CN114913452A - Office place-based violation detection system and method - Google Patents
Office place-based violation detection system and method Download PDFInfo
- Publication number
- CN114913452A CN114913452A CN202210415037.5A CN202210415037A CN114913452A CN 114913452 A CN114913452 A CN 114913452A CN 202210415037 A CN202210415037 A CN 202210415037A CN 114913452 A CN114913452 A CN 114913452A
- Authority
- CN
- China
- Prior art keywords
- violation
- behavior
- network
- detection
- identification
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Abstract
The invention discloses a violation detection system and method based on office places, wherein the system comprises a multi-channel video reading and distributing module, a multi-channel video processing module and a video processing module, wherein the multi-channel video reading and distributing module is used for reading input multi-channel video streams and preprocessing the input multi-channel video streams to obtain monitoring images; the violation detection and identification module is used for detecting and identifying the violation of the monitored image; the violation behavior triggering and stopping identification module is used for storing a behavior judgment result of the single-frame monitoring image with a certain time length by using the time sequence dynamic buffer pool, and judging whether the violation behavior is triggered or the triggered violation behavior is stopped according to whether the violation behavior duration of the single-frame monitoring image reaches a specified threshold value; and the identity recognition module is used for recognizing the identity information of the personnel in the monitoring image according to the face and the appearance characteristics. The invention judges and records whether the violation behaviors occur or not and the identity information of the violation actor by detecting and analyzing the real-time video picture of the monitoring camera accessed to the office, thereby effectively supervising the violation behaviors of the staff and improving the working efficiency.
Description
Technical Field
The invention relates to the field of image recognition, in particular to a violation detection system and method based on an office place.
Background
The human behavior recognition technology can fully utilize equipment resources, extract relevant data information from the existing camera video and assist behavior recognition. However, because behavior recognition belongs to the problem of multi-person form recognition in a complex scene, the accuracy of the current universal recognition algorithm is low, so that the matching accuracy of the human behavior recognition result and the identity recognition result is low, and the human behavior recognition result can only be roughly recognized, and can not be refined and accurate and correspond to each person with definite identity.
Disclosure of Invention
Aiming at the problems, the invention provides an office-based violation detection system and method, which are used for helping managers to monitor the violation of staff and improving the working efficiency.
In a first aspect of the present invention, an office-based violation detection system, the system comprising:
the multi-channel video reading and distributing module is used for reading and preprocessing the input multi-channel video stream to obtain a monitoring image;
the violation behavior detection and identification module is used for carrying out violation behavior detection and identification on the monitoring image, the violation behavior detection and identification on the monitoring image is based on a specific target detection neural network, the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
The further technical scheme of the invention is as follows: the system also comprises an identity recognition module which is used for recognizing the identity information of the personnel in the designated area of the monitored image according to the human face and the appearance characteristics.
The further technical scheme of the invention is as follows: the system also comprises an illegal action triggering and termination identification module which is used for storing the action judgment result of the single-frame monitoring image with a certain time length by utilizing the time sequence dynamic buffer pool and judging whether the illegal action is triggered or terminated according to whether the duration time of the illegal action of the single-frame monitoring image reaches a specified threshold value.
The further technical scheme of the invention is as follows: the identity recognition module detects and recognizes the face in the monitored image by using a face recognition network, if the identity information is successfully recognized, the appearance characteristics of the human body are extracted through a feature network and stored in an appearance vector retrieval library, and the identity information corresponding to the current camera station is recorded; and if the face identity information is not successfully obtained and the current monitoring image has illegal behaviors, starting identity re-identification, extracting the appearance characteristics of the illegal behavior through an identity re-identification network, searching the most similar appearance characteristic vector in an appearance vector retrieval library, and obtaining corresponding identity information.
The further technical scheme of the invention is as follows: the target detection network is Yolov5, and the classification network is Vgg 16.
The further technical scheme of the invention is as follows: before the violation detection and identification module detects and identifies the violation of the monitored image, a certain number of training set pictures are collected, the training set pictures are manually marked as specific target behaviors, and then the specific target detection neural network is trained.
In a second aspect of the present invention, a method for detecting an illegal action based on an office space includes the following steps:
reading input multi-channel video streams and preprocessing the input multi-channel video streams to obtain monitoring images;
and carrying out illegal behavior detection and identification on the monitored image, wherein the illegal behavior detection and identification on the monitored image is based on a specific target detection neural network, the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
The further technical scheme of the invention is as follows: the method further comprises the step of identifying the personnel identity information of the designated area of the monitored image according to the human face and the appearance characteristics.
