CN112733630A - Channel gate detection method, device, equipment and storage medium - Google Patents
Channel gate detection method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN112733630A CN112733630A CN202011584043.0A CN202011584043A CN112733630A CN 112733630 A CN112733630 A CN 112733630A CN 202011584043 A CN202011584043 A CN 202011584043A CN 112733630 A CN112733630 A CN 112733630A
- Authority
- CN
- China
- Prior art keywords
- channel gate
- sample set
- training
- network
- gate detection
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 82
- 238000012549 training Methods 0.000 claims abstract description 70
- 238000000034 method Methods 0.000 claims abstract description 42
- 238000002372 labelling Methods 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 3
- 208000027418 Wounds and injury Diseases 0.000 abstract description 5
- 230000006378 damage Effects 0.000 abstract description 5
- 208000014674 injury Diseases 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 230000009471 action Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a channel gate detection method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring a channel gate video; selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set; training the centernet _ mobilenet _ v2 network by using a training sample set, and establishing a channel gate detection model; and detecting the target to be detected in the channel gate according to the trained channel gate detection model. The channel gate detection model established by the invention takes a lightweight mobilent _ v2 network as a backbone network, has higher real-time performance, can be deployed on an embedded platform, effectively detects various targets such as adults, children, bicycles, electric vehicles, baby carriages, traveling cases and the like in the channel gate, solves the problems of personnel injury, gate collision and the like caused by a blind area of infrared opposite emission of the channel gate, and has superior detection speed.
Description
Technical Field
The invention relates to the field of intelligent security monitoring, in particular to a channel gate detection method, a channel gate detection device, channel gate detection equipment and a storage medium.
Background
The channel gate machine is a common device in import and export management, and the channel gate is often used in community access & exit, station access & exit, building access & exit etc. department, along with the rise of wisdom community, wisdom building, has obtained extensive application. The existing channel gate detection system mainly uses infrared grating correlation sensors which are distributed at different positions of a gate machine to monitor the passing of personnel.
However, for some special crowd scenes, such as scenes that an adult rides an electric vehicle or a bicycle, the adult carries an oversized trunk or carries a large parcel, the adult carries a child or holds a baby or pushes a baby carriage, and the child does not reach a specified height, the traditional infrared grating correlation detection has a blind area, and safety accidents such as people clamping injuries, collision of a pedestrian passageway gate and the like cannot be avoided.
Therefore, how to effectively detect the special crowd scene and solve a series of problems of personnel pinching, collision and the like caused by the dead zone of the infrared grating correlation existing in the current channel gate, and the technical problem to be solved by the technical personnel in the field is urgently needed.
Disclosure of Invention
In view of the above, the present invention provides a method, an apparatus, a device and a storage medium for detecting a tunnel gate, which can effectively detect multiple types of targets of the tunnel gate and have high real-time performance and excellent detection speed. The specific scheme is as follows:
a channel gate detection method, comprising:
acquiring a channel gate video;
selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set;
training the centernet _ mobilenet _ v2 network by using the training sample set, and establishing a channel gate detection model;
and detecting a target to be detected in the channel gate according to the trained channel gate detection model.
Preferably, in the method for detecting a channel gate provided in the embodiment of the present invention, while training the centernet _ mobilenet _ v2 network, the method further includes:
and accessing an RFB module for training after a feature extraction layer of the centernet _ mobilene _ v2 network.
Preferably, in the method for detecting a channel gate provided in the embodiment of the present invention, while training the centernet _ mobilenet _ v2 network, the method further includes:
and connecting an SE module to train after the feature extraction layer of the centrenetet _ mobilenet _ v2 network.
Preferably, in the above method for detecting a tunnel gate provided in the embodiment of the present invention, a calculation formula of a multitasking loss function of the tunnel gate detection model is as follows:
Ldet=Lk+λsizeLsize+λoffLoff
wherein L isdetAs a multitask penalty function, LkAs a center point thermodynamic loss function, LsizeAs a function of the regression loss of the target dimension, λsizeWeight of the target size regression loss, LoffAs a function of center point offset loss, λoffThe weight of the regression loss of the target center point.
Preferably, in the method for detecting a tunnel gate provided in the embodiment of the present invention, after labeling the target in each picture in the training sample set, the method further includes:
performing data cleaning on the training sample set;
randomly adjusting the corresponding proportion of brightness, contrast, hue, saturation and size of the pictures in the training sample set;
and carrying out mirror image operation and random clipping on the pictures in the training sample set.
Preferably, in the method for detecting a channel gate provided in the embodiment of the present invention, acquiring a channel gate video specifically includes:
and acquiring channel gate videos through cameras arranged on the left side and the right side of the channel gate.
The embodiment of the present invention further provides a channel gate detection device, including:
the video acquisition module is used for acquiring a channel gate video;
the sample set generation module is used for selecting a plurality of pictures from the acquired channel gate video as a training sample set and marking targets in the pictures in the training sample set;
the network training module is used for training the centernet _ mobilenet _ v2 network by using the training sample set to establish a channel gate detection model;
and the target detection module is used for detecting a target to be detected in the channel gate according to the trained channel gate detection model.
Preferably, in the above channel gate detection apparatus provided in the embodiment of the present invention, the network training module is further configured to access an RFB module for training after a feature extraction layer of the centernet _ mobilene _ v2 network, and is further configured to access an SE module for training after the feature extraction layer of the centernet _ mobilene _ v2 network.
The embodiment of the invention also provides channel gate detection equipment, which comprises a processor and a memory, wherein the channel gate detection method provided by the embodiment of the invention is realized when the processor executes the computer program stored in the memory.
The embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the above channel gate detection method provided in the embodiment of the present invention.
According to the above technical solution, the channel gate detection method provided by the present invention includes: acquiring a channel gate video; selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set; training the centernet _ mobilenet _ v2 network by using a training sample set, and establishing a channel gate detection model; and detecting the target to be detected in the channel gate according to the trained channel gate detection model.
The centernet-based channel gate detection model established by the invention takes a lightweight mobilenet _ v2 network as a backbone network, has higher real-time performance, can be deployed on an embedded platform with limited resources and the like, effectively detects various targets such as adults, children, bicycles, electric vehicles, baby carriages, traveling cases and the like in the channel gate, solves the problems of personnel clamping injury, collision with a gate machine and the like caused by a blind area of infrared correlation of the channel gate, and has superior detection speed. In addition, the invention also provides a corresponding device, equipment and a computer readable storage medium aiming at the channel gate detection method, so that the method has higher practicability, and the device, the equipment and the computer readable storage medium have corresponding advantages.
Drawings
In order to more clearly illustrate the embodiments of the present invention or technical solutions in related arts, the drawings used in the description of the embodiments or related arts will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flowchart of a channel gate detection method according to an embodiment of the present invention;
FIG. 2 is a schematic view of a tunnel gate camera installation provided in an embodiment of the present invention;
FIG. 3 is a schematic diagram of a centernet detection network according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an RFB module according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an SE module according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a tunnel gate detection apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a channel gate detection method, as shown in fig. 1, comprising the following steps:
s101, acquiring a