CN111866449B - Intelligent video acquisition system and method - Google Patents

Intelligent video acquisition system and method Download PDF

Info

Publication number
CN111866449B
CN111866449B CN202010553474.4A CN202010553474A CN111866449B CN 111866449 B CN111866449 B CN 111866449B CN 202010553474 A CN202010553474 A CN 202010553474A CN 111866449 B CN111866449 B CN 111866449B
Authority
CN
China
Prior art keywords
video acquisition
acquisition terminal
server
video
acu
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.)
Active
Application number
CN202010553474.4A
Other languages
Chinese (zh)
Other versions
CN111866449A (en
Inventor
赵洪钢
时晨
张晓�
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.)
National University of Defense Technology
Original Assignee
National University of Defense Technology
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 National University of Defense Technology filed Critical National University of Defense Technology
Priority to CN202010553474.4A priority Critical patent/CN111866449B/en
Publication of CN111866449A publication Critical patent/CN111866449A/en
Application granted granted Critical
Publication of CN111866449B publication Critical patent/CN111866449B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/66—Remote control of cameras or camera parts, e.g. by remote control devices

Abstract

The invention provides an intelligent video acquisition system and method, which comprises a video acquisition terminal, a LoRa gateway and a server, wherein self-adaptive communication connection is established between the video acquisition terminal and the server; the video acquisition terminal and the server perform self-adaptive communication interaction; the server sends an instruction to the video acquisition terminal, specifies the object type to be identified by the video acquisition terminal, specifies the mode of sending acquisition information to the server by the video acquisition terminal as a self-adaptive mode or a small data mode, and specifies the video acquisition mode of the video acquisition terminal as a continuous video acquisition mode or a non-continuous video acquisition mode. The video acquisition system and the method disclosed by the invention can acquire the video, can identify the object type in the acquired video, can adopt a proper video acquisition mode and a proper acquisition information sending mode according to the actual situation, and are suitable for a plurality of application fields such as public safety monitoring, field surgery research and the like.