The further technical scheme of the invention is as follows: the method further comprises triggering and terminating the violation, and specifically comprises the following steps: and storing the behavior judgment result of the single-frame monitoring image with a certain time length by using a time sequence dynamic buffer pool, and judging whether the violation is triggered or the triggered violation is terminated according to whether the violation duration of the single-frame monitoring image reaches a specified threshold value.
The further technical scheme of the invention is as follows: the identifying of the personnel identity information of the designated area of the monitored image according to the face and the appearance characteristics specifically comprises the following steps: detecting and identifying the face in the monitored image by using a face identification network, if the identity information is successfully identified, extracting the appearance characteristics of the human body through a characteristic network and storing the appearance characteristics in an appearance vector retrieval library, and recording the identity information corresponding to the current camera station; and if the face identity information is not successfully obtained and the current monitoring image has an illegal action, identity re-identification is started, and after the appearance features of the illegal action are extracted through an identity re-identification network, the most similar appearance feature vector is searched in an appearance vector search library to obtain corresponding identity information.
The invention provides an office-based violation detection system and method, which can effectively help managers to monitor violation behaviors of employees and improve working efficiency by detecting and analyzing real-time video pictures returned by monitoring cameras of accessed offices or open office halls and judging and recording whether the violation behaviors occur in corresponding stations and identity information of violation actors. The beneficial effects obtained finally are as follows:
(1) according to the invention, a deep learning technology is applied to an indoor violation detection task, so that the detection efficiency of the violation of the staff in an office environment is greatly improved, and the management staff can conveniently and accurately obtain the violation and identity information of the staff in time;
(2) the invention provides a novel two-stage detection and identification network on the violation detection method, firstly, a video frame is subjected to target detection and identification to obtain an image area where possible violations are located, and then the image area is input into a secondary identification network to carry out secondary classification on corresponding violations, so that the obtained violation detection result is stable and reliable.
(3) The invention provides a double recognition strategy aiming at the identity recognition of an illegal actor, namely a double structure which is mainly based on a face detection and recognition network and assisted by an identity re-recognition network. When the high-quality face cannot be captured, the final identity information is obtained by identifying and retrieving the appearance characteristics of the human body, the identity information of the illegal actor can be accurately provided under the condition of face identification failure caused by picture angles or other interference factors, manual checking by managers is not needed, and the usability of the system is greatly improved.
Drawings
FIG. 1 is a schematic structural diagram of an office-based violation detection system according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of an implementation of a multi-channel video reading and distributing module according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of an implementation flow of an illegal behavior detection and identification module according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of an implementation flow of an violation triggering and termination identifying module according to an embodiment of the present invention;
fig. 5 is a schematic flow chart of a violation detection method based on an office according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical scheme of the present invention in detail, the present embodiment is implemented on the premise of the technical scheme of the present invention, and detailed implementation modes and specific steps are given.
The embodiment of the invention provides the following embodiments aiming at a violation detection system and method based on an office place:
example 1 based on the invention
This embodiment is used to explain an office-based violation detection system, and as shown in fig. 1, the system includes:
the multi-channel video reading and distributing module is used for reading the input multi-channel video stream and carrying out preprocessing to obtain a monitoring image;
and the violation behavior detection and identification module is used for carrying out violation behavior detection and identification on the monitored image, and carrying out violation behavior detection and identification on the monitored image based on a specific target detection neural network, wherein the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
In this embodiment, the multi-channel video reading and distributing module is designed to process multi-channel monitoring videos concurrently, the multi-channel video reading and distributing module inputs streaming media videos defined as a multi-channel RTSP or RTMP format, and after all video streams are synchronously accessed, an independent process is allocated to each video stream for decoding and preprocessing image data. The specific flow is shown in fig. 2, wherein, considering that the monitoring video stream is mostly 25 frames per second, and the difference exists between the subsequent detection and identification processing and the video stream decoding speed, this embodiment creates an independent buffer pool for each video stream to synchronize the reading and processing sub-processes as much as possible; image down-sampling down-scales the original high resolution 1080P or 2K surveillance video image to 1280 x 720 resolution to reduce the amount of computation; the histogram equalization aims at balancing the illumination condition of the image and reducing the influence of overexposure or over-darkness of the image on the judgment of subsequent illegal actions.
