WO2020220951A1 - 一种录像数据存储方法及装置 - Google Patents

一种录像数据存储方法及装置 Download PDF

Info

Publication number
WO2020220951A1
WO2020220951A1 PCT/CN2020/083732 CN2020083732W WO2020220951A1 WO 2020220951 A1 WO2020220951 A1 WO 2020220951A1 CN 2020083732 W CN2020083732 W CN 2020083732W WO 2020220951 A1 WO2020220951 A1 WO 2020220951A1
Authority
WO
WIPO (PCT)
Prior art keywords
channel
data
video data
recording data
video
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2020/083732
Other languages
English (en)
French (fr)
Inventor
高在伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Hikvision Digital Technology Co Ltd
Original Assignee
Hangzhou Hikvision Digital Technology Co Ltd
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 Hangzhou Hikvision Digital Technology Co Ltd filed Critical Hangzhou Hikvision Digital Technology Co Ltd
Publication of WO2020220951A1 publication Critical patent/WO2020220951A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/76—Television signal recording
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/76—Television signal recording
    • H04N5/78—Television signal recording using magnetic recording
    • H04N5/781—Television signal recording using magnetic recording on disks or drums

Definitions

  • This application relates to the field of monitoring technology, in particular to a method and device for storing video data.
  • the video capture device collects the video data, it needs to send the video data to electronic equipment such as DVR (Digital Video Recorder), NVR (Network Video Recorder), central storage device, etc. storage.
  • electronic equipment such as DVR (Digital Video Recorder), NVR (Network Video Recorder), central storage device, etc. storage.
  • DVR Digital Video Recorder
  • NVR Network Video Recorder
  • central storage device etc. storage.
  • the electronic device directly stores the video data after receiving the video data.
  • monitoring channels such as IPC (Internet Protocol Camera) channels and analog channels under a monitoring system
  • IPC Internet Protocol Camera
  • analog channels under a monitoring system
  • the purpose of the embodiments of the present application is to provide a method and device for storing video data, so as to save the hardware cost of the electronic device storing video data.
  • the specific technical solutions are as follows:
  • an embodiment of the present application provides a recording data storage method, which includes:
  • the recording data of this channel will be stored.
  • the recording data includes audio data
  • the steps of performing target recognition on the recording data of the channel and identifying the designated target in the recording data of the channel include:
  • the recording data includes video data
  • the steps of performing target recognition on the recording data of the channel and identifying the designated target in the recording data of the channel include:
  • each image data to be recognized it is determined whether there is a designated target in the video data of the channel.
  • the method further includes:
  • the step of storing the recording data of the channel includes:
  • an embodiment of the present application provides a video data storage device, which includes:
  • the acquisition module is used to sequentially acquire the recording data of each channel in a polling manner
  • the recognition module is used to identify the target of the recording data of a channel after acquiring the recording data of a channel each time, and identify the designated target in the recording data of the channel;
  • the storage module is used to store the video data of the channel if there is a designated target in the video data of the channel.
  • the recording data includes audio data
  • Identification module specifically used for:
  • the recording data includes video data
  • Identification module specifically used for:
  • each image data to be recognized it is determined whether there is a designated target in the video data of the channel.
  • the device further includes:
  • the stop module is used to stop the storage of the channel's video data if there is no designated target in the channel's video data.
  • storage module specifically used for:
  • an embodiment of the present application provides an electronic device including a processor and a memory, where the memory stores machine executable instructions that can be executed by the processor, and the machine executable instructions are loaded and executed by the processor to achieve The method provided in the first aspect of the embodiments of the present application.
  • an embodiment of the present application provides a machine-readable storage medium, and the machine-readable storage medium stores machine-executable instructions.
  • the machine-executable instructions When the machine-executable instructions are loaded and executed by a processor, they implement the first On the one hand the method provided.
  • an embodiment of the present application provides an application program for execution at runtime: the method provided in the first aspect of the embodiment of the present application.
  • the video data storage method and device acquire the video data of each channel in turn in a polling manner, and perform target identification on the video data of the channel after acquiring the video data of a channel each time .
  • To identify the specified target in the recording data of the channel if the specified target exists in the recording data of the channel, the recording data of the channel will be stored.
  • Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel. If it exists, the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure
  • the electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of
  • FIG. 1 is a schematic flowchart of a method for storing video data according to an embodiment of the application
  • FIG. 2 is a schematic flowchart of a recording data storage method according to another embodiment of the application.
  • FIG. 3 is a schematic structural diagram of an electronic device in an application scenario of an embodiment of the application.
  • FIG. 4 is a schematic flowchart of a method for storing video data executed by a DVR according to an embodiment of the application
  • FIG. 5 is a schematic flowchart of a video data storage method executed by an NVR according to an embodiment of the application
  • FIG. 6 is a schematic diagram of the structure of a video data storage device according to an embodiment of the application.
  • FIG. 7 is a schematic structural diagram of an electronic device according to an embodiment of the application.
  • embodiments of the present application provide a video data storage method, device, electronic device, and machine-readable storage medium.
  • the video data storage method provided by the embodiment of the present application is first introduced.
  • the video data storage method provided by the embodiments of the present application can be applied to electronic devices with video data storage functions such as DVR, NVR, and central storage devices.
  • the method for implementing the recording data storage method provided in the embodiments of the present application may be at least one of software, hardware circuit, and logic circuit provided in the above electronic device.
  • the method for storing video data may include the following steps.
  • S101 Obtain video data of each channel in sequence according to a polling manner.
  • the monitoring system includes channels for collecting video data such as IPC channels and analog channels. Each channel is responsible for collecting video data within a certain monitoring range. Each channel sends the collected video data to the electronic device, and the electronic device stores the video data .
  • the electronic device obtains the recording data of each channel in turn according to the polling method.
  • the polling method is to first obtain the recording data of one channel, and after an interval of several frames or a period of time, obtain the recording data of the next channel.
  • the polling sequence for each channel can be preset, and the polling time interval can also be preset. Normally, the polling time interval can be set according to the time required for subsequent target identification, for example, The time for one target identification is 20ms, and the set polling interval can be greater than or equal to 20ms.
  • S102 After acquiring the video recording data of a channel each time, perform target recognition on the video recording data of the channel, and identify the designated target in the video recording data of the channel.
