的世界坐标，包括：In some embodiments, the obtaining the world coordinates of the detection target through a calibration algorithm includes:

获取所述相机的内部参数；Obtain internal parameters of the camera;

计算出相机模型垂直于车厢检测背景的平移向量；Calculate the translation vector of the camera model perpendicular to the car detection background;

计算世界坐标系绕图像坐标系的旋转矩阵；Calculate the rotation matrix of the world coordinate system around the image coordinate system;

结合所述相机的实际位置得到图像坐标和世界坐标之间的坐标转换关系；Combining the actual position of the camera to obtain the coordinate transformation relationship between the image coordinates and the world coordinates;

基于所述坐标转换关系得到所述检测目标的世界坐标。The world coordinates of the detection target are obtained based on the coordinate transformation relationship.

在一些实施例，所述识别所述检测目标在视频中发生的动作，包括：In some embodiments, the identifying the action of the detected target in the video includes:

采用特征标注法对不同场景的检测目标行为数据进行标注；The feature labeling method is used to label the detection target behavior data in different scenes;

将标注的检测目标行为数据输入到联邦学习模型中，联邦学习模型输出隐私数据，在将加密后的检测目标行为数据输入到预设的车厢异常事件检测模型中进行算法训练，得到异常事件检测模型；Input the marked detection target behavior data into the federated learning model, and the federated learning model outputs private data, and input the encrypted detection target behavior data into the preset car abnormal event detection model for algorithm training, and obtain the abnormal event detection model ;

将所述检测目标的实时视频数据输入到所述车厢异常事件检测模型中，以识别所述检测目标在视频中发生的动作。The real-time video data of the detection target is input into the vehicle compartment abnormal event detection model to identify the action of the detection target in the video.

在一些实施例，所述车厢异常事件检测模型的训练方法如下：In some embodiments, the training method of the compartment abnormal event detection model is as follows:

构建联邦学习模型；Build a federated learning model;

实地采集检测数据；Collect test data on the spot;

确定需要检测的车厢异常事件种类；Determine the type of abnormal event in the car that needs to be detected;

通过联邦学习训练所述联邦学习模型，直至达到目标检测精度，得到所述车厢异常事件检测模型。The federated learning model is trained through federated learning until the target detection accuracy is achieved, and the compartment abnormal event detection model is obtained.

在一些实施例，所述根据所述动作判断是否存在车厢异常事件，包括：In some embodiments, the judging whether there is an abnormal event in the compartment according to the action includes:

检测是否出现某一列车车厢出现人员聚集现象时，通过统计列车车厢的人员数量来确定某一车厢是否出现人员聚集，在人员数量超过预设的数量阈值时，则判断存在人员聚集的车厢异常事件；When detecting whether there is a crowding phenomenon in a certain train compartment, it is determined whether there is a crowding in a certain compartment by counting the number of people in the train compartment. When the number of people exceeds the preset number threshold, it is judged that there is an abnormal event in the compartment where people gather ;

检测列车车厢是否有人员跌倒的情况时，如果标注的人体肢体接触地面的时间超过预设的时间阈值，则判断存在人员跌倒的车厢异常事件；When detecting whether there is a person falling in the train car, if the marked human limbs touch the ground for more than the preset time threshold, it is judged that there is an abnormal event in the car that the person fell;

检测列车车厢是否有人员发生肢体冲突，在检测人员是否发生肢体冲突时，标记冲突的基本动作，在满足冲突的基本动作的情况下，则判断存在人员发生肢体冲突的车厢异常事件；Detect whether there is a physical conflict in the train compartment. When detecting whether there is a physical conflict between the personnel, mark the basic movement of the conflict. If the basic movement of the conflict is satisfied, it is judged that there is an abnormal event in the carriage where the personnel have a physical conflict;

检测列车车厢行李架部位是否有易脱落的行李物品，确定列车车架的位置以及列车车架摆放行李的范围，当摆放的行李超过行李架的位置时，则判断存在超过行李架位置的行李物品的车厢异常事件；并检测列车行李架物品的种类，如果检测到不属于行李架摆放的物品种类，则判断存在不属于行李架摆放的物品种类的的车厢异常事件；Detect whether there are luggage items that are easy to fall off on the luggage rack of the train compartment, determine the position of the train frame and the range of luggage placed on the train rack, and when the luggage placed exceeds the position of the luggage rack, it is judged that there is a luggage that exceeds the position of the luggage rack. Carriage abnormal event of luggage items; and detect the type of luggage rack items on the train, if it is detected that it does not belong to the type of items placed on the luggage rack, it is judged that there is an abnormal event in the compartment that does not belong to the type of items placed on the luggage rack;

检测列车车厢过道是否有大件物品阻塞通道时，首先通过场景定义确定列车车厢通道的位置，然后通过检测算法检测存在列车通道的物品，当出现阻塞通道的物品时，则判断存在大件物品阻塞列车车厢通道的车厢异常事件。When detecting whether there are large items blocking the aisle of the train carriage, first determine the position of the aisle of the train carriage through the scene definition, and then use the detection algorithm to detect the items that exist in the aisle of the train. When there is an item blocking the passage, it is judged that there is a blockage of large items Carriage anomalies in the passage of train carriages.

在一些实施例，在所述根据所述动作判断是否存在车厢异常事件之后，还包括：In some embodiments, after determining whether there is an abnormal event in the compartment according to the action, it further includes:

对所述车厢异常事件进行预警和上报。Carry out early warning and report on the abnormal event of the carriage.

为实现上述目的，本申请实施例的第二方面提出了一种车厢异常事件检测装置，所述装置包括：In order to achieve the above purpose, the second aspect of the embodiment of the present application proposes a device for detecting abnormal events in the compartment, the device includes:

获取模块，用于获取相机拍摄列车车厢场景的视频；Obtaining module, is used for obtaining the video of camera shooting train compartment scene;

确定模块，用于确定列车车厢场景内的检测目标；A determining module, configured to determine the detection target in the train compartment scene;

标定模块，用于通过标定算法得到所述检测目标的世界坐标；A calibration module, configured to obtain the world coordinates of the detection target through a calibration algorithm;

加密模块，用于通过联邦学习算法模型对所述检测目标的视频隐私信息进行加密；An encryption module, configured to encrypt the video privacy information of the detection target through a federated learning algorithm model;

识别模块，用于识别所述检测目标在视频中发生的动作；A recognition module, configured to recognize the action of the detected target in the video;

判断模块，用于根据所述动作判断是否存在车厢异常事件。A judging module, configured to judge whether there is an abnormal event in the carriage according to the action.

为实现上述目的，本申请实施例的第三方面提出了一种电子设备，所述电子设备包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现上述第一方面所述的方法。In order to achieve the above purpose, the third aspect of the embodiments of the present application proposes an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned computer program when executing the computer program. The method described in the first aspect.

为实现上述目的，本申请实施例的第四方面提出了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行时实现上述第一方面所述的方法。In order to achieve the above purpose, the fourth aspect of the embodiments of the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the above-mentioned first aspect. described method.

本申请提出的车厢异常事件检测方法和装置、电子设备及存储介质,获取相机拍摄列车车厢场景的视频；确定列车车厢场景内的检测目标；通过标定算法得到检测目标的世界坐标；通过联邦学习模型对检测目标的视频隐私信息进行加密；识别检测目标在视频中发生的动作；根据动作判断是否存在车厢异常事件。基于此，本申请实施例相比于现有的人力检测方法，采用机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作，从而实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够及时判断在列车车厢场景出现的异常事件，以确保列车的运行安全。因此，本申请实施例通过采用联邦学习算法模型来达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。The method and device, electronic equipment, and storage medium proposed in this application for detecting an abnormal event in a carriage obtain the video of the train carriage scene captured by the camera; determine the detection target in the train carriage scene; obtain the world coordinates of the detection target through a calibration algorithm; use the federated learning model Encrypt the video privacy information of the detection target; identify the motion of the detection target in the video; judge whether there is an abnormal event in the car according to the motion. Based on this, compared with the existing human detection method, the embodiment of the present application adopts machine vision technology and federated learning model to capture the target inside the train compartment while performing video analysis on the scene of the train compartment, and obtain the target through the calibration algorithm The world coordinates of the target, using the federated learning algorithm model to protect the privacy of the target's video information, while detecting the target's actions in the video, thus realizing a machine vision-based detection method for abnormal events in train carriages from the perspective of video, It can timely judge the abnormal events in the train compartment scene to ensure the safety of the train operation. Therefore, the embodiment of the present application uses a federated learning algorithm model to reduce the risk of user privacy leakage and at the same time detect abnormal events in the car scene.

附图说明Description of drawings

图1是本申请实施例提供的车厢异常事件检测方法的流程图；Fig. 1 is a flow chart of the method for detecting an abnormal event in the carriage provided by the embodiment of the present application;

图2是本申请实施例提供的车厢异常事件检测方法的子流程图；Fig. 2 is a sub-flow chart of the method for detecting an abnormal event in the carriage provided by the embodiment of the present application;

图3是本申请实施例提供的车厢异常事件检测方法的子流程图；Fig. 3 is a sub-flow chart of the method for detecting an abnormal event in the carriage provided by the embodiment of the present application;

图4是本申请实施例提供的车厢异常事件检测方法的子流程图；Fig. 4 is a sub-flow chart of a method for detecting an abnormal event in a carriage provided by an embodiment of the present application;

图5是本申请实施例提供的车厢异常事件检测模型训练方法的流程图；FIG. 5 is a flow chart of a method for training a car abnormality event detection model provided by an embodiment of the present application;

图6是本申请实施例提供的车厢异常事件检测装置的结构示意图；FIG. 6 is a schematic structural diagram of a vehicle abnormal event detection device provided by an embodiment of the present application;

图7是本申请实施例提供的电子设备的硬件结构示意图。FIG. 7 is a schematic diagram of a hardware structure of an electronic device provided by an embodiment of the present application.

具体实施方式Detailed ways

为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, not to limit the present application.

需要说明的是，虽然在装置示意图中进行了功能模块划分，在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于装置中的模块划分，或流程图中的顺序执行所示出或描述的步骤。说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。It should be noted that although the functional modules are divided in the schematic diagram of the device, and the logical sequence is shown in the flowchart, in some cases, it can be executed in a different order than the module division in the device or the flowchart in the flowchart. steps shown or described. The terms "first", "second" and the like in the specification and claims and the above drawings are used to distinguish similar objects, and not necessarily used to describe a specific sequence or sequence.

针对现有技术中人力检测效率低下且存在用户隐私泄露的风险的技术问题，本申请实施例提供了一种车厢异常事件检测方法和装置、电子设备及存储介质，获取相机拍摄列车车厢场景的视频；确定列车车厢场景内的检测目标；通过标定算法得到检测目标的世界坐标；通过联邦学习模型对检测目标的视频隐私信息进行加密；识别检测目标在视频中发生的动作；根据动作判断是否存在车厢异常事件。基于此，本申请实施例相比于现有的人力检测方法，采用机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作，从而实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够及时判断在列车车厢场景出现的异常事件，以确保列车的运行安全。因此，本申请实施例通过采用联邦学习算法模型来达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。Aiming at the technical problems of low human detection efficiency and the risk of user privacy leakage in the prior art, the embodiment of the present application provides a method and device for detecting an abnormal event in a carriage, electronic equipment, and a storage medium to obtain a video of a train carriage scene captured by a camera ; Determine the detection target in the scene of the train compartment; obtain the world coordinates of the detection target through the calibration algorithm; encrypt the video privacy information of the detection target through the federated learning model; identify the action of the detection target in the video; judge whether there is a compartment according to the action unusual event. Based on this, compared with the existing human detection method, the embodiment of the present application adopts machine vision technology and federated learning model to capture the target inside the train compartment while performing video analysis on the scene of the train compartment, and obtain the target through the calibration algorithm The world coordinates of the target, using the federated learning algorithm model to protect the privacy of the target's video information, while detecting the target's actions in the video, thus realizing a machine vision-based detection method for abnormal events in train carriages from the perspective of video, It can timely judge the abnormal events in the train compartment scene to ensure the safety of the train operation. Therefore, the embodiment of the present application uses a federated learning algorithm model to reduce the risk of user privacy leakage and at the same time detect abnormal events in the car scene.

本申请实施例提供的车厢异常事件检测方法和装置、电子设备及存储介质，具体通过如下实施例进行说明，首先描述本申请实施例中的车厢异常事件检测方法。The method and device, electronic device, and storage medium for detecting an abnormal event in a carriage provided in the embodiments of the present application are specifically described through the following embodiments. First, the method for detecting an abnormal event in a carriage in the embodiment of the present application is described.

图1是本申请实施例提供的车厢异常事件检测方法的一个可选的流程图，图1中的方法可以包括但不限于包括步骤S101至步骤S106。Fig. 1 is an optional flow chart of a method for detecting an abnormal event in a carriage provided by an embodiment of the present application. The method in Fig. 1 may include, but is not limited to, step S101 to step S106.

步骤S101，获取相机拍摄列车车厢场景的视频；Step S101, obtaining the video of the scene of the train car taken by the camera;

步骤S102，确定列车车厢场景内的检测目标；Step S102, determining the detection target in the train compartment scene;

步骤S103，通过标定算法得到检测目标的世界坐标；Step S103, obtaining the world coordinates of the detection target through a calibration algorithm;

步骤S104，通过联邦学习模型对检测目标的视频隐私信息进行加密；Step S104, encrypting the video privacy information of the detection target through the federated learning model;

步骤S105，识别检测目标在视频中发生的动作；Step S105, identifying the action of the detection target in the video;

步骤S106，根据动作判断是否存在车厢异常事件。Step S106, judging whether there is an abnormal event in the compartment according to the action.

在一些实施例中，由于列车车厢异常事件检测涉及到的场景多，不同的场景对应不同的异常事件，因此可以先确定该场景的类别以及该场景对应的异常事件。In some embodiments, since there are many scenarios involved in the detection of an abnormal event in a train car, and different scenarios correspond to different abnormal events, the category of the scenario and the corresponding abnormal event may be determined first.

在一些实施例中，检测的列车车厢场景包括但不限于：列车车厢内部、相邻列车车厢链接部位、列车行李架部位、卧铺以及软卧车厢过道等。In some embodiments, the detected train car scenes include, but are not limited to: inside the train car, connecting parts of adjacent train cars, parts of train luggage racks, sleeper berths, and aisles of soft sleeper cars.

在一些实施例中，不同的场景对应的异常事件包括但不限于：统计不同列车车厢的人数，看是否出现某一列车车厢出现人员聚集现象、检测列车车厢是否有人员跌倒、是否有人员发生肢体冲突、列车车厢连接处是否出现人员聚集、列车车厢行李架部位是否有易脱落的行李物品、列车车厢过道是否有大件物品阻塞通道等。In some embodiments, the abnormal events corresponding to different scenarios include but are not limited to: counting the number of people in different train cars, checking whether there is a phenomenon of people gathering in a certain train car, detecting whether a person falls in a train car, and whether a person has a physical injury. Conflicts, whether there is a gathering of people at the junction of the train carriages, whether there are luggage items that are easy to fall off on the luggage rack of the train carriages, whether there are large objects blocking the passage in the aisles of the train carriages, etc.

在一些实施例中，每一个场景都包含人员数量检测事件，不同的车辆运行状况会规定车厢的人员数量，如没有站票的车厢人员数量应该是额定人数，有站票的列车车厢人员数量要根据不同的车厢型号来确定，具体的场景需要根据实际的情况来确定人员的数量是否超过规定的人员数量。In some embodiments, each scene includes a number of people detection event, and different vehicle operating conditions will stipulate the number of people in the car. For example, the number of people in a car without a station ticket should be the rated number of people, and the number of people in a train car with a station ticket should be the number of people. It is determined according to different car models, and the specific scene needs to determine whether the number of personnel exceeds the specified number of personnel according to the actual situation.

在一些实施例中，为了提高列车车厢异常事件的检测精度，本申请实施例对相机进行了标定，通过标定算法来确定被检测目标的实际位置。首先获取到相机的内部参数，再计算出相机模型垂直于车厢检测背景的平移向量，再计算世界坐标系绕图像坐标系的旋转矩阵，结合相机的实际位置便可以得到图像坐标和世界坐标之间的坐标转换关系。需要说明的是，世界坐标的位置就是目标的现实空间的位置。In some embodiments, in order to improve the detection accuracy of abnormal events in the train compartment, the embodiment of the present application calibrates the camera, and uses a calibration algorithm to determine the actual position of the detected object. First obtain the internal parameters of the camera, then calculate the translation vector of the camera model perpendicular to the car detection background, and then calculate the rotation matrix of the world coordinate system around the image coordinate system, combined with the actual position of the camera, the distance between the image coordinates and the world coordinates can be obtained coordinate transformation relationship. It should be noted that the position of the world coordinate is the position of the real space of the target.

在一些实施例中，在获取目标时采用可以检测多个目标的YOLOV7检测算法模型，该模型可以同时识别多个不同的目标，相较于其他检测算法模型，该模型在视频检测方面可以提供更加实时的检测信息，可以实时将监控视频内的信息传输到检测模型中。In some embodiments, the YOLOV7 detection algorithm model that can detect multiple targets is used when acquiring the target. This model can recognize multiple different targets at the same time. Compared with other detection algorithm models, this model can provide more in video detection. Real-time detection information, which can transmit the information in the surveillance video to the detection model in real time.

在一些实施例中，在选择检测模型时，为了最大限度保护用户数据的隐私，本申请选取了联邦学习模型来作为检测模型的检测框架。联邦学习也称为协同学习，它可以在产生数据的设备上进行大规模的训练，并且这些敏感数据保留在数据的所有者那里，本地收集、本地训练。在本地训练后，中央的训练协调器通过获取分布模型的更新获得每个节点的训练贡献，但是不访问实际的敏感数据。In some embodiments, when selecting a detection model, in order to protect the privacy of user data to the greatest extent, this application selects a federated learning model as the detection framework of the detection model. Federated learning is also called collaborative learning. It can perform large-scale training on the device that generates the data, and the sensitive data is kept with the data owner for local collection and local training. After local training, the central training coordinator obtains each node's training contribution by fetching updates to the distributed model, but does not have access to the actual sensitive data.

在一些实施例中，联邦学习有横向、纵向、迁移等不同的学习模型种类，横向联邦学习的本质是样本的联合，适用于参与者间业态相同但触达客户不同，即特征重叠多，用户重叠少时的场景；纵向联邦学习的本质是特征的联合，适用于用户重叠多，特征重叠少的场景；当参与者间特征和样本重叠都很少时可以考虑使用联邦迁移学习；本申请涉及列车车厢，由于不同的列车车厢相互重叠，所以本申请实施例以横向联邦学习为基准来进行模型训练。In some embodiments, federated learning has different types of learning models such as horizontal, vertical, and migration. The essence of horizontal federated learning is the combination of samples, which is suitable for participants with the same format but different customers, that is, there are many overlapping features, and users Scenarios with little overlap; the essence of vertical federated learning is the combination of features, which is suitable for scenarios with more user overlap and less feature overlap; federated transfer learning can be considered when there is little overlap between features and samples between participants; this application involves trains Cars, since different train cars overlap each other, the embodiment of the present application uses horizontal federated learning as a benchmark to perform model training.

请参阅图2，在一些实施例中，在步骤S101之前可以包括但不限于包括步骤S201至步骤S202：Please refer to Fig. 2, in some embodiments, before step S101 may include but not limited to include step S201 to step S202:

步骤S201，对不同的列车车厢场景进行定义；Step S201, defining different train carriage scenes;

步骤S202，确定不同的列车车厢场景对应车厢异常事件的种类。Step S202, determining the types of car abnormalities corresponding to different train car scenes.

在一些实施例中，利用相机拍摄需要检测的列车车厢场景，由于不同的检测场景对应不同的异常事件，所以要对不同的场景进行定义。In some embodiments, the camera is used to capture the scene of the train compartment to be detected. Since different detection scenes correspond to different abnormal events, different scenes need to be defined.

