CN114359976B - Intelligent security method and device based on person identification - Google Patents
Intelligent security method and device based on person identification Download PDFInfo
- Publication number
- CN114359976B CN114359976B CN202210269953.2A CN202210269953A CN114359976B CN 114359976 B CN114359976 B CN 114359976B CN 202210269953 A CN202210269953 A CN 202210269953A CN 114359976 B CN114359976 B CN 114359976B
- Authority
- CN
- China
- Prior art keywords
- target
- identity information
- graph
- track
- difference degree
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000010586 diagram Methods 0.000 claims description 80
- 230000004044 response Effects 0.000 claims description 25
- 238000005516 engineering process Methods 0.000 claims description 19
- 238000012216 screening Methods 0.000 claims description 19
- 230000002159 abnormal effect Effects 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 7
- 238000012546 transfer Methods 0.000 claims description 5
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000013473 artificial intelligence Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 210000000746 body region Anatomy 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
Abstract
The invention provides an intelligent security method and device based on character recognition, wherein the method comprises the following steps: the method comprises the steps of collecting shot pictures through a person identification and collection device, identifying identity information of each person in the shot pictures, determining target identity information of persons who do not live in a target area system where the main control device is located, collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information, generating a person track graph of the person corresponding to the target identity information, calculating the difference degree between the track graph and tracks of other residents, and judging whether the person corresponding to the target identity information needs to be pre-warned according to the difference degree. The invention has the beneficial effects that: the method and the device have the advantages that the figure is analyzed and early-warned based on the shot picture, so that the safety in the region is guaranteed, and the safety of the region can be effectively improved.
Description
Technical Field
The invention relates to the field of intelligent security, in particular to an intelligent security method and device based on character recognition.
Background
The monitoring video is widely applied to various fields such as security, business and industrial production, the object needing attention in the monitoring video is a person, the important significance is achieved on danger early warning, at present, the identification of the person is greatly limited, and the person cannot be analyzed and early warned based on a shot picture generally.
Disclosure of Invention
The invention mainly aims to provide an intelligent security method, an intelligent security device, an intelligent security equipment and a storage medium based on character recognition, and aims to solve the problem that the character recognition cannot analyze and early warn characters based on shot pictures.
The invention provides an intelligent security method based on person identification, which is applied to each pre-divided area system, wherein each area system comprises a main control device and a plurality of person identification and acquisition devices, each person identification and acquisition device at least comprises a camera, a shot picture of a person is acquired through the camera and uploaded to the main control device, the main control device stores position information corresponding to each person identification and acquisition device in the area system and resident information of each resident in the area system, a block chain is formed among the main control devices in each area system, each main control device is a block chain node of the block chain, and the method comprises the following steps:
The main control equipment acquires a plurality of shot pictures according to a time sequence through the figure identification acquisition equipment;
identifying the identity information of each figure in the shot picture through a preset face identification technology;
screening target identity information of people who do not live in a target area system where the master control equipment is located from the identity information based on the resident information; the target area system is one of the area systems divided in advance;
based on the position information between the target area system and other area systems, the target identity information is sent to target main control equipment of the area system in a preset interval range in the block chain;
receiving feedback information of the target master control equipment;
if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, performing character marking on the shot pictures based on the target identity information, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
sequencing the shot images in the shot image set according to the time sequence to respectively obtain a first characteristic diagram, a second characteristic diagram, an N characteristic diagram;
Calculating a figure track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph, and calculating the difference degree between the figure track graph and the tracks of other residents in the target area system;
judging whether the figure corresponding to the target identity information is abnormal or not based on the difference degree;
and if the target identity information is abnormal, pre-warning the figure corresponding to the target identity information.
Further, the step of calculating a person trajectory diagram corresponding to the target identity information according to the first feature diagram, the second feature diagram, and the nth feature diagram, and calculating a difference degree between the person trajectory diagram and trajectories of other residents in the target area system includes:
inputting each shot image into a preset CNN backbone network to obtain a first standard characteristic tensor map corresponding to the first characteristic map, a second standard characteristic tensor map, a standard characteristic tensor map and an Nth standard characteristic tensor map, wherein the second standard characteristic tensor map and the Nth standard characteristic tensor map respectively correspond to the second characteristic map; the sizes of all the standard characteristic tensor maps are C multiplied by H multiplied by W;
recording the standard feature tensor graphs adjacent in time sequence as a group of data, and selecting (H + 1-K) x (W + 1-K) target blocks with K x K size from one sliding window of the standard feature tensor graphs;
Performing cross-correlation operation on each target block relative to another characteristic tensor diagram in a group of corresponding data to obtain each target blockA response graph of the target blocks relative to the characteristic tensor map; wherein the manner of the cross-correlation operation is
，
A response diagram of the ith target block of the jth standard feature diagram is shown,
showing the ith target block of the jth standard feature diagram,
representing another characteristic tensor map;
according to the formula
Fusing each response graph in a group of data to obtain an attention graph;
generating a person track graph of a person corresponding to the target identity information based on each attention graph;
and expressing the figure track graph according to a pixel coordinate sequence, and calculating the difference degree between the figure track graph and other resident tracks in the target area system based on the pixel coordinate sequence.
Further, the step of representing the person trajectory graph according to a pixel coordinate sequence and calculating the difference between the person trajectory graph and the trajectories of other residents in the target area system based on the pixel coordinate sequence includes:
screening and intercepting a plurality of target track graph sets which are the same as or similar to the starting point and the end point of the figure track graph from other resident tracks;
Calculating the position difference degree, the speed difference degree and the direction difference degree of the character track map and the target track map set according to the pixel coordinate sequence; wherein the formula for calculating the position difference degree is
The formula for calculating the velocity difference degree is
，
The formula for calculating the direction difference degree is
，
and calculating the difference degree of the person track graph and the track graph set according to the position difference degree, the speed difference degree and the direction difference degree.
Further, after the step of representing the person trajectory graph according to a pixel coordinate sequence and calculating the difference between the person trajectory graph and the trajectories of other residents in the target area system based on the pixel coordinate sequence, the method further includes:
adjusting the risk level of the target identity information based on the difference degree, and uploading the risk level to a block chain;
judging whether the risk level reaches a preset risk level or not;
if yes, recording the target identity information into a blacklist, and storing the figure track graph so as to facilitate subsequent calling and analysis.
Further, after the step of calculating the difference degree of the positions, the difference degree of the speeds, and the difference degree of the directions of the human track map and the target track map set according to the pixel coordinate sequence, the method includes:
acquiring a first three-dimensional data set of the position, the speed and the direction of each pixel point of the figure track graph and a second three-dimensional data set of the position, the speed and the direction of the resident track graph;
calculating the similarity of the first three-dimensional data set and each second three-dimensional data set;
pre-associating the target households corresponding to the household track graphs with the similarity greater than the preset similarity;
Calling first identity information of the target resident, and calling a shot picture simultaneously having the target resident and a character corresponding to the target identity information based on the first identity information and the target identity information;
judging whether the number of pictures in the shot pictures, in which the distance between the target resident and the figure corresponding to the target identity information is smaller than a preset value, is larger than the preset value or not;
and if so, associating the target resident with the figure corresponding to the target identity information.
Further, if the step of warning the person corresponding to the target identity information is abnormal, the method further includes:
monitoring whether the time length for the person corresponding to the target identity information to break away from the shot picture reaches a preset time length or not through the camera;
if yes, first position information of a camera of a figure corresponding to the target identity information is obtained through the last shooting;
and feeding back to a similar regional system in the block chain based on the first position information.
The invention also provides an intelligent security device based on person identification, which is applied to each pre-divided area system, the area system comprises a main control device and a plurality of person identification and acquisition devices, the person identification and acquisition devices at least comprise cameras, shooting pictures of persons are acquired through the cameras and uploaded to the main control device, the main control device stores position information corresponding to each person identification and acquisition device in the area system and resident information of each resident in the area system, a block chain is formed among the main control devices in each area system, each main control device is a block chain node of the block chain, and the intelligent security device based on person identification comprises:
The acquisition module is used for acquiring a plurality of shot pictures according to a time sequence through the person identification acquisition equipment;
the identification module is used for identifying the identity information of each figure in the shot picture through a preset face identification technology;
the screening module is used for screening target identity information of people who do not live in the target area system where the main control equipment is located from the identity information based on the resident information; the target area system is one of the area systems divided in advance;
a sending module, configured to send the target identity information to a target master control device of a regional system within a preset interval range in the block chain based on location information between the target regional system and another regional system;
the receiving module is used for receiving the feedback information of the target main control equipment;
the marking module is used for marking people in the shot pictures based on the target identity information if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
The sorting module is used for sorting the shot images in the shot image set according to the time sequence to respectively obtain a first feature map, a second feature map, an Nth feature map;
the difference degree calculation module is used for calculating a person track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph and calculating the difference degree between the person track graph and the tracks of other residents in the target area system;
the judging module is used for judging whether the person corresponding to the target identity information is abnormal or not based on the difference degree;
and the early warning module is used for early warning the figure corresponding to the target identity information if the target identity information is abnormal.
The invention has the beneficial effects that: the method comprises the steps of collecting shot pictures through a person identification and collection device, identifying identity information of each person in the shot pictures, determining target identity information of persons who do not live in a target area system where the main control device is located, collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information, generating a person track graph of the person corresponding to the target identity information, calculating the difference degree between the track graph and tracks of other residents, and judging whether the person corresponding to the target identity information needs to be pre-warned according to the difference degree. The method and the device have the advantages that the people are analyzed and early warned based on the shot picture, so that the safety in the region is guaranteed, the safety of the region can be effectively improved, and the loss of residents is reduced.
Drawings
Fig. 1 is a schematic flowchart of an intelligent security method based on person identification according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of an intelligent security method based on person identification according to an embodiment of the present invention.
The implementation, functional features and advantages of the present invention will be further described with reference to the accompanying drawings.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
It should be noted that all directional indicators (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used for explaining the relative position relationship between the components, the movement situation, etc. in a specific posture (as shown in the attached drawings), and if the specific posture is changed, the directional indicator is also changed accordingly, and the connection may be a direct connection or an indirect connection.
The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and B, may mean: a exists alone, A and B exist simultaneously, and B exists alone.
In addition, the descriptions related to "first", "second", etc. in the present invention are only for descriptive purposes and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In addition, technical solutions between various embodiments may be combined with each other, but must be realized by a person skilled in the art, and when the technical solutions are contradictory or cannot be realized, such a combination should not be considered to exist, and is not within the protection scope of the present invention.
Referring to fig. 1, the present invention provides an intelligent security method based on person identification, which is applied to each pre-divided area system, wherein the area system includes a main control device and a plurality of person identification and acquisition devices, each person identification and acquisition device at least includes a camera, a shooting picture of a person is acquired through the camera, and the shooting picture is uploaded to the main control device, the main control device stores position information corresponding to each person identification and acquisition device in the area system and household information of each household in the area system, a block chain is formed between the main control devices in each area system, each main control device is a block chain node of the block chain, and the method includes:
S1: the main control equipment acquires a plurality of shot pictures according to a time sequence through the figure identification acquisition equipment;
s2: identifying the identity information of each figure in the shot picture through a preset face identification technology;
s3: screening target identity information of people who do not live in a target area system where the master control equipment is located from the identity information based on the resident information; the target area system is one of the area systems divided in advance;
s4: based on the position information between the target area system and other area systems, the target identity information is sent to the target main control equipment of the area system in a preset interval range in the block chain;
s5: receiving feedback information of the target master control equipment;
s6: if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, performing character marking on the shot pictures based on the target identity information, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
s7: sequencing the shot images in the shot image set according to the time sequence to respectively obtain a first characteristic diagram, a second characteristic diagram, a.
S8: calculating a figure track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph, and calculating the difference degree between the figure track graph and the tracks of other residents in the target area system;
s9: judging whether the figure corresponding to the target identity information is abnormal or not based on the difference degree;
s10: and if the target identity information is abnormal, pre-warning the person corresponding to the target identity information.
As described in the above steps S1-S4, the main control device captures a plurality of photographed pictures in chronological order by the character recognition capturing device; identifying the identity information of each person in the shot picture through a preset face identification technology; screening out target identity information of people who do not live in a target area system where the main control equipment is located from the identity information based on the resident information; the target area system is one of the pre-divided area systems; and sending the target identity information to target main control equipment of the regional system in a preset interval range in the block chain based on the position information between the target regional system and other regional systems.
The reason why the shot pictures are collected according to the time sequence in the present application is that, in general, a pedestrian appears in the pictures continuously, so that it is impossible to collect the shot pictures in a jumping manner, and in addition, the time can be recorded according to the time sequence, of course, in the prior art, the shot pictures have the corresponding time, so that the shot time of each shot picture can be obtained accordingly, the preset face recognition technology is not limited, the face recognition technology mainly recognizes some features on the face, such as the recognition technologies of thousands of people and thousands of faces of the public security system, and the present application does not describe this much, and the identity information of each person in the shot pictures can be recognized according to the face recognition technology, and it should be noted that, if the public security system is not accessed, the identity information here can be the identity of each person marked by oneself, for example, the first passerby and the second passerby do not need detailed information of all people, but all areas have information in all areas, and in order to prevent privacy leakage, the information in all area systems is not shared, so that target identity information of people who do not live in a target area system where the main control equipment is located can be screened out based on resident information prestored in the area systems, and the target identity information is sent to other area systems to perform secondary judgment.
Receiving the target master device feedback information as described in the above steps S5-S8; if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, performing character marking on the shot pictures based on the target identity information, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information; sequencing the shot images in the shot image set according to the time sequence to respectively obtain a first characteristic diagram, a second characteristic diagram, an N characteristic diagram; inputting each shot image into a preset CNN backbone network to obtain a first standard feature tensor diagram corresponding to the first feature diagram, a second standard feature tensor diagram corresponding to the second feature diagram, and an Nth standard feature tensor diagram corresponding to the Nth feature diagram; the dimensions of each standard feature tensor map are C × H × W.
When other regional systems feed back information of people who do not have any correlation, people corresponding to the target identity information need to be marked to facilitate analysis of tracks of the people, so that the shot pictures of the people are collected to obtain a target shot picture set corresponding to each target identity information, then the target shot picture sets are sorted according to time sequence to obtain a plurality of feature maps, the feature maps are input into a CNN backbone network, the CNN backbone network is initialized by adopting a network weight pre-trained by ImageNet, and of course, the number and the size of the input pictures need to be kept consistent, and thus the obtained standard feature tensor maps are consistent.
In one embodiment, the step S8 of calculating a person trajectory diagram corresponding to the target identity information according to the first feature diagram, the second feature diagram, and the nth feature diagram, and calculating the degree of difference from trajectories of other residents in the target area system further includes:
s801: recording the standard feature tensor images adjacent in time sequence as a group of data, and selecting (H + 1-K) × (W + 1-K) target blocks with the size of K × K from one standard feature tensor image sliding window;
s802: performing cross-correlation operation on each target block relative to another characteristic tensor diagram in a group of corresponding data to obtain a response diagram of each target block relative to the characteristic tensor diagram; wherein the manner of the cross-correlation operation is
，
A response diagram showing the ith target block of the jth second standard feature diagram,
the ith target block of the jth second standard feature diagram is shown,
representing another characteristic tensor map;
s803: according to the formula
Fusing each response graph in a group of data to obtain an attention graph;
s804: generating a person track graph of a person corresponding to the target identity information based on each attention graph;
s805: expressing the figure track graph according to a pixel coordinate sequence, and calculating the difference degree between the figure track graph and other resident tracks in the target area system based on the pixel coordinate sequence;
As described in the above steps S801 to S805, the standard feature tensor maps adjacent in time sequence are regarded as a set of data, and one of the standard feature tensor map sliding windows selects (H + 1-K) × (W + 1-K) target blocks of K × K size; performing cross-correlation operation on each target block relative to another characteristic tensor diagram in a group of corresponding data to obtain a response diagram of each target block relative to the characteristic tensor diagram; wherein the manner of the cross-correlation operation is
，
A response diagram showing the ith target block of the jth second standard feature diagram,
the ith target block of the jth second standard feature diagram is shown,
representing another characteristic tensor map; according to the formula
And fusing each response graph in a group of data to obtain an attention graph.
The sliding window operation is to sequentially take target blocks with the size of K × K in the sequence in the feature map, for example, a region with the size of 2 × 2 in a squared figure needs to be sequentially moved by 4 sliding window operations to take a value. The cross-correlation operation mode can be realized through standard convolution, so that a single-channel tensor response graph can be obtained, as the higher the similarity of convolution calculation is, the larger the response value is, the response graph can find a region similar to a target block, so that the response values of an occlusion region and a background region can be weakened, each response graph represents the result of a local region, in order to obtain the complete response result of the two graphs, the two graphs need to be fused, so that the final attention diagram A is obtained, and as the response of the occlusion region and the background region is weaker, the attention diagram can focus on the shared unoccluded human body region between the two images. Generating a person track graph of a person corresponding to the target identity information based on each attention graph; expressing the figure track graph according to a pixel coordinate sequence, and calculating the difference degree between the figure track graph and other resident tracks in the target area system based on the pixel coordinate sequence; judging whether the figure corresponding to the target identity information is abnormal or not based on the difference degree; and if the target identity information is abnormal, pre-warning the person corresponding to the target identity information.
The method for generating the person track graph can be directly generated in the attention map or generated in the shot picture, and since the person corresponding to the target identity information is marked, the method can return to any one of the shot pictures to mark the position where the relevant feature appears, so as to generate the track graph, and the track graph is expressed according to a pixel coordinate sequence, so that the difference degree between the person track graph and the tracks of other residents in the target area system can be calculated subsequently. The way of calculating the difference is described in detail later, and is not described herein again. If the difference degree is large, the behavior track of the target person is different from most of the behavior tracks, and the target person should be pre-warned here, so that the danger in the area can be prevented, related security personnel only need to monitor the target person in a key mode, and the workload of checking and monitoring of the security personnel is reduced.
In one embodiment, the step S805 of representing the person trajectory graph according to a pixel coordinate sequence and calculating a difference between the person trajectory graph and other resident trajectories in the target area system based on the pixel coordinate sequence includes:
S8051: screening and intercepting a plurality of target track graph sets which are the same as or similar to the starting point and the end point of the figure track graph from other resident tracks;
s8052: calculating the position difference degree, the speed difference degree and the direction difference degree of the character track map and the target track map set according to the pixel coordinate sequence; wherein, the formula for calculating the position difference degree is
The formula for calculating the velocity difference degree is
，
The formula for calculating the direction difference degree is
，
；
s8053: and calculating the difference degree of the person track graph and the track graph set according to the position difference degree, the speed difference degree and the direction difference degree.
As described in the above steps S8051 to S8053, the calculation of the difference degree is realized. And screening and intercepting a plurality of target track map sets which are the same as or similar to the starting point and the end point of the figure track map from other resident tracks. The screening mode is to calculate the resident track of the same time point and the starting point of the figure track graph within the preset range and the resident track of the terminal point of the figure track graph within the preset range. And comprehensively selecting to obtain a plurality of target tracks to form a target track set, and then calculating the position difference degree, the speed difference degree and the direction difference degree. Generally speaking, whether the trajectory is different from the resident trajectory or not can be detected from the position, the movement speed and the trajectory movement direction, so as to determine the difference degree, and further determine whether the early warning is needed or not. The difference between the person trajectory graph and the trajectory graph set is calculated according to the position difference, the speed difference, and the direction difference, and the calculation of the difference may be specifically a weighted calculation, or may be performed in another manner, for example, an averaging manner.
In one embodiment, after the step S8053 of representing the person trajectory diagram according to a pixel coordinate sequence and calculating the difference between the person trajectory diagram and the trajectories of other residents in the target area system based on the pixel coordinate sequence, the method further includes:
s80531: adjusting the risk level of the target identity information based on the difference degree, and uploading the risk level to a block chain;
s80532: judging whether the risk level reaches a preset risk level or not;
s80533: if yes, recording the target identity information into a blacklist, and storing the figure track graph so as to facilitate subsequent calling and analysis.
As described in the above steps S80531-S80533, the risk level is adjusted, and whether the target identity information satisfies the black name condition is determined according to the risk level. The specific judgment method can be that a preset risk level is set, when the risk level is greater than the preset risk level, the visitor is judged to meet the condition of adding into a preset blacklist database, the visitor is uploaded to a block chain, other regional systems are reminded to perform early warning at the same time, prevention to a greater extent can be achieved, a figure track graph of the visitor is stored, and if the visitor has a criminal activity, related monitoring data can be called in time to be checked.
In one embodiment, after the step S8053 of calculating a difference in position, a difference in speed, and a difference in direction between the human trajectory diagram and the target trajectory diagram set according to the pixel coordinate sequence, the method includes:
s8054: acquiring a first three-dimensional data set of the position, the speed and the direction of each pixel point of the figure track graph and a second three-dimensional data set of the position, the speed and the direction of the resident track graph;
s8055: calculating the similarity of the first three-dimensional data set and each second three-dimensional data set;
s8056: pre-associating the target households corresponding to the household track graphs with the similarity greater than the preset similarity;
s8057: calling first identity information of the target resident, and calling a shot picture which simultaneously has the target resident and a person corresponding to the target identity information based on the first identity information and the target identity information;
s8058: judging whether the number of pictures in the shot picture, of which the distance between the target resident and the person corresponding to the target identity information is smaller than a preset value, is larger than the preset value or not;
s8059: and if so, associating the target resident with the figure corresponding to the target identity information.
As described above in steps S8051-S8059, associating the target identity information with the relevant household is achieved. Since the above calculation process already has the position, speed and direction of each pixel point of the character track map, the similarity is calculated by taking the first three-dimensional data set as the first three-dimensional data set of the character track graph and the second three-dimensional data set of the resident track, the similarity can be calculated by calculating Euclidean distance, because the three-dimensional data of each point is known, the Euclidean distance can be directly calculated and taken as the similarity, then the target householder corresponding to the householder track graph with the similarity greater than the preset similarity is selected for pre-association, at the moment, the condition that the target householder has the same path is only determined, further judgment is needed, therefore, pre-association is performed first, the first identity information of the target resident is called, and a shot picture which simultaneously has the target resident and the person corresponding to the target identity information is called based on the first identity information and the target identity information. When the number of the pictures of which the distance between the target resident and the character corresponding to the target identity information is smaller than the preset value is larger than the preset value, the relation between the target resident and the pictures is shown, and the pictures are possibly friends of the target resident, so the pictures are related to be further analyzed subsequently, namely if the character corresponding to the target identity information carries out criminal activities, the partner or the victim can be quickly positioned.
In an embodiment, after the step S10 of performing an early warning on the person corresponding to the target identity information if the abnormality occurs, the method further includes:
s1001: monitoring whether the time length for the person corresponding to the target identity information to break away from the shot picture reaches a preset time length or not through the camera;
s1002: if so, acquiring first position information of a camera which shoots the figure corresponding to the target identity information for the last time;
s1003: and feeding back to a similar regional system in the block chain based on the first position information.
As described in the above steps S1001 to S1003, the warning to the other regional systems is realized. Whether the time length for which the person corresponding to the target identity information departs from the shot picture reaches a preset time length or not is monitored through the camera, namely when the person corresponding to the target identity information is lost in the area system, the person may enter other area systems, certainly, the person may leave by bus, but the person also has a certain probability of leaving in an adjacent area range, and at the moment, in order to avoid the other area systems to calculate the information again, the information can be fed back to the adjacent area systems in the block chain based on the first position information.
The invention has the beneficial effects that: the method comprises the steps of collecting shot pictures through a person identification and collection device, identifying identity information of all persons in the shot pictures, determining target identity information of persons who do not live in a target area system where the main control device is located, collecting the shot pictures with the same target identity information, obtaining a target shot picture set corresponding to each target identity information, generating a person track graph of the persons corresponding to the target identity information, calculating the difference degree between the track graph and tracks of other residents, and judging whether the persons corresponding to the target identity information need to be pre-warned or not according to the difference degree. The method and the device have the advantages that the analysis and early warning of the figures based on the shot pictures are realized, so that the safety in the region is ensured, the safety of the region can be effectively improved, and the loss of residents is reduced.
Referring to fig. 2, the present invention further provides an intelligent security device based on person identification, which is applied to each pre-divided area system, the area system includes a main control device and a plurality of person identification and acquisition devices, the person identification and acquisition devices at least include a camera, a shot picture of a person is acquired by the camera and uploaded to the main control device, the main control device stores position information corresponding to each person identification and acquisition device in the area system and resident information of each resident in the area system, a block chain is formed between the main control devices in each area system, each main control device is a block chain node of the block chain, and the intelligent security device based on person identification includes:
The acquisition module is used for acquiring a plurality of shot pictures according to a time sequence through the person identification acquisition equipment;
the identification module is used for identifying the identity information of each figure in the shot picture through a preset face identification technology;
the screening module is used for screening target identity information of people who do not live in a target area system where the main control equipment is located from the identity information based on the resident information; the target area system is one of the area systems divided in advance;
a sending module, configured to send the target identity information to a target master control device of a regional system within a preset interval range in the block chain based on location information between the target regional system and another regional system;
the receiving module is used for receiving the feedback information of the target main control equipment;
the marking module is used for marking people in the shot pictures based on the target identity information if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
The sorting module is used for sorting the shot images in the shot image set according to the time sequence to respectively obtain a first characteristic diagram, a second characteristic diagram, an Nth characteristic diagram;
the difference degree calculation module is used for calculating a person track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph and calculating the difference degree between the person track graph and the tracks of other residents in the target area system;
the judging module is used for judging whether the figure corresponding to the target identity information is abnormal or not based on the difference degree;
and the early warning module is used for early warning the figure corresponding to the target identity information if the target identity information is abnormal.
In one embodiment, the disparity calculation module further includes:
the input sub-module is used for inputting each shot image into a preset CNN backbone network to obtain a first standard characteristic tensor map corresponding to the first characteristic map, a second standard characteristic tensor map, a standard image; wherein the sizes of the standard characteristic tensor maps are C multiplied by H multiplied by W;
The selection submodule is used for recording the standard feature tensor graphs adjacent in time sequence into a group of data, and selecting (H + 1-K) × (W + 1-K) target blocks with the size of K × K from one standard feature tensor graph sliding window;
the cross-correlation submodule is used for carrying out cross-correlation operation on each target block relative to another characteristic tensor diagram in a group of corresponding data to obtain a response diagram of each target block relative to the characteristic tensor diagram; wherein the manner of the cross-correlation operation is
，
A response diagram showing the ith target block of the jth second standard feature diagram,
the ith target block of the jth second standard feature diagram is shown,
representing another characteristic tensor map;
a fusion submodule for expressing
Fusing each response graph in a group of data to obtain an attention graph;
the generation submodule is used for generating a person track graph of a person corresponding to the target identity information based on each attention graph;
and the representing submodule is used for representing the person track graph according to the pixel coordinate sequence and calculating the difference degree between the person track graph and other resident tracks in the target area system based on the pixel coordinate sequence.
In one embodiment, the representation submodule includes:
The screening unit is used for screening and intercepting a plurality of target track graph sets which are the same as or close to the starting point and the end point of the figure track graph from other resident tracks;
a calculating unit, configured to calculate a position difference degree, a speed difference degree and a speed difference degree of the person track map and the target track map set according to the pixel coordinate sequenceThe degree of difference in direction; wherein, the formula for calculating the position difference degree is
The formula for calculating the velocity difference degree is
，
The formula for calculating the direction difference degree is
，
and the difference degree calculation unit is used for calculating the difference degree between the person track map and the track map set according to the position difference degree, the speed difference degree and the direction difference degree.
In one embodiment, the intelligent security device based on person identification further includes:
the adjusting module is used for adjusting the risk level of the target identity information based on the difference degree and uploading the risk level to a block chain;
the risk grade judging module is used for judging whether the risk grade reaches a preset risk grade or not;
and the blacklist recording module is used for recording the target identity information into a blacklist and storing the figure track map if the target identity information is in the blacklist, so that subsequent calling and analysis are facilitated.
The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.
The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (10)
1. The utility model provides an intelligent security and protection method based on people discernment, its characterized in that is applied to each regional system of dividing in advance, regional system includes master control equipment and a plurality of people discernment collection equipment, people discernment collection equipment includes the camera at least, through the shooting picture of people is gathered to the camera, and will the shooting picture is uploaded to master control equipment, master control equipment has the position information that corresponds each people discernment collection equipment in the regional system and the resident family information of each resident family in the regional system, each constitute the block chain between the master control equipment in the regional system, each master control equipment is the block chain node of block chain, the method includes:
The main control equipment acquires a plurality of shot pictures according to a time sequence through the figure identification acquisition equipment;
identifying the identity information of each person in the shot picture through a preset face identification technology;
screening out target identity information of people who do not live in a target area system where the main control equipment is located from the identity information based on the resident information; the target area system is one of the pre-divided area systems;
based on the position information between the target area system and other area systems, the target identity information is sent to the target main control equipment of the area system in a preset interval range in the block chain;
receiving feedback information of the target master control equipment;
if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, performing character marking on the shot pictures based on the target identity information, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
sequencing the shot images in the shot image set according to the time sequence to respectively obtain a first characteristic diagram, a second characteristic diagram, a.
Calculating a figure track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph, and calculating the difference degree between the figure track graph and the tracks of other residents in the target area system;
judging whether the figure corresponding to the target identity information is abnormal or not based on the difference degree;
and if the target identity information is abnormal, pre-warning the figure corresponding to the target identity information.
2. The intelligent security method based on person identification as claimed in claim 1, wherein the step of calculating a person trajectory graph corresponding to the target identity information according to the first feature graph, the second feature graph, the nth feature graph and calculating a degree of difference between the person trajectory graph and trajectories of other residents in the target area system comprises:
inputting each shot image into a preset CNN backbone network to obtain a first standard feature tensor diagram corresponding to the first feature diagram, a second standard feature tensor diagram corresponding to the second feature diagram, and an Nth standard feature tensor diagram corresponding to the Nth feature diagram; the sizes of all the standard characteristic tensor maps are C multiplied by H multiplied by W;
Recording the standard feature tensor images adjacent in time sequence as a group of data, and selecting (H + 1-K) × (W + 1-K) target blocks with the size of K × K from one standard feature tensor image sliding window;
performing cross-correlation operation on each target block relative to another characteristic tensor diagram in a group of corresponding data to obtain a response diagram of each target block relative to the characteristic tensor diagram; wherein the manner of the cross-correlation operation is
，
A response diagram of the ith target block of the jth standard feature diagram is shown,
showing the ith target block of the jth standard feature diagram,
representing another characteristic tensor map;
according to the formula
Fusing each response graph in a group of data to obtain an attention graph;
generating a person track graph of a person corresponding to the target identity information based on each attention graph;
and expressing the figure track graph according to a pixel coordinate sequence, and calculating the difference degree between the figure track graph and other resident tracks in the target area system based on the pixel coordinate sequence.
3. The intelligent security method based on human recognition as claimed in claim 2, wherein the step of representing the human trajectory graph according to a pixel coordinate sequence and calculating the difference degree between the human trajectory graph and the trajectories of other residents in the target area system based on the pixel coordinate sequence comprises:
Screening and intercepting a plurality of target track graph sets which are the same as or similar to the starting point and the end point of the figure track graph from other resident tracks;
calculating the position difference degree, the speed difference degree and the direction difference degree of the character track map and the target track map set according to the pixel coordinate sequence; wherein the formula for calculating the position difference degree is
The formula for calculating the velocity difference degree is
，
The formula for calculating the direction difference degree is
，
and calculating the difference degree between the person track graph and the track graph set according to the position difference degree, the speed difference degree and the direction difference degree.
4. The intelligent security method based on human recognition as claimed in claim 1, wherein after the step of representing the human trajectory graph according to a pixel coordinate sequence and calculating the difference between the human trajectory graph and the trajectories of other residents in the target area system based on the pixel coordinate sequence, the method further comprises:
adjusting the risk level of the target identity information based on the difference degree, and uploading the risk level to a block chain;
judging whether the risk level reaches a preset risk level or not;
if yes, recording the target identity information into a blacklist, and storing the figure track graph so as to facilitate subsequent calling and analysis.
5. The intelligent security method based on human recognition as claimed in claim 3, wherein the step of calculating the difference degree of the position, the difference degree of the speed and the difference degree of the direction between the human trajectory graph and the target trajectory graph set according to the pixel coordinate sequence comprises:
Acquiring a first three-dimensional data set of the position, the speed and the direction of each pixel point of the figure track graph and a second three-dimensional data set of the position, the speed and the direction of the resident track graph;
calculating the similarity of the first three-dimensional data set and each second three-dimensional data set;
pre-associating the target households corresponding to the household track graphs with the similarity greater than the preset similarity;
calling first identity information of the target resident, and calling a shot picture which simultaneously has the target resident and a person corresponding to the target identity information based on the first identity information and the target identity information;
judging whether the number of pictures in the shot picture, of which the distance between the target resident and the person corresponding to the target identity information is smaller than a preset value, is larger than the preset value or not;
and if so, associating the target resident with the figure corresponding to the target identity information.
6. The intelligent security method based on person identification as claimed in claim 1, wherein after the step of performing early warning on the person corresponding to the target identity information if the person is abnormal, the method further comprises:
monitoring whether the time length of the person corresponding to the target identity information, which is separated from the shot picture, reaches a preset time length or not through the camera;
If yes, first position information of a camera of a figure corresponding to the target identity information is obtained through the last shooting;
and feeding back to a similar regional system in the block chain based on the first position information.
7. The utility model provides an intelligent security device based on personage discernment, its characterized in that is applied to each regional system of dividing in advance, regional system includes main control equipment and a plurality of personage discernment collection equipment, personage discernment collection equipment includes the camera at least, through the shooting picture of personage is gathered to the camera, and will it uploads to shoot the picture main control equipment, main control equipment has the correspondence each personage discernment collection equipment's in the regional system positional information and the resident family information of each resident family in the regional system, each constitute the block chain between the main control equipment in the regional system, each main control equipment does the block chain node of block chain, intelligent security device based on personage discernment includes:
the acquisition module is used for acquiring a plurality of shot pictures according to a time sequence through the person identification acquisition equipment;
the identification module is used for identifying the identity information of each figure in the shot picture through a preset face identification technology;
The screening module is used for screening target identity information of people who do not live in the target area system where the main control equipment is located from the identity information based on the resident information; the target area system is one of the pre-divided area systems;
a sending module, configured to send the target identity information to a target master control device of a regional system within a preset interval range in the block chain based on location information between the target regional system and another regional system;
the receiving module is used for receiving the feedback information of the target main control equipment;
the marking module is used for marking people in the shot pictures based on the target identity information if the feedback information indicates that the target identity information is not in a target area system where the target master control equipment is located, and collecting the shot pictures with the same target identity information to obtain a target shot picture set corresponding to each target identity information;
the sorting module is used for sorting the shot images in the shot image set according to the time sequence to respectively obtain a first feature map, a second feature map, an Nth feature map;
The difference degree calculation module is used for calculating a person track graph corresponding to the target identity information according to the first feature graph, the second feature graph, the Nth feature graph and calculating the difference degree between the person track graph and the tracks of other residents in the target area system;
the judging module is used for judging whether the person corresponding to the target identity information is abnormal or not based on the difference degree;
and the early warning module is used for early warning the figure corresponding to the target identity information if the target identity information is abnormal.
8. The intelligent security device based on human recognition as claimed in claim 7, wherein the degree of difference calculation module further comprises:
the input sub-module is used for inputting each shot image into a preset CNN backbone network to obtain a first standard characteristic tensor map corresponding to the first characteristic map, a second standard characteristic tensor map, a standard image; the sizes of all the standard characteristic tensor maps are C multiplied by H multiplied by W;
the selection submodule is used for recording the standard feature tensor graphs adjacent in time sequence into a group of data, and selecting (H + 1-K) × (W + 1-K) target blocks with the size of K × K from one standard feature tensor graph sliding window;
A cross-correlation sub-module for correlating each target block against another feature in its corresponding set of dataPerforming cross-correlation operation on the tensor map to obtain a response map of each target block relative to the characteristic tensor map; wherein the manner of the cross-correlation operation is
，
A response diagram showing the ith target block of the jth second standard feature diagram,
the ith target block of the jth second standard feature diagram is shown,
representing another characteristic tensor map;
a fusion submodule for generating a formula based on
Fusing each response graph in a group of data to obtain an attention graph;
the generation submodule is used for generating a person track graph of a person corresponding to the target identity information based on each attention graph;
and the representing submodule is used for representing the person track graph according to the pixel coordinate sequence and calculating the difference degree between the person track graph and other resident tracks in the target area system based on the pixel coordinate sequence.
9. The smart security device based on human recognition as claimed in claim 8, wherein the representation sub-module comprises:
the screening unit is used for screening and intercepting a plurality of target track map sets which are the same as or close to the starting point and the end point of the figure track map from other resident tracks;
A calculation unit for calculating the character track map and the target track according to the pixel coordinate sequenceThe position difference degree, the speed difference degree and the direction difference degree of the trace graph set; wherein the formula for calculating the position difference degree is
The formula for calculating the speed difference degree is
，
The formula for calculating the direction difference degree is
，
and the difference degree calculation unit is used for calculating the difference degree of the person track graph and the track graph set according to the position difference degree, the speed difference degree and the direction difference degree.
10. The security device according to claim 8, wherein the security device further comprises:
the adjusting module is used for adjusting the risk level of the target identity information based on the difference degree and uploading the risk level to a block chain;
the risk grade judging module is used for judging whether the risk grade reaches a preset risk grade or not;
and the blacklist recording module is used for recording the target identity information into a blacklist and storing the figure track map if the target identity information is in the blacklist, so that subsequent calling and analysis are facilitated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210269953.2A CN114359976B (en) | 2022-03-18 | 2022-03-18 | Intelligent security method and device based on person identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210269953.2A CN114359976B (en) | 2022-03-18 | 2022-03-18 | Intelligent security method and device based on person identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114359976A CN114359976A (en) | 2022-04-15 |
| CN114359976B true CN114359976B (en) | 2022-06-14 |
Family
ID=81094354
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210269953.2A Active CN114359976B (en) | 2022-03-18 | 2022-03-18 | Intelligent security method and device based on person identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114359976B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114903472B (en) * | 2022-06-02 | 2024-09-27 | 深圳市博铭维系统工程有限公司 | Hospital management method, system, terminal and storage medium based on Internet of Things positioning |
| CN114973153B (en) * | 2022-07-27 | 2022-11-04 | 广州宏途数字科技有限公司 | Smart campus security detection method, device, equipment and storage medium |
| CN115861316B (en) * | 2023-02-27 | 2023-09-29 | 深圳佑驾创新科技股份有限公司 | Training method and device for pedestrian detection model and pedestrian detection method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102523370A (en) * | 2011-11-22 | 2012-06-27 | 上海交通大学 | Multi-camera video abnormal behavior detection method based on network transmission algorithm |
| CN106097223A (en) * | 2016-06-15 | 2016-11-09 | 广州市华标科技发展有限公司 | A kind of certificate declaration management system and declare management method |
| CN110766895A (en) * | 2019-09-17 | 2020-02-07 | 重庆特斯联智慧科技股份有限公司 | Intelligent community abnormity alarm system and method based on target trajectory analysis |
| WO2020239210A1 (en) * | 2019-05-28 | 2020-12-03 | Gottfried Wilhelm Leibniz Universität Hannover | Method, apparatus and computer program for tracking of moving objects |
| CN112243110A (en) * | 2020-10-15 | 2021-01-19 | 成都易瞳科技有限公司 | Panoramic target track recording device and recording method |
-
2022
- 2022-03-18 CN CN202210269953.2A patent/CN114359976B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102523370A (en) * | 2011-11-22 | 2012-06-27 | 上海交通大学 | Multi-camera video abnormal behavior detection method based on network transmission algorithm |
| CN106097223A (en) * | 2016-06-15 | 2016-11-09 | 广州市华标科技发展有限公司 | A kind of certificate declaration management system and declare management method |
| WO2020239210A1 (en) * | 2019-05-28 | 2020-12-03 | Gottfried Wilhelm Leibniz Universität Hannover | Method, apparatus and computer program for tracking of moving objects |
| CN110766895A (en) * | 2019-09-17 | 2020-02-07 | 重庆特斯联智慧科技股份有限公司 | Intelligent community abnormity alarm system and method based on target trajectory analysis |
| CN112243110A (en) * | 2020-10-15 | 2021-01-19 | 成都易瞳科技有限公司 | Panoramic target track recording device and recording method |
Non-Patent Citations (1)
| Title |
|---|
| 5G时代人工智能识别技术在视频侦查中的应用;王鹏等;《河北公安警察职业学院学报》;20200318(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114359976A (en) | 2022-04-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114359976B (en) | Intelligent security method and device based on person identification | |
| KR101850286B1 (en) | A deep learning based image recognition method for CCTV | |
| CN111898514B (en) | Multi-target visual supervision method based on target detection and action recognition | |
| CN113256924A (en) | Monitoring system, monitoring method and monitoring device for rail train | |
| JP5459674B2 (en) | Moving object tracking system and moving object tracking method | |
| KR102172239B1 (en) | Method and system for abnormal situation monitoring based on video | |
| CN110428449A (en) | Target detection tracking method, device, equipment and storage medium | |
| CN111898581A (en) | Animal detection method, device, electronic equipment and readable storage medium | |
| KR102511287B1 (en) | Image-based pose estimation and action detection method and appratus | |
| CN112329691A (en) | Monitoring video analysis method and device, electronic equipment and storage medium | |
| KR20200052418A (en) | Automated Violence Detecting System based on Deep Learning | |
| CN109255360B (en) | Target classification method, device and system | |
| KR20190088087A (en) | method of providing categorized video processing for moving objects based on AI learning using moving information of objects | |
| CN110599129A (en) | Campus attendance checking method, device, identification terminal and system based on image tracking | |
| Rezaee et al. | Deep-Transfer-learning-based abnormal behavior recognition using internet of drones for crowded scenes | |
| KR20230097854A (en) | Method and system for recognizing dangerous behavior of workers in power plant | |
| KR20210013865A (en) | Abnormal behavior detection system and method using generative adversarial network | |
| CN113128414A (en) | Personnel tracking method and device, computer readable storage medium and electronic equipment | |
| CN117593792A (en) | Abnormal gesture detection method and device based on video frame | |
| KR20220019377A (en) | Drone detection apparatus using deep learning and method thereof | |
| CN111144260A (en) | Detection method, device and system of crossing gate | |
| CN109815921A (en) | The prediction technique and device of the class of activity in hydrogenation stations | |
| TWI730795B (en) | Multi-target human body temperature tracking method and system | |
| Amrutha et al. | A robust system for video classification: identification and tracking of suspicious individuals from surveillance videos | |
| Subbarayudu et al. | Detection of anomalous behaviour of a student in examination hall using deep learning techniques |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: An Intelligent Security Method and Device Based on Character Recognition Granted publication date: 20220614 Pledgee: Guanggu Branch of Wuhan Rural Commercial Bank Co.,Ltd. Pledgor: WUHAN PKU HIGH-TECH SOFT Co.,Ltd. Registration number: Y2024980009351 |