  • the electronic device After acquiring the video data of a channel, the electronic device performs target recognition on the video data of the channel, and recognizes the car (vehicle brand, car model, license plate, etc. attributes) that the user cares about, and people (men, women, top color, bottom clothes) Color, whether to ride a bicycle or other attributes) and other designated targets, determine whether there is a designated target in the video data of the channel.
  • car vehicle brand, car model, license plate, etc. attributes
  • people men, women, top color, bottom clothes
  • Color whether to ride a bicycle or other attributes
  • the video data of the channel can be stored.
  • the recording data of each channel is obtained in turn according to the polling method.
  • the recording data of a channel is identified by the target, and the recording data of the channel is identified Specify target, if there is a specified target in the recording data of this channel, the recording data of this channel will be stored.
  • Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel. If it exists, the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure The electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device is adopted to reduce the demand for computing resources, and at the same time, it can achieve the purpose of target detection in all channels.
  • the recording data may include audio data.
  • S102 can be specifically implemented by the following steps: preprocessing the audio data of the channel to obtain the audio data to be identified; using the sliding window method to obtain audio units in different time domains from the audio data to be identified; using the first Preset the deep learning model, perform audio recognition on each audio unit, and obtain the recognition result of each audio unit; use the pre-built language model library to match the recognition results of each audio unit with similarity; according to the matching result of each audio unit To determine whether there is a specified target in the audio data of the channel.
  • the process of preprocessing audio data can be to set sampling parameters such as audio sampling rate, bit width, etc., regularize the audio sampling, and also use noise filtering to filter out the noise.
  • the obtained audio data to be identified is Noise-free regular audio data.
  • a sliding window method can be used to obtain audio units in different time domains from the audio data to be recognized.
  • RNN Recurrent Neural Network, cyclic neural network
  • other first preset deep learning models perform voice recognition on each audio unit, and obtain the recognition result of each audio unit.
  • the recognition result is the probability of what the audio content in the audio unit is, generally on electronic devices
  • a language model library is established in advance.
  • the language model library stores audio type, content and other information. Using the language model library to match the similarity of the recognition results of each audio unit, you can determine the audio collected by one channel Whether there is a designated target in the data, the higher the matching degree, the greater the possibility of the designated target in the audio data.
  • the recording data may include video data.
  • S102 can be specifically implemented through the following steps: preprocessing each image data in the video data of the channel to obtain each image data to be recognized; using the second preset deep learning model to perform Target recognition: According to the recognition result of each image data to be recognized, it is judged whether there is a designated target in the video data of the channel.
  • the process of preprocessing each image data in the video data is mainly to uniformly input the image data of the preset deep learning model, for example, the resolution and image color space can be unified, and the filtering technology can also be used to filter out the noise signal in the image , Using a second preset deep learning model such as FRCNN (Fast Region-based Convolutional Neural Network, fast convolutional neural network based on candidate regions) to perform target recognition on each image to be recognized, and the recognition results are obtained.
  • FRCNN Fast convolutional neural network based on candidate regions
  • the recognition of the specified target can only recognize the audio target, or only the video target, or both the audio target and the video target, which is not limited here.
  • audio target recognition and video target recognition in addition to the above-mentioned deep neural network methods such as RNN and FRCNN, methods such as feature comparison and pixel matching can also be used, which will not be repeated here.
  • S103 can be specifically implemented through the following steps:
  • the recording data of the channel If there is a specified target in the recording data of the channel, read the recording data in the preset time period from the buffer corresponding to the channel, where the buffer area corresponding to the channel stores the preset time period before the current time
  • the video data collected by the channel; the video data within the preset time period and the video data acquired in the channel are stored.
  • the electronic device can open up a buffer for Cache the video data collected by the channel in the preset time period before the current time.
  • the preset time period can be set according to the time it takes for the electronic device to complete a complete polling process for all channels, for example, a total of 5 If the polling interval for each channel is 20ms, the preset time period can be set to a time period greater than or equal to 100ms. In this way, it is equivalent to setting up the pre-recording function.
  • the embodiment of the present application also provides a recording data storage method, as shown in FIG. 2, which may include the following steps.
  • S201 Obtain video data of each channel in sequence according to a polling manner.
  • S202 After acquiring the video recording data of a channel each time, perform target recognition on the video recording data of the channel, and identify the designated target in the video recording data of the channel.
  • S201-S203 are the same as S101-S103 in the embodiment shown in FIG. 1, and will not be repeated here.
  • the steps of S201-S203 need to be executed in a loop. If for a certain channel, during a certain recognition, it is recognized that the specified target does not exist in the recording data of the channel , You need to stop storing the video data of this channel.
  • the recording data of each channel is obtained in turn according to the polling method.
  • the recording data of a channel is identified by the target, and the recording data of the channel is identified Specify target, if there is a specified target in the recording data of this channel, the recording data of this channel will be stored.
  • Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel. If it exists, the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure The electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device Provides polling and pre-recording functions, and performs one channel identification and storage operations at a time. Compared with full channels, it requires less computing resources while ensuring full channel identification. And, if for a certain channel, it is recognized that there is no specified target in the recording data of the channel during a certain recognition, stop continuing to store the recording data of the channel to ensure that the electronic device does not store too much non-existent
  • the video data of the designated target is stored as far as possible and only the video data of the designated target is stored, which further saves the hardware cost of the electronic equipment.
  • the electronic equipment mainly includes several software and/or hardware units: video capture unit, code stream packaging unit, video unit, storage unit, configuration unit and deep learning processing unit.
  • video capture unit mainly includes several software and/or hardware units: video capture unit, code stream packaging unit, video unit, storage unit, configuration unit and deep learning processing unit.
  • code stream packaging unit mainly includes several software and/or hardware units: video capture unit, code stream packaging unit, video unit, storage unit, configuration unit and deep learning processing unit.
  • the connection relationship is shown in Figure 3.
  • the video acquisition unit is mainly responsible for the access of video analog signals or digital signals;