Description

Intelligent video acquisition system and method
Technical Field
The invention belongs to the field of Internet of things, relates to video acquisition, and particularly relates to an intelligent video acquisition system and method.
Background
With the rapid development of image processing, embedded and internet of things technologies, miniaturized, intelligent and wireless transmission-supported video acquisition systems are continuously emerging and are used in various fields such as public safety monitoring, field surgery research and the like.
However, the existing video acquisition system and method are generally only specific to some application scenarios, and when the situations of difficulty in power supply of the commercial power, insufficient wireless signal coverage capability or unstable network service are encountered, it is difficult to adopt a proper video acquisition mode and a proper acquisition information transmission mode according to the actual situation, so that it is difficult to timely and effectively transmit the field situation back to the server and supply power by using the battery device for as long as possible, thereby limiting the environmental adaptability and the application expandability of the existing video acquisition system.
The card computer is a microcomputer mainboard which is only slightly larger than a credit card in recent years, such as raspberry pie, banana pie and the like, and comprises a processor, a memory, a USB interface, a network interface, a GPIO interface and the like, and supports various operating systems such as Linux, ubuntu and the like. Due to the characteristics of the card computer, the card computer not only has good hardware expandability, but also is suitable for loading lightweight deep learning models represented by MobileNet, Shufflenet and the like; taking MobileNet as an example, it is not only a model for various visual recognition tasks such as object class recognition, but also can be efficiently run on a mobile device. Meanwhile, as a new internet of things technology, the LoRa has the characteristics of wide coverage, large connection, low power consumption, low cost and the like, and is very suitable for providing flexible and low-speed network access service in areas which cannot be covered by operator networks such as the field and do not have commercial power supply conditions.
In conclusion, the card computer, the deep learning model for object type identification, the LoRa technology and the like have the advantages of being very suitable for improving the intelligent level of the video acquisition system; the advantages are fully utilized, and the intelligent video acquisition system and the intelligent video acquisition method are provided, which are beneficial to solving the problems of poor environmental adaptability and the like of the existing video acquisition system.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide an intelligent video acquisition system and method, which can acquire a video and identify the object type in the acquired video, can further improve the identification accuracy on the basis of the existing identification model in the identification process, and can adopt a proper video acquisition mode and a proper acquisition information sending mode according to the actual situation, thereby having better adaptability to different environments such as urban areas, fields and the like.
In order to solve the technical problems, the invention adopts the following technical scheme:
an intelligent video acquisition method adopts a video acquisition terminal, a LoRa gateway and a server, and is characterized by specifically comprising the following steps:
step one, self-adaptive communication connection is established between a video acquisition terminal and a server, wherein the self-adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
The direct communication mode CommdiThe video acquisition terminal and the server are directly communicated with each other without forwarding through an LoRa gateway;
the indirect communication mode ComminThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
step two, the video acquisition terminal and the server carry out self-adaptive communication interaction;
step three, the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
step four, the server sends an instruction to the video acquisition terminal, and the mode of sending acquisition information to the server by the video acquisition terminal is designated as a self-adaptive mode or a small data mode;
step five, the server sends an instruction to the video acquisition terminal, and the video acquisition mode of the video acquisition terminal is designated to be a continuous video acquisition mode or a non-continuous video acquisition mode;
and sixthly, the video acquisition terminal performs video acquisition and object type identification in the acquired video locally and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode.
The invention also has the following technical characteristics:
specifically, in the first step, in the process of establishing the adaptive communication connection, the video acquisition terminal performs the following steps:
step 1.1.1, creating a timer T1;
step 1.1.2, checking whether a TCP connection which is successfully established exists between the video acquisition terminal and the server, if so, executing step 1.1.6, otherwise, executing step 1.1.3;
step 1.1.3, the video acquisition terminal initiates a TCP connection request to the server, if the TCP connection is successfully established, step 1.1.6 is executed, otherwise step 1.1.4 is executed;
step 1.1.4, the video acquisition terminal sends an indirect connection request REQ to the server through the forwarding of the LoRa gatewaynodeIf the video acquisition terminal receives the indirect connection response REP of the serverserverIf not, executing step 1.1.5, otherwise, executing step 1.1.2;
step 1.1.5, set up Comm of video acquisition terminal itselftypIs ComminStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.6, set up Comm of video acquisition terminal itselftypIs CommdiStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.7, after waiting for the timer T1 to trigger, step 1.1.2 is executed.
Specifically, in the first step, in the process of establishing the adaptive communication connection, the server performs the following steps:
step 1.2.1, create timers T21 and T22;
step 1.2.2, checking whether a TCP connection which is successfully established exists between the server and the video acquisition terminal, if so, executing step 1.2.8, otherwise, executing step 1.2.3;
step 1.2.3, checking whether the server is in a LISTEN state of the TCP connection, if so, executing step 1.2.4, otherwise, setting the TCP connection to be in the LISTEN state and then executing step 1.2.4;
step 1.2.4, the serverSending an indirect connection request REQ to a video acquisition terminal through forwarding of a LoRa gatewayserverIf receiving the indirect connection response REP of the video acquisition terminalnodeIf not, executing step 1.2.7, otherwise, executing step 1.2.5;
step 1.2.5, starting a timer T21;
step 1.2.6, if the server receives the TCP connection request sent by the video acquisition terminal and the TCP connection is successfully established before the timer T21 is triggered, immediately closing the timer T21 and executing step 1.2.8, otherwise, executing step 1.2.3 again after the timer T21 is triggered;
step 1.2.7, set up Comm of server itselftypIs ComminStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.8, set Comm of the server itselftypIs CommdiStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.9, after waiting for the timer T22 to trigger, step 1.2.2 is executed.
Specifically, in the second step, in the adaptive communication interaction process, the video acquisition terminal and the server all perform the following steps:
step 2.1, check its CommtypIf CommtypIs CommdiStep 2.2 is executed if CommtypIs ComminThen step 2.3 is executed;
2.2, the video acquisition terminal and the server carry out direct communication interaction;
and 2.3, the video acquisition terminal and the server are forwarded through the LoRa gateway to perform communication interaction.
Specifically, in the fourth step, the adaptive method is performed according to the following steps:
step 4.1, the video acquisition terminal checks Comm of the video acquisition terminaltypIf CommtypIs CommdiStep 4.2 is executed if CommtypIs ComminThen step 4.3 is executed;
step 4.2, the video acquisition terminal directly sends the acquired video and the object type identification result in the acquired video to the server;
4.3, the video acquisition terminal only sends the object type identification result in the acquired video to the LoRa gateway, and then the LoRa gateway forwards the object type identification result to the server;
