CN213715948U - Lightweight and multi-pose face detection and recognition system based on deep learning - Google Patents
Lightweight and multi-pose face detection and recognition system based on deep learning Download PDFInfo
- Publication number
- CN213715948U CN213715948U CN202022623118.3U CN202022623118U CN213715948U CN 213715948 U CN213715948 U CN 213715948U CN 202022623118 U CN202022623118 U CN 202022623118U CN 213715948 U CN213715948 U CN 213715948U
- Authority
- CN
- China
- Prior art keywords
- wireless signal
- face
- signal receiver
- signal transmitter
- alarm
- 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.)
- Expired - Fee Related
Links
Images
Abstract
The invention provides a light-weight and multi-pose face detection and recognition system based on deep learning, which has less manpower requirement, higher accuracy, quicker response time and more uniform scheduling management, and comprises: the system comprises a plurality of cameras, a plurality of sensors and a control unit, wherein the cameras comprise door diameter cameras arranged at door diameter openings of a high-voltage electric operation area, control ball cameras arranged in the high-voltage electric operation area and high-voltage chamber cameras arranged in a high-voltage chamber of the high-voltage electric operation area; the first wireless signal transmitter is connected with the camera; a first wireless signal receiver associated with the first wireless signal transmitter; the plurality of alarm devices are arranged at different positions of the high-voltage electric operation area; the second wireless signal receiver is connected with the alarm device; a second wireless signal transmitter associated with a second wireless signal receiver; the wireless switch is connected with the first wireless signal receiver and the second wireless signal transmitter; a face management platform and a face video analysis box.
Description
Technical Field
The utility model belongs to the face identification field, concretely relates to building site security protection system, based on light-weight and multi-pose face detection and identification system of degree of depth study.
Background
The automatic face detection and recognition system is a novel engineering full-life-cycle management concept, and is realized in the industry of the intelligent earth concept in the engineering field. The existing security system adopts a camera and a manual mode to monitor and dispatch, and more labor cost is needed. Meanwhile, the accuracy of manual monitoring is uncontrollable, a prevention mechanism is not provided for accidents, a millisecond notification mechanism is not provided for field workers and non-field responsible persons, and the timeliness of notification and processing during accidents is not guaranteed.
Meanwhile, the wireless transmission camera is far away from a monitoring point, signal loss is large, images are blurred, the shot videos are stored in the independent storage space of the camera, unified backup is not available, and the images cannot be accurately, live and rapidly checked.
In addition, different systems are often used for different devices of different scenes in a construction site, which also does not facilitate uniform scheduling management.
SUMMERY OF THE UTILITY MODEL
For solving above-mentioned problem, provide a human demand still less, the degree of accuracy is higher, response time is rapider, more unified face detection and identification system of dispatch management, the utility model discloses a following technical scheme:
the utility model provides a lightweight and many gestures face detection and identification system based on degree of depth study for monitor and real-time early warning, a serial communication port, include the staff of operation in the building site that has the high-tension electricity operation district: the system comprises a plurality of cameras, a plurality of cameras and a control device, wherein the cameras comprise door diameter cameras which are arranged in door diameter openings of a high-voltage electric operation area and are used for recording videos of entering and exiting of door diameter operators, ball distribution and control cameras which are arranged in the high-voltage electric operation area and are used for recording videos of working of the operators, and high-voltage chamber cameras which are arranged in a high-voltage chamber of the high-voltage electric operation area and are used for recording videos of activities of the operators in the high-voltage chamber; the first wireless signal transmitter is connected with the camera; a first wireless signal receiver associated with the first wireless signal transmitter; the plurality of alarm devices are arranged at different positions of the high-voltage electric operation area; the second wireless signal receiver is connected with the alarm device; a second wireless signal transmitter associated with a second wireless signal receiver; the wireless switch is connected with the first wireless signal receiver and the second wireless signal transmitter; the face management platform is used for receiving videos transmitted by the camera sequentially through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch; and the face video analysis box is connected with the face management platform and is used for carrying out face video analysis on the video received by the face management platform.
The utility model provides a lightweight and many gestures face detection and identification system based on degree of depth study can also have such technical feature, wherein, still deploys in the high-tension electricity operation district and is used for patrolling and examining the robot, and the camera is still including setting up the robot camera on patrolling and examining the robot, and alarm device is including setting up the robot alarm on patrolling and examining the robot.