  • the code stream encapsulation unit is mainly responsible for encapsulating video data into RTP (Reliable Transport Protocol) formats;
  • the storage unit is mainly responsible for the storage of video data;
  • the configuration unit It is mainly responsible for the configuration and management of the video recording unit;
  • the deep learning processing unit is mainly responsible for identifying the input video data, identifying people, cars or other objects of interest to users in the video data.
  • the main body that implements the video data storage method provided in the embodiments of the present application is a DVR or NVR.
  • the flow of the DVR's implementation of the video data storage method is shown in Figure 4. Since the input of the DVR is analog data, there is no need to decode the video data before target recognition.
  • the left side of Figure 4 is the video data purification flow, including the video capture unit. Collect video data and use FRCNN to identify the target. If the specified target is recognized, the storage notification of the corresponding channel will be opened through the configuration unit. Otherwise, the storage stop notification will be initiated through the configuration unit. After a test, it is determined whether to poll the next channel, and if yes, then receive The video data collected by the next channel.
  • the right side of Figure 4 shows the video data storage process.
  • the corresponding channel collects video data, encodes H264 or H265, and then encodes the video data.
  • the video pre-recording unit pre-records the video, and the pre-recording unit receives the video data purification process.
  • the video data sent is stored for video data. This completes the video data storage process of one channel.
  • the flow of the NVR's implementation of the video data storage method is shown in Figure 5. Since the NVR input is IPC data, the video data needs to be decoded before target recognition, which is different from the DVR processing flow.
  • the video data on the left side of Figure 5 In the purification process, video decoding only decodes the I frame of the video, so that the specified target in the video data can be decoded and identified quickly and efficiently, and in the video data storage process on the right side of Figure 5, video encoding is not required.
  • an embodiment of the present application provides a video data storage device.
  • the device may include:
  • the obtaining module 610 is configured to sequentially obtain the video recording data of each channel in a polling manner
  • the recognition module 620 is configured to perform target recognition on the recording data of a channel after acquiring the recording data of a channel each time, and identify the designated target in the recording data of the channel;
  • the storage module 630 is configured to store the video data of the channel if there is a designated target in the video data of the channel.
  • the recording data may include audio data
  • the identification module 620 may be specifically used for:
  • the recording data includes video data
  • the identification module 620 may be specifically used for:
  • each image data to be recognized it is determined whether there is a designated target in the video data of the channel.
  • the device may also include:
  • the stop module is used to stop the storage of the channel's video data if there is no designated target in the channel's video data.
  • the storage module 630 can be specifically used for:
  • the recording data of each channel is obtained in turn according to the polling method.
  • the recording data of a channel is identified by the target, and the recording data of the channel is identified Specify target, if there is a specified target in the recording data of this channel, the recording data of this channel will be stored.
  • Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel. If it exists, the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure
  • the electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video The hardware cost of the data electronic device.
  • An embodiment of the present application provides an electronic device, as shown in FIG. 7, including a processor 701 and a memory 702, where the memory 702 stores machine executable instructions that can be executed by the processor 701, and the machine Executable instructions are loaded and executed by the processor 701 to implement the recording data storage method provided in the embodiment of the present application.
  • the foregoing memory may include RAM (Random Access Memory, random access memory), and may also include NVM (Non-volatile Memory, non-volatile memory), such as at least one disk storage.
  • NVM Non-volatile Memory, non-volatile memory
  • the memory may also be at least one storage device located far away from the foregoing processor.
  • the above-mentioned processor may be a general-purpose processor, including CPU (Central Processing Unit), NP (Network Processor, network processor), etc.; it may also be DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array, field programmable gate array) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.
  • CPU Central Processing Unit
  • NP Network Processor, network processor
  • DSP Digital Signal Processor
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array, field programmable gate array
  • other programmable logic devices discrete gates or transistor logic devices, discrete hardware components.
  • the memory 702 and the processor 701 may perform data transmission through a wired connection or a wireless connection, and the electronic device and other devices may communicate through a wired communication interface or a wireless communication interface. What is shown in FIG. 7 is only an example of data transmission through the bus, and is not intended to limit the specific connection mode.
  • the processor reads the machine executable instructions stored in the memory, and loads and executes the machine executable instructions, so as to achieve: according to the polling method, the video data of each channel can be obtained in turn, and every time After acquiring the video data of a channel, perform target recognition on the video data of the channel, identify the designated target in the video data of the channel, and store the video data of the channel if the designated target exists in the video data of the channel. Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel.
  • the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure
  • the electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video The hardware cost of the data electronic device.
  • an embodiment of the present application also provides a machine-readable storage medium that stores machine-executable instructions in the machine-readable storage medium.
  • the present application is implemented.
  • the recording data storage method provided by the embodiment.
  • the machine-readable storage medium stores machine executable instructions that execute the recording data storage method provided by the embodiment of the present application at runtime, so it can be realized: according to the polling method, the recording of each channel is obtained in turn Data, each time the recording data of a channel is obtained, target recognition is performed on the recording data of the channel, and the specified target in the recording data of the channel is identified. If the specified target exists in the recording data of the channel, the channel is stored Video data. Target recognition is performed on the video data collected by a channel, and it is judged whether there is a specified target in the video data of the channel.
  • the video data of the channel is stored, and the above operations are performed on each channel by polling to ensure
  • the electronic device stores the video data collected by each channel with the specified target.
  • the amount of stored video data is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video
  • the hardware cost of the data electronic device is less than the amount of data stored in the traditional way, and the storage space required is less than the storage space required by the traditional way, saving the storage of video The hardware cost of the data electronic device.
  • the embodiment of the present application also provides an application program for executing at runtime: the recording data storage method provided by the embodiment of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Television Signal Processing For Recording (AREA)
  • Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)

Abstract

本申请实施例提供了一种录像数据存储方法及装置，按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。

Description

一种录像数据存储方法及装置
本申请要求于2019年04月29日提交中国专利局、申请号为201910355683.5发明名称为“一种录像数据存储方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及监控技术领域，特别是涉及一种录像数据存储方法及装置。
背景技术
在监控系统中，视频采集设备在采集到录像数据后，需要将录像数据发送至DVR(Digital Video Recorder，硬盘录像机)、NVR(Network Video Recorder，网络硬盘录像机)、中心存储设备等电子设备中进行存储。当前的录像数据存储方法中，电子设备在收到录像数据后，直接存储录像数据。
然而，随着监控系统的不断扩大，一个监控系统下的IPC(Internet Protocol Camera，网络摄像机)通道、模拟通道等监控通道的数目越来越多，导致电子设备需要存储的录像数据的数据量越来越庞大，则要求电子设备必须有足够的存储空间，增加了存储录像数据的电子设备的硬件成本
发明内容
本申请实施例的目的在于提供一种录像数据存储方法及装置，以节约存储录像数据的电子设备的硬件成本。具体技术方案如下：
第一方面，本申请实施例提供了一种录像数据存储方法，该方法包括：
按照轮询的方式，依次获取各通道的录像数据；
在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标；
若该通道的录像数据中存在指定目标，则存储该通道的录像数据。
可选的，录像数据包括音频数据；
对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标的步骤，包括：
对该通道的音频数据进行预处理，得到待识别音频数据；
采用滑动窗口方式，从待识别音频数据中，获取不同时域的音频单元；
采用第一预设深度学习模型，对各音频单元进行音频识别，得到各音频单元的识别结果；
采用预先建立的语言模型库，对各音频单元的识别结果进行相似度匹配；
根据各音频单元对应的匹配结果，判断该通道的音频数据中是否存在指定目标。
可选的，录像数据包括视频数据；
对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标的步骤，包括：
对该通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；
采用第二预设深度学习模型，对各待识别图像数据进行目标识别；
根据各待识别图像数据的识别结果，判断该通道的视频数据中是否存在指定目标。
可选的，在对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标的步骤之后，该方法还包括：
若该通道的录像数据中不存在指定目标，则停止对该通道的录像数据的存储。
可选的，若该通道的录像数据中存在指定目标，则存储该通道的录像数据的步骤，包括：
若该通道的录像数据中存在指定目标，则从该通道对应的缓冲区中读取预设时段内的录像数据，其中，该通道对应的缓存区中存储的是当前时刻之前的预设时段内该通道采集的录像数据；
存储预设时段内的录像数据及获取到的该通道的录像数据。
第二方面，本申请实施例提供了一种录像数据存储装置，该装置包括：
获取模块，用于按照轮询的方式，依次获取各通道的录像数据；
识别模块，用于在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标；
存储模块，用于若该通道的录像数据中存在指定目标，则存储该通道的录像数据。
可选的，录像数据包括音频数据；
识别模块，具体用于：
对该通道的音频数据进行预处理，得到待识别音频数据；
采用滑动窗口方式，从待识别音频数据中，获取不同时域的音频单元；
采用第一预设深度学习模型，对各音频单元进行音频识别，得到各音频单元的识别结果；
采用预先建立的语言模型库，对各音频单元的识别结果进行相似度匹配；
根据各音频单元对应的匹配结果，判断该通道的音频数据中是否存在指定目标。
可选的，录像数据包括视频数据；
识别模块，具体用于：
对该通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；
采用第二预设深度学习模型，对各待识别图像数据进行目标识别；
根据各待识别图像数据的识别结果，判断该通道的视频数据中是否存在指定目标。
可选的，该装置还包括：
停止模块，用于若该通道的录像数据中不存在指定目标，则停止对该通道的录像数据的存储。
可选的，存储模块，具体用于：
若该通道的录像数据中存在指定目标，则从该通道对应的缓冲区中读取预设时段内的录像数据，其中，该通道对应的缓存区中存储的是当前时刻之前的预设时段内该通道采集的录像数据；
存储所述预设时段内的录像数据及获取到的该通道的录像数据。
第三方面，本申请实施例提供了一种电子设备，包括处理器和存储器，其中，存储器存储有能够被处理器执行的机器可执行指令，机器可执行指令由处理器加载并执行，以实现本申请实施例第一方面所提供的方法。
第四方面，本申请实施例提供了一种机器可读存储介质，机器可读存储介质内存储有机器可执行指令，机器可执行指令在被处理器加载并执行时，实现本申请实施例第一方面所提供的方法。
第五方面，本申请实施例提供了一种应用程序，用于在运行时执行：本申请实施例第一方面所提供的方法。
本申请实施例提供的一种录像数据存储方法及装置，按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。
附图说明
为了更清楚地说明本申请实施例和现有技术的技术方案，下面对实施例和现有技术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申请的一些实施例，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为本申请一实施例的录像数据存储方法的流程示意图；
图2为本申请另一实施例的录像数据存储方法的流程示意图
图3为本申请实施例的应用场景下电子设备的结构示意图；
图4为本申请实施例的DVR执行视频数据存储方法的流程示意图；
图5为本申请实施例的NVR执行视频数据存储方法的流程示意图；
图6为本申请实施例的录像数据存储装置的结构示意图；
图7为本申请实施例的电子设备的结构示意图。
具体实施方式
为使本申请的目的、技术方案、及优点更加清楚明白，以下参照附图并举实施例，对本申请进一步详细说明。显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
为了节约存储录像数据的电子设备的硬件成本，本申请实施例提供了一种录像数据存储方法、装置、电子设备及机器可读存储介质。下面，首先对本申请实施例所提供的录像数据存储方法进行介绍。
本申请实施例所提供的录像数据存储方法，可以应用于DVR、NVR、中心存储设备等具有录像数据存储功能的电子设备。实现本申请实施例所提供的录像数据存储方法的方式可以为设置于上述电子设备中的软件、硬件电路和逻辑电路中的至少一种。
如图1所示，本申请实施例所提供的一种录像数据存储方法，可以包括如下步骤。
S101，按照轮询的方式，依次获取各通道的录像数据。
监控系统中包括IPC通道、模拟通道等采集录像数据的通道，每个通道负责采集一定监控范围内的录像数据，每个通道将采集的录像数据发送至电子设备，由电子设备对录像数据进行存储。
电子设备按照轮询的方式，依次获取各通道的录像数据，轮询的方式就 是先获取一个通道的录像数据，间隔若干帧或者一段时间后，再获取下一个通道的录像数据。对各通道的轮询顺序可以是预先设置的，轮询的时间间隔也可以是预先设置的，通常情况下，轮询的时间间隔，可以根据后续进行目标识别所需的时间设定，例如，进行一次目标识别的时间为20ms，则设置的轮询的时间间隔可以大于或等于20ms。
S102，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标。
电子设备在每获取到一个通道的录像数据之后，对该通道的录像数据进行目标识别，识别用户关心的车(车辆品牌、车型、车牌等属性)、人(男人、女人、上衣颜色、下衣颜色、是否骑车等属性)等指定目标，判断该通道的录像数据中是否存在指定目标。
S103，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。
如果识别出一通道的录像数据中存在指定目标，由于指定目标是用户关心的目标内容，则可以存储该通道的录像数据。
应用本申请实施例，按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。并且，采用轮询加缓存的方式，减少计算资源的需求量，同时能达到全通道进行目标检测的目的。
下面，对图1所示实施例中各步骤的实现方式分别进行介绍。
在本申请的一种实现方式中，录像数据可以包括音频数据。相应的，S102具体可以通过如下步骤实现：对该通道的音频数据进行预处理，得到待识别音频数据；采用滑动窗口方式，从待识别音频数据中，获取不同时域的音频 单元；采用第一预设深度学习模型，对各音频单元进行音频识别，得到各音频单元的识别结果；采用预先建立的语言模型库，对各音频单元的识别结果进行相似度匹配；根据各音频单元对应的匹配结果，判断该通道的音频数据中是否存在指定目标。
对音频数据进行预处理的过程，可以是设置例如音频的采样率、位宽等采样参数，规整音频的采样，还可以利用噪声滤波的方式，将噪声过滤掉，所得到的待识别音频数据是无噪声的规则音频数据，由于音频数据往往是连续的，为了提高音频识别的准确率，可以采用滑动窗口方式，从待识别音频数据中，获取不同时域的音频单元，采用例如RNN(Recurrent Neural Network，循环神经网络)等第一预设深度学习模型，对各音频单元进行语音识别，得到各音频单元的识别结果，识别结果是音频单元中的音频是什么内容的概率，一般在电子设备上预先建立有一个语言模型库，该语言模型库中存储着音频的类型、内容等信息，采用该语言模型库，对各音频单元的识别结果进行相似度匹配，则可以判断出一个通道采集的音频数据中是否存在指定目标，匹配度越高，则说明音频数据中存在指定目标的可能性越大。
在本申请的另一种实现方式中，录像数据可以包括视频数据。相应的，S102具体可以通过如下步骤实现：对该通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；采用第二预设深度学习模型，对各待识别图像数据进行目标识别；根据各待识别图像数据的识别结果，判断该通道的视频数据中是否存在指定目标。
对视频数据中的各图像数据进行预处理的过程，主要是统一输入预设深度学习模型的图像数据，例如可以统一分辨率、图像颜色空间等，还可以采用滤波技术滤掉图像中的噪声信号，采用例如FRCNN(Fast Region-based Convolutional Neural Network，快速的基于候选区域的卷积神经网络)等第二预设深度学习模型，对各待识别图像进行目标识别，得到识别结果，识别结果是各待识别图像数据中是否存在指定目标的概率，综合得到一个通道采集的视频数据中是否存在指定目标的识别结果。
在实际应用中，对指定目标的识别，可以只识别音频目标，也可以只识别视频目标，还可以既识别音频目标又识别视频目标，这里不做限定。进行 音频目标识别、视频目标识别的方式，除了可以采用上述的RNN、FRCNN等深度神经网络方法以外，还可以采用特征比对、像素匹配等方法，这里不再一一赘述。
在本申请的再一种实现方式中，S103具体可以通过如下步骤实现：
若该通道的录像数据中存在指定目标，则从该通道对应的缓冲区中读取预设时段内的录像数据，其中，该通道对应的缓存区中存储的是当前时刻之前的预设时段内该通道采集的录像数据；存储预设时段内的录像数据及获取到的该通道的录像数据。
在进行目标识别时，识别到的实际指定目标可能已在监控场景下出现了一段时间，为了跟踪指定目标从出现到消失的整个过程，针对每一个通道，电子设备可以开辟一个缓冲区，用于缓存当前时刻之前的预设时段内该通道采集的录像数据，通常情况下，预设时段可以根据电子设备对所有通道完成一次完整的轮询过程所花费的时间进行设定，例如，一共有5个通道，每个通道轮询的间隔为20ms，则预设时段可以设定为大于或等于100ms的时间段。这样，相当于设置了预录功能，则在存储录像数据时，不仅仅存储获取到的该通道的录像数据，还可以存储从缓冲区中读取到的该通道在预设时段内采集的录像数据。由于音频数据的连续性较差，为了保证数据的连续性，在进行预录时，可以只缓存通道采集的视频数据。
基于图1所示实施例，本申请实施例还提供了一种录像数据存储方法，如图2所示，可以包括如下步骤。
S201，按照轮询的方式，依次获取各通道的录像数据。
S202，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标。
S203，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。
S201-S203与图1所示实施例中的S101-S103相同，这里不再赘述。
S204，若该通道的录像数据中不存在指定目标，则停止对该通道的录像 数据的存储。
为了保证存储的始终是存在指定目标的录像数据，则需要一直循环执行S201-S203的步骤，如果对于某一个通道来讲，在某一次识别时，识别出该通道的录像数据中不存在指定目标，则需要停止继续存储该通道的录像数据。
一旦识别出某一个通道的录像数据中不存在指定目标，则停止对该通道的录像数据的存储，保证电子设备中不会存储过多的不存在指定目标的录像数据，尽可能的只存储存在指定目标的录像数据，进一步节约了电子设备的硬件成本。
应用本申请实施例，按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。提供轮询和预录的功能，每次进行一个通道的识别、存储操作，相比全通道，在保证全通道识别的情况下，需要的计算资源更少。并且，如果对于某一个通道来讲，在某一次识别时，识别出该通道的录像数据中不存在指定目标，停止继续存储该通道的录像数据，保证电子设备中不会存储过多的不存在指定目标的录像数据，尽可能的只存储存在指定目标的录像数据，进一步节约了电子设备的硬件成本。