In this embodiment, the violation detection and identification module analyzes whether a violation exists in a designated area in the monitored image in real time based on a deep learning technology. The method comprises the steps of firstly reading the latest image frame from an image buffer pool, then detecting whether a specific violation action exists in the current image frame by using a specific target detection neural network, and updating a time sequence record. The specific behavior detection and identification process is as follows:
step 1: and reading the latest image frame, namely the monitoring image, from the image buffer pool.
Step 2: and sending the monitoring image into a specific target detection neural network, and outputting the frame coordinates and the category of the specific target behavior detection.
And step 3: and (3) performing double non-maximum suppression on the detection result of the specific target behavior in the step (2), and removing repeated frames.
And 4, step 4: if the mobile phone is detected in the step 3 and the mobile phone target frame is at least in one person target frame, the detection result of 'playing the mobile phone' of the current frame is abnormal, and if the sleeping target frame is detected in the step 3, the detection result of 'sleeping' of the current frame is abnormal.
And 5: and writing the violation detection result of the current frame and the current timestamp into a time sequence dynamic buffer pool of the corresponding camera, checking the state of the buffer pool and responding.
Preferably, the specific target behaviors in the step 2 are three types of target behaviors including "mobile phone", "non-sleeping person" and "sleeping person", wherein the "mobile phone" and the "non-sleeping person" form a "mobile phone playing" violation behavior, and the violation behavior can be prevented from being mistakenly reported by separately detecting the mobile phone.
Preferably, the specific target detection neural network comprises a target detection network and a classification network, wherein the target detection network is Yolov5, and the classification network is Vgg 16. Before the violation detection and identification module detects and identifies the violation of the monitored image, a certain number of training set pictures are collected, the training set pictures are manually marked as specific target behaviors, and then the specific target detection neural network is trained. Specifically, in order to train the Yolov5 and Vgg16 networks, a certain number of training sets need to be collected first, images of the training sets are shot by cameras in a real monitoring scene, three types of targets to be detected need to be marked manually, and after training of the neural networks is completed, the training sets can be deployed for detecting behaviors of specific targets. As shown in fig. 3, the monitoring image frames are input into a Yolov5 target detection network, and after the detection of three types of target behaviors is completed, secondary classification is performed, that is, the detection result of each type of target is input into a Vgg16 classification network for secondary judgment, so as to reduce target misjudgment.
Example 2 based on the invention
On the basis of embodiment 1, this embodiment provides an office-based violation detection system, where the system further includes a violation triggering and termination determining module, configured to store a behavior determination result of a single-frame monitoring image for a certain duration by using a time-series dynamic buffer pool, and determine, according to whether a violation duration of the single-frame monitoring image reaches a specified threshold, whether to trigger a violation or terminate a triggered violation.
In this embodiment, a flowchart of the violation triggering and termination recognizing module further recognizing the beginning and the end of the violation according to the time-sequence dynamic buffer pool is shown in fig. 4. In the embodiment 1, because the detection result of the illegal target of only a single image frame is obtained in steps 3 and 4, and considering the time continuity of the illegal action and the possible false detection of the neural network for detecting a specific target, the embodiment uses the time sequence buffer pool to store the detection result of the single image frame for a certain time, and judges whether to trigger the illegal action or terminate the triggered illegal action according to whether the detection result of the single image frame in the continuous time reaches a specified threshold value. Specifically, as shown in fig. 4:
s1, for the input current frame detection result, firstly determining the buffer pool state:
when the detection result is abnormal and the buffer pool state is normal, S2 is executed;
when the detection result is abnormal and the buffer pool state is also abnormal, modifying the buffer pool state to be abnormal, triggering violation abnormity, and entering identity recognition;
when the detection result is normal and the buffer pool state is abnormal, modifying the buffer pool state into abnormal stop, and determining whether an abnormal state ending condition is met, if so, executing S5, otherwise, ending the judgment;
when the detection result is normal and the buffer pool state is normal, ending the judgment;
s2, determining whether a preposed violation behavior exists, if so, executing S3, otherwise, recording the current violation behavior, and updating the violation behavior starting time;
s3, determining whether the duration time of the violation exceeds a threshold value, if so, executing S4, and if not, finishing the judgment;
s4, modifying the buffer pool state to be abnormal, triggering violation abnormity, and entering identity recognition;
and S5, modifying the state of the buffer pool to be normal, updating the violation ending time, sending a violation ending signal, and emptying the buffer pool.
The buffer pool state as the abnormal state ending condition is in abnormal stop and the duration time exceeds a threshold value.