specifically, in the fourth step, the small data mode is that the video acquisition terminal only sends the object type identification result in the acquired video to the server.
In the fifth step, the continuous video acquisition mode is that the video acquisition terminal continuously performs local video acquisition and object type identification in the acquired video, and transmits the acquisition information to the server;
specifically, in the fifth step, the discontinuous video acquisition mode is performed according to the following steps:
step 5.1, the video acquisition terminal establishes timers T31 and T32;
step 5.2, the video acquisition terminal starts local video acquisition and object type identification in the acquired video, and transmits acquisition information to the server;
step 5.3, the video acquisition terminal starts a timer T31;
step 5.4, if the object type needing to be identified is identified before the timer T31 is triggered, immediately closing the timer T31 and executing the step 5.3 again, otherwise, executing the step 5.5 after the timer T31 is triggered;
step 5.5, the video acquisition terminal stops local video acquisition, acquires object type identification in the video, sends acquisition information to the server and starts a timer T32;
and 5.6, if the video acquisition terminal detects animals and people before the timer T32 is triggered, or the ultrasonic ranging module detects that an object enters and moves, immediately closing the timer T32 and re-executing the step 5.2, or else, re-executing the step 5.2 after the timer T32 is triggered.
Specifically, the object type identification in the collected video is performed according to the following steps:
step 6.1, the video acquisition terminal establishes a timer T4, and the video acquisition terminal sets a threshold valuep、VAL1And VAL2；
Step 6.2, m object classes are set in the model for object class identification, and the ith object class CAT isiSet variable Numi、ACU_SUMiAnd ACU _ AVEi，i＝1,2,3,…,m；
Step 6.3, let Numi＝0，ACU_SUMi＝0，ACU_AVEi0, i-1, 2,3, …, m, and let j-0;
6.4, the video acquisition terminal acquires a real-time single-frame image in the acquired video, identifies the object type in the single-frame image, obtains the first N object types with the maximum identification probability value, and sequentially obtains CAT from large to smalla，CATb，CATc，CATdAnd CATeWherein a, b, c, d, e ∈ {1,2,3, …, m }, that is, the a-th, b-th, c-th, d-th and e-th object classes in the model for object class identification are the top N object classes with the highest identification probability values in the single-frame image identification result, and the obtained corresponding identification probability values are ACU in turna，ACUb，ACUc，ACUd，ACUeAnd make Numi＝Numi+1，ACU_SUMi＝ACU_SUMi+ACUi，i＝a,b,c,d,e；
Step 6.5, if ACUa≥VAL1If not, executing step 6.11, otherwise, executing step 6.6;
step 6.6, let j equal j +1, and start timer T4, and execute step 6.7 after timer T4 triggers;
step 6.7, if j < p, re-executing step 6.4, otherwise executing step 6.8;
step 6.8, let ACU _ AVEi＝ACU_SUMi/NumiSelecting the first N object categories with the largest ACU _ AVE value to form a set S;
step 6.9, judging whether the object type to be identified belongs to the set S, if so, executing step 6.10, otherwise, executing step 6.3;
step 6.10, if the object type to be identified corresponds to the ACU _ AVE is greater than or equal to VAL2If the object type is not identified, the video acquisition terminal identifies the object type to be identified in the acquired video, and then the step 6.3 is skipped;
step 6.11, judging the object type and ACU to be identifiedaAnd (4) whether the corresponding object types are consistent or not, if so, identifying the object type to be identified in the acquired video by the video acquisition terminal, otherwise, not identifying, and then skipping to the step 6.3.
Further preferably, in step 6.4, the identifying the object type in the single-frame image is performed according to the following steps:
step 6.4.1, loading a model for identifying the object type, wherein the loading comprises the structure and the weight of the model;
step 6.4.2, image preprocessing is carried out on the acquired single-frame image, size scaling and pixel value scaling are carried out on the image according to the interface standard of the model, and the image is adjusted to be in an RGB independent three-channel format;
step 6.4.3, inputting the preprocessed image data into a model to obtain the probability value of each classification;
and 6.4.4, selecting the top N categories with the maximum probability values as recognition results, and recording the corresponding recognition probability values.
The invention also protects an intelligent video acquisition system, which comprises a video acquisition terminal, a LoRa gateway and a server, wherein the video acquisition terminal comprises a card computer, and a camera module, a wireless internet access card module, a LoRa wireless data transmission module, an infrared pyroelectric module and an ultrasonic ranging module which are respectively connected with the card computer; the video acquisition terminal is connected with an IP network through a built-in network card or a wireless network card module of the card computer;
the video acquisition terminal and the server establish self-adaptive communication connection, and the self-adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
The direct communication mode CommdiFor video acquisition terminals and clothesThe servers are directly communicated with each other without forwarding through an LoRa gateway;
the indirect communication mode ComminThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
the video acquisition terminal and the server carry out self-adaptive communication interaction;
the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
the server sends an instruction to the video acquisition terminal, and the mode of sending acquisition information to the server by the video acquisition terminal is designated as a self-adaptive mode or a small data mode;
the server sends an instruction to the video acquisition terminal, and designates the video acquisition mode of the video acquisition terminal as a continuous video acquisition mode or a non-continuous video acquisition mode;
the video acquisition terminal carries out local video acquisition and object type identification in the acquired video and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode.
Compared with the prior art, the invention has the following technical effects:
the system and the method provided by the invention can not only collect the video, but also identify the object type in the collected video, can further improve the identification accuracy on the basis of the existing identification model in the identification process, and can also adopt a proper video collection mode and a proper collection information sending mode according to the actual situation, thereby having better adaptability to different environments such as urban areas, fields and the like.
The invention (II) makes full use of the characteristics of small volume and strong performance of the card computer, and improves the identification accuracy by operating the identification method based on the existing identification model, so that the video acquisition terminal can not only acquire the video, but also identify the object type in the acquired video, and the utilization value of the acquired video is effectively improved.
The invention (III) makes full use of the characteristics of wide coverage, large connection and flexible deployment of the LoRa gateway of the LoRa technology, when the video acquisition terminal can not access the IP network through the built-in network card or the wireless network card module, the LoRa gateway receives the object type identification result in the acquired video and forwards the object type identification result to the server, so that the system is not only suitable for the situation of network service sudden failure, but also suitable for the areas which can not be covered by the operator networks such as original forests, remote mountain areas and the like, and the server monitoring personnel can acquire the field situation as much as possible.
(IV) the invention fully utilizes the characteristic of strong expansibility of a card computer, and can detect the change condition of a video acquisition area by connecting the infrared pyroelectric module and the ultrasonic ranging module, so that the video acquisition terminal can adopt a discontinuous video acquisition working mode, unnecessary energy consumption can be reduced, the video acquisition terminal can utilize battery equipment to supply power for as long as possible, and the whole system can adapt to more complex and changeable environments.