The utility model provides a light-weight and many gestures face detection and identification system based on degree of depth study can also have such technical feature, and wherein, alarm device carries out the door footpath manager managed, deploys at the door footpath alarm of door footpath mouth, deploys the cloth accuse ball alarm in the high-tension electricity operation district and deploys the hyperbaric chamber alarm in the hyperbaric chamber including being used for the business turn over to the door footpath mouth.
The utility model provides a light-weight and many gestures face detection and identification system based on degree of depth study can also have such technical feature, wherein, alarm device is for being used for carrying out face video analysis according to face video analysis box and obtaining and loop through the analysis result that face management platform, wireless switch, second radio signal transmitter and first radio signal receiver transmitted and carry out the alarm device that corresponds the warning.
The utility model provides a light-weight and many gestures face detection and identification system based on degree of depth study can also have such technical feature, and wherein, face video analysis box includes transmission controller and a plurality of degree of depth study module, and degree of depth study module includes face detection module, people's face alignment module and face identification module at least, and transmission control module is connected with face management platform and is connected with each degree of depth study module respectively.
The utility model provides a light-weight and many gestures face detection and identification system based on degree of depth study can also have such technical characteristics, still includes: and the storage server is used for receiving and storing the video transmitted by the camera sequentially through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch.
The utility model provides a light-weight and many gestures face detection and identification system based on degree of depth study can also have such technical characteristics, still includes: the monitoring center and the smart phone are both used for receiving videos transmitted by the camera sequentially through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch, and monitoring and displaying the videos.
Utility model with the functions and effects
According to the light-weight and multi-pose human face detection and recognition system based on deep learning, as the building site personnel in the high-voltage field are dynamically monitored through the camera arranged on the building site, and are sent to the human face management platform through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch in sequence, the human face video analysis box is enabled to process the human face video analysis box; further, when the face video analysis box generates an analysis result, the alarm signal is sent to the alarm device to alarm through the face management platform, the wireless switch, the second wireless signal transmitter and the second wireless signal receiver in sequence. Therefore, through the utility model discloses a light-weight and many gestures face detection and identification system based on degree of depth study can carry out dynamic detection and discernment to whole high-pressure area operating personnel in the building site in real time, accomplishes real-time early warning, and intellectual detection system and pursuit, the standard management can improve reliability, security and the traceability of intelligent security work.
Drawings
Fig. 1 is the embodiment of the present invention provides a schematic diagram of a light-weight and multi-pose face detection and recognition system based on deep learning.
Detailed Description
As a new application field, Machine Learning (ML) technology has made a breakthrough in the fields of face detection and recognition. A typical deep learning model is a Convolutional Neural Network (CNN), and the weight value of the CNN shares a Network structure to make the CNN more similar to a biological Neural Network, so that the complexity of the Network model is reduced, and the number of the weight values is reduced. The advantage is more obvious when the input of the network is a multi-dimensional image, so that the image can be directly used as the input of the network, and the complex characteristic extraction and data reconstruction process in the traditional recognition algorithm is avoided. Convolutional networks are a multi-layered perceptron specifically designed to recognize two-dimensional shapes, the structure of which is highly invariant to translation, scaling, tilting, or other forms of deformation.
The overall superiority and inferiority of the face detection and recognition technology depends on the selection and operation speed of a core algorithm, and the technology determines the detection speed, the correctness and the system stability of the system in an actual application scene. Although the conventional feature extraction algorithm commonly used in the face detection at present uses an integral graph to accelerate feature extraction calculation amount, the face feature extraction is simpler, the system calculation complexity is increased in the face of complex feature extraction, the real-time detection speed is slowed, and the real-time detection requirement of intelligent security cannot be met. The improved Retinaface algorithm generates a target candidate frame through the algorithm, then performs classification regression on the candidate frame, adopts the technology of a feature pyramid, realizes the fusion of multi-scale information, greatly improves the speed and the precision, and particularly has better landmark robustness under the large-angle (roll, yaw and pitch) human face.
The utility model relates to a light-weight and many gestures face detection and identification system based on degree of depth learning, the face identification algorithm that this system adopted has carried out following optimization on original algorithm basis: 1) the Retinaface algorithm is selected as the feature extraction in the face detection algorithm, the operator has strong robustness to the image gray level change caused by illumination change, and the face detection error rate of the system in the actual complex background can be reduced. 2) A backtracking mechanism algorithm is introduced into forward search of deformable constraint and dense regression loss classification learning, so that the system is helped to select the optimal strong classifier, and the performance of the overall combined classifier is improved. 3) The human face is trained by a large-scale sample set, and the detection of more than 50 persons from a single crowd picture is supported at one time.
The Arcface network is used as an important one-stage deep learning model in face recognition, the network can learn face targets from a large number of face pictures to train faces with image characteristics, the algorithm speed is higher, normalization and additive angle intervals of feature vectors are improved on the basis of a convolutional neural network, and intra-class compactness and inter-class difference are improved while inter-class separability is improved.
And simultaneously, the utility model discloses a cloud engine frame is compared to real-time high concurrent people's face has still been established to the system in the server. The framework can fuse load balancing servers and the like through a network, and the servers support the people flow dense area N on line: and N, high-speed face comparison. Compared with the traditional comparison engine single machine structure, the internal structure adopts high-speed memory comparison and quick response technology, thereby solving the system comparison and real-time high concurrency requirements under large computation amount and really realizing the quick response of the system.
Therefore, the utility model discloses based on be equipped with above-mentioned Arcface network and people's face and compare the server of cloud engine frame, this server is face video analysis box in fact, comes to carry out real-time processing to the video that the camera of setting for in the predetermined place was shot in the building site to realize real-time early warning.
In order to make the utility model discloses the technological means, creation characteristic, achievement purpose and efficiency that realize are easily understood and are known, and it is right below to combine embodiment and attached drawing the utility model discloses a light-weight and multi-attitude face detection and identification system based on degree of depth learning does specifically expounded.
< example >
The utility model discloses a lightweight and many gestures face detection and identification system setting are in a building site based on degree of depth study for monitor and real-time early warning in order to avoid the operation dangerous to the staff, have high-tension electricity operation district in this building site, high-tension electricity operation district is equipped with the door footpath mouth that is used for letting the staff business turn over and deposits high-tension circuit's hyperbaric chamber, simultaneously, still deploys in the high-tension electricity operation district and has been used for making a round trip to patrol and examine the robot.
Fig. 1 is the embodiment of the present invention provides a schematic diagram of a light-weight and multi-pose face detection and recognition system based on deep learning.
As shown in fig. 1, the light-weight and multi-pose face detection and recognition system 100 based on deep learning includes a plurality of cameras 101, a plurality of alarm devices 102, a first wireless signal transmitter 103, a first wireless signal receiver 104, a second wireless signal transmitter 105, a second wireless signal receiver 106, a wireless switch 107, a face management platform 108, a face video analysis box 109, a storage server 110, a monitoring center 111, and a smart phone 112.
The cameras 101 are deployed at a plurality of predetermined locations in the worksite, and in the present embodiment, the cameras are divided into a door diameter camera 11, a cloth control ball camera 12, a high-pressure room camera 13, and a robot camera 14. Specifically, the method comprises the following steps:
the door diameter camera 11 is disposed at a door diameter of the high-voltage electric working area and used for recording videos of workers entering and exiting the door diameter. The door is arranged at the door diameter opening, the camera 11 at the door diameter opening is arranged at the edge of the door, and the lens faces the direction in which the staff walk, so that the face of the staff can be shot in the shot video.
The ball distribution control camera 12 is disposed in the high-voltage electric operation area and used for recording working videos of workers.
The high-voltage chamber camera 13 is disposed in a high-voltage chamber of the high-voltage electric working area and used for recording videos of activities of workers in the high-voltage chamber.
The robot camera 14 is disposed on each inspection robot and used for shooting videos corresponding to inspection pictures when the inspection robot performs inspection.
In this embodiment, each camera is connected to the first wireless signal transmitter 103, and can transmit the shot video to the first wireless signal transmitter 103 in real time.
The alarm device 102 is used for confirming that the identity is opened, the door cannot be opened, the door is opened and the entry and exit time is recorded, the door cannot be opened and the early warning is performed, the early warning of a suspect is performed, the suspect is controlled and the like, and specifically, the alarm device 102 comprises a door diameter manager 21, a door diameter alarm 22, a control ball alarm 23, a high-pressure chamber alarm 24 and a robot alarm 25.
The door diameter manager 21 is disposed at a door of a door diameter opening for managing opening and closing of the door.
The door diameter alarm 22 is disposed at the door diameter opening and used for giving an early warning when the door diameter manager 21 cannot open the door.
The ball distribution alarm 23 is arranged in a high-voltage electric operation area, the high-voltage chamber camera 13 is arranged in a high-voltage chamber of the high-voltage electric operation area, the robot alarm 25 is arranged on the inspection robot, and the three are used for alarming according to signals, for example, when a suspect appears in a construction site or a human face video analysis box 109 analyzes that a worker is in danger, the alarm is given.
In this embodiment, each of the alarm devices 102 is connected to the second wireless signal receiver 106.
The first wireless signal transmitter 103 is associated with a first wireless signal receiver 104 and the second wireless signal transmitter 105 is associated with a second wireless signal receiver 106. Wherein the first wireless signal receiver 104 and the second wireless signal receiver 106 are connected to the wireless switch 107.
In this embodiment, when the camera 101 captures a video, the video is transmitted to the wireless switch 107 connected to the first wireless signal receiver 104 through the first wireless signal transmitter 103, and similarly, when the human face video analysis box 109 analyzes a danger and an alarm signal is transmitted to the wireless switch 107, the alarm signal is transmitted to the alarm device 102 connected to the second wireless signal receiver 106 through the second wireless signal transmitter 105.
The wireless switch 107 is connected to the first wireless signal receiver 104 and the second wireless signal receiver 106, and is also configured to communicate with the face management platform 108, the storage server 110, the monitoring center 111, and the smart phone 112.
In this embodiment, when the wireless switch 107 receives the videos shot by the camera 101, the videos are respectively sent to the face management platform 108, the storage server 110, the monitoring center 111, and the smart phone 112.
The face management platform 108 is connected to the face video analysis box 109, and when the face management platform 108 receives the video, the video is transmitted to the face video analysis box 109 to perform face video analysis and obtain corresponding feedback information. The feedback information includes the analysis result of the face video analysis box 109 and alarm information.
In this embodiment, the face management platform 108 and the face video analysis box 109 are both disposed in a general control room of the construction site.
The face video analysis box 109 includes a transmission control module 91 and a plurality of different depth learning modules 92. The deep learning modules 92 are memories storing corresponding deep learning algorithms, and the face video analysis box 109 can be provided with the corresponding deep learning modules 92 according to actual needs.
The transmission control module 91 is connected with the face management platform 108 and is respectively connected with each deep learning module 92.
The deep learning module 92 includes a face detection module 921, a face tracking module 922, a face attribute analysis module 923, and a face recognition module 924.
The face detection module 921 is a pixel-level-based face positioning method, where a mobile is used for selecting a backbone network to achieve model lightweight, and the method adopts a multitask learning strategy, detects and positions a face at the same time, supports complex environments such as strong light, weak light, dark night, backlight, and detects multi-angle face positions such as a front face and a side face.
The face tracking module 922 accurately locates and tracks the position of the face region through the detected face frame, has high precision, wide adaptation angle and high speed, cuts the face into a 120 × 120 picture, and adopts five feature pyramids which are connected from top to bottom and transversely to improve the receptive field and increase the capability of context modeling, thereby tracking the face features.
The face attribute analysis module 923 is configured to detect attributes related to the face according to the features exhibited by the face. The method comprises the steps of obtaining information such as age, gender and head posture, connecting random boxes of all feature points together to generate a global feature, and then using the global feature to perform global linear regression.
The face recognition module 924 calculates similarity by comparing 512-dimensional feature values extracted from two faces, and provides a corresponding similarity ratio, where a value greater than a certain threshold is considered as a white list, and a value less than a certain threshold is considered as a black list.
Through the deep learning module 92, the face video analysis box 109 can analyze the face in the video, quickly judge whether the video shot by each camera 101 is dangerous or not, and generate corresponding alarm information to feed back to the face management platform 108 when the video is dangerous.
In this embodiment, when the wireless switch 107 receives the analysis results of the face video platform, the analysis results are also sent to the face management platform 108, the storage server 110, the monitoring center 111, and the smart phone 112, respectively.
The storage server 110 is used to store video transmitted by the wireless switch 107.
The monitoring center 111 and the smart phone 112 are used for relevant personnel to monitor and check the state of the construction site, the monitoring center 111 at least has a plurality of monitoring display screens for displaying videos, and the smart phone 112 is held by the relevant personnel.
In specific implementation, the camera 102 captures videos of various positions in the construction site in real time, and sends the videos to the face management platform 108 through the first wireless signal transmitter 103, the first wireless signal receiver 104 and the wireless switch 107, and the face management platform pushes the videos to the face video analysis box 109.
Next, in the face video analysis box 109, when the face detection module 921 analyzes the "face snapshot" information in a specific video, a corresponding analysis result is generated and fed back to the face management platform 108 by the transmission control module 91, and further transmitted to the wireless switch 107, on one hand, the wireless switch 107 sends the "face" information to the storage server 110, the monitoring center 111, and the smart phone 112 for storage, monitoring, and checking, on the other hand, the second wireless signal transmitter 105 and the second wireless signal receiver 106 feed back signals to the alarm device 102, and an alarm corresponding to the alarm device 102 gives an alarm, such as confirming that the identity opens the door, the door cannot be opened, the door opening and entry time is recorded, the door cannot be opened and an early warning is given, the early warning of a suspect, and the control of the suspect. Similarly, when the other deep learning modules 92 in the face video analysis box 109, i.e. the face tracking module 922, the face attribute analysis module 923 and the face recognition module 924 analyze the corresponding results, the corresponding working principle is the same as above, and will not be described herein again.
Examples effects and effects
According to the light-weight and multi-pose face detection and recognition system based on deep learning, a camera arranged on a construction site is used for dynamically monitoring construction site personnel in a high-voltage place, and the construction site personnel are sequentially sent to a face management platform through a first wireless signal transmitter, a first wireless signal receiver and a wireless switch so as to be processed by a face video analysis box; further, when the face video analysis box generates an analysis result, the alarm signal is sent to the alarm device to alarm through the face management platform, the wireless switch, the second wireless signal transmitter and the second wireless signal receiver in sequence. Therefore, through the light-weight and multi-pose face detection and recognition system based on deep learning, dynamic detection and recognition can be performed on the whole high-voltage area operators in the construction site in real time, real-time early warning, intelligent detection and tracking and standard management are achieved, and reliability, safety and traceability of intelligent security work can be improved.
The above embodiments are merely illustrative of specific embodiments of the present invention, and the present invention is not limited to the description of the above embodiments.
For example, in the above embodiment, the deep learning module includes four modules, namely a face detection module, a face tracking module, a face attribute analysis module and a face recognition module. The utility model discloses an in other embodiments, the degree of depth learning module also can be according to the field application demand, customizes different degree of depth learning network, chooses for use different frames and selects different reasoning equipment to realize multiple demands such as ageing, accuracy, removal deployment, data interaction.
Claims (6)
1. A light-weight and multi-pose face detection and recognition system based on deep learning for monitoring and real-time early warning of workers operating in a construction site with a high-voltage electric operating area, comprising:
the cameras comprise door diameter cameras which are arranged in door diameter openings of the high-voltage electric operating area and used for recording videos of entrance and exit of door diameter workers, ball distribution control cameras which are arranged in the high-voltage electric operating area and used for recording videos of the workers, and high-voltage chamber cameras which are arranged in a high-voltage chamber of the high-voltage electric operating area and used for recording videos of the workers in the high-voltage chamber;
the first wireless signal transmitter is connected with the camera;
a first wireless signal receiver associated with the first wireless signal transmitter;
the plurality of alarm devices are arranged at different positions of the high-voltage electric operation area;
the second wireless signal receiver is connected with the alarm device;
a second wireless signal transmitter associated with the second wireless signal receiver;
the wireless switch is connected with the first wireless signal receiver and the second wireless signal transmitter;
the face management platform is used for receiving the video transmitted by the camera sequentially through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch; and
and the face video analysis box is connected with the face management platform and is used for carrying out face video analysis on the video received by the face management platform.
2. The deep learning based lightweight and multi-pose face detection and recognition system according to claim 1, wherein:
wherein, the high-voltage electric operating area is also provided with an inspection robot for inspection,
the camera also comprises a robot camera arranged on the inspection robot,
the alarm device comprises a robot alarm arranged on the inspection robot.
3. The deep learning based lightweight and multi-pose face detection and recognition system according to claim 1, wherein:
the alarm device comprises a door diameter manager for managing the entrance and exit of the door diameter port, a door diameter alarm deployed at the door diameter port, a ball distribution and control alarm deployed in the high-voltage electric operation area and a high-voltage chamber alarm deployed in the high-voltage chamber.
4. The deep learning based lightweight and multi-pose face detection and recognition system of claim 3, wherein:
the alarm device is used for carrying out corresponding alarm according to the analysis result which is obtained by carrying out the face video analysis by the face video analysis box and is transmitted by the face management platform, the wireless switch, the second wireless signal transmitter and the first wireless signal receiver in sequence.
5. The deep learning based lightweight and multi-pose face detection and recognition system of claim 1, further comprising:
and the storage server is used for receiving and storing the video transmitted by the camera sequentially through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch.
6. The deep learning based lightweight and multi-pose face detection and recognition system of claim 1, further comprising:
the monitoring center and the smart phone are both used for receiving the video transmitted by the camera through the first wireless signal transmitter, the first wireless signal receiver and the wireless switch in sequence, and monitoring and displaying the video.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202022623118.3U CN213715948U (en) | 2020-11-13 | 2020-11-13 | Lightweight and multi-pose face detection and recognition system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202022623118.3U CN213715948U (en) | 2020-11-13 | 2020-11-13 | Lightweight and multi-pose face detection and recognition system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN213715948U true CN213715948U (en) | 2021-07-16 |
Family
ID=76805278
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202022623118.3U Expired - Fee Related CN213715948U (en) | 2020-11-13 | 2020-11-13 | Lightweight and multi-pose face detection and recognition system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN213715948U (en) |
-
2020
- 2020-11-13 CN CN202022623118.3U patent/CN213715948U/en not_active Expired - Fee Related
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105979210B (en) | A kind of pedestrian's identifying system based on the more ball video camera arrays of multiple gun | |
| CN105894702B (en) | A kind of intrusion detection warning system and its detection method based on multiple-camera data fusion | |
| CN101119482B (en) | Overall view monitoring method and apparatus | |
| CN109607031B (en) | Intelligent warehousing system and method based on unmanned aerial vehicle panorama | |
| CN106412501B (en) | A kind of the construction safety behavior intelligent monitor system and its monitoring method of video | |
| EP4105101A1 (en) | Monitoring system, monitoring method, and monitoring device for railway train | |
| CN110428522A (en) | A kind of intelligent safety and defence system of wisdom new city | |
| KR102149832B1 (en) | Automated Violence Detecting System based on Deep Learning | |
| CN111242025B (en) | Real-time action monitoring method based on YOLO | |
| CN109887040A (en) | The moving target actively perceive method and system of facing video monitoring | |
| CN110516529A (en) | It is a kind of that detection method and system are fed based on deep learning image procossing | |
| CN111679695B (en) | Unmanned aerial vehicle cruising and tracking system and method based on deep learning technology | |
| MX2012009579A (en) | Moving object tracking system and moving object tracking method. | |
| CN101093603A (en) | Module set of intellective video monitoring device, system and monitoring method | |
| CN105701945B (en) | Subdistrict intelligent monitoring warning positioning system and method | |
| CN108500992A (en) | A kind of multi-functional mobile security robot | |
| Lim et al. | Gun detection in surveillance videos using deep neural networks | |
| CN113963315A (en) | Real-time video multi-user behavior recognition method and system in complex scene | |
| CN110288801A (en) | Electric field video monitoring method, device, computer equipment and storage medium | |
| CN112184773A (en) | Helmet wearing detection method and system based on deep learning | |
| CN109544870A (en) | Alarm decision method and intelligent monitor system for intelligent monitor system | |
| US20240046701A1 (en) | Image-based pose estimation and action detection method and apparatus | |
| CN116664359A (en) | Intelligent fire early warning decision system and method based on multi-sensor fusion | |
| CN113505704B (en) | Personnel safety detection method, system, equipment and storage medium for image recognition | |
| CN111160149A (en) | Vehicle-mounted face recognition system and method based on motion scene and deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210716 Termination date: 20211113 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |