WO2022105609A1 - 一种高空抛物检测方法、装置、计算机设备及存储介质 - Google Patents

一种高空抛物检测方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2022105609A1
WO2022105609A1 PCT/CN2021/128478 CN2021128478W WO2022105609A1 WO 2022105609 A1 WO2022105609 A1 WO 2022105609A1 CN 2021128478 W CN2021128478 W CN 2021128478W WO 2022105609 A1 WO2022105609 A1 WO 2022105609A1
Authority
WO
WIPO (PCT)
Prior art keywords
altitude
moving
classification
lstm
altitude parabolic
Prior art date
Application number
PCT/CN2021/128478
Other languages
English (en)
French (fr)
Inventor
徐光耀
季翔宇
Original Assignee
中科智云科技有限公司
上海点泽智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中科智云科技有限公司, 上海点泽智能科技有限公司 filed Critical 中科智云科技有限公司
Priority to US18/265,870 priority Critical patent/US20240062518A1/en
Publication of WO2022105609A1 publication Critical patent/WO2022105609A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
    • G06V10/993—Evaluation of the quality of the acquired pattern
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/02—Alarms for ensuring the safety of persons

Definitions

  • the present application relates to the technical field of image recognition, and in particular, to a high-altitude parabolic detection method, device, computer equipment and storage medium.
  • High-altitude projectiles and falling objects have occurred from time to time.
  • High-altitude projectiles are an uncivilized behavior, which not only pollutes the environment, but also seriously endangers the public safety of residents and disturbs social peace.
  • the specific location of the perpetrator is located from many households so as to be held accountable, and the pedestrians on the ground cannot be reminded to avoid in time, resulting in frequent high-altitude parabolic safety accidents. Therefore, if the high-altitude parabolic floor can be accurately and timely located, and the early warning and interception can be carried out in time, it can greatly protect the safety of people's life and property, which has always been a problem that the society is concerned about and needs to be solved urgently.
  • a computer vision-based high-altitude parabola detection method has appeared in the related art.
  • a Chinese patent (CN111476973A) provides a method for detecting high-altitude parabolas using an infrared sensing system, but the infrared sensing system is expensive and difficult to implement;
  • Chinese patent (CN205982657U) A high-altitude parabolic alarm method based on ultrasonic reflection detection is provided, but the system is expensive, difficult to deploy, easily interfered by the surrounding environment, and not accurate enough.
  • the present invention proposes a kind of high-altitude parabola detection method, comprising the following steps:
  • the detected moving object is tracked by means of Kalman filtering, and the trajectory characteristics and related parameters of the moving object are obtained, including the trajectory curve trend, the object movement acceleration, the frame intersection ratio before and after the object, and the object shape and pixel size change;
  • Rule-based high-altitude parabola recognition to determine whether the recognition is a high-altitude parabola
  • the long short-term memory network (LSTM) classification model is used for classification, and parameters such as trajectory curve trend, object moving acceleration, frame intersection ratio before and after the object, object shape, and pixel size change are input into the LSTM as feature data to obtain the classification results, and determine whether for false positives;
  • the alarm information is pushed to the monitoring center.
  • the rule-based high-altitude parabola identification adopts the setting of the outer contour line of the building and the special position line. Once the trajectory of the moving object crosses the contour line from the inside to the outside, or crosses the special position line from the top to the bottom, it can be A parabola is thought to have occurred.
  • the background modeling algorithm is a single Gaussian model method or a mixed Gaussian model method to detect the moving object.
  • the embodiment of the present application also provides a high-altitude parabolic detection device, including:
  • the camera module is used to collect target video images in real time
  • the detection module is used to detect all moving objects in the target video image by using the background modeling algorithm
  • the tracking module is used to track the detected moving object by means of Kalman filtering, and obtain the trajectory characteristics and related parameters of the moving object, including the trajectory curve trend, the acceleration of the movement of the object, the ratio of the frame before and after the object, the shape and pixel size changes;
  • the identification module is used for rule-based high-altitude parabola recognition, and judges and identifies whether the moving object is a high-altitude parabola;
  • the filtering module is used for classification using the long short-term memory network (LSTM) classification model.
  • LSTM long short-term memory network
  • the parameters such as trajectory curve trend, object moving acceleration, frame intersection ratio before and after the object, object shape, and pixel size change are input into LSTM as feature data to obtain. Classification results to determine whether it is a false positive;
  • the alarm module is used to push alarm information to the monitoring center.
  • Embodiments of the present application further provide a computer device, including a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, when the processor executes the computer program, the processor executes the following steps: method described above.
  • Embodiments of the present application further provide a storage medium on which a computer program is stored, characterized in that, when the computer program is run by a processor, the method as described above is executed.
  • the high-altitude parabola detection method used in this application first finds out the trajectory of the moving object to be judged; then analyzes and filters all the trajectories to find the trajectory of the high-altitude parabola in the scene; and then issues an alarm to the location where the high-altitude parabola occurs. ; Finally, by filtering out occasional false detections and missed detections, false positives are reduced.
  • the method applied in the present application has high real-time performance and low missed detection rate, and can not only remotely schedule on-site monitoring in real time, but also obtain alarm pictures and accurate locations where alarm events occur.
  • FIG. 1 is a schematic flowchart of a method for detecting high-altitude parabola provided by an embodiment of the present application
  • FIG. 2 is a schematic diagram of a classification flow of a long short-term memory network (LSTM) classification model provided by an embodiment of the present application;
  • LSTM long short-term memory network
  • FIG. 3 is a schematic structural diagram of a high-altitude parabolic detection device provided by an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of a computer device provided by an embodiment of the present application.
  • a high-altitude parabolic detection method comprises the following steps:
  • Step S110 setting a video surveillance area, and collecting target video images in real time
  • Step S120 use a background modeling algorithm to detect all moving objects in the target video image
  • a single Gaussian model method, a mixed Gaussian model method, or an optical flow method can be used to detect moving objects.
  • the single-Gaussian model method calculates the mean and variance of N samples of pixel values in N frames of video images for fixed pixels in multiple video frames, and the single-Gaussian background model can be uniquely determined by using the mean and variance.
  • the subtracted value is compared with the threshold (three times the variance) to judge the foreground or background and determine whether there is a moving object.
  • the Gaussian mixture model extracts N frames of video images in turn, and iteratively models each pixel each time. Comparing the current picture with the assumed static background and finding an area with obvious changes, it can be considered that there is a moving object in this area.
  • the optical flow method detects moving objects, and dynamically analyzes the image according to the velocity vector characteristics of each pixel in the video frame. If there is no moving target in the image, the optical flow vector changes continuously in the entire image area, and when there is relative motion between the object and the image background, the velocity vector formed by the moving object must be different from the velocity of the neighborhood background. vector to detect the position of the moving object. Since the optical flow method needs to perform iterative operations, the higher the precision, the greater the amount of computation.
  • the single Gaussian model method, the mixed Gaussian model method or the optical flow method for detecting moving objects are in the prior art, and details are not described in this embodiment.
  • Step S130 use Kalman filtering to track the detected moving object, and obtain the trajectory characteristics and related parameters of the moving object, including trajectory curve trend, object movement acceleration, frame intersection ratio before and after the object, object shape and pixel size changes;
  • the present application adopts the Kalman filtering algorithm to achieve accurate tracking of the moving target, and obtain the motion trajectory of the object in multiple frames.
  • Step S140 based on the rule-based high-altitude parabola recognition, determine whether the recognition is a high-altitude parabola;
  • the rule-based high-altitude parabola recognition adopts the setting of the outer contour line and special position line of the building. Once the trajectory of the moving object crosses the contour line from the inside to the outside, or crosses the special position line from the top to the bottom, it can be considered that the occurrence has occurred. high-altitude parabola;
  • Step S150 Use a long short-term memory network (LSTM) classification model for classification, and input parameters such as trajectory curve trend, object moving acceleration, frame intersection ratio before and after the object, object shape, and pixel size change as feature data into the LSTM to obtain the classification result. , to judge whether it is a false positive;
  • LSTM long short-term memory network
  • LSTM is a special type of RNN that is used to solve problems that RNNs cannot rely on for long periods of time.
  • FIG. 2 a schematic diagram of a classification process of a long short-term memory network (LSTM) classification model provided by an embodiment of the present application.
  • the trajectory curve trend, object movement acceleration, frame intersection ratio before and after the object, and object shape obtained by the Kalman filter algorithm in this embodiment are obtained.
  • the pixel size change take the above data as the feature data that can best represent the content of the moving object, input the feature data into the LSTM network model for training and learning, and finally output the classification result to determine whether the object is a high-altitude parabola.
  • the input layer of the LSTM network model is the trajectory curve trend of the extracted feature data, the object moving acceleration, the frame intersection ratio before and after the object, the object shape and pixel size changes.
  • the number of neurons in the first LSTM hidden layer is 128, and the second The number of neurons in the hidden layer of LSTM is 32, and the final output layer is 1 neuron, which represents the probability of high-altitude parabola.
  • Step S160 If the classification result of the long short-term memory network (LSTM) classification model is not a false alarm, push the alarm information to the monitoring center
  • LSTM long short-term memory network
  • the embodiment of the present application provides a high-altitude parabolic detection device 300, including:
  • a camera module 310 configured to collect target video images in real time
  • the detection module 320 is used for detecting all moving objects in the target video image by using a background modeling algorithm
  • the tracking module 330 is used to track the detected moving object by means of Kalman filtering, and obtain the trajectory characteristics and related parameters of the moving object, including the trajectory curve trend, the object movement acceleration, the frame intersection ratio before and after the object, Object shape and pixel size change;
  • the identification module 340 is used for rule-based high-altitude parabola recognition, and judges and identifies whether the moving object is a high-altitude parabola;
  • the filtering module 350 is used for classification using a long short-term memory network (LSTM) classification model, and parameters such as trajectory curve trend, object moving acceleration, frame intersection ratio before and after the object, object shape, and pixel size change are input into the LSTM as feature data Obtain the classification result and judge whether it is a false positive;
  • LSTM long short-term memory network
  • the alarm module 360 is used to push alarm information to the monitoring center.
  • the device corresponds to the above-mentioned embodiments of the high-altitude parabola detection method, and can perform various steps involved in the above-mentioned method embodiments.
  • the device includes at least one software function module that can be stored in a memory in the form of software or firmware or fixed in an operating system (OS) of the device.
  • OS operating system
  • FIG. 4 is a schematic structural diagram of a computer device provided by an embodiment of the present application.
  • a computer device 400 provided by an embodiment of the present application includes: a processor 410 and a memory 420, where the memory 420 stores a computer program executable by the processor 410, and the computer program is executed by the processor 410 to execute the above method.
  • the embodiment of the present application also provides a storage medium 430, where a computer program is stored on the storage medium 430, and the computer program is executed by the processor 410 to execute the above method.
  • the storage medium 430 can be realized by any type of volatile or non-volatile storage device or their combination, such as static random access memory (Static Random Access Memory, SRAM for short), electrically erasable programmable read-only Memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), Erasable Programmable Read Only Memory (Erasable Programmable Read Only Memory, referred to as EPROM), Programmable Read-Only Memory (Programmable Red-Only Memory, referred to as PROM), only Read-Only Memory (ROM for short), magnetic memory, flash memory, magnetic disk or optical disk.
  • static random access memory Static Random Access Memory, SRAM for short
  • EEPROM Electrically erasable programmable read-only Memory
  • EPROM Erasable Programmable Read Only Memory
  • PROM Programmable Read-Only Memory
  • ROM Only Read-Only Memory
  • first and second are only used for the purpose of description, and cannot be understood as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature defined as “first” or “second” may expressly or implicitly include one or more of that feature. “Plurality” means two or more, unless expressly specifically limited otherwise.
  • the terms “installed”, “connected”, “connected”, “fixed” and other terms should be understood in a broad sense, for example, it may be a fixed connection or a detachable connection , or integrated; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal connection of the two elements or the interaction relationship between the two elements.
  • installed may be a fixed connection or a detachable connection , or integrated; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal connection of the two elements or the interaction relationship between the two elements.
  • a first feature "on” or “under” a second feature may be in direct contact between the first and second features, or the first and second features indirectly through an intermediary touch.
  • the first feature being “above”, “over” and “above” the second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is level higher than the second feature.
  • the first feature being “below”, “below” and “below” the second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature has a lower level than the second feature.
  • a "computer-readable medium” can be any device that can contain, store, communicate, propagate, or transport the program for use by or in connection with an instruction execution system, apparatus, or apparatus.
  • computer readable media include the following: electrical connections with one or more wiring (electronic devices), portable computer disk cartridges (magnetic devices), random access memory (RAM), Read Only Memory (ROM), Erasable Editable Read Only Memory (EPROM or Flash Memory), Fiber Optic Devices, and Portable Compact Disc Read Only Memory (CDROM).
  • the computer readable medium may even be paper or other suitable medium on which the program may be printed, as the paper or other medium may be optically scanned, for example, followed by editing, interpretation, or other suitable medium as necessary process to obtain the program electronically and then store it in computer memory.
  • various parts of the present invention may be implemented in hardware, software, firmware or a combination thereof.
  • various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system.
  • a suitable instruction execution system For example, if implemented in hardware, as in another embodiment, it can be implemented by any one or a combination of the following techniques known in the art: Discrete logic circuits, application specific integrated circuits with suitable combinational logic gates, Programmable Gate Arrays (PGA), Field Programmable Gate Arrays (FPGA), etc.
  • each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may exist physically alone, or two or more units may be integrated into one module.
  • the above-mentioned integrated modules can be implemented in the form of hardware, and can also be implemented in the form of software function modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.
  • the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, and the like.

Abstract

本申请提供一种高空抛物检测方法、装置、计算机设备及存储介质，该方法包括以下步骤：设置视频监控区域，实时采集目标视频图像；采用背景建模算法检测目标视频画面内所有移动物体；采用卡尔曼滤波的方式对检测出的所述移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数；基于规则的高空抛物识别，判断识别所述移动物体是否为高空抛物；采用长短期记忆网络(LSTM)分类模型进行分类，将移动物体参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报。本申请所采用的高空抛物检测方法，抗干扰能力强，能有效过滤掉非高空抛物的运动物体；能够实时检测出现在监控摄像头视场内的高空抛物事件，实时性高，漏检率低。

Description

一种高空抛物检测方法、装置、计算机设备及存储介质 技术领域
本申请涉及图像识别技术领域，具体而言，涉及一种高空抛物检测方法、装置、计算机设备及存储介质。
背景技术
近年来高空抛物、坠物事件时有发生，高空抛物是一种不文明的行为，在污染环境的同时也严重危害居民的公共安全、扰乱社会安宁，一旦发生高空抛物伤人事件时，很难从众多住户中定位肇事者的具体位置从而追究责任，且无法及时提醒地面上的行人注意躲避，导致高空抛物安全事故频发。因此，若能准确及时定位高空抛物楼层，并及时预警与拦截，能极大保护人民的生命及财产安全，一直是社会所关注且亟待解决的问题。
相关技术中出现了基于计算机视觉的高空抛物检测方法，具体地，中国专利(CN111476973A)提供了一种利用红外感应系统进行高空抛物检测，但红外感应系统价格昂贵，实施不易；中国专利(CN205982657U)提供了一种基于超声波反射检测的高空抛物报警方法，但系统造价昂贵，部署不易，且易受周边环境干扰，精准度不够。此外，在实际情况中，存在单路摄像头无法监控整个楼体、自然物体的运动复杂多变受到风力、气候、空气阻力影响、非抛物的自然物体持续干扰等问题，出现较高的误检率和漏检率，也即，在监控画面复杂且存在自然物体持续来回运动时，相关技术中的高空抛物检测算法不能达到理想效果。
发明内容
为解决上述技术问题，本发明提出一种高空抛物检测方法，包括以下步骤：
设置视频监控区域，实时采集目标视频图像；
采用背景建模算法检测出目标视频图像内所有移动物体；
采用卡尔曼滤波的方式对检测出的移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化；
基于规则的高空抛物识别，判断识别是否为高空抛物；
采用长短期记忆网络(LSTM)分类模型进行分类，将轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
若经所述长短期记忆网络(LSTM)分类模型分类结果为非误报，推送报警信息至监控中心。
优选地，所述基于规则的高空抛物识别，采用设定建筑外轮廓线及特殊位置线，一旦出现运动物体轨迹由内而外穿过轮廓线，或从上至下穿越特殊位置线，即可认为发生高空抛物。
优选地，所述背景建模算法为单高斯模型的方法，或混合高斯模型的方法，实现对所述移动物体的检测。
本申请实施例还提供了一种高空抛物检测装置，包括：
摄像模块，用于实时采集目标视频图像；
检测模块，用于采用背景建模算法检测目标视频图像内所有移动物体；
跟踪模块，用于采用卡尔曼滤波的方式对检测出的所述移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化；
识别模块，用于基于规则的高空抛物识别，判断识别所述移动物体是否为高空抛物；
过滤模块，用于采用长短期记忆网络(LSTM)分类模型进行分类，将轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
报警模块，用于推送报警信息至监控中心。
本申请实施例还提供了一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述计算机程序时，执行如上面描述的方法。
本申请实施例还提供了一种存储介质，其上存储有计算机程序，其特征在于，该计算机程序被处理器运行时执行如上面描述的方法。
通过上述技术方案，本发明的有益效果是：
本申请所采用的高空抛物检测方法，首先找出待判断的移动物体的轨迹；再对所有轨迹进行分析与过滤，找出符合该场景下高空抛物的轨迹；然后对发生高空抛物发生地点发出警报；最后，通过过滤掉偶然出现的误检、漏检情况，从而减少误报。本申请所应用的方法，实时性高，漏检率低，既可以远程实时调度现场监控，又可以获取报警图片和报警事件发生的准确位置。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明中记载的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的高空抛物检测方法的流程示意图；
图2为本申请实施例提供的长短期记忆网络(LSTM)分类模型分类流程示意图；
图3为本申请实施例提供的高空抛物检测装置的结构示意图；
图4为本申请实施例提供的计算机设备的结构示意图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。
除非另有定义，本文所使用的所有的技术和科学术语与属于本发明的技术领域的技术人员通常理解的含义相同。本文中在本发明的说明书中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本发明。本文所使用的术语“及/或”包括一个或多个相关的所列项目的任意的和所有的组合。
请参见图1本申请实施例提供的高空抛物检测方法的流程示意图；一 种高空抛物检测方法包括以下步骤：
步骤S110：设置视频监控区域，实时采集目标视频图像；
步骤S120：采用背景建模算法检测目标视频图像内所有移动物体；
具体地，本实施例可采用单高斯模型方法、混合高斯模型方法或光流法实现对移动物体的检测。
单高斯模型方法针对多个视频帧中固定的像素点，计算N帧视频图像中该点的像素值的N个样本的均值和方差，用均值和方差即可唯一确定单高斯背景模型，背景相减后的值与阈值(取三倍的方差)比较，即可判断前景或背景，确定是否出现移动物体。
混合高斯模型依次提取N帧视频图像，每次对每个像素点迭代建模。将当前画面与假设是静态背景进行比较发现有明显的变化的区域，就可以认为该区域出现移动的物体。
光流法检测运动目标，根据视频帧中各像素点的速度矢量特征对图像进行动态的分析。若图像中不存在运动目标，那么光流矢量在整个图像区域则是连续变化的，而当物体和图像背景中存在相对运动时，运动物体所形成的速度矢量则必然不同于邻域背景的速度矢量，从而将运动物体的位置检测出来。由于光流法要进行迭代运算，精度越高计算量就越大。
关于单高斯模型方法、混合高斯模型方法或光流法检测运动目标为现有技术，本实施例不再赘述。
步骤S130：采用卡尔曼滤波的方式对检测出的移动物体进行跟踪，获取移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、 物体前后帧交并比、物体形状及像素大小变化；
本申请采用卡尔曼滤波算法实现对移动目标进行准确跟踪，得到多帧的物体运动轨迹。
步骤S140：基于规则的高空抛物识别，判断识别是否为高空抛物；
具体地，基于规则的高空抛物识别，采用设定建筑外轮廓线及特殊位置线，一旦出现运动物体轨迹由内而外穿过轮廓线，或从上至下穿越特殊位置线，即可认为发生高空抛物；
步骤S150：采用长短期记忆网络(LSTM)分类模型进行分类，将轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
在深度学习中，LSTM是一种特殊类型的RNN，用来解决RNN不能长期依赖的问题。请参见图2本申请实施例提供的长短期记忆网络(LSTM)分类模型分类流程示意图，本实施例通过卡尔曼滤波算法获取的轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化，将上述数据作为最能代表运动物体内容的特征数据，将特征数据输入到LSTM网络模型中进行训练学习，最后输出分类结果，判断所述物体是否为高空抛物。
LSTM网络模型的输入层是提取的特征数据轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化，第一个LSTM隐藏层的神经元个数为128，第二个LSTM隐藏层的神经元个数为32，最后输出层是1个神经元，代表高空抛物的概率。
通过测试可知，当使用轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化五种特征数据作为输入数据时，模型的泛化效果较好，分类的准确率较高，实现了高空抛物的判断。
步骤S160：若经长短期记忆网络(LSTM)分类模型分类结果为非误报，推送报警信息至监控中心
请参见图3示出的本申请实施例提供的高空抛物检测装置的结构示意图；本申请实施例提供了一种高空抛物检测装置300，包括：
摄像模块310，用于实时采集目标视频图像；
检测模块320，用于采用背景建模算法检测目标视频图像内所有移动物体；
跟踪模块330，用于采用卡尔曼滤波的方式对检测出的所述移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化；
识别模块340，用于基于规则的高空抛物识别，判断识别所述移动物体是否为高空抛物；
过滤模块350，用于采用长短期记忆网络(LSTM)分类模型进行分类，将轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
报警模块360，用于推送报警信息至监控中心。
应理解的是，该装置与上述的高空抛物检测方法实施例对应，能够执 行上述方法实施例涉及的各个步骤，该装置具体的功能可以参见上文中的描述，为避免重复，此处适当省略详细描述。该装置包括至少一个能以软件或固件(firmware)的形式存储于存储器中或固化在装置的操作系统(operating system，OS)中的软件功能模块。
请参见图4示出的本申请实施例提供的计算机设备的结构示意图。本申请实施例提供的一种计算机设备400，包括：处理器410和存储器420，存储器420存储有处理器410可执行的计算机程序，计算机程序被处理器410执行时执行如上的方法。
本申请实施例还提供了一种存储介质430，该存储介质430上存储有计算机程序，该计算机程序被处理器410运行时执行如上的方法。
其中，存储介质430可以由任何类型的易失性或非易失性存储设备或者它们的组合实现，如静态随机存取存储器(Static Random Access Memory,简称SRAM)，电可擦除可编程只读存储器(Electrically Erasable Programmable Read-Only Memory,简称EEPROM)，可擦除可编程只读存储器(Erasable Programmable Read Only Memory,简称EPROM)，可编程只读存储器(Programmable Red-Only Memory,简称PROM)，只读存储器(Read-Only Memory,简称ROM)，磁存储器，快闪存储器，磁盘或光盘。
在本发明的描述中，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者更多个该特征。“多个”的含义是两个或两个以上，除非另有明确具体的限定。
在本发明中，除非另有明确的规定和限定，术语“安装”、“相连”、“连接”、“固定”等术语应做广义理解，例如，可以是固定连接，也可以是可拆卸连接，或成一体；可以是机械连接，也可以是电连接；可以是直接相连，也可以通过中间媒介间接相连，可以是两个元件内部的连通或两个元件的相互作用关系。对于本领域的普通技术人员而言，可以根据具体情况理解上述术语在本发明中的具体含义。
在本发明中，除非另有明确的规定和限定，第一特征在第二特征“上”或“下”可以是第一和第二特征直接接触，或第一和第二特征通过中间媒介间接接触。而且，第一特征在第二特征“之上”、“上方”和“上面”可是第一特征在第二特征正上方或斜上方，或仅仅表示第一特征水平高度高于第二特征。第一特征在第二特征“之下”、“下方”和“下面”可以是第一特征在第二特征正下方或斜下方，或仅仅表示第一特征水平高度小于第二特征。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不必针对相同的实施例或示例。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
流程图中或在此以其他方式描述的任何过程或方法描述可以被理解为，表示包括一个或更多个用于实现特定逻辑功能或过程的步骤的可执行指令的代码的模块、片段或部分，并且本发明的优选实施方式的范围包括另外的实现，其中可以不按所示出或讨论的顺序，包括根据所涉及的功能按基本同时的方式或按相反的顺序，来执行功能，这应被本发明的实施例所属技术领域的技术人员所理解。
在流程图中表示或在此以其他方式描述的逻辑和/或步骤，例如，可以被认为是用于实现逻辑功能的可执行指令的定序列表，可以具体实现在任何计算机可读介质中，以供指令执行系统、装置或设备(如基于计算机的系统、包括处理器的系统或其他可以从指令执行系统、装置或设备取指令并执行指令的系统)使用，或结合这些指令执行系统、装置或设备而使用。就本说明书而言，“计算机可读介质”可以是任何可以包含、存储、通信、传播或传输程序以供指令执行系统、装置或设备或结合这些指令执行系统、装置或设备而使用的装置。计算机可读介质的更具体的示例(非穷尽性列表)包括以下：具有一个或多个布线的电连接部(电子装置)，便携式计算机盘盒(磁装置)，随机存取存储器(RAM)，只读存储器(ROM)，可擦除可编辑只读存储器(EPROM或闪速存储器)，光纤装置，以及便携式光盘只读存储器(CDROM)。另外，计算机可读介质甚至可以是可在其上打印所述程序的纸或其他合适的介质，因为可以例如通过对纸或其他介质进行光学扫描，接着进行编辑、解译或必要时以其他合适方式进行处理来以电子方式获得所述程序，然后将其存储在计算机存储器中。
应当理解，本发明的各部分可以用硬件、软件、固件或它们的组合来实现。在上述实施方式中，多个步骤或方法可以用存储在存储器中且由合适的指令执行系统执行的软件或固件来实现。例如，如果用硬件来实现，和在另一实施方式中一样，可用本领域公知的下列技术中的任一项或他们的组合来实现：具有用于对数据信号实现逻辑功能的逻辑门电路的离散逻辑电路，具有合适的组合逻辑门电路的专用集成电路，可编程门阵列(PGA)，现场可编程门阵列(FPGA)等。
本技术领域的普通技术人员可以理解实现上述实施例方法携带的全部或部分步骤是可以通过程序来指令相关的硬件完成，所述的程序可以存储于一种计算机可读存储介质中，该程序在执行时，包括方法实施例的步骤之一或其组合。
此外，在本发明各个实施例中的各功能单元可以集成在一个处理模块中，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模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。
上述提到的存储介质可以是只读存储器，磁盘或光盘等。尽管上面已经示出和描述了本发明的实施例，可以理解的是，上述实施例是示例性的，不能理解为对本发明的限制，本领域的普通技术人员在本发明的范围内可以对上述实施例进行变化、修改、替换和变型。

Claims (6)

  1. 一种高空抛物检测方法，其特征在于，包括以下步骤：
    设置视频监控区域，实时采集目标视频图像；
    采用背景建模算法检测所述目标视频图像内所有移动物体；
    采用卡尔曼滤波的方式对检测出的所述移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化；
    基于规则的高空抛物识别，判断识别所述移动物体是否为高空抛物；
    采用长短期记忆网络(LSTM)分类模型进行分类，将所述轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
    若经所述长短期记忆网络(LSTM)分类模型分类结果为非误报，推送报警信息至监控中心。
  2. 根据权利要求1所述的一种高空抛物检测方法，其特征在于，所述基于规则的高空抛物识别，判断识别是否为高空抛物，包括：
    采用设定建筑外轮廓线及特殊位置线，一旦出现运动物体轨迹由内而外穿过所述轮廓线，或从上至下穿越所述特殊位置线，即可认定是高空抛物。
  3. 根据权利要求1所述的一种高空抛物检测方法，其特征在于，所述背景建模算法为单高斯模型方法、混合高斯模型方法或光流法，实现对所述移动物体的检测。
  4. 一种高空抛物检测装置，其特征在于，包括：
    摄像模块，用于实时采集目标视频图像；
    检测模块，用于采用背景建模算法检测目标视频图像内所有移动物体；
    跟踪模块，用于采用卡尔曼滤波的方式对检测出的所述移动物体进行跟踪，获取所述移动物体轨迹特点及相关参数，包括轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状及像素大小变化；
    识别模块，用于基于规则的高空抛物识别，判断识别所述移动物体是否为高空抛物；
    过滤模块，用于采用长短期记忆网络(LSTM)分类模型进行分类，将轨迹曲线趋势、物体移动加速度、物体前后帧交并比、物体形状、像素大小变化等参数作为特征数据输入到LSTM中获取分类结果，判断是否为误报；
    报警模块，用于推送报警信息至监控中心。
  5. 一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述计算机程序时，实现根据权利要求1-3中任一项所述的高空抛物检测方法。
  6. 一种存储介质，其上存储有计算机程序，其特征在于，该计算机程序被处理器执行时实现根据权利要求1-3中任一项所述的高空抛物检测方法。
PCT/CN2021/128478 2020-11-19 2021-11-03 一种高空抛物检测方法、装置、计算机设备及存储介质 WO2022105609A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US18/265,870 US20240062518A1 (en) 2020-11-19 2021-11-03 Method, apparatus, computer device and storage medium for detecting objects thrown from height

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011306953.2A CN112418069B (zh) 2020-11-19 2020-11-19 一种高空抛物检测方法、装置、计算机设备及存储介质
CN202011306953.2 2020-11-19

Publications (1)

Publication Number Publication Date
WO2022105609A1 true WO2022105609A1 (zh) 2022-05-27

Family

ID=74774667

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/128478 WO2022105609A1 (zh) 2020-11-19 2021-11-03 一种高空抛物检测方法、装置、计算机设备及存储介质

Country Status (3)

Country Link
US (1) US20240062518A1 (zh)
CN (1) CN112418069B (zh)
WO (1) WO2022105609A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114863370A (zh) * 2022-07-08 2022-08-05 合肥中科类脑智能技术有限公司 一种复杂场景高空抛物识别方法及系统
CN115294744A (zh) * 2022-07-29 2022-11-04 杭州海康威视数字技术股份有限公司 一种图像显示系统、方法、装置及设备
CN116012368A (zh) * 2023-02-16 2023-04-25 江西惜能照明有限公司 基于智慧灯杆的安防监测方法、系统、存储介质及计算机
CN116453065A (zh) * 2023-06-16 2023-07-18 云途信息科技(杭州)有限公司 路面异物抛洒识别方法、装置、计算机设备及存储介质
CN116543013A (zh) * 2023-04-19 2023-08-04 北京拙河科技有限公司 一种球类运动轨迹分析方法及装置
CN116994201A (zh) * 2023-07-20 2023-11-03 山东产研鲲云人工智能研究院有限公司 对高空抛物进行溯源监测的方法及计算设备
CN117372427A (zh) * 2023-12-06 2024-01-09 南昌中展数智科技有限公司 基于视频分析的工程施工监管方法及系统

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112418069B (zh) * 2020-11-19 2021-05-11 中科智云科技有限公司 一种高空抛物检测方法、装置、计算机设备及存储介质
CN113076809B (zh) * 2021-03-10 2023-07-21 海纳云物联科技有限公司 一种基于视觉Transformer的高空坠物检测方法
CN112926538B (zh) * 2021-04-08 2022-06-24 广州绿简智能科技有限公司 一种基于监控视频的高空抛物识别方法
CN113239781B (zh) * 2021-05-10 2024-04-19 福建汇川物联网技术科技股份有限公司 一种高空抛物的检测方法及装置
CN113297949B (zh) * 2021-05-20 2024-02-20 科大讯飞股份有限公司 高空抛物检测方法、装置、计算机设备和存储介质
CN113409362B (zh) * 2021-06-29 2023-02-21 深圳市商汤科技有限公司 高空抛物检测方法和装置、设备及计算机存储介质
CN113408550B (zh) * 2021-08-17 2021-11-26 中国建筑第五工程局有限公司 基于图像处理的智能称重管理系统
CN114332154B (zh) * 2022-03-04 2022-06-14 英特灵达信息技术(深圳)有限公司 一种高空抛物检测方法及系统
CN114693556B (zh) * 2022-03-25 2023-06-27 英特灵达信息技术(深圳)有限公司 一种高空抛物帧差法运动目标检测去拖影方法
CN115223103B (zh) * 2022-09-13 2022-11-22 深圳市研超科技有限公司 基于数字图像处理的高空抛物检测方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102547244A (zh) * 2012-01-17 2012-07-04 深圳辉锐天眼科技有限公司 视频监控方法及系统
CN105163067A (zh) * 2015-08-12 2015-12-16 武汉大学 一种基于数字图像处理技术的高空抛物取证系统
CN111898511A (zh) * 2020-07-23 2020-11-06 北京以萨技术股份有限公司 基于深度学习的高空抛物检测方法、装置及介质
CN111931719A (zh) * 2020-09-22 2020-11-13 苏州科达科技股份有限公司 高空抛物检测方法以及装置
CN112418069A (zh) * 2020-11-19 2021-02-26 中科智云科技有限公司 一种高空抛物检测方法、装置、计算机设备及存储介质

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109309811B (zh) * 2018-08-31 2021-02-19 中建三局智能技术有限公司 一种基于计算机视觉的高空抛物检测系统与方法
CN110647822A (zh) * 2019-08-30 2020-01-03 重庆博拉智略科技有限公司 高空抛物行为识别方法、装置、存储介质及电子设备
CN111079663B (zh) * 2019-12-19 2022-01-11 深圳云天励飞技术股份有限公司 高空抛物的监测方法、装置、电子设备及存储介质
CN111553256A (zh) * 2020-04-26 2020-08-18 上海天诚比集科技有限公司 基于物体轨迹识别的高空抛物预警识别方法
CN111627049B (zh) * 2020-05-29 2023-09-26 北京中科晶上科技股份有限公司 高空抛物的确定方法、装置、存储介质及处理器

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102547244A (zh) * 2012-01-17 2012-07-04 深圳辉锐天眼科技有限公司 视频监控方法及系统
CN105163067A (zh) * 2015-08-12 2015-12-16 武汉大学 一种基于数字图像处理技术的高空抛物取证系统
CN111898511A (zh) * 2020-07-23 2020-11-06 北京以萨技术股份有限公司 基于深度学习的高空抛物检测方法、装置及介质
CN111931719A (zh) * 2020-09-22 2020-11-13 苏州科达科技股份有限公司 高空抛物检测方法以及装置
CN112418069A (zh) * 2020-11-19 2021-02-26 中科智云科技有限公司 一种高空抛物检测方法、装置、计算机设备及存储介质

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114863370A (zh) * 2022-07-08 2022-08-05 合肥中科类脑智能技术有限公司 一种复杂场景高空抛物识别方法及系统
CN115294744A (zh) * 2022-07-29 2022-11-04 杭州海康威视数字技术股份有限公司 一种图像显示系统、方法、装置及设备
CN115294744B (zh) * 2022-07-29 2024-03-22 杭州海康威视数字技术股份有限公司 一种图像显示系统、方法、装置及设备
CN116012368A (zh) * 2023-02-16 2023-04-25 江西惜能照明有限公司 基于智慧灯杆的安防监测方法、系统、存储介质及计算机
CN116543013A (zh) * 2023-04-19 2023-08-04 北京拙河科技有限公司 一种球类运动轨迹分析方法及装置
CN116453065A (zh) * 2023-06-16 2023-07-18 云途信息科技(杭州)有限公司 路面异物抛洒识别方法、装置、计算机设备及存储介质
CN116453065B (zh) * 2023-06-16 2023-09-19 云途信息科技(杭州)有限公司 路面异物抛洒识别方法、装置、计算机设备及存储介质
CN116994201A (zh) * 2023-07-20 2023-11-03 山东产研鲲云人工智能研究院有限公司 对高空抛物进行溯源监测的方法及计算设备
CN116994201B (zh) * 2023-07-20 2024-03-29 山东产研鲲云人工智能研究院有限公司 对高空抛物进行溯源监测的方法及计算设备
CN117372427A (zh) * 2023-12-06 2024-01-09 南昌中展数智科技有限公司 基于视频分析的工程施工监管方法及系统
CN117372427B (zh) * 2023-12-06 2024-03-22 南昌中展数智科技有限公司 基于视频分析的工程施工监管方法及系统

Also Published As

Publication number Publication date
CN112418069B (zh) 2021-05-11
CN112418069A (zh) 2021-02-26
US20240062518A1 (en) 2024-02-22

Similar Documents

Publication Publication Date Title
WO2022105609A1 (zh) 一种高空抛物检测方法、装置、计算机设备及存储介质
WO2019179024A1 (zh) 机场跑道智能监控方法、应用服务器及计算机存储介质
CN109977782B (zh) 基于目标位置信息推理的跨店经营行为检测方法
Wang et al. Automated sewer pipe defect tracking in CCTV videos based on defect detection and metric learning
Ellis Performance metrics and methods for tracking in surveillance
Venetianer et al. Performance evaluation of an intelligent video surveillance system–A case study
US9805256B2 (en) Method for setting a tridimensional shape detection classifier and method for tridimensional shape detection using said shape detection classifier
US20150071492A1 (en) Abnormal behaviour detection
CN100565557C (zh) 基于粒子动态采样模型的红外人体目标跟踪系统
CN109598700A (zh) 采用卷积神经网络的早期火灾检测方法
CN104219488B (zh) 目标图像的生成方法和装置、以及视频监控系统
CN111145475A (zh) 基于振动光纤和深度学习的入侵报警系统、方法及介质
CN112633174B (zh) 一种基于改进的YOLOv4高穹顶场景下火灾检测方法、存储介质
Tang et al. Hybrid blob and particle filter tracking approach for robust object tracking
CN113362374A (zh) 一种基于目标跟踪网络的高空抛物检测方法及系统
CN111627049A (zh) 高空抛物的确定方法、装置、存储介质及处理器
Lichtenauer et al. Influence of the observation likelihood function on particle filtering performance in tracking applications
Nodehi et al. Multi-metric re-identification for online multi-person tracking
WO2023273010A1 (zh) 高空抛物检测方法和装置、设备，及计算机存储介质
Liu et al. A review of traffic visual tracking technology
CN112070033A (zh) 一种基于有限状态机分析的视频遗留物检测方法
Wang et al. A new approach for real-time detection of abandoned and stolen objects
CN112541403B (zh) 一种利用红外摄像头的室内人员跌倒检测方法
JPWO2020250296A1 (ja) 画像処理装置、画像処理方法及びプログラム
Leykin Visual human tracking and group activity analysis: A video mining system for retail marketing

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21893756

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 18265870

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21893756

Country of ref document: EP

Kind code of ref document: A1