请参阅图3，在一些实施例中，步骤S103可以包括但不限于包括步骤S301至步骤S305：Referring to FIG. 3, in some embodiments, step S103 may include but not limited to include steps S301 to S305:

步骤S301，获取相机的内部参数；Step S301, acquiring internal parameters of the camera;

步骤S302，计算出相机模型垂直于车厢检测背景的平移向量；Step S302, calculating the translation vector of the camera model perpendicular to the car detection background;

步骤S303，计算世界坐标系绕图像坐标系的旋转矩阵；Step S303, calculating the rotation matrix of the world coordinate system around the image coordinate system;

步骤S304，结合相机的实际位置得到图像坐标和世界坐标之间的坐标转换关系；Step S304, combining the actual position of the camera to obtain the coordinate transformation relationship between the image coordinates and the world coordinates;

步骤S305，基于坐标转换关系得到检测目标的世界坐标。In step S305, the world coordinates of the detection target are obtained based on the coordinate transformation relationship.

在一些实施例中，为了提高列车车厢异常事件的检测精度，本申请实施例对相机进行了标定，通过标定算法来确定被检测目标的实际位置。首先获取到相机的内部参数，再计算出相机模型垂直于车厢检测背景的平移向量，再计算世界坐标系绕图像坐标系的旋转矩阵，结合相机的实际位置便可以得到图像坐标和世界坐标之间的坐标转换关系。需要说明的是，世界坐标的位置就是目标的现实空间的位置。In some embodiments, in order to improve the detection accuracy of abnormal events in the train compartment, the embodiment of the present application calibrates the camera, and uses a calibration algorithm to determine the actual position of the detected object. First obtain the internal parameters of the camera, then calculate the translation vector of the camera model perpendicular to the car detection background, and then calculate the rotation matrix of the world coordinate system around the image coordinate system, combined with the actual position of the camera, the distance between the image coordinates and the world coordinates can be obtained coordinate transformation relationship. It should be noted that the position of the world coordinate is the position of the real space of the target.

请参阅图4，在一些实施例中，步骤S105可以包括但不限于包括步骤S401至步骤S403：Referring to FIG. 4, in some embodiments, step S105 may include but not limited to include steps S401 to S403:

步骤S401，采用特征标注法对不同场景的检测目标行为数据进行标注；Step S401, using the feature labeling method to label the detection target behavior data in different scenes;

步骤S402，将标注的检测目标行为数据输入到联邦学习模型中，联邦学习模型输出隐私数据，在将加密后的检测目标行为数据输入到预设的车厢异常事件检测模型中进行算法训练，得到异常事件检测模型；Step S402, input the marked detection target behavior data into the federated learning model, and the federated learning model outputs private data, input the encrypted detection target behavior data into the preset car abnormality event detection model for algorithm training, and obtain abnormal event detection model;

步骤S403，将检测目标的实时视频数据输入到车厢异常事件检测模型中，以识别检测目标在视频中发生的动作。Step S403, inputting the real-time video data of the detection target into the abnormal event detection model in the compartment, so as to identify the actions of the detection target in the video.

在一些实施例中，本申请利用联邦学习算法模型来保护用户的隐私安全，同时采用特征标注法对不同场景的数据进行标注，最后将标注的乘客行为数据输入到联邦学习模型中，联邦学习模型输出隐私数据，在将加密后的数据输入到异常事件检测模型中，进行算法训练，最后得到异常事件检测模型，实时将发生的事件传递到相关部门，起到实时检测、实时预警的作用。In some embodiments, this application uses the federated learning algorithm model to protect the privacy of users, and uses the feature labeling method to mark the data of different scenarios, and finally inputs the marked passenger behavior data into the federated learning model, and the federated learning model Output private data, input the encrypted data into the abnormal event detection model, carry out algorithm training, and finally obtain the abnormal event detection model, and transmit the occurred events to relevant departments in real time, which plays the role of real-time detection and real-time early warning.

请参阅图5，在一些实施例中，车厢异常事件检测模型的训练方法可以包括但不限于包括步骤S501至步骤S504：Please refer to FIG. 5 , in some embodiments, the training method of the compartment abnormal event detection model may include but not limited to steps S501 to S504:

步骤S501，构建联邦学习模型；Step S501, constructing a federated learning model;

步骤S502，实地采集检测数据；Step S502, collecting detection data on the spot;

步骤S503，确定需要检测的车厢异常事件种类；Step S503, determining the type of abnormal event in the car that needs to be detected;

步骤S504，通过联邦学习训练联邦学习模型，直至达到目标检测精度，得到车厢异常事件检测模型。In step S504, the federated learning model is trained through federated learning until the target detection accuracy is achieved, and an abnormal event detection model of the compartment is obtained.

在一些实施例中，本申请选取联邦学习模型来作为检测模型的检测框架，构建完毕检测模型后，将数据输入联邦学习模型中进行数据训练，得到检测模型的检测精度。在对模型进行数据训练时，数据的80％用来训练，20％用来测试，如果达不到需要的检测精度，那么继续通过数据来训练模型，直到达到要求的精度。In some embodiments, the application selects the federated learning model as the detection framework of the detection model. After the detection model is constructed, data is input into the federated learning model for data training to obtain the detection accuracy of the detection model. When training the model with data, 80% of the data is used for training and 20% for testing. If the required detection accuracy cannot be achieved, then continue to use the data to train the model until the required accuracy is achieved.

在一些实施例中，步骤S106可以包括但不限于包括步骤S601至步骤S605：In some embodiments, step S106 may include, but is not limited to, step S601 to step S605:

步骤S601，检测是否出现某一列车车厢出现人员聚集现象时，通过统计列车车厢的人员数量来确定某一车厢是否出现人员聚集，在人员数量超过预设的数量阈值时，则判断存在人员聚集的车厢异常事件；Step S601, when detecting whether there is a gathering of people in a certain train car, determine whether there is a crowd of people in a certain car by counting the number of people in the train car, and when the number of people exceeds the preset number threshold, it is judged that there is a gathering of people Abnormal events in the carriage;

步骤S602，检测列车车厢是否有人员跌倒的情况时，如果标注的人体肢体接触地面的时间超过预设的时间阈值，则判断存在人员跌倒的车厢异常事件；Step S602, when detecting whether there is a person falling in the train car, if the marked human limbs contact the ground for more than a preset time threshold, it is judged that there is an abnormal event in the car that the person fell;

步骤S603，检测列车车厢是否有人员发生肢体冲突，在检测人员是否发生肢体冲突时，标记冲突的基本动作，在满足冲突的基本动作的情况下，则判断存在人员发生肢体冲突的车厢异常事件；Step S603, detecting whether there is a physical conflict between persons in the train compartment, marking the basic movement of the conflict when detecting whether there is a physical conflict between the personnel, and judging that there is an abnormal event in the carriage in which the physical conflict occurs;

步骤S604，检测列车车厢行李架部位是否有易脱落的行李物品，确定列车车架的位置以及列车车架摆放行李的范围，当摆放的行李超过行李架的位置时，则判断存在超过行李架位置的行李物品的车厢异常事件；并检测列车行李架物品的种类，如果检测到不属于行李架摆放的物品种类，则判断存在不属于行李架摆放的物品种类的的车厢异常事件；Step S604: Detect whether there are luggage items that are easy to fall off in the luggage rack of the train compartment, determine the position of the train rack and the range of luggage placed on the train rack, and when the luggage placed exceeds the position of the luggage rack, it is judged that there is excess luggage. Carriage abnormal events of luggage items at the rack position; and detect the type of luggage rack items on the train. If it is detected that it does not belong to the type of items placed on the luggage rack, it is judged that there is an abnormal event in the compartment that does not belong to the type of items placed on the luggage rack;

步骤S605，检测列车车厢过道是否有大件物品阻塞通道时，首先通过场景定义确定列车车厢通道的位置，然后通过检测算法检测存在列车通道的物品，当出现阻塞通道的物品时，则判断存在大件物品阻塞列车车厢通道的车厢异常事件。Step S605, when detecting whether there is a large item blocking the aisle of the train carriage, first determine the position of the aisle of the train carriage through the scene definition, and then detect the items existing in the aisle of the train through the detection algorithm, and when there is an item blocking the aisle, it is judged that there is a large item blocking the aisle. An abnormal event in a carriage where an item blocks the passage of a train carriage.