Example 3 based on the invention
On the basis of the foregoing embodiment 1 or embodiment 2, this embodiment provides an office-based violation detection system, and the system further includes an identity recognition module configured to recognize identity information of a person in a designated area of the monitored image according to a face and an appearance feature.
In this embodiment, the identity recognition module detects and recognizes a face in a monitored image by using a face recognition network, and if identity information is successfully recognized, extracts human body appearance features through a feature network and stores the human body appearance features in an appearance vector retrieval library, and records identity information corresponding to a current camera station; and if the face identity information is not successfully obtained and the current monitoring image has an illegal action, identity re-identification is started, and after the appearance features of the illegal action are extracted through an identity re-identification network, the most similar appearance feature vector is searched in an appearance vector search library to obtain corresponding identity information.
Specifically, after the violation is detected, further detecting and identifying identity information of the violation actor greatly facilitates real-time monitoring and post-incident check of statistical data of the violation by a manager. And the illegal behavior identity recognition module carries out identity recognition based on the human body detection result of the illegal behavior detection and identification module. Due to external factors such as the installation angle of a camera, the posture of a human body and the like, the latest identity information cannot be obtained in real time only by using a face recognition mode, and in order to improve the accuracy and reliability of identity recognition, the identity recognition module provided by the invention comprises two identity recognition modes which mainly adopt face recognition and secondarily adopt identity re-recognition. And when the illegal behavior is detected, the identity recognition module continuously detects and recognizes the identity information, preferentially adopts the face recognition result of the current frame of the illegal behavior, and if the face recognition result with high reliability cannot be obtained, obtains the identity of the current illegal behavior through identity re-recognition. For example, when a person lowers his head, the overhead camera cannot shoot the front face at a downward shooting angle, and an effective face recognition result cannot be obtained at this time, but in consideration of the temporal continuity of the human body appearance information, the identity information is indirectly obtained by retrieving the most similar appearance features with the identity information.
In a preferred example of the embodiment, the violation detection and identification module transmits the detection results of "sleeping person" and "non-sleeping person" and the video image frame to the identity identification module in real time. The face recognition network detects and recognizes faces in the video image frames, if identity information is successfully recognized, human body appearance features are extracted through the feature network, stored in an appearance vector retrieval library, and identity information corresponding to the current camera station is recorded; if the face identity information cannot be successfully obtained and the current frame has illegal behaviors, identity re-identification is started, and after the appearance characteristics of the illegal behaviors are extracted by the identity re-identification network, the most similar appearance vector is searched in an appearance vector search library to obtain corresponding identity information.
Preferably, the face recognition network is ArcFace, the feature network is RetinaFace, the appearance vector retrieval base is milvus, and the identity re-recognition network is a ResNet50 neural network, after the face target frame is detected by ArcFace, the face target frame is input into the RetinaFace face vector extraction network to obtain a 512-dimensional face feature vector, and the most similar face is retrieved in the milvus vector retrieval base to obtain identity information. Wherein, the confidence of the face detected by the ArcFace and the similarity of the vector detected in the milvus both need to exceed a threshold value to be used as an effective detection result. The identity re-identification network adopts a classical ResNet50 neural network as a feature extraction network to obtain a 2048-dimensional human body appearance feature vector, and the similarity is calculated with the human body appearance vector stored in the milvus to indirectly obtain the identity information of the detected person.
Example 4 based on the invention
This embodiment describes a method corresponding to the system according to embodiments 1 to 3 of the present disclosure with reference to fig. 5, and a method for detecting an illegal action based on an office, where the method includes the following steps: reading input multi-channel video streams and preprocessing the input multi-channel video streams to obtain monitoring images; and carrying out illegal behavior detection and judgment on the monitored image, and carrying out illegal behavior detection and judgment on the monitored image based on a specific target detection neural network, wherein the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
Further, the method further comprises the step of identifying the personnel identity information of the designated area of the monitoring image according to the human face and the appearance characteristics.
Further, the method further comprises triggering and terminating the violation, and specifically comprises: and storing the behavior judgment result of the single-frame monitoring image with a certain time length by using a time sequence dynamic buffer pool, and judging whether to trigger the illegal behavior or terminate the triggered illegal behavior according to whether the duration of the illegal behavior of the single-frame monitoring image reaches a specified threshold value.
Further, identifying the personnel identity information of the designated area of the monitored image according to the human face and the appearance characteristics specifically comprises the following steps: detecting and identifying the face in the monitored image by using a face identification network, if the identity information is successfully identified, extracting the appearance characteristics of the human body through a characteristic network and storing the appearance characteristics in an appearance vector retrieval library, and recording the identity information corresponding to the current camera station; and if the face identity information is not successfully obtained and the current monitoring image has illegal behaviors, starting identity re-identification, extracting the appearance characteristics of the illegal behavior through an identity re-identification network, searching the most similar appearance characteristic vector in an appearance vector retrieval library, and obtaining corresponding identity information.
The specific working process of the office-based violation detection method in this embodiment refers to the description of the above office-based violation detection system in embodiments 1 to 3, and is not described again.
The invention provides an office-based violation detection system and method, which take an image processing technology as a core and mainly comprise violation identification and identity identification. The violation behavior judgment mainly analyzes whether a person in a specific video area has a specific violation behavior; and the identity recognition part further recognizes the identity of the agent when detecting that the illegal action occurs so as to help the manager to acquire complete abnormal information. Whether the violation behaviors occur in the corresponding stations and the identity information of the violation behaviors are judged and recorded by detecting and analyzing the real-time video pictures returned by the monitoring cameras of the accessed offices or the open office hall, so that the management personnel can be effectively helped to supervise the violation behaviors of the staff, and the working efficiency is improved. The beneficial effects obtained finally are as follows: according to the invention, the deep learning technology is applied to the indoor violation detection task, so that the detection efficiency of the violation of the staff in the office environment is greatly improved, and the management staff can conveniently and accurately obtain the violation of the staff and the identity information in time; the invention provides a novel two-stage detection and identification network on the violation detection method, firstly, a video frame is subjected to target detection and identification to obtain an image area where possible violations are located, and then the image area is input into a secondary identification network to carry out secondary classification on corresponding violations, so that the obtained violation detection result is stable and reliable; the invention provides a double recognition strategy aiming at the identity recognition of an illegal actor, namely a double structure which is mainly based on a face detection and recognition network and assisted by an identity re-recognition network. When the high-quality face cannot be captured, the final identity information is obtained by identifying and retrieving the appearance characteristics of the human body, the identity information of the illegal actor can be accurately provided under the condition of face identification failure caused by picture angles or other interference factors, manual checking by managers is not needed, and the usability of the system is greatly improved.
In this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process or method.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (10)
1. An office-based violation detection system, the system comprising:
the multi-channel video reading and distributing module is used for reading the input multi-channel video stream and carrying out preprocessing to obtain a monitoring image;
the illegal behavior detection and identification module is used for carrying out illegal behavior detection and identification on the monitored image, the illegal behavior detection and identification on the monitored image is based on a specific target detection neural network, the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
2. The office-based violation behavior detection system according to claim 1, further comprising an identity recognition module for recognizing personnel identity information in the monitored image according to the human face and the appearance features.
3. The office-based violation detection system according to claim 1, further comprising a violation triggering and termination identification module, configured to store a behavior determination result of a single-frame monitoring image for a certain duration by using a time-series dynamic buffer pool, and determine whether to trigger a violation or terminate a triggered violation according to whether a violation duration of the single-frame monitoring image reaches a specified threshold.
4. The office-based violation behavior detection system according to claim 2, wherein the identity recognition module detects and recognizes a face in the monitored image by using a face recognition network, extracts human appearance features through a feature network and stores the human appearance features in an appearance vector retrieval library if the identity information is successfully recognized, and records identity information corresponding to a current camera station; and if the face identity information is not successfully obtained and the current monitoring image has illegal behaviors, starting identity re-identification, extracting the appearance characteristics of the illegal behavior through an identity re-identification network, searching the most similar appearance characteristic vector in an appearance vector retrieval library, and obtaining corresponding identity information.
5. The office-based violation detection system according to claim 1, wherein said target detection network is Yolov5 and said classification network is Vgg 16.
6. The office-based violation detection system according to claim 1, wherein the violation detection and identification module collects a certain number of training set pictures before performing violation detection and identification on the monitored image, and trains the specific target detection neural network after manually marking the training set pictures as specific target behaviors.
7. An office-based violation detection method, comprising the steps of:
reading input multi-channel video streams and preprocessing the input multi-channel video streams to obtain monitoring images;
and carrying out illegal behavior detection and identification on the monitored image, wherein the illegal behavior detection and identification on the monitored image is based on a specific target detection neural network, the specific target detection neural network comprises a target detection network and a classification network, the target detection network is used for detecting the marked specific target behavior, and the classification network is used for carrying out secondary classification on the finished specific target behavior detection result and judging the behavior result.
8. The office-based violation behavior detection method according to claim 7, further comprising identifying personnel identity information in the monitored image according to the human face and the appearance features.
9. The office-based violation detection method according to claim 7, further comprising triggering and terminating the violation, specifically comprising: and storing the behavior judgment result of the single-frame monitoring image with a certain time length by using a time sequence dynamic buffer pool, and judging whether the violation is triggered or the triggered violation is terminated according to whether the violation duration of the single-frame monitoring image reaches a specified threshold value.
10. The office violation behavior detection method based on claim 8, wherein the identifying of the identity information of the personnel in the designated area of the monitored image according to the face and the appearance features specifically comprises: detecting and identifying the face in the monitored image by using a face identification network, if the identity information is successfully identified, extracting the appearance characteristics of the human body through a characteristic network and storing the appearance characteristics in an appearance vector retrieval library, and recording the identity information corresponding to the current camera station; and if the face identity information is not successfully obtained and the current monitoring image has illegal behaviors, starting identity re-identification, extracting the appearance characteristics of the illegal behavior through an identity re-identification network, searching the most similar appearance characteristic vector in an appearance vector retrieval library, and obtaining corresponding identity information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210415037.5A CN114913452A (en) | 2022-04-20 | 2022-04-20 | Office place-based violation detection system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210415037.5A CN114913452A (en) | 2022-04-20 | 2022-04-20 | Office place-based violation detection system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114913452A true CN114913452A (en) | 2022-08-16 |
Family
ID=82764219
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210415037.5A Pending CN114913452A (en) | 2022-04-20 | 2022-04-20 | Office place-based violation detection system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114913452A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117152419A (en) * | 2023-10-11 | 2023-12-01 | 中国矿业大学 | Method and system for detecting illegal carrying articles of personnel of mine overhead manned device |
-
2022
- 2022-04-20 CN CN202210415037.5A patent/CN114913452A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117152419A (en) * | 2023-10-11 | 2023-12-01 | 中国矿业大学 | Method and system for detecting illegal carrying articles of personnel of mine overhead manned device |
| CN117152419B (en) * | 2023-10-11 | 2024-03-29 | 中国矿业大学 | Method and system for detecting illegal carrying articles of personnel of mine overhead manned device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222640B (en) | Method, device and method for identifying suspect in monitoring site and storage medium | |
| CN110569720B (en) | Audio and video intelligent identification processing method based on audio and video processing system | |
| CN109299683B (en) | Security protection evaluation system based on face recognition and behavior big data | |
| CN110264493B (en) | Method and device for tracking multiple target objects in motion state | |
| CN110381366B (en) | Automatic event reporting method, system, server and storage medium | |
| CN108269333A (en) | Face identification method, application server and computer readable storage medium | |
| CN105844659B (en) | The tracking and device of moving component | |
| CN106204815A (en) | A kind of gate control system based on human face detection and recognition | |
| WO2020094088A1 (en) | Image capturing method, monitoring camera, and monitoring system | |
| CN104883548B (en) | Monitor video face captures processing method and its system | |
| KR20050057586A (en) | Enhanced commercial detection through fusion of video and audio signatures | |
| CN106951846A (en) | A kind of face 3D models typing and recognition methods and device | |
| CN110852147A (en) | Security alarm method, security alarm device, server and computer readable storage medium | |
| KR101372860B1 (en) | System for searching video and server for analysing video | |
| CN112184771B (en) | Method and device for tracking personnel track of community | |
| CN112634561A (en) | Safety alarm method and system based on image recognition | |
| CN114913452A (en) | Office place-based violation detection system and method | |
| CN110334568B (en) | Track generation and monitoring method, device, equipment and storage medium | |
| CN113269127B (en) | Face recognition and pedestrian re-recognition monitoring method and system for real-time automatic database establishment | |
| CN114022942A (en) | Dynamic personnel monitoring method, system, medium and equipment based on face recognition | |
| CN106572324A (en) | Energy-efficient smart monitoring device | |
| CN114819110B (en) | Method and device for identifying speaker in video in real time | |
| JP3612220B2 (en) | Person monitoring method | |
| CN115988164A (en) | Conference room multimedia control method, system and computer equipment | |
| CN109145758A (en) | A kind of recognizer of the face based on video monitoring |
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 |