Drawings
Fig. 1 is a schematic diagram of application example 1.
Fig. 2 is a schematic diagram of application example 2.
Fig. 3 is a schematic diagram of a connection relationship of a video capture terminal.
The present invention will be explained in further detail with reference to examples.
Detailed Description
All modules in the video acquisition terminal are known modules sold in the market, wherein the infrared pyroelectric module is used for sending a trigger signal to a card computer when detecting animals and people; the ultrasonic ranging module is used for sending a trigger signal to the card computer when detecting that an object enters and moves.
The present invention is not limited to the following embodiments, and all equivalent changes based on the technical solutions of the present invention fall within the protection scope of the present invention.
Example 1:
the embodiment provides an intelligent video acquisition system, as shown in fig. 1 to 3, the system includes a video acquisition terminal, a LoRa gateway and a server, the video acquisition terminal includes a card computer, and a camera module, a wireless internet access card module, a LoRa wireless data transmission module, an infrared pyroelectric module and an ultrasonic ranging module which are respectively connected with the card computer; the video acquisition terminal is connected with the IP network through a built-in network card or a wireless internet card module of the card computer;
an adaptive communication connection is established between the video acquisition terminal and the server, and the adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
Comm in direct communication modediThe video acquisition terminal and the server are directly communicated with each other without forwarding through an LoRa gateway;
comm of indirect communication modeinThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
the video acquisition terminal and the server perform self-adaptive communication interaction;
the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
the method comprises the steps that a server sends an instruction to a video acquisition terminal, and the mode that the video acquisition terminal sends acquisition information to the server is designated to be a self-adaptive mode or a small data mode;
the method comprises the steps that a server sends an instruction to a video acquisition terminal, and the video acquisition mode of the video acquisition terminal is designated to be a continuous video acquisition mode or a non-continuous video acquisition mode;
the video acquisition terminal carries out local video acquisition and object type identification in the acquired video and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode.
Example 2:
the embodiment provides an intelligent video acquisition method, which adopts a video acquisition terminal, a LoRa gateway and a server, and specifically comprises the following steps:
step one, establishing self-adaptive communication connection between a video acquisition terminal and a server, wherein the self-adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
Comm in direct communication modediThe video acquisition terminal and the server are directly communicated with each other without forwarding through an LoRa gateway;
comm of indirect communication modeinThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
in the first step, in the process of establishing the self-adaptive communication connection, the video acquisition terminal performs the following steps:
step 1.1.1, creating a timer T1;
step 1.1.2, checking whether a TCP connection which is successfully established exists between the video acquisition terminal and the server, if so, executing step 1.1.6, otherwise, executing step 1.1.3;
step 1.1.3, the video acquisition terminal initiates a TCP connection request to the server, if the TCP connection is successfully established, step 1.1.6 is executed, otherwise step 1.1.4 is executed;
step 1.1.4, the video acquisition terminal sends an indirect connection request REQ to the server through the forwarding of the LoRa gatewaynodeIf the video acquisition terminal receives the indirect connection response REP of the serverserverIf not, executing step 1.1.5, otherwise, executing step 1.1.2;
step 1.1.5, set up Comm of video acquisition terminal itselftypIs ComminStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.6, set up Comm of video acquisition terminal itselftypIs CommdiStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.7, after waiting for the timer T1 to trigger, step 1.1.2 is executed.
In the first step, in the process of establishing the self-adaptive communication connection, the server performs the following steps:
step 1.2.1, create timers T21 and T22;
step 1.2.2, checking whether a TCP connection which is successfully established exists between the server and the video acquisition terminal, if so, executing step 1.2.8, otherwise, executing step 1.2.3;
step 1.2.3, checking whether the server is in a LISTEN state of the TCP connection, if so, executing step 1.2.4, otherwise, setting the TCP connection to be in the LISTEN state and then executing step 1.2.4;
step 1.2.4, the server sends an indirect connection request REQ to the video acquisition terminal through the forwarding of the LoRa gatewayserverIf receiving the indirect connection response REP of the video acquisition terminalnodeIf not, executing step 1.2.7, otherwise, executing step 1.2.5;
step 1.2.5, starting a timer T21;
step 1.2.6, if the server receives the TCP connection request sent by the video acquisition terminal and the TCP connection is successfully established before the timer T21 is triggered, immediately closing the timer T21 and executing step 1.2.8, otherwise, executing step 1.2.3 again after the timer T21 is triggered;
step 1.2.7, set up Comm of server itselftypIs ComminStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.8, set Comm of the server itselftypIs CommdiStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.9, after waiting for the timer T22 to trigger, step 1.2.2 is executed.
Step two, the video acquisition terminal and the server perform self-adaptive communication interaction;
in the second step, in the process of self-adaptive communication interaction, the video acquisition terminal and the server are carried out according to the following steps:
step 2.1, examine its own CommtypIf CommtypIs CommdiStep 2.2 is executed if CommtypIs ComminThen step 2.3 is executed;
2.2, the video acquisition terminal and the server carry out direct communication interaction;
and 2.3, the video acquisition terminal and the server are forwarded through the LoRa gateway to perform communication interaction.
Step three, the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
step four, the server sends an instruction to the video acquisition terminal, and the mode of sending acquisition information to the server by the video acquisition terminal is designated as a self-adaptive mode or a small data mode;
in the fourth step, the self-adaptive mode is carried out according to the following steps:
step 4.1, the video acquisition terminal checks Comm of the video acquisition terminaltypIf CommtypIs CommdiStep 4.2 is executed if CommtypIs ComminThen step 4.3 is executed;
step 4.2, the video acquisition terminal directly sends the acquired video and the object type identification result in the acquired video to the server;
4.3, the video acquisition terminal only sends the object type identification result in the acquired video to the LoRa gateway, and then the LoRa gateway forwards the object type identification result to the server;
in the fourth step, the small data mode is that the video acquisition terminal only sends the object type identification result in the acquired video to the server.
Step five, the server sends an instruction to the video acquisition terminal, and the video acquisition mode of the video acquisition terminal is designated to be a continuous video acquisition mode or a non-continuous video acquisition mode;
in the fifth step, the continuous video acquisition mode is that the video acquisition terminal continuously performs local video acquisition and object type identification in the acquired video, and transmits acquisition information to the server;
in the fifth step, the discontinuous video acquisition mode is carried out according to the following steps:
step 5.1, the video acquisition terminal establishes timers T31 and T32;
step 5.2, the video acquisition terminal starts local video acquisition and object type identification in the acquired video, and transmits acquisition information to the server;
step 5.3, the video acquisition terminal starts a timer T31;