为了便于理解，下面结合具体的应用场景，对本申请实施例所提供的录像数据存储方法进行介绍。以录像数据仅包括视频数据为例，电子设备中主要包括视频采集单元、码流封装单元、录像单元、存储单元、配置单元和深度学习处理单元几个软件和/或硬件单元，这些单元之间的连接关系如图3所示。视频采集单元主要负责视频模拟信号或者数字信号的接入；码流封装单元主要负责将视频数据封装成RTP(Reliable Transport Protocol，可靠传输协议)等 格式；存储单元主要负责视频数据的存储；配置单元主要负责对录像单元进行配置管理；深度学习处理单元主要负责对输入的视频数据进行识别，识别视频数据中的人、车或其他用户感兴趣的目标。
实现本申请实施例所提供的视频数据存储方法的执行主体主要是DVR或者NVR。DVR执行视频数据存储方法的流程如图4所示，由于DVR输入的是模拟数据，在进行目标识别之前，不需要对视频数据进行解码，图4左侧为视频数据提纯流程，包括视频采集单元采集视频数据，通过FRCNN，进行目标识别，如果识别到指定目标，通过配置单元开启对应通道的存储通知，否则通过配置单元发起停止存储通知，一次检测后判断是否轮询下一个通道，是则接收下一个通道采集的视频数据。图4右侧为视频数据存储的流程，对应通道通过采集视频数据，进行H264或者H265等编码，编码后进行视频数据封装，录像预录单元对录像进行预录，预录单元接收视频数据提纯流程发来的视频数据，进行视频数据的存储。这样就完成了一个通道的视频数据存储过程。
NVR执行视频数据存储方法的流程如图5所示，由于NVR输入的是IPC数据，在进行目标识别之前，需要对视频数据进行解码，不同于DVR的处理流程，在图5左侧的视频数据提纯流程中，视频解码仅解码视频的I帧，这样就可以快速高效的解码并识别视频数据中的指定目标，并且在图5右侧的视频数据存储流程中，不需要进行视频编码。
相应于上述方法实施例，本申请实施例提供了一种录像数据存储装置，如图6所示，该装置可以包括：
获取模块610，用于按照轮询的方式，依次获取各通道的录像数据；
识别模块620，用于在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标；
存储模块630，用于若该通道的录像数据中存在指定目标，则存储该通道的录像数据。
可选的，录像数据可以包括音频数据；
识别模块620，具体可以用于：
对该通道的音频数据进行预处理，得到待识别音频数据；
采用滑动窗口方式，从待识别音频数据中，获取不同时域的音频单元；
采用第一预设深度学习模型，对各音频单元进行音频识别，得到各音频单元的识别结果；
采用预先建立的语言模型库，对各音频单元的识别结果进行相似度匹配；
根据各音频单元对应的匹配结果，判断该通道的音频数据中是否存在指定目标。
可选的，录像数据包括视频数据；
识别模块620，具体可以用于：
对该通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；
采用第二预设深度学习模型，对各待识别图像数据进行目标识别；
根据各待识别图像数据的识别结果，判断该通道的视频数据中是否存在指定目标。
可选的，该装置还可以包括：
停止模块，用于若该通道的录像数据中不存在指定目标，则停止对该通道的录像数据的存储。
可选的，存储模块630，具体可以用于：
若该通道的录像数据中存在指定目标，则从该通道对应的缓冲区中读取预设时段内的录像数据，其中，该通道对应的缓存区中存储的是当前时刻之前的预设时段内该通道采集的录像数据；
存储所述预设时段内的录像数据及获取到的该通道的录像数据。
应用本申请实施例，按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识 别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。
本申请实施例提供了一种电子设备，如图7所示，包括处理器701和存储器702，其中，所述存储器702存储有能够被所述处理器701执行的机器可执行指令，所述机器可执行指令由所述处理器701加载并执行，以实现本申请实施例所提供的录像数据存储方法。
上述存储器可以包括RAM(Random Access Memory，随机存取存储器)，也可以包括NVM(Non-volatile Memory，非易失性存储器)，例如至少一个磁盘存储器。可选的，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述处理器可以是通用处理器，包括CPU(Central Processing Unit，中央处理器)、NP(Network Processor，网络处理器)等；还可以是DSP(Digital Signal Processor，数字信号处理器)、ASIC(Application Specific Integrated Circuit，专用集成电路)、FPGA(Field-Programmable Gate Array，现场可编程门阵列)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
存储器702与处理器701之间可以通过有线连接或者无线连接的方式进行数据传输，并且电子设备与其他设备之间可以通过有线通信接口或者无线通信接口进行通信。图7所示的仅为通过总线进行数据传输的示例，不作为具体连接方式的限定。
本申请实施例中，处理器通过读取存储器中存储的机器可执行指令，并通过加载和执行机器可执行指令，能够实现：按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数 据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。
另外，本申请实施例还提供了一种机器可读存储介质，所述机器可读存储介质内存储有机器可执行指令，所述机器可执行指令在被处理器加载并执行时，实现本申请实施例所提供的录像数据存储方法。
本申请实施例中，机器可读存储介质存储有在运行时执行本申请实施例所提供的录像数据存储方法的机器可执行指令，因此能够实现：按照轮询的方式，依次获取各通道的录像数据，在每次获取到一个通道的录像数据后，对该通道的录像数据进行目标识别，识别该通道的录像数据中的指定目标，若该通道的录像数据中存在指定目标，则存储该通道的录像数据。针对于一个通道采集的录像数据进行目标识别，判断该通道的录像数据中是否存在指定目标，如果存在，则存储该通道的录像数据，并且采用轮询的方式对每一个通道执行以上操作，保证电子设备存储的是各通道采集的存在指定目标的录像数据，存储的录像数据的数据量少于传统方式存储的数据量，所需要的存储空间小于传统方式所需的存储空间，节约了存储录像数据的电子设备的硬件成本。
本申请实施例还提供了一种应用程序，用于在运行时执行：本申请实施例所提供的录像数据存储方法。
对于电子设备、机器可读存储介质及应用程序实施例而言，由于其涉及的方法内容基本相似于前述的方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、 “包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置、电子设备、机器可读存储介质以及应用程序实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (13)