在一些实施例中，检测是否出现某一列车车厢出现人员聚集现象异常事件，可以通过统计不同列车车厢的人员数量来确定某一车厢是否出现人员聚集，由于不同列车的载客量以及不同列车的售票情况不同，需要更据实际运行情况来确定是否出现人员聚集情况。In some embodiments, to detect whether there is an abnormal event of people gathering in a certain train car, it can be determined whether a certain car has a crowd by counting the number of people in different train cars. Due to the passenger capacity of different trains and the number of different trains The situation of ticket sales is different, and it is necessary to determine whether there is a gathering of people according to the actual operation situation.

在一些实施例中，在检测列车车厢是否有人员跌倒的情况时，由于人体跌倒后的形状不同，为了提高检测的精度，可以通过人工加入标注的数据来增加模型的准确性。人工标注数据指的是标注人体的上下肢，以及人体的头颅，确定人体下肢接触地面的时间阈值，如果超过了设定的时间阈值，则判断为跌倒。In some embodiments, when detecting whether a person falls in a train carriage, since the human body has a different shape after the fall, in order to improve the accuracy of the detection, the accuracy of the model can be increased by manually adding labeled data. Manually labeling data refers to labeling the upper and lower limbs of the human body, as well as the head of the human body, and determining the time threshold for the lower limbs of the human body to touch the ground. If it exceeds the set time threshold, it will be judged as a fall.

在一些实施例中，检测列车车厢是否有人员发生肢体冲突，在检测人员是否发生肢体冲突时，首先要标注人体的手臂以及人体的下肢，而且要标注特定的动作，因为有时候不同个体之间的肢体触碰不是冲突的表现，所以需要标记冲突的基本动作，在满足冲突的基本动作后确定该人员发生肢体冲突。In some embodiments, to detect whether there is a physical conflict between people in the train car, when detecting whether there is a physical conflict between people, firstly, the arms and lower limbs of the human body must be marked, and specific actions must be marked, because sometimes different individuals Physical contact is not a manifestation of conflict, so it is necessary to mark the basic movements of the conflict, and determine that the person has a physical conflict after satisfying the basic movements of the conflict.

在一些实施例中，检测列车车厢行李架部位是否有易脱落的行李物品，首先确定列车车架的位置以及列车车架可以摆放行李的范围，当摆放的行李超过行李架的位置时，检测算法会检测到超过行李架位置的行李物品，并提醒列车管理人员，同时也可以检测列车行李架物品的种类，如果检测到不属于行李架摆放的物品种类，也会及时发出警报。In some embodiments, to detect whether there are luggage items that are easy to fall off on the luggage rack of the train compartment, first determine the position of the train frame and the range where the luggage can be placed on the train rack. When the luggage placed exceeds the position of the luggage rack, The detection algorithm will detect luggage items that exceed the position of the luggage rack and remind the train management personnel. At the same time, it can also detect the type of luggage rack items on the train. If it detects that it does not belong to the type of luggage rack, it will also issue an alarm in time.

在一些实施例中，检测列车车厢过道是否有大件物品阻塞通道时，首先通过场景定义，确定列车车厢通道的位置，然后通过检测算法检测存在列车通道的物品，当出现阻塞通道的物品时，列车会发出警报，由于不同车型列车车厢的尺寸以及用途不同，需要根据实际情况来判断该物品是否堵塞通道，所以首先要通过场景标记来确定列车车厢的实际使用情况。In some embodiments, when detecting whether there is a large item blocking the passage in the aisle of the train carriage, the position of the aisle of the train carriage is first determined through the scene definition, and then the items in the aisle of the train passage are detected by a detection algorithm. When there is an article blocking the passage, The train will issue an alarm. Due to the different sizes and uses of the train cars of different models, it is necessary to judge whether the item is blocking the passage according to the actual situation. Therefore, the actual use of the train cars must first be determined through the scene mark.

需要指出的是，不同的场景对应的异常事件包括但不限于：统计不同列车车厢的人数，看是否出现某一列车车厢出现人员聚集现象、检测列车车厢是否有人员跌倒、是否有人员发生肢体冲突、列车车厢连接处是否出现人员聚集、列车车厢行李架部位是否有易脱落的行李物品、列车车厢过道是否有大件物品阻塞通道等。It should be pointed out that the abnormal events corresponding to different scenarios include but are not limited to: counting the number of people in different train cars, checking whether there is a phenomenon of people gathering in a certain train car, detecting whether there is a person falling in the train car, and whether there is a physical conflict between people , Whether there is a gathering of people at the junction of the train carriages, whether there are luggage items that are easy to fall off on the luggage rack of the train carriages, whether there are large objects blocking the passage in the aisles of the train carriages, etc.

在一些实施例中，在步骤S106之后可以包括但不限于包括步骤S107：In some embodiments, after step S106, step S107 may be included but not limited to:

步骤S107，对车厢异常事件进行预警和上报。Step S107, giving early warning and reporting of abnormal events in the carriage.

在一些实施例中，通过机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作；实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够在列车车厢场景出现异常行驶情况时及时进行预警和上报，确保列车的运行安全。In some embodiments, through machine vision technology and federated learning model, while performing video analysis on the scene of the train carriage, the target inside the train carriage is captured, the world coordinates of the target are obtained through the calibration algorithm, and the federated learning algorithm model is used to protect the target While maintaining the privacy of the video information, it detects the actions of the target in the video; it realizes a machine vision-based detection method for abnormal events in the train carriage from the perspective of video, which can detect abnormal events in the train carriage scene in time. Early warning and reporting to ensure the safety of train operation.

基于此，本申请提出一种采用机器视觉检测多种列车车厢室内异常事件的方法，该方法采用联邦学习算法，通过联邦学习模型对用户的数据进行加密来保证用户数据的隐私安全；在检测不同场景时对不同的场景进行了标注，可以实现不同场景不同种类的异常事件检测。本申请实施例还建立针对列车车厢室内异常事件检测的算法模型，通过摄像机拍摄结合本申请提出的算法模型来实现不同场景下不同种类的列车车厢室内的异常事件检测。Based on this, this application proposes a method of using machine vision to detect abnormal events in various train compartments. The method uses a federated learning algorithm to encrypt user data through a federated learning model to ensure the privacy and security of user data; Different scenarios are marked in the scene, which can realize the detection of different types of abnormal events in different scenarios. The embodiment of the present application also establishes an algorithm model for detecting abnormal events in train compartments, and realizes detection of abnormal events in different types of train compartments in different scenarios through camera shooting combined with the algorithm model proposed in this application.

请参阅图6，本申请实施例还提供一种车厢异常事件检测装置，可以实现上述车厢异常事件检测方法，该装置包括：Please refer to FIG. 6. The embodiment of the present application also provides a device for detecting an abnormal event in a carriage, which can implement the above method for detecting an abnormal event in a carriage. The device includes:

获取模块610，用于获取相机拍摄列车车厢场景的视频；Acquisition module 610, for obtaining the video of camera shooting train compartment scene;

确定模块620，用于确定列车车厢场景内的检测目标；Determiningmodule 620, used to determine the detection target in the train compartment scene;

标定模块630，用于通过标定算法得到检测目标的世界坐标；Acalibration module 630, configured to obtain the world coordinates of the detection target through a calibration algorithm;

加密模块640，用于通过联邦学习算法模型对检测目标的视频隐私信息进行加密；Theencryption module 640 is used to encrypt the video privacy information of the detection target through the federated learning algorithm model;

识别模块650，用于识别检测目标在视频中发生的动作；Arecognition module 650, configured to recognize the action of the detection target in the video;

判断模块660，用于根据动作判断是否存在车厢异常事件。A judgingmodule 660, configured to judge whether there is an abnormal event in the compartment according to the action.

基于此，本申请实施例的车厢异常事件检测装置，获取模块610获取相机拍摄列车车厢场景的视频；确定模块620确定列车车厢场景内的检测目标；标定模块630通过标定算法得到检测目标的世界坐标；加密模块640通过联邦学习算法模型对检测目标的视频隐私信息进行加密；识别模块650识别检测目标在视频中发生的动作；判断模块660根据动作判断是否存在车厢异常事件。本申请实施例通过获取相机拍摄列车车厢场景的视频；确定列车车厢场景内的检测目标；通过标定算法得到检测目标的世界坐标；通过联邦学习模型对检测目标的视频隐私信息进行加密；识别检测目标在视频中发生的动作；根据动作判断是否存在车厢异常事件。基于此，本申请实施例相比于现有的人力检测方法，采用机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作，从而实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够及时判断在列车车厢场景出现的异常事件，以确保列车的运行安全。因此，本申请实施例通过采用联邦学习算法模型来达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。Based on this, in the vehicle abnormal event detection device of the embodiment of the present application, theacquisition module 610 acquires the video of the train compartment scene captured by the camera; thedetermination module 620 determines the detection target in the train compartment scene; thecalibration module 630 obtains the world coordinates of the detection target through a calibration algorithm Theencryption module 640 encrypts the video privacy information of the detection target through the federated learning algorithm model; theidentification module 650 identifies the actions of the detection target in the video; thejudgment module 660 judges whether there is an abnormal event in the carriage according to the action. In the embodiment of the present application, the camera captures the video of the train compartment scene; determines the detection target in the train compartment scene; obtains the world coordinates of the detection target through a calibration algorithm; encrypts the video privacy information of the detection target through a federated learning model; identifies the detection target Actions that occur in the video; judge whether there is an abnormal event in the compartment based on the action. Based on this, compared with the existing human detection method, the embodiment of the present application adopts machine vision technology and federated learning model to capture the target inside the train compartment while performing video analysis on the scene of the train compartment, and obtain the target through the calibration algorithm The world coordinates of the target, using the federated learning algorithm model to protect the privacy of the target's video information, while detecting the target's actions in the video, thus realizing a machine vision-based detection method for abnormal events in train carriages from the perspective of video, It can timely judge the abnormal events in the train compartment scene to ensure the safety of the train operation. Therefore, the embodiment of the present application uses a federated learning algorithm model to reduce the risk of user privacy leakage and at the same time detect abnormal events in the car scene.

该车厢异常事件检测装置的具体实施方式与上述车厢异常事件检测方法的具体实施例基本相同，在此不再赘述。The specific implementation of the device for detecting an abnormal event in a carriage is basically the same as the specific embodiment of the method for detecting an abnormal event in a carriage, and will not be repeated here.

本申请实施例还提供了一种电子设备，电子设备包括存储器和处理器，存储器存储有计算机程序，处理器执行计算机程序时实现上述车厢异常事件检测方法。该电子设备可以为包括平板电脑、车载电脑等任意智能终端。The embodiment of the present application also provides an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the above method for detecting an abnormal event in the compartment is implemented. The electronic device may be any intelligent terminal including a tablet computer, a vehicle-mounted computer, and the like.

请参阅图7，图7示意了另一实施例的电子设备的硬件结构，电子设备包括：Please refer to FIG. 7. FIG. 7 illustrates a hardware structure of an electronic device in another embodiment. The electronic device includes:

处理器701，可以采用通用的CPU(CentralProcessingUnit，中央处理器)、微处理器、应用专用集成电路(ApplicationSpecificIntegratedCircuit，ASIC)、或者一个或多个集成电路等方式实现，用于执行相关程序，以实现本申请实施例所提供的技术方案。Theprocessor 701 may be implemented by a general-purpose CPU (Central Processing Unit, central processing unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute related programs to realize The technical solutions provided by the embodiments of the present application.

存储器702，可以采用只读存储器(ReadOnlyMemory，ROM)、静态存储设备、动态存储设备或者随机存取存储器(RandomAccessMemory，RAM)等形式实现。存储器702可以存储操作系统和其他应用程序，在通过软件或者固件来实现本说明书实施例所提供的技术方案时，相关的程序代码保存在存储器702中，并由处理器701来调用执行本申请实施例的车厢异常事件检测方法，即通过获取相机拍摄列车车厢场景的视频；确定列车车厢场景内的检测目标；通过标定算法得到检测目标的世界坐标；通过联邦学习模型对检测目标的视频隐私信息进行加密；识别检测目标在视频中发生的动作；根据动作判断是否存在车厢异常事件。基于此，本申请实施例相比于现有的人力检测方法，采用机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作，从而实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够及时判断在列车车厢场景出现的异常事件，以确保列车的运行安全。因此，本申请实施例通过采用联邦学习算法模型来达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。Thememory 702 may be implemented in the form of a read-only memory (ReadOnlyMemory, ROM), a static storage device, a dynamic storage device, or a random access memory (RandomAccessMemory, RAM). Thememory 702 can store operating systems and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in thememory 702 and called by theprocessor 701 to execute the implementation of the present application. In this example, the method for detecting an abnormal event in a carriage is to capture the video of the train carriage scene by the camera; to determine the detection target in the train carriage scene; to obtain the world coordinates of the detection target through a calibration algorithm; Encryption; identify the action of the detection target in the video; judge whether there is an abnormal event in the car according to the action. Based on this, compared with the existing human detection method, the embodiment of the present application adopts machine vision technology and federated learning model to capture the target inside the train compartment while performing video analysis on the scene of the train compartment, and obtain the target through the calibration algorithm The world coordinates of the target, using the federated learning algorithm model to protect the privacy of the target's video information, while detecting the target's actions in the video, thus realizing a machine vision-based detection method for abnormal events in train carriages from the perspective of video, It can timely judge the abnormal events in the train compartment scene to ensure the safety of the train operation. Therefore, the embodiment of the present application uses a federated learning algorithm model to reduce the risk of user privacy leakage and at the same time detect abnormal events in the car scene.

输入/输出接口703，用于实现信息输入及输出。The input/output interface 703 is used to realize information input and output.

通信接口704，用于实现本设备与其他设备的通信交互，可以通过有线方式(例如USB、网线等)实现通信，也可以通过无线方式(例如移动网络、WIFI、蓝牙等)实现通信。Thecommunication interface 704 is used to realize the communication and interaction between this device and other devices, and the communication can be realized through a wired method (such as USB, network cable, etc.), or can be realized through a wireless method (such as a mobile network, WIFI, Bluetooth, etc.).

总线，在设备的各个组件(例如处理器701、存储器702、输入/输出接口703和通信接口704)之间传输信息。A bus that transfers information between various components of the device (eg,processor 701,memory 702, input/output interface 703, and communication interface 704).

其中处理器701、存储器702、输入/输出接口703和通信接口704通过总线实现彼此之间在设备内部的通信连接。Theprocessor 701 , thememory 702 , the input/output interface 703 and thecommunication interface 704 are connected to each other within the device through a bus.

本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序，该计算机程序被处理器执行时实现上述车厢异常事件检测方法。The embodiment of the present application also provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the above method for detecting an abnormal event in a carriage is implemented.

存储器作为一种非暂态计算机可读存储介质，可用于存储非暂态软件程序以及非暂态性计算机可执行程序。此外，存储器可以包括高速随机存取存储器，还可以包括非暂态存储器，例如至少一个磁盘存储器件、闪存器件、或其他非暂态固态存储器件。在一些实施方式中，存储器可选包括相对于处理器远程设置的存储器，这些远程存储器可以通过网络连接至该处理器。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。As a non-transitory computer-readable storage medium, memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

本申请实施例提供的车厢异常事件检测方法、车厢异常事件检测装置、电子设备及存储介质，通过获取相机拍摄列车车厢场景的视频；确定列车车厢场景内的检测目标；通过标定算法得到检测目标的世界坐标；通过联邦学习模型对检测目标的视频隐私信息进行加密；识别检测目标在视频中发生的动作；根据动作判断是否存在车厢异常事件。基于此，本申请实施例相比于现有的人力检测方法，采用机器视觉技术以及联邦学习模型，在对列车车厢场景进行视频分析的同时，捕捉到列车车厢内部的目标，通过标定算法得到目标的世界坐标，利用联邦学习算法模型保护目标的视频信息隐私的同时，检测到目标在视频中发生的动作，从而实现了从视频角度出发的一种基于机器视觉的列车车厢异常事件的检测方法，能够及时判断在列车车厢场景出现的异常事件，以确保列车的运行安全。因此，本申请实施例通过采用联邦学习算法模型来达到降低用户隐私泄露的风险的同时检测车厢场景下的异常事件。The method for detecting an abnormal event in a carriage, the device for detecting an abnormal event in a carriage, the electronic device, and the storage medium provided in the embodiment of the present application obtain a camera to take a video of a train carriage scene; determine the detection target in the train carriage scene; obtain the detection target through a calibration algorithm World coordinates; Encrypt the video privacy information of the detection target through the federated learning model; Identify the actions of the detection target in the video; Judge whether there is an abnormal event in the car according to the action. Based on this, compared with the existing human detection method, the embodiment of the present application adopts machine vision technology and federated learning model to capture the target inside the train compartment while performing video analysis on the scene of the train compartment, and obtain the target through the calibration algorithm The world coordinates of the target, using the federated learning algorithm model to protect the privacy of the target's video information, while detecting the target's actions in the video, thus realizing a machine vision-based detection method for abnormal events in train carriages from the perspective of video, It can timely judge the abnormal events in the train compartment scene to ensure the safety of the train operation. Therefore, the embodiment of the present application uses a federated learning algorithm model to reduce the risk of user privacy leakage and at the same time detect abnormal events in the car scene.

本领域普通技术人员可以理解，上文中所公开方法中的全部或某些步骤、系统可以被实施为软件、固件、硬件及其适当的组合。某些物理组件或所有物理组件可以被实施为由处理器，如中央处理器、数字信号处理器或微处理器执行的软件，或者被实施为硬件，或者被实施为集成电路，如专用集成电路。这样的软件可以分布在计算机可读介质上，计算机可读介质可以包括计算机存储介质(或非暂时性介质)和通信介质(或暂时性介质)。如本领域普通技术人员公知的，术语计算机存储介质包括在用于存储信息(诸如计算机可读程序、数据结构、程序模块或其他数据)的任何方法或技术中实施的易失性和非易失性、可移除和不可移除介质。计算机存储介质包括但不限于RAM、ROM、EEPROM、闪存或其他存储器技术、CD-ROM、数字多功能盘(DVD)或其他光盘存储、磁盒、磁带、磁盘存储或其他磁存储装置、或者可以用于存储期望的信息并且可以被计算机访问的任何其他的介质。此外，本领域普通技术人员公知的是，通信介质通常包含计算机可读程序、数据结构、程序模块或者诸如载波或其他传输机制之类的调制数据信号中的其他数据，并且可包括任何信息递送介质。Those skilled in the art can understand that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware and an appropriate combination thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit . Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those of ordinary skill in the art, the term computer storage media includes both volatile and nonvolatile media implemented in any method or technology for storage of information, such as computer readable programs, data structures, program modules, or other data. permanent, removable and non-removable media. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cartridges, tape, magnetic disk storage or other magnetic storage devices, or can Any other medium used to store desired information and which can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable programs, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media .

本申请实施例描述的实施例是为了更加清楚的说明本申请实施例的技术方案，并不构成对于本申请实施例提供的技术方案的限定，本领域技术人员可知，随着技术的演变和新应用场景的出现，本申请实施例提供的技术方案对于类似的技术问题，同样适用。The embodiments described in the embodiments of the present application are to illustrate the technical solutions of the embodiments of the present application more clearly, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of technology and new For the emergence of application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

本领域技术人员可以理解的是，图中示出的技术方案并不构成对本申请实施例的限定，可以包括比图示更多或更少的步骤，或者组合某些步骤，或者不同的步骤。Those skilled in the art can understand that the technical solution shown in the figure does not constitute a limitation to the embodiment of the present application, and may include more or less steps than those shown in the figure, or combine some steps, or different steps.

以上所描述的装置实施例仅仅是示意性的，其中作为分离部件说明的单元可以是或者也可以不是物理上分开的，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

本领域普通技术人员可以理解，上文中所公开方法中的全部或某些步骤、系统、设备中的功能模块/单元可以被实施为软件、固件、硬件及其适当的组合。Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, the functional modules/units in the system, and the device can be implemented as software, firmware, hardware, and an appropriate combination thereof.

本申请的说明书及上述附图中的术语“第一”、“第二”、“第三”、“第四”等(如果存在)是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。The terms "first", "second", "third", "fourth", etc. (if any) in the description of the present application and the above drawings are used to distinguish similar objects and not necessarily to describe specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to the expressly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

应当理解，在本申请中，“至少一个(项)”是指一个或者多个，“多个”是指两个或两个以上。“和/或”，用于描述关联对象的关联关系，表示可以存在三种关系，例如，“A和/或B”可以表示：只存在A，只存在B以及同时存在A和B三种情况，其中A，B可以是单数或者复数。字符“/”一般表示前后关联对象是一种“或”的关系。“以下至少一项(个)”或其类似表达，是指这些项中的任意组合，包括单项(个)或复数项(个)的任意组合。例如，a，b或c中的至少一项(个)，可以表示：a，b，c，“a和b”，“a和c”，“b和c”，或“a和b和c”，其中a，b，c可以是单个，也可以是多个。It should be understood that in this application, "at least one (item)" means one or more, and "multiple" means two or more. "And/or" is used to describe the association relationship of associated objects, indicating that there can be three types of relationships, for example, "A and/or B" can mean: only A exists, only B exists, and A and B exist at the same time , where A and B can be singular or plural. The character "/" generally indicates that the contextual objects are an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one item (piece) of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c ", where a, b, c can be single or multiple.

在本申请所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，上述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。In the several embodiments provided in this application, it should be understood that the disclosed devices and methods may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the above units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or can be Integrate into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

上述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

另外，在本申请各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括多指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例的方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，简称ROM)、随机存取存储器(Random Access Memory，简称RAM)、磁碟或者光盘等各种可以存储程序的介质。If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including multiple instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM for short), random access memory (Random Access Memory, RAM for short), magnetic disk or optical disk, etc., which can store programs. medium.

以上参照附图说明了本申请实施例的优选实施例，并非因此局限本申请实施例的权利范围。本领域技术人员不脱离本申请实施例的范围和实质内所作的任何修改、等同替换和改进，均应在本申请实施例的权利范围之内。The preferred embodiments of the embodiments of the present application have been described above with reference to the accompanying drawings, which does not limit the scope of rights of the embodiments of the present application. Any modifications, equivalent replacements and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall fall within the scope of rights of the embodiments of the present application.

Claims (10)

1. A car anomaly detection method, the method comprising:

acquiring a video of a scene of a train carriage shot by a camera;

determining a detection target in a train carriage scene;

obtaining world coordinates of the detection target through a calibration algorithm;

encrypting the video privacy information of the detection target through a federal learning model;

identifying an action of the detection target occurring in the video;

and judging whether a carriage abnormal event exists according to the action.

2. The method of claim 1, further comprising, prior to the capturing video of the railcar scene by the camera:

defining different ones of said railcar scenes;

and determining the types of the abnormal events of the corresponding carriage of different train carriage scenes.

3. The method according to claim 1, wherein the obtaining world coordinates of the detection target by a calibration algorithm includes:

acquiring internal parameters of the camera;

calculating a translation vector of the camera model perpendicular to the carriage detection background;

calculating a rotation matrix of the world coordinate system around the image coordinate system;

combining the actual position of the camera to obtain a coordinate conversion relation between the image coordinate and the world coordinate;

And obtaining world coordinates of the detection target based on the coordinate conversion relation.

4. The method of claim 1, wherein the act of identifying the detection target as occurring in video comprises:

marking the detection target behavior data of different scenes by adopting a characteristic marking method;

inputting the marked detection target behavior data into a federal learning model, outputting privacy data by the federal learning model, and inputting the encrypted detection target behavior data into a preset carriage abnormal event detection model for algorithm training to obtain an abnormal event detection model;

and inputting the real-time video data of the detection target into the carriage abnormal event detection model to identify the action of the detection target in the video.

5. The method of claim 4, wherein the training method of the car anomaly detection model is as follows:

constructing a federal learning model;

collecting detection data in the field;

determining the types of abnormal events of the carriage to be detected;

training the federal learning model through federal learning until reaching target detection precision, and obtaining the carriage abnormal event detection model.

6. The method of claim 1, wherein said determining whether a car anomaly exists based on said action comprises:

when detecting whether a certain train carriage has a person gathering phenomenon, determining whether the certain carriage has the person gathering by counting the number of the persons in the train carriage, and judging that a carriage abnormal event of the person gathering exists when the number of the persons exceeds a preset number threshold;

when detecting whether a person falls down in a carriage of the train, if the time of the marked human body limb contacting the ground exceeds a preset time threshold, judging that a carriage abnormal event of the falling of the person exists;

detecting whether a train carriage has a limb conflict or not, marking basic actions of the conflict when detecting whether the limb conflict occurs to the personnel, and judging that a carriage abnormal event of the limb conflict occurs to the personnel under the condition that the basic actions of the conflict are met;

detecting whether a luggage rack part of a train carriage has luggage articles which are easy to fall off, determining the position of a train frame and the range of the train frame for placing luggage, and judging that the carriage abnormal event of the luggage articles exceeding the position of the luggage rack exists when the placed luggage exceeds the position of the luggage rack; detecting the types of articles of the luggage rack of the train, and judging that a carriage abnormal event which does not belong to the types of articles placed on the luggage rack exists if the types of articles which do not belong to the luggage rack are detected;

When detecting whether a large article blocks a passage or not in a train carriage aisle, firstly determining the position of the train carriage aisle through scene definition, then detecting articles in the train carriage aisle through a detection algorithm, and judging that a carriage abnormal event of the large article blocking the train carriage aisle exists when the articles blocking the passage occur.

7. The method of claim 1, further comprising, after said determining from said action whether a car anomaly event exists:

and carrying out early warning and reporting on the carriage abnormal event.

8. A car abnormal event detection device, characterized in that the device comprises:

the acquisition module is used for acquiring videos of the scene of the train carriage shot by the camera;

a determination module for determining a detection target within a train car scene;

the calibration module is used for obtaining world coordinates of the detection target through a calibration algorithm;

the encryption module is used for encrypting the video privacy information of the detection target through the federal learning algorithm model;

the identification module is used for identifying the action of the detection target in the video;

and the judging module is used for judging whether a carriage abnormal event exists according to the action.

9. An electronic device comprising a memory storing a computer program and a processor that when executing the computer program implements the car anomaly event detection method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the cabin abnormal event detection method according to any one of claims 1 to 7.

Images