step 5.4, if the object type needing to be identified is identified before the timer T31 is triggered, immediately closing the timer T31 and executing the step 5.3 again, otherwise, executing the step 5.5 after the timer T31 is triggered;
step 5.5, the video acquisition terminal stops local video acquisition, acquires object type identification in the video, sends acquisition information to the server and starts a timer T32;
and 5.6, if the video acquisition terminal detects animals and people before the timer T32 is triggered, or the ultrasonic ranging module detects that an object enters and moves, immediately closing the timer T32 and re-executing the step 5.2, or else, re-executing the step 5.2 after the timer T32 is triggered.
And sixthly, the video acquisition terminal performs local video acquisition and object type identification in the acquired video and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode.
In the method, the object type identification in the collected video is carried out according to the following steps:
step 6.1, the video acquisition terminal establishes a timer T4, and the video acquisition terminal sets thresholds p and VAL1And VAL2；
Step 6.2, m object classes are set in the model for object class identification, and the ith object class CAT isiSet variable Numi、ACU_SUMiAnd ACU _ AVEi，i＝1,2,3,…,m；
Step 6.3, let Numi＝0，ACU_SUMi＝0，ACU_AVEi＝0，i＝1,2,3,…M, and let j equal 0;
6.4, the video acquisition terminal acquires a real-time single-frame image in the acquired video, identifies the object type in the single-frame image, obtains the first N object types with the maximum identification probability value, and sequentially obtains CAT from large to smalla，CATb，CATc，CATdAnd CATeWherein a, b, c, d, e ∈ {1,2,3, …, m }, that is, the a-th, b-th, c-th, d-th and e-th object classes in the model for object class identification are the top N object classes with the highest identification probability values in the single-frame image identification result, and the obtained corresponding identification probability values are ACU in turna，ACUb，ACUc，ACUd，ACUeAnd make Numi＝Numi+1，ACU_SUMi＝ACU_SUMi+ACUi，i＝a,b,c,d,e；
In step 6.4, the object type in the single frame image is identified according to the following steps:
step 6.4.1, loading a model for identifying the object type, wherein the loading comprises the structure and the weight of the model;
step 6.4.2, image preprocessing is carried out on the acquired single-frame image, size scaling and pixel value scaling are carried out on the image according to the interface standard of the model, and the image is adjusted to be in an RGB independent three-channel format;
step 6.4.3, inputting the preprocessed image data into a model to obtain the probability value of each classification;
and 6.4.4, selecting the top N categories with the maximum probability values as recognition results, and recording the corresponding recognition probability values.
Step 6.5, if ACUa≥VAL1If not, executing step 6.11, otherwise, executing step 6.6;
step 6.6, let j equal j +1, and start timer T4, and execute step 6.7 after timer T4 triggers;
step 6.7, if j < p, re-executing step 6.4, otherwise executing step 6.8;
step 6.8, let ACU _ AVEi＝ACU_SUMi/NumiSelecting the first N object categories with the largest ACU _ AVE value to form a set S;
step 6.9, judging whether the object type to be identified belongs to the set S, if so, executing step 6.10, otherwise, executing step 6.3;
step 6.10, if the ACU _ AVE corresponding to the object type to be identified is greater than or equal to VAL2If the object type is not identified, the video acquisition terminal identifies the object type to be identified in the acquired video, and then the step 6.3 is skipped;
step 6.11, judging the object type and ACU to be identifiedaAnd (4) whether the corresponding object types are consistent or not, if so, identifying the object type to be identified in the acquired video by the video acquisition terminal, otherwise, not identifying, and then skipping to the step 6.3.
As a preferred scheme of this embodiment, the method employs the intelligent video capture system in embodiment 1.
Application example 1:
the example 2 of the present application is applied to a scene in the field of security monitoring, and as shown in fig. 1, the video capture terminal is deployed in an urban environment with better power supply conditions and network access conditions, so that the video capture terminal can access an IP network through a built-in network card or a wireless network card module according to the network access conditions of a site to directly communicate with a server.
In the scenario shown in fig. 1, in step four, the server sends an instruction to the video capture terminal, and the mode in which the video capture terminal sends capture information to the server is designated as an adaptive mode.
In the scene shown in fig. 1, in step five, the server sends an instruction to the video capture terminal, and designates the video capture mode of the video capture terminal as a continuous video capture mode.
In the scene shown in fig. 1, the model for identifying the object type is preferably a MobileNet model, and the value of m is 1000.
In the scenario shown in FIG. 1, p is preferably 10, VAL1Preferably 70%, VAL2Is 10%, and N is preferably takenThe value was 5.
In the scene shown in fig. 1, in step 6.4.2, image preprocessing is performed on the acquired single frame image, the image size is adjusted to 224 × 224 according to the interface standard of the MobileNet model, each pixel value range is adjusted to [ -1,1], and the image is adjusted to an RGB independent three-channel format.
In the scenario shown in fig. 1, when some problems occur and the video capture terminal cannot access the IP network temporarily, the video capture terminal will immediately switch to a standby LoRa communication means, and only send the object class identification result in the captured video to the LoRa gateway and then forward the object class identification result to the server by the LoRa gateway while performing local video capture and object class identification in the captured video, so that the entire system has better adaptability and reliability.
Application example 2:
the scene that embodiment 2 was applied to the wild animal observation field is given in this application example, as shown in fig. 2, the field region that video acquisition terminal deployed can't provide mains supply and operator's network signal can't cover, and video acquisition terminal is in the coverage of LoRa gateway, and LoRa gateway adopts mains supply power supply and inserts IP network and server communication.
In the scenario shown in fig. 2, in step four, the server sends an instruction to the video capture terminal, and the mode in which the video capture terminal sends capture information to the server is designated as a small data mode.
In the scene shown in fig. 2, in the fifth step, the server sends an instruction to the video capture terminal, and the video capture mode of the video capture terminal is designated as a discontinuous video capture mode.
In the scene shown in fig. 2, the model for identifying the object type is preferably a MobileNet model, and the value of m is 1000.
In the scenario shown in FIG. 2, p is preferably 10, VAL1Preferably 70%, VAL2Is 10%, and N is preferably 5.
In the scene shown in fig. 2, in step 6.4.2, image preprocessing is performed on the acquired single frame image, the image size is adjusted to 224 × 224 according to the interface standard of the MobileNet model, each pixel value range is adjusted to [ -1,1], and the image is adjusted to an RGB independent three-channel format.
In the scene shown in fig. 2, the video acquisition terminal adopts a discontinuous video acquisition working mode and can acquire video in time after detecting that the situation of the video acquisition area changes, so that the video acquisition terminal can better reduce energy consumption and effectively acquire the video of wild animals although the battery is adopted for power supply; and because the video acquisition terminal adopts a small data transmission mode to transmit the acquisition information to the server, monitoring personnel of the server can acquire the field situation as much as possible.
In conclusion, the intelligent video acquisition method provided by the invention can acquire videos and identify the object types in the acquired videos, can further improve the identification accuracy on the basis of the existing identification model in the identification process, and can adopt a proper video acquisition mode and a proper acquisition information sending mode according to the actual situation, so that the intelligent video acquisition method has better adaptability to different environments such as urban areas, fields and the like.

Claims (8)

1. An intelligent video acquisition method adopts a video acquisition terminal, a LoRa gateway and a server, and is characterized by specifically comprising the following steps:
step one, self-adaptive communication connection is established between the video acquisition terminal and the server, and the self-adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
The direct communication mode CommdiThe video acquisition terminal and the server are directly communicated with each other without forwarding through an LoRa gateway;
the indirect communication mode ComminThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
step two, the video acquisition terminal and the server carry out self-adaptive communication interaction;
step three, the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
step four, the server sends an instruction to the video acquisition terminal, and the mode of sending acquisition information to the server by the video acquisition terminal is designated as a self-adaptive mode or a small data mode;
step five, the server sends an instruction to the video acquisition terminal, and the video acquisition mode of the video acquisition terminal is designated to be a continuous video acquisition mode or a non-continuous video acquisition mode;
step six, the video acquisition terminal carries out local video acquisition and object type identification in the acquired video, and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode;
the method comprises the following steps of:
step 6.1, the video acquisition terminal establishes a timer T4, and the video acquisition terminal sets thresholds p and VAL1And VAL2；
Step 6.2, m object classes are set in the model for object class identification, and the ith object class CAT isiSet variable Numi、ACU_SUMiAnd ACU _ AVEi，i＝1,2,3,…,m；
Step 6.3, let Numi＝0，ACU_SUMi＝0，ACU_AVEi0, i-1, 2,3, …, m, and let j-0;
6.4, the video acquisition terminal acquires a real-time single-frame image in the acquired video, identifies the object type in the single-frame image, obtains the first N object types with the maximum identification probability value, and sequentially obtains CAT from large to smalla，CATb，CATc，CATdAnd CATeWhere a, b, c, d, e ∈ {1,2,3, …, m }, i.e., forThe a-th, b-th, c-th, d-th and e-th object categories in the object category identification model are the first N object categories with the maximum identification probability values in the single-frame image identification result, and the obtained corresponding identification probability values are ACU in sequencea，ACUb，ACUc，ACUd，ACUeAnd make Numi＝Numi+1，ACU_SUMi＝ACU_SUMi+ACUi，i＝a,b,c,d,e；
Step 6.5, if ACUa≥VAL1If not, executing step 6.11, otherwise, executing step 6.6;
step 6.6, let j equal j +1, and start timer T4, and execute step 6.7 after timer T4 triggers;
step 6.7, if j < p, re-executing step 6.4, otherwise executing step 6.8;
step 6.8, let ACU _ AVEi＝ACU_SUMi/NumiSelecting the first N object categories with the largest ACU _ AVE value to form a set S;
step 6.9, judging whether the object type to be identified belongs to the set S, if so, executing step 6.10, otherwise, executing step 6.3;
step 6.10, if the ACU _ AVE corresponding to the object type to be identified is greater than or equal to VAL2If the object type is not identified, the video acquisition terminal identifies the object type to be identified in the acquired video, and then the step 6.3 is skipped;
step 6.11, judging the object type and ACU to be identifiedaAnd (4) whether the corresponding object types are consistent or not, if so, identifying the object type to be identified in the acquired video by the video acquisition terminal, otherwise, not identifying, and then skipping to the step 6.3.
2. The intelligent video acquisition method according to claim 1, wherein in the first step, in the process of establishing the adaptive communication connection, the video acquisition terminal performs the following steps:
step 1.1.1, creating a timer T1;
step 1.1.2, checking whether a TCP connection which is successfully established exists between the video acquisition terminal and the server, if so, executing step 1.1.6, otherwise, executing step 1.1.3;
step 1.1.3, the video acquisition terminal initiates a TCP connection request to the server, if the TCP connection is successfully established, step 1.1.6 is executed, otherwise step 1.1.4 is executed;
step 1.1.4, the video acquisition terminal sends an indirect connection request REQ to the server through the forwarding of the LoRa gatewaynodeIf the video acquisition terminal receives the indirect connection response REP of the serverserverIf not, executing step 1.1.5, otherwise, executing step 1.1.2;
step 1.1.5, set up Comm of video acquisition terminal itselftypIs ComminStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.6, set up Comm of video acquisition terminal itselftypIs CommdiStarting a timer T1, and jumping to the step 1.1.7;
step 1.1.7, after waiting for the timer T1 to trigger, step 1.1.2 is executed.
3. The intelligent video capture method of claim 1, wherein in step one, in the process of establishing the adaptive communication connection, the server performs the following steps:
step 1.2.1, create timers T21 and T22;
step 1.2.2, checking whether a TCP connection which is successfully established exists between the server and the video acquisition terminal, if so, executing step 1.2.8, otherwise, executing step 1.2.3;
step 1.2.3, checking whether the server is in a LISTEN state of the TCP connection, if so, executing step 1.2.4, otherwise, setting the TCP connection to be in the LISTEN state and then executing step 1.2.4;
step 1.2.4, the server sends an indirect connection request REQ to the video acquisition terminal through the forwarding of the LoRa gatewayserverIf receiving the indirect connection response REP of the video acquisition terminalnodeIf yes, execute step 1.2.7, otherwise executeStep 1.2.5;
step 1.2.5, starting a timer T21;
step 1.2.6, if the server receives the TCP connection request sent by the video acquisition terminal and the TCP connection is successfully established before the timer T21 is triggered, immediately closing the timer T21 and executing step 1.2.8, otherwise, executing step 1.2.3 again after the timer T21 is triggered;
step 1.2.7, set up Comm of server itselftypIs ComminStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.8, set Comm of the server itselftypIs CommdiStarting a timer T22, and jumping to the step 1.2.9;
step 1.2.9, after waiting for the timer T22 to trigger, step 1.2.2 is executed.
4. The intelligent video collection method according to claim 1, wherein in the second step, in the adaptive communication interaction process, the video collection terminal and the server are both performed according to the following steps:
step 2.1, check its CommtypIf CommtypIs CommdiStep 2.2 is executed if CommtypIs ComminThen step 2.3 is executed;
2.2, the video acquisition terminal and the server carry out direct communication interaction;
and 2.3, the video acquisition terminal and the server are forwarded through the LoRa gateway to perform communication interaction.
5. The intelligent video capture method of claim 1, wherein in step four, the adaptive mode is performed according to the following steps:
step 4.1, the video acquisition terminal checks Comm of the video acquisition terminaltypIf CommtypIs CommdiStep 4.2 is executed if CommtypIs ComminThen step 4.3 is executed;
step 4.2, the video acquisition terminal directly sends the acquired video and the object type identification result in the acquired video to the server;