  1. 一种录像数据存储方法，其特征在于，所述方法包括：
    按照轮询的方式，依次获取各通道的录像数据；
    在每次获取到一个通道的录像数据后，对所述一个通道的录像数据进行目标识别，识别所述一个通道的录像数据中的指定目标；
    若所述一个通道的录像数据中存在所述指定目标，则存储所述一个通道的录像数据。
  2. 根据权利要求1所述的方法，其特征在于，所述录像数据包括音频数据；
    所述对所述一个通道的录像数据进行目标识别，识别所述一个通道的录像数据中的指定目标，包括：
    对所述一个通道的音频数据进行预处理，得到待识别音频数据；
    采用滑动窗口方式，从所述待识别音频数据中，获取不同时域的音频单元；
    采用第一预设深度学习模型，对各音频单元进行音频识别，得到所述各音频单元的识别结果；
    采用预先建立的语言模型库，对所述各音频单元的识别结果进行相似度匹配；
    根据所述各音频单元对应的匹配结果，判断所述一个通道的音频数据中是否存在指定目标。
  3. 根据权利要求1所述的方法，其特征在于，所述录像数据包括视频数据；
    所述对所述一个通道的录像数据进行目标识别，识别所述一个通道的录像数据中的指定目标，包括：
    对所述一个通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；
    采用第二预设深度学习模型，对所述各待识别图像数据进行目标识别；
    根据所述各待识别图像数据的识别结果，判断所述一个通道的视频数据中是否存在指定目标。
  4. 根据权利要求1所述的方法，其特征在于，在所述在每次获取到一个通道的录像数据后，对所述一个通道的录像数据进行目标识别，识别所述一个通道的录像数据中的指定目标之后，所述方法还包括：
    若所述一个通道的录像数据中不存在所述指定目标，则停止对所述一个通道的录像数据的存储。
  5. 根据权利要求1所述的方法，其特征在于，所述若所述一个通道的录像数据中存在所述指定目标，则存储所述一个通道的录像数据，包括：
    若所述一个通道的录像数据中存在所述指定目标，则从所述一个通道对应的缓冲区中读取预设时段内的录像数据，所述一个通道对应的缓存区中存储的是当前时刻之前的预设时段内所述一个通道采集的录像数据；
    存储所述预设时段内的录像数据及获取到的所述一个通道的录像数据。
  6. 一种录像数据存储装置，其特征在于，所述装置包括：
    获取模块，用于按照轮询的方式，依次获取各通道的录像数据；
    识别模块，用于在每次获取到一个通道的录像数据后，对所述一个通道的录像数据进行目标识别，识别所述一个通道的录像数据中的指定目标；
    存储模块，用于若所述一个通道的录像数据中存在所述指定目标，则存储所述一个通道的录像数据。
  7. 根据权利要求6所述的装置，其特征在于，所述录像数据包括音频数据；
    所述识别模块，具体用于：
    对所述一个通道的音频数据进行预处理，得到待识别音频数据；
    采用滑动窗口方式，从所述待识别音频数据中，获取不同时域的音频单元；
    采用第一预设深度学习模型，对各音频单元进行音频识别，得到所述各音频单元的识别结果；
    采用预先建立的语言模型库，对所述各音频单元的识别结果进行相似度匹配；
    根据所述各音频单元对应的匹配结果，判断所述一个通道的音频数据中是否存在指定目标。
  8. 根据权利要求6所述的装置，其特征在于，所述录像数据包括视频数据；
    所述识别模块，具体用于：
    对所述一个通道的视频数据中的各图像数据分别进行预处理，得到各待识别图像数据；
    采用第二预设深度学习模型，对所述各待识别图像数据进行目标识别；
    根据所述各待识别图像数据的识别结果，判断所述一个通道的视频数据中是否存在指定目标。
  9. 根据权利要求6所述的装置，其特征在于，所述装置还包括：
    停止模块，用于若所述一个通道的录像数据中不存在所述指定目标，则停止对所述一个通道的录像数据的存储。
  10. 根据权利要求6所述的装置，其特征在于，所述存储模块，具体用于：
    若所述一个通道的录像数据中存在所述指定目标，则从所述一个通道对应的缓冲区中读取预设时段内的录像数据，所述一个通道对应的缓存区中存储的是当前时刻之前的预设时段内所述一个通道采集的录像数据；
    存储所述预设时段内的录像数据及获取到的所述一个通道的录像数据。
  11. 一种电子设备，其特征在于，包括处理器和存储器，其中，所述存储器存储有能够被所述处理器执行的机器可执行指令，所述机器可执行指令由所述处理器加载并执行，以实现权利要求1-5任一项所述的方法。
  12. 一种机器可读存储介质，其特征在于，所述机器可读存储介质内存 储有机器可执行指令，所述机器可执行指令在被处理器加载并执行时，实现权利要求1-5任一项所述的方法。
  13. 一种应用程序，其特征在于，用于在运行时执行：权利要求1-5任一项所述的方法。
PCT/CN2020/083732 2019-04-29 2020-04-08 一种录像数据存储方法及装置 Ceased WO2020220951A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910355683.5A CN111866444A (zh) 2019-04-29 2019-04-29 一种录像数据存储方法及装置
CN201910355683.5 2019-04-29

Publications (1)

Publication Number Publication Date
WO2020220951A1 true WO2020220951A1 (zh) 2020-11-05

Family

ID=72966254

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/083732 Ceased WO2020220951A1 (zh) 2019-04-29 2020-04-08 一种录像数据存储方法及装置

Country Status (2)

Country Link
CN (1) CN111866444A (zh)
WO (1) WO2020220951A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116418943A (zh) * 2021-12-30 2023-07-11 浙江宇视科技有限公司 一种视频管理方法、装置、存储介质及系统

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007124526A (ja) * 2005-10-31 2007-05-17 Hitachi Ltd 画像監視記録装置及び画像監視記録システム
CN101299812A (zh) * 2008-06-25 2008-11-05 北京中星微电子有限公司 视频分析和存储方法、系统,及视频检索方法、系统
CN101867786A (zh) * 2009-04-20 2010-10-20 中兴通讯股份有限公司 一种视频监控方法及装置
CN102665054A (zh) * 2012-05-10 2012-09-12 江苏友上科技实业有限公司 一种用人脸快速检索的网络视频录像机系统
WO2013069565A1 (ja) * 2011-11-09 2013-05-16 ステラグリーン株式会社 撮影記録装置
CN103188468A (zh) * 2011-12-30 2013-07-03 支录奎 有用视频识别录像机的研究
CN104202575A (zh) * 2014-09-17 2014-12-10 广州中国科学院软件应用技术研究所 一种视频监控处理方法及系统
CN104836992A (zh) * 2015-05-08 2015-08-12 无锡天脉聚源传媒科技有限公司 一种监控视频录制方法及装置
CN106157952A (zh) * 2016-08-30 2016-11-23 北京小米移动软件有限公司 声音识别方法及装置
CN109344688A (zh) * 2018-08-07 2019-02-15 江苏大学 一种基于卷积神经网络的监控视频中人的自动识别方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007241377A (ja) * 2006-03-06 2007-09-20 Sony Corp 検索システム、撮像装置、データ保存装置、情報処理装置、撮像画像処理方法、情報処理方法、プログラム
CN101998138A (zh) * 2009-08-25 2011-03-30 北京达鸣慧科技有限公司 电视频道监控系统及其实时监控方法
JP5962916B2 (ja) * 2012-11-14 2016-08-03 パナソニックＩｐマネジメント株式会社 映像監視システム
CN104038717B (zh) * 2014-06-26 2017-11-24 北京小鱼在家科技有限公司 一种智能录制系统
CN108198545B (zh) * 2017-12-19 2021-11-02 安徽建筑大学 一种基于小波变换的语音识别方法

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007124526A (ja) * 2005-10-31 2007-05-17 Hitachi Ltd 画像監視記録装置及び画像監視記録システム
CN101299812A (zh) * 2008-06-25 2008-11-05 北京中星微电子有限公司 视频分析和存储方法、系统,及视频检索方法、系统
CN101867786A (zh) * 2009-04-20 2010-10-20 中兴通讯股份有限公司 一种视频监控方法及装置
WO2013069565A1 (ja) * 2011-11-09 2013-05-16 ステラグリーン株式会社 撮影記録装置
CN103188468A (zh) * 2011-12-30 2013-07-03 支录奎 有用视频识别录像机的研究
CN102665054A (zh) * 2012-05-10 2012-09-12 江苏友上科技实业有限公司 一种用人脸快速检索的网络视频录像机系统
CN104202575A (zh) * 2014-09-17 2014-12-10 广州中国科学院软件应用技术研究所 一种视频监控处理方法及系统
CN104836992A (zh) * 2015-05-08 2015-08-12 无锡天脉聚源传媒科技有限公司 一种监控视频录制方法及装置
CN106157952A (zh) * 2016-08-30 2016-11-23 北京小米移动软件有限公司 声音识别方法及装置
CN109344688A (zh) * 2018-08-07 2019-02-15 江苏大学 一种基于卷积神经网络的监控视频中人的自动识别方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116418943A (zh) * 2021-12-30 2023-07-11 浙江宇视科技有限公司 一种视频管理方法、装置、存储介质及系统

Also Published As

Publication number Publication date
CN111866444A (zh) 2020-10-30

Similar Documents

Publication Publication Date Title
US20210337217A1 (en) Video analytics encoding for improved efficiency of video processing and compression
US11025942B2 (en) Progressive compressed domain computer vision and deep learning systems
US10917704B1 (en) Automated video preview generation
US11930063B2 (en) Content completion detection for media content
US9842258B2 (en) System and method for video preview
EP3725088B1 (en) Systems and methods for editing a video
CN111355977B (zh) 一种网络直播视频的优化保存方法及装置
CN113947376B (zh) 基于多重生物特征的c/s打卡方法和装置
CN111784750A (zh) 视频图像中移动物体追踪方法、装置、设备以及存储介质
CN110457974B (zh) 图像叠加方法、装置、电子设备及可读存储介质
CN113438468B (zh) 视频质量的动态控制方法、装置、存储介质及电子设备
WO2020220951A1 (zh) 一种录像数据存储方法及装置
CN110335626A (zh) 基于音频的年龄识别方法及装置、存储介质
CN111866522A (zh) 一种录像数据编码方法及装置
CN111405358A (zh) 基于缓存的视频帧提取方法、装置、介质及系统
US10083720B2 (en) Method and system for video data stream storage
CN111147954A (zh) 一种缩略图抽取方法及装置
CN120656488A (zh) 情绪识别方法、装置、电子设备、存储介质及程序产品
WO2022048129A1 (zh) 对象识别方法、装置及系统
CN111866428B (zh) 一种历史录像数据处理方法及装置
CN114740975B (zh) 目标内容的获取方法及相关设备
CN111857551B (zh) 一种录像数据老化方法及装置
US20210152782A1 (en) Data Transmission Method, Camera and Electronic Device
CN115914742A (zh) 一种视频字幕的人物识别方法、装置、设备及存储介质
CN106603840A (zh) 基于移动终端的音频数据处理方法及装置

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: 20799148

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20799148

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24-05-2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20799148

Country of ref document: EP

Kind code of ref document: A1