CN114494350A - Personnel gathering detection method and device - Google Patents
Personnel gathering detection method and device Download PDFInfo
- Publication number
- CN114494350A CN114494350A CN202210107678.4A CN202210107678A CN114494350A CN 114494350 A CN114494350 A CN 114494350A CN 202210107678 A CN202210107678 A CN 202210107678A CN 114494350 A CN114494350 A CN 114494350A
- Authority
- CN
- China
- Prior art keywords
- personnel
- gathering
- people
- suspicious
- real
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of intelligent identification, in particular to a personnel gathering detection method and a personnel gathering detection device, which utilize target coordinates and characteristic values identified by a structured algorithm to provide high-level target identification rate for personnel gathering detection; the target trajectory prediction is carried out on the basis of the target structure identified by the structural algorithm by using the DeepsORT algorithm, so that the overall accuracy of personnel gathering detection is further improved, and the problem of target loss of the structural algorithm is effectively solved; and calibrating suspicious personnel based on the calculation result of the DeepSORT algorithm, and accurately obtaining the maximum number of the suspicious personnel aggregates by circularly traversing the information of the suspicious personnel to divide the aggregation circle. Therefore, the method can effectively improve the target recognition rate, reduce the false alarm abnormal conditions in a complex scene, and comprehensively improve the accuracy of personnel gathering detection.
Description
Technical Field
The invention relates to the technical field of intelligent identification, in particular to a personnel gathering detection method and a personnel gathering detection device.
Background
Along with the acceleration of the construction pace of the smart city, the requirements of people on intelligent security and protection are higher and higher. The method has the advantages that the detection requirements for the people gathering scene exist in part of key places, people target identification can be carried out on the monitored scene by using a structured algorithm, but due to mutual superposition or shielding of people in the scene, part of targets can be lost or flickered. Eventually, a large number of false alarm exceptions may be generated, which all cause difficulties in the application of intelligent people gathering detection techniques.
For this reason, in the prior art, a Multi-Object Tracking technique, i.e. MOT (Multi-Object Tracking), is often used, and as the name implies, multiple objects are tracked simultaneously in a piece of video. The main application scenes of the MOT are security monitoring, automatic driving and the like, and in the scenes, people often need to track a plurality of targets simultaneously. And the DeepsORT algorithm is the multi-target tracking algorithm which is most applied at present. Although the DeepsORT algorithm can accurately give a multi-target motion track, the final accuracy achieved only by the algorithm is still difficult to meet the requirement of people gathering detection in a complex scene. Therefore, a people gathering detection method capable of effectively improving the target recognition rate and reducing false alarm abnormal conditions in complex scenes is needed in the art.
Disclosure of Invention
The present invention aims to provide a people group detection method, thereby solving the aforementioned problems in the prior art.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
the invention provides a personnel gathering detection method, which comprises the following steps:
sampling the decoded real-time video image, and acquiring a current frame image as a sampling frame;
analyzing the current frame image through a personnel structuring algorithm to obtain a current frame personnel information vector;
merging the current frame personnel information vector into a historical tracking matrix containing historical frame personnel information vectors through a DeepsORT algorithm to obtain a real-time tracking matrix;
traversing personnel information vectors in the real-time tracking matrix to obtain movement track vectors of different personnel, marking the personnel with continuous multiframe movement distances smaller than a movement threshold value and duration time exceeding an aggregation time threshold value as suspicious personnel based on the movement track vectors, and obtaining a suspicious personnel tracking matrix;
and circularly traversing the personnel information vectors in the suspicious personnel tracking matrix, acquiring the maximum suspicious personnel gathering number based on a personnel gathering distance threshold, and judging that the personnel gathering phenomenon occurs when the maximum suspicious personnel gathering number reaches a gathering number threshold.
Preferably, the current frame personnel information vector comprises personnel information vectors of a plurality of personnel, and the personnel information vectors comprise target coordinates and characteristic values; all tracking matrices are formed by personnel information vectors of different personnel in successive sampling frames.
Preferably, when the number of people in the real-time tracking matrix or the suspicious people tracking matrix is smaller than the threshold value of the number of people gathering, the decoded real-time video image is directly prepared to be sampled next time, and the real-time tracking matrix is used as a historical tracking matrix required by next detection.
Preferably, when a person whose continuous non-occurrence time exceeds a temporal separation time threshold exists in the real-time tracking matrix, deleting the person information vector of the person in the real-time tracking matrix, and using the real-time tracking matrix as a historical tracking matrix required by the next detection.
Preferably, the sampling of the decoded real-time video image specifically includes:
decoding the real-time video image and acquiring an initial frame number of a current frame image;
and determining the number of sampling interval frames based on the sampling interval time, and if the initial frame number variable quantity between the current frame image and the last sampling frame is greater than the sampling interval frame number, acquiring the current frame image as a sampling frame and coding the sampling frame number.
Preferably, the obtaining the maximum number of suspicious people gathered based on the people gathering distance threshold specifically includes:
based on the personnel information vectors in the suspicious personnel tracking matrix, respectively taking each suspicious personnel as an aggregation circle, and acquiring a minimum distance matrix between personnel in any aggregation circle and personnel in other aggregation circles;
judging whether the minimum distance smaller than the personnel gathering distance threshold exists in the minimum distance matrix;
if the minimum distance exists, merging the two clustering circles corresponding to the minimum distance to serve as a new clustering circle, updating the minimum distance matrix, and judging whether the minimum distance smaller than the personnel clustering distance threshold exists in the minimum distance matrix again;
if not, the number of people in the cluster with the most people is taken as the maximum number of suspicious people.
Correspondingly, the invention also provides a personnel gathering detection device, which comprises:
the acquisition module is used for sampling the decoded real-time video image and acquiring a current frame image as a sampling frame;
the first processing module is used for analyzing the current frame image through a personnel structuring algorithm to obtain a current frame personnel information vector;
the second processing module is used for merging the current frame personnel information vector into a historical tracking matrix containing historical frame personnel information vectors through a DeepsORT algorithm to obtain a real-time tracking matrix;
the first determining module is used for traversing the personnel information vectors in the real-time tracking matrix, acquiring the movement track vectors of different personnel, marking the personnel with continuous multiframe movement distances smaller than a movement threshold value and duration time exceeding an aggregation time threshold value as suspicious personnel based on the movement track vectors, and acquiring a suspicious personnel tracking matrix;
and the second confirmation module is used for circularly traversing the personnel information vectors in the suspicious personnel tracking matrix, acquiring the maximum suspicious personnel gathering number based on the personnel gathering distance threshold, and judging that the personnel gathering phenomenon occurs when the maximum suspicious personnel gathering number reaches the gathering number threshold.
Correspondingly, the invention also provides a people group detection device, which comprises a processor and a memory, wherein: the memory having stored therein computer instructions; the processor executes the computer instructions to implement the method of any of the above.
Accordingly, the present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions which, when run on a computer, cause the computer to perform the method as described in any one of the above.
The invention has the beneficial effects that:
the invention provides a personnel gathering detection method and a device, which utilize a target coordinate and a characteristic value identified by a structured algorithm to provide a high-level target identification rate for personnel gathering detection; the target track prediction is carried out on the basis of the target structure identified by the structural algorithm by using the DeepSORT algorithm, so that the overall accuracy of personnel gathering detection is further improved, and the problem of target loss of the structural algorithm is effectively solved; and calibrating suspicious personnel based on the calculation result of the DeepSORT algorithm, and accurately obtaining the maximum number of the suspicious personnel aggregates by circularly traversing the information of the suspicious personnel to divide the aggregation circle. Therefore, the method can effectively improve the target recognition rate, reduce the false alarm abnormal conditions in a complex scene, and comprehensively improve the accuracy of personnel gathering detection.
Drawings
Fig. 1 is a schematic flow chart of a people group detection method provided in an embodiment of the present invention;
FIG. 2 is a schematic diagram of tracking people associations provided in an embodiment of the invention;
FIG. 3 is a schematic diagram of a suspect collection of persons provided in an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a people gathering detection device provided in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
As shown in fig. 1, the present invention provides a people gathering detection method, including:
s101, sampling the decoded real-time video image, and acquiring a current frame image as a sampling frame.
And S102, analyzing the current frame image through a personnel structuring algorithm to obtain a current frame personnel information vector.
S103, merging the current frame personnel information vector into a historical tracking matrix containing historical frame personnel information vectors through a DeepsORT algorithm to obtain a real-time tracking matrix.
S104, traversing the personnel information vectors in the real-time tracking matrix, obtaining the moving track vectors of different personnel, marking the personnel with continuous multi-frame moving distance smaller than a moving threshold and duration exceeding an aggregation time threshold as suspicious personnel based on the moving track vectors, and obtaining a suspicious personnel tracking matrix.
And S105, circularly traversing the personnel information vectors in the suspicious personnel tracking matrix, acquiring the maximum suspicious personnel gathering number based on a personnel gathering distance threshold, and judging that the personnel gathering phenomenon occurs when the maximum suspicious personnel gathering number reaches a gathering number threshold.
In this embodiment, the sampling of the decoded real-time video image specifically includes:
decoding a real-time video image and acquiring an initial frame number vf (vf ═ 1,2, 3.) of a current frame image;
determining a sampling interval frame number f (video playback frame rate PR x sampling interval time TI) based on the sampling interval time TI and the video playback frame rate PR;
if the initial frame number variation Δ vf between the current frame image and the last sampling frame is greater than the sampling interval frame number f, acquiring the current frame image as a sampling frame and encoding a sampling frame number vi (vi is 1, 2.. multidot.m), where the sampling frame number of the current frame is m.
In this embodiment, the person structuring algorithm may identify target coordinates and feature values of n1 persons in the current frame image to form a current frame person information vector I' x (x ═ 1, 2.., n 1). The historical tracking matrix TM' x, y (x 1, 2., n 2; y 1, 2., m-1), the real-time tracking matrix TMx, y (x 1, 2., n; y 1, 2., m) and the suspect tracking matrix AMx, y (x 1, 2., p; y 1, 2., m) are each constructed from personnel information vectors of different personnel in successive sampled frames. For example, the suspicious person tracking matrix AMx, y uses the person information vectors of p suspicious persons in consecutive sampling frames as matrix elements, the same row represents the same suspicious person, the same column represents the same sampling frame, the suspicious persons represented by different rows are different, the sampling frames represented by different columns are different, and the sampling frames should be arranged in sequence based on the sampling order.
In this embodiment, the history tracking matrix TM' x, y includes the personnel information vectors of all history frames (not including the current frame m) of n2 history tracking personnel; the real-time tracking matrix TMx, y contains personnel information vectors of all frames (including the current frame m) of n tracked personnel; the suspicious person tracking matrix AMx, y contains the person information vectors for all frames of p suspicious persons. The real-time tracking matrix TMx, y only increases the personnel information vectors of all the trackers at the current frame m (i.e. all the trackers information vectors Ix at the current frame) compared with the historical tracking matrix TM' x, y. The current frame person information vector I' x analyzed by the structural algorithm does not necessarily include all trackers, nor is it necessarily all related to historical trackers, i.e., n1 is not necessarily equal to n2, and n is not necessarily equal to n 2. The number p of suspicious persons is less than or equal to the number n of all tracked persons.
The DeepSORT used in the embodiment comprises Kalman filtering and Hungarian algorithms, and the DeepSORT can quickly find the optimal matching solution of a plurality of targets of two adjacent frames. The Kalman filtering can optimize the scene that the target overlapped target is lost temporarily through high-reliability target track prediction, and effectively solves the problem that the target of the structured algorithm is lost.
As shown in fig. 2, in this embodiment, a current frame person information vector I 'x and a historical tracking matrix TM' x, y are used as input through a DeepSORT algorithm, all tracked person information vectors Ix of a current frame are calculated, and a real-time tracking matrix TMx, y is obtained. In the figure, different shape patterns represent different tracked persons, and different fill patterns represent different sampling frames. The current frame personnel number n1 analyzed by a general video image through a personnel structuring algorithm is different from the historical tracking personnel number n2, and different processing is required. The method specifically comprises the following steps:
if n1 current frame persons can be associated with the historical tracking persons, the current frame person information vector needs to be recorded in the current frame m of the real-time tracking matrix TMx, y, and the person information vector of the unassociated historical tracking persons in the current frame m is recorded as empty, and the number n of the real-time tracking persons is the same as the number n2 of the historical tracking persons. For example, expression (2) is obtained from expression (1), where in expression (2), x is 4 to 5, y is TMx corresponding to 5, y information is null, n1 is 3, n2 is 5, and n is 5.
If u persons are not associated with the historical tracking persons in the n1 current frame persons, the current frame person information vector needs to be recorded in the current frame m of the real-time tracking matrix TMx, y, the person information vector of the unassociated historical tracking persons in the current frame m is recorded as null, the person information vector of the u persons in the historical frame is recorded as null, and the number n of the real-time tracking persons is n2+ u. For example, expression (3) is obtained from expression (1), where in expression (2), x is 3 to 5, y is TMx corresponding to 5, y information is null, and x is 6, y is TMx corresponding to 1 to 4, y information is null, n1 is 3, n2 is 5, u is 1, and n is 6.
In order to increase the detection efficiency and avoid performing invalid detection, it is necessary to skip an unnecessary detection process and eliminate tracking people that do not appear for a long time, and in this embodiment, the method specifically includes:
and when the number of people in the real-time tracking matrix TMx, y or the suspicious people tracking matrix AMx, y is less than the threshold value CL of the number of gathered people, directly preparing to sample the decoded real-time video image for the next time, updating the historical tracking matrix TM 'x, y, and taking the real-time tracking matrix TMx, y as the historical tracking matrix TM' x, y required by the next detection.
And when the personnel with continuous non-occurrence time exceeding the temporary leaving time threshold TH exist in the real-time tracking matrix TMx, y, deleting the personnel information vector of the personnel in the real-time tracking matrix TMx, y, updating the historical tracking matrix TM 'x, y, and taking the real-time tracking matrix TMx, y as the historical tracking matrix TM' x, y required by the next detection.
In order to determine whether a person aggregation phenomenon exists, in this embodiment, the real-time tracking matrix TMx, y is decomposed into n person movement trajectory vectors My (y is 1, 2.. multidot.m), and if the person reverse-order movement trajectory vectors My have a continuous multi-frame movement distance smaller than a movement threshold ML and a duration exceeding an aggregation time threshold TL, the person reverse-order movement trajectory vectors are calibrated as suspicious persons, so as to form a suspicious person tracking matrix AMx, y.
In this embodiment, based on the person information vectors in the suspicious person tracking matrices AMx and y, each suspicious person is taken as a gathering circle, and a minimum distance matrix ADx, y (x, y is 1,2, p) between a person in any gathering circle and a person in other gathering circles is obtained, where when x is y, the value of ADx and y is zero.
And judging whether the minimum distance ADi, j smaller than the personnel gathering distance threshold DL exists in the minimum distance matrix ADx, y.
If yes, merging the two clustering circles corresponding to the minimum distance ADi, j to be used as a new clustering circle, updating the minimum distance matrix ADx, y (the minimum distance of people in the same clustering circle is zero), judging whether the minimum distance matrix ADx, y has the minimum distance ADi, j smaller than the people clustering distance threshold value DL again, and repeating the steps for k times. And when the maximum suspicious people gathering number AC reaches the gathering number threshold CL, judging that people gathering occurs, and synchronously pushing and alarming.
As shown in fig. 3, when k is 1, person 1 and person 2 are one tuck C1; person 3, who exceeds the threshold DL with other persons, is a single colony C3; person 4 forms two tucks C5, C6 with person 5 and person 6, respectively. When k is 2, the two tucks C5 and C6 merge into a new tuck C4 due to sharing person 4; after which no other focus circles merge and the cycle ends.
It should be noted that, in the present invention, it is necessary to initialize relevant parameters during the initial detection, including a threshold CL of the number of people gathered, a threshold DL of the distance of people gathered, a threshold TL of the gathering time, a moving threshold ML, a sampling interval time TI, a threshold TH of the temporal distance time, and other parameters. At the moment, the real-time tracking matrix TMx and y and the historical tracking matrix TM' x and y are empty tables, no personnel information vector exists, and personnel information can be gradually increased or removed in the following circulation process.
As shown in fig. 4, the present invention also provides a people gathering detection apparatus, including:
an obtaining module 401, configured to sample the decoded real-time video image, and obtain a current frame image as a sampling frame.
The first processing module 402 is configured to analyze the current frame image through a person structuring algorithm, and obtain a current frame person information vector.
And a second processing module 403, configured to merge the current frame staff information vector into a historical tracking matrix containing historical frame staff information vectors through a DeepSORT algorithm, so as to obtain a real-time tracking matrix.
The first determining module 404 is configured to traverse the staff information vectors in the real-time tracking matrix, obtain movement track vectors of different staff, mark, based on the movement track vectors, a staff with a continuous multi-frame movement distance smaller than a movement threshold and a duration time exceeding an aggregation time threshold as a suspicious staff, and obtain a suspicious staff tracking matrix.
The second confirming module 405 is configured to cycle through the people information vectors in the suspicious people tracking matrix, obtain a maximum suspicious people aggregation number based on a people aggregation distance threshold, and determine that a people aggregation phenomenon occurs when the maximum suspicious people aggregation number reaches an aggregation number threshold.
By adopting the technical scheme disclosed by the invention, the following beneficial effects are obtained:
the invention provides a personnel gathering detection method and a device, which utilize a target coordinate and a characteristic value identified by a structured algorithm to provide a high-level target identification rate for personnel gathering detection; the target trajectory prediction is carried out on the basis of the target structure identified by the structural algorithm by using the DeepsORT algorithm, so that the overall accuracy of personnel gathering detection is further improved, and the problem of target loss of the structural algorithm is effectively solved; and calibrating suspicious personnel based on the calculation result of the DeepSORT algorithm, and accurately obtaining the maximum number of the suspicious personnel aggregates by circularly traversing the information of the suspicious personnel to divide the aggregation circle. The method can effectively improve the target recognition rate, reduce the false alarm abnormal conditions in a complex scene, and comprehensively improve the accuracy of personnel gathering detection.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, many modifications and adaptations can be made without departing from the principle of the present invention, and such modifications and adaptations should also be considered to be within the scope of the present invention.
Claims (9)
1. A people gathering detection method, comprising:
sampling the decoded real-time video image, and acquiring a current frame image as a sampling frame;
analyzing the current frame image through a personnel structured algorithm to obtain a current frame personnel information vector;
merging the current frame personnel information vector into a historical tracking matrix containing historical frame personnel information vectors through a DeepsORT algorithm to obtain a real-time tracking matrix;
traversing personnel information vectors in the real-time tracking matrix to obtain movement track vectors of different personnel, marking the personnel with continuous multiframe movement distances smaller than a movement threshold value and duration time exceeding an aggregation time threshold value as suspicious personnel based on the movement track vectors, and obtaining a suspicious personnel tracking matrix;
and circularly traversing the personnel information vectors in the suspicious personnel tracking matrix, acquiring the maximum suspicious personnel gathering number based on a personnel gathering distance threshold, and judging that the personnel gathering phenomenon occurs when the maximum suspicious personnel gathering number reaches a gathering number threshold.
2. The people gathering detection method according to claim 1, wherein the current frame people information vector includes people information vectors of a plurality of people, the people information vectors including target coordinates and eigenvalues; all tracking matrices are formed by personnel information vectors of different personnel in successive sampling frames.
3. The people gathering detection method as claimed in claim 1, wherein when the number of people in the real-time tracking matrix or the suspicious people tracking matrix is less than the threshold value of the number of people gathering, the decoded real-time video image is directly prepared for next sampling, and the real-time tracking matrix is used as a history tracking matrix required for next detection.
4. The people group detection method according to claim 1, wherein when there exists a person whose continuous non-occurrence time exceeds a temporal separation time threshold in the real-time tracking matrix, the person information vector of the person in the real-time tracking matrix is deleted, and the real-time tracking matrix is used as a historical tracking matrix required for the next detection.
5. The people gathering detection method as claimed in claim 1, wherein the sampling of the decoded real-time video image specifically comprises:
decoding the real-time video image and acquiring an initial frame number of a current frame image;
and determining the number of sampling interval frames based on the sampling interval time, and if the initial frame number variable quantity between the current frame image and the last sampling frame is greater than the sampling interval frame number, acquiring the current frame image as a sampling frame and coding the sampling frame number.
6. The people gathering detection method according to claim 1, wherein the obtaining of the maximum number of suspicious people gathering based on the people gathering distance threshold specifically comprises:
based on the personnel information vectors in the suspicious personnel tracking matrix, respectively taking each suspicious personnel as an aggregation circle, and acquiring a minimum distance matrix between personnel in any aggregation circle and personnel in other aggregation circles;
judging whether the minimum distance smaller than the personnel gathering distance threshold exists in the minimum distance matrix;
if the minimum distance exists, merging the two clustering circles corresponding to the minimum distance to serve as a new clustering circle, updating the minimum distance matrix, and judging whether the minimum distance smaller than the personnel clustering distance threshold exists in the minimum distance matrix again;
if not, the number of people in the cluster with the most people is taken as the maximum number of suspicious people.
7. A people gathering detection device, comprising:
the acquisition module is used for sampling the decoded real-time video image and acquiring a current frame image as a sampling frame;
the first processing module is used for analyzing the current frame image through a personnel structuring algorithm to obtain a current frame personnel information vector;
the second processing module is used for merging the current frame personnel information vector into a historical tracking matrix containing historical frame personnel information vectors through a DeepsORT algorithm to obtain a real-time tracking matrix;
the first determining module is used for traversing the personnel information vectors in the real-time tracking matrix, acquiring the movement track vectors of different personnel, marking the personnel with continuous multiframe movement distances smaller than a movement threshold value and duration time exceeding an aggregation time threshold value as suspicious personnel based on the movement track vectors, and acquiring a suspicious personnel tracking matrix;
and the second confirmation module is used for circularly traversing the personnel information vectors in the suspicious personnel tracking matrix, acquiring the maximum suspicious personnel gathering number based on the personnel gathering distance threshold, and judging that the personnel gathering phenomenon occurs when the maximum suspicious personnel gathering number reaches the gathering number threshold.
8. People gathering detection device, characterized in that the people gathering detection device comprises a processor and a memory, wherein: the memory having stored therein computer instructions; the processor executes the computer instructions to implement the method of any one of claims 1-6.
9. A computer-readable storage medium having stored thereon computer instructions which, when executed on a computer, cause the computer to perform the method of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210107678.4A CN114494350B (en) | 2022-01-28 | 2022-01-28 | Personnel gathering detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210107678.4A CN114494350B (en) | 2022-01-28 | 2022-01-28 | Personnel gathering detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114494350A true CN114494350A (en) | 2022-05-13 |
| CN114494350B CN114494350B (en) | 2022-10-14 |
Family
ID=81476602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210107678.4A Active CN114494350B (en) | 2022-01-28 | 2022-01-28 | Personnel gathering detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114494350B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079192A (en) * | 2023-10-12 | 2023-11-17 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101325690A (en) * | 2007-06-12 | 2008-12-17 | 上海正电科技发展有限公司 | Method and system for detecting human flow analysis and crowd accumulation process of monitoring video flow |
| CN104658008A (en) * | 2015-01-09 | 2015-05-27 | 北京环境特性研究所 | Personnel gathering detection method based on video images |
| WO2016019973A1 (en) * | 2014-08-04 | 2016-02-11 | Bitmonlab S.L.U | Method for determining stationary crowds |
| US20160335491A1 (en) * | 2015-05-14 | 2016-11-17 | Ricoh Company, Ltd. | Method and device for detecting gathering of objects based on stereo vision as well as non-transitory computer-readable medium |
| JP2019106631A (en) * | 2017-12-12 | 2019-06-27 | セコム株式会社 | Image monitoring device |
| CN110147743A (en) * | 2019-05-08 | 2019-08-20 | 中国石油大学(华东) | Real-time online pedestrian analysis and number system and method under a kind of complex scene |
| WO2020104254A1 (en) * | 2018-11-20 | 2020-05-28 | Signify Holding B.V. | A people counting system with aggregated detection regions |
| CN111325048A (en) * | 2018-12-13 | 2020-06-23 | 杭州海康威视数字技术股份有限公司 | Personnel gathering detection method and device |
| CN111666821A (en) * | 2020-05-12 | 2020-09-15 | 深圳力维智联技术有限公司 | Personnel gathering detection method, device and equipment |
| CN112232316A (en) * | 2020-12-11 | 2021-01-15 | 科大讯飞(苏州)科技有限公司 | Crowd gathering detection method and device, electronic equipment and storage medium |
| CN112668432A (en) * | 2020-12-22 | 2021-04-16 | 上海幻维数码创意科技股份有限公司 | Human body detection tracking method in ground interactive projection system based on YoloV5 and Deepsort |
| CN113887372A (en) * | 2021-09-27 | 2022-01-04 | 厦门汇利伟业科技有限公司 | Target aggregation detection method and computer-readable storage medium |
-
2022
- 2022-01-28 CN CN202210107678.4A patent/CN114494350B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101325690A (en) * | 2007-06-12 | 2008-12-17 | 上海正电科技发展有限公司 | Method and system for detecting human flow analysis and crowd accumulation process of monitoring video flow |
| WO2016019973A1 (en) * | 2014-08-04 | 2016-02-11 | Bitmonlab S.L.U | Method for determining stationary crowds |
| CN104658008A (en) * | 2015-01-09 | 2015-05-27 | 北京环境特性研究所 | Personnel gathering detection method based on video images |
| US20160335491A1 (en) * | 2015-05-14 | 2016-11-17 | Ricoh Company, Ltd. | Method and device for detecting gathering of objects based on stereo vision as well as non-transitory computer-readable medium |
| JP2019106631A (en) * | 2017-12-12 | 2019-06-27 | セコム株式会社 | Image monitoring device |
| WO2020104254A1 (en) * | 2018-11-20 | 2020-05-28 | Signify Holding B.V. | A people counting system with aggregated detection regions |
| CN111325048A (en) * | 2018-12-13 | 2020-06-23 | 杭州海康威视数字技术股份有限公司 | Personnel gathering detection method and device |
| CN110147743A (en) * | 2019-05-08 | 2019-08-20 | 中国石油大学(华东) | Real-time online pedestrian analysis and number system and method under a kind of complex scene |
| CN111666821A (en) * | 2020-05-12 | 2020-09-15 | 深圳力维智联技术有限公司 | Personnel gathering detection method, device and equipment |
| CN112232316A (en) * | 2020-12-11 | 2021-01-15 | 科大讯飞(苏州)科技有限公司 | Crowd gathering detection method and device, electronic equipment and storage medium |
| CN112668432A (en) * | 2020-12-22 | 2021-04-16 | 上海幻维数码创意科技股份有限公司 | Human body detection tracking method in ground interactive projection system based on YoloV5 and Deepsort |
| CN113887372A (en) * | 2021-09-27 | 2022-01-04 | 厦门汇利伟业科技有限公司 | Target aggregation detection method and computer-readable storage medium |
Non-Patent Citations (3)
| Title |
|---|
| SHAILENDER KUMAR等: "Object tracking and counting in a zone using YOLOv4, DeepSORT and TensorFlow", 《PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND SMART SYSTEMS (ICAIS-2021)》 * |
| 李荔等: "视频监控场景下基于深度学习的多人脸跟踪", 《计算机与多媒体技术》 * |
| 陈冲等: "基于视频分析的人群密集场所客流监控预警研究", 《中国安全生产科学技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079192A (en) * | 2023-10-12 | 2023-11-17 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
| CN117079192B (en) * | 2023-10-12 | 2024-01-02 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114494350B (en) | 2022-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6759474B2 (en) | Vessel automatic tracking methods and systems based on deep learning networks and average shifts | |
| US8929588B2 (en) | Object tracking | |
| Bai et al. | Traffic anomaly detection via perspective map based on spatial-temporal information matrix. | |
| US11816881B2 (en) | Multiple object detection method and apparatus | |
| CN115830075A (en) | Hierarchical association matching method for pedestrian multi-target tracking | |
| CN114494350B (en) | Personnel gathering detection method and device | |
| Bhagya et al. | An overview of deep learning based object detection techniques | |
| CN110781733A (en) | Image duplicate removal method, storage medium, network equipment and intelligent monitoring system | |
| CN111767847A (en) | Pedestrian multi-target tracking method integrating target detection and association | |
| CN112434566B (en) | Passenger flow statistics method and device, electronic equipment and storage medium | |
| CN116403139A (en) | Visual tracking and positioning method based on target detection | |
| CN108241150B (en) | Method for detecting and tracking moving object in three-dimensional sonar point cloud environment | |
| Alhothali et al. | Anomalous event detection and localization in dense crowd scenes | |
| CN110766715A (en) | Multi-target tracking method combined with single target track | |
| Chauhan et al. | Study of moving object detection and tracking for video surveillance | |
| CN113052139A (en) | Deep learning double-flow network-based climbing behavior detection method and system | |
| Wu et al. | A Dense Litchi Target Recognition Algorithm for Large Scenes | |
| Gad et al. | Crowd density estimation using multiple features categories and multiple regression models | |
| CN112767438A (en) | Multi-target tracking method combining space-time motion | |
| CN109885560B (en) | Cleaning method and device for infrared thermal image monitoring data of shale gas fracturing equipment | |
| CN113673383A (en) | Time-space domain obstacle detection method and system for complex road scene | |
| Xu et al. | Fall detection based on person detection and multi-target tracking | |
| Xu et al. | Abnormal behavior detection in crowd scenes | |
| CN110309698A (en) | Movement human abnormal behaviour automatic identifying method | |
| CN114820714B (en) | Visual multi-target tracking method based on multi-feature similarity matching algorithm |
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 |