4.3, the video acquisition terminal only sends the object type identification result in the acquired video to the LoRa gateway, and then the LoRa gateway forwards the object type identification result to the server;
in the fourth step, the small data mode is that the video acquisition terminal only sends the object type identification result in the acquired video to the server.
6. The intelligent video acquisition method according to claim 1, wherein in step five, the continuous video acquisition mode is that the video acquisition terminal continuously performs local video acquisition and object type identification in the acquired video, and sends acquisition information to the server;
in the fifth step, the discontinuous video acquisition mode is carried out according to the following steps:
step 5.1, the video acquisition terminal establishes timers T31 and T32;
step 5.2, the video acquisition terminal starts local video acquisition and object type identification in the acquired video, and transmits acquisition information to the server;
step 5.3, the video acquisition terminal starts a timer T31;
step 5.4, if the object type needing to be identified is identified before the timer T31 is triggered, immediately closing the timer T31 and executing the step 5.3 again, otherwise, executing the step 5.5 after the timer T31 is triggered;
step 5.5, the video acquisition terminal stops local video acquisition, acquires object type identification in the video, sends acquisition information to the server and starts a timer T32;
and 5.6, if the video acquisition terminal detects animals and people before the timer T32 is triggered, or the ultrasonic ranging module detects that an object enters and moves, immediately closing the timer T32 and re-executing the step 5.2, or else, re-executing the step 5.2 after the timer T32 is triggered.
7. The intelligent video capture method of claim 1, wherein in step 6.4, the identification of the object class in the single frame image is performed according to the following steps:
step 6.4.1, loading a model for identifying the object type, wherein the loading comprises the structure and the weight of the model;
step 6.4.2, image preprocessing is carried out on the acquired single-frame image, size scaling and pixel value scaling are carried out on the image according to the interface standard of the model, and the image is adjusted to be in an RGB independent three-channel format;
step 6.4.3, inputting the preprocessed image data into a model to obtain the probability value of each classification;
and 6.4.4, selecting the top N categories with the maximum probability values as recognition results, and recording the corresponding recognition probability values.
8. An intelligent video acquisition system comprises a video acquisition terminal, a LoRa gateway and a server, and is characterized in that the video acquisition terminal comprises a card computer, and a camera module, a wireless internet access card module, a LoRa wireless data transmission module, an infrared pyroelectric module and an ultrasonic ranging module which are respectively connected with the card computer; the video acquisition terminal is connected with an IP network through a built-in network card or a wireless network card module of the card computer;
the video acquisition terminal and the server establish self-adaptive communication connection, and the self-adaptive communication connection specifically comprises two communication modes Commtyp: comm in direct communication modediAnd indirect communication mode Commin；
The direct communication mode CommdiThe video acquisition terminal and the server are directly communicated with each other without forwarding through an LoRa gateway;
the indirect communication mode ComminThe video acquisition terminal and the server are communicated with each other through forwarding of the LoRa gateway;
the video acquisition terminal and the server carry out self-adaptive communication interaction;
the server sends an instruction to the video acquisition terminal to specify the object type to be identified by the video acquisition terminal;
the server sends an instruction to the video acquisition terminal, and the mode of sending acquisition information to the server by the video acquisition terminal is designated as a self-adaptive mode or a small data mode;
the server sends an instruction to the video acquisition terminal, and designates the video acquisition mode of the video acquisition terminal as a continuous video acquisition mode or a non-continuous video acquisition mode;
the video acquisition terminal carries out local video acquisition and object type identification in the acquired video and sends acquisition information to the server, and the video acquisition terminal receives an instruction sent by the server at any time so as to update the object type to be identified or send the acquisition information to the server or the video acquisition mode;
the method comprises the following steps of:
step 6.1, the video acquisition terminal establishes a timer T4, and the video acquisition terminal sets thresholds p and VAL1And VAL2；
Step 6.2, m object classes are set in the model for object class identification, and the ith object class CAT isiSet variable Numi、ACU_SUMiAnd ACU _ AVEi，i＝1,2,3,…,m；
Step 6.3, let Numi＝0，ACU_SUMi＝0，ACU_AVEi0, i-1, 2,3, …, m, and let j-0;
6.4, the video acquisition terminal acquires a real-time single-frame image in the acquired video, identifies the object type in the single-frame image, obtains the first N object types with the maximum identification probability value, and sequentially obtains CAT from large to smalla，CATb，CATc，CATdAnd CATeWherein a, b, c, d, e ∈ {1,2,3, …, m }, that is, the a-th, b-th, c-th, d-th and e-th object classes in the model for object class identification are the top N object classes with the highest identification probability values in the single-frame image identification result, and are obtainedTheir corresponding recognition probability values are in turn ACUa，ACUb，ACUc，ACUd，ACUeAnd make Numi＝Numi+1，ACU_SUMi＝ACU_SUMi+ACUi，i＝a,b,c,d,e；
Step 6.5, if ACUa≥VAL1If not, executing step 6.11, otherwise, executing step 6.6;
step 6.6, let j equal j +1, and start timer T4, and execute step 6.7 after timer T4 triggers;
step 6.7, if j < p, re-executing step 6.4, otherwise executing step 6.8;
step 6.8, let ACU _ AVEi＝ACU_SUMi/NumiSelecting the first N object categories with the largest ACU _ AVE value to form a set S;
step 6.9, judging whether the object type to be identified belongs to the set S, if so, executing step 6.10, otherwise, executing step 6.3;
step 6.10, if the ACU _ AVE corresponding to the object type to be identified is greater than or equal to VAL2If the object type is not identified, the video acquisition terminal identifies the object type to be identified in the acquired video, and then the step 6.3 is skipped;
step 6.11, judging the object type and ACU to be identifiedaAnd (4) whether the corresponding object types are consistent or not, if so, identifying the object type to be identified in the acquired video by the video acquisition terminal, otherwise, not identifying, and then skipping to the step 6.3.
CN202010553474.4A 2020-06-17 2020-06-17 Intelligent video acquisition system and method Active CN111866449B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010553474.4A CN111866449B (en) 2020-06-17 2020-06-17 Intelligent video acquisition system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010553474.4A CN111866449B (en) 2020-06-17 2020-06-17 Intelligent video acquisition system and method

Publications (2)

Publication Number Publication Date
CN111866449A CN111866449A (en) 2020-10-30
CN111866449B true CN111866449B (en) 2022-03-29

Family

ID=72986814

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010553474.4A Active CN111866449B (en) 2020-06-17 2020-06-17 Intelligent video acquisition system and method

Country Status (1)

Country Link
CN (1) CN111866449B (en)

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101098199A (en) * 2006-06-29 2008-01-02 中国科学院半导体研究所 Intelligent monitor system
CN203134031U (en) * 2013-01-04 2013-08-14 中国电子科技集团公司第二十八研究所 Perimeter protecting system based on wireless sensor network
CN206742969U (en) * 2017-05-16 2017-12-12 四川省东宇信息技术有限责任公司 A kind of intelligent terminal based on electric power property
CN107728680A (en) * 2017-09-11 2018-02-23 江苏大学 A kind of remote pig house environment multiparameter measurement and control system and its method based on LoRa
WO2018121018A1 (en) * 2016-12-30 2018-07-05 腾讯科技（深圳）有限公司 Picture identification method and device, server and storage medium
CN108683576A (en) * 2018-05-14 2018-10-19 成都零点六八科技有限公司 A kind of intelligent household communication system based on dual mode communication
CN108830252A (en) * 2018-06-26 2018-11-16 哈尔滨工业大学 A kind of convolutional neural networks human motion recognition method of amalgamation of global space-time characteristic
CN208156681U (en) * 2018-01-20 2018-11-27 南京铁道职业技术学院 A kind of video monitoring image identifying system
CN109444536A (en) * 2018-11-23 2019-03-08 广东电网有限责任公司 A kind of electric power consumption detection device of adaptive signal transmission
US10289910B1 (en) * 2014-07-10 2019-05-14 Hrl Laboratories, Llc System and method for performing real-time video object recognition utilizing convolutional neural networks
CN109919008A (en) * 2019-01-23 2019-06-21 平安科技（深圳）有限公司 Moving target detecting method, device, computer equipment and storage medium
CN109963046A (en) * 2017-12-26 2019-07-02 原盛科技股份有限公司 Movement detection device and related mobile detection method
CN110113575A (en) * 2019-05-14 2019-08-09 嘉应学院 A kind of agriculture feelings information real-time monitoring platform based on NB-IoT
CN111225178A (en) * 2018-11-23 2020-06-02 杭州眼云智家科技有限公司 Video monitoring method and system based on object detection

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9336456B2 (en) * 2012-01-25 2016-05-10 Bruno Delean Systems, methods and computer program products for identifying objects in video data
US10417501B2 (en) * 2017-12-06 2019-09-17 International Business Machines Corporation Object recognition in video

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101098199A (en) * 2006-06-29 2008-01-02 中国科学院半导体研究所 Intelligent monitor system
CN203134031U (en) * 2013-01-04 2013-08-14 中国电子科技集团公司第二十八研究所 Perimeter protecting system based on wireless sensor network
US10289910B1 (en) * 2014-07-10 2019-05-14 Hrl Laboratories, Llc System and method for performing real-time video object recognition utilizing convolutional neural networks
WO2018121018A1 (en) * 2016-12-30 2018-07-05 腾讯科技（深圳）有限公司 Picture identification method and device, server and storage medium
CN206742969U (en) * 2017-05-16 2017-12-12 四川省东宇信息技术有限责任公司 A kind of intelligent terminal based on electric power property
CN107728680A (en) * 2017-09-11 2018-02-23 江苏大学 A kind of remote pig house environment multiparameter measurement and control system and its method based on LoRa
CN109963046A (en) * 2017-12-26 2019-07-02 原盛科技股份有限公司 Movement detection device and related mobile detection method
CN208156681U (en) * 2018-01-20 2018-11-27 南京铁道职业技术学院 A kind of video monitoring image identifying system
CN108683576A (en) * 2018-05-14 2018-10-19 成都零点六八科技有限公司 A kind of intelligent household communication system based on dual mode communication
CN108830252A (en) * 2018-06-26 2018-11-16 哈尔滨工业大学 A kind of convolutional neural networks human motion recognition method of amalgamation of global space-time characteristic
CN109444536A (en) * 2018-11-23 2019-03-08 广东电网有限责任公司 A kind of electric power consumption detection device of adaptive signal transmission
CN111225178A (en) * 2018-11-23 2020-06-02 杭州眼云智家科技有限公司 Video monitoring method and system based on object detection
CN109919008A (en) * 2019-01-23 2019-06-21 平安科技（深圳）有限公司 Moving target detecting method, device, computer equipment and storage medium
CN110113575A (en) * 2019-05-14 2019-08-09 嘉应学院 A kind of agriculture feelings information real-time monitoring platform based on NB-IoT

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
赵浩如等."基于RPN与B-CNN的细粒度图像分类算法研究".《计算机应用与软件》.2019,第36卷(第3期), *
钮赛赛等."基于YOLO智能网络的红外弱小多目标检测技术".《上海航天》.2019,第36卷(第5期), *

Also Published As

Publication number Publication date
CN111866449A (en) 2020-10-30

Similar Documents

Publication Publication Date Title
US8171129B2 (en) Smart endpoint and smart monitoring system having the same
CN110730327A (en) Method for monitoring power transmission line and acquisition front end
JP2023524243A (en) Smart livestock management system and method
CN111898524A (en) 5G edge computing gateway and application thereof
CN105139471A (en) Inspection method for electrical equipment
CN105701945A (en) Intelligent monitoring alarm positioning system and method for community
CN106447873A (en) 5G base station on-line management system based on Internet of Things
CN109544870A (en) Alarm decision method and intelligent monitor system for intelligent monitor system
US11257355B2 (en) System and method for preventing false alarms due to display images
CN115312052A (en) Processing system based on digital people
CN111866449B (en) Intelligent video acquisition system and method
CN112016380B (en) Wild animal monitoring method and system
CN208314959U (en) A kind of intelligent safety and defence system
CN115083132B (en) Research and judgment method for reducing false alarm rate of fire alarm
CN111447422A (en) Audio and video data acquisition method and device and computer equipment
CN106829658B (en) A kind of elevator device with three-level monitoring network
US20220335725A1 (en) Monitoring presence or absence of an object using local region matching
US20210274133A1 (en) Pre-generating video event notifications
CN113592003B (en) Picture transmission method, device, equipment and storage medium
CN116740874A (en) Intrusion detection method and related device
CN209928591U (en) Based on network science and technology control early warning system
CN111768534A (en) Service control method based on biological recognition, acquisition equipment and recognition equipment
CN212181614U (en) Accurate old person system at home of being convenient for operation
CN107765700A (en) Use householder method, system and the aircraft of mobile terminal
CN113128297A (en) Equipment docking method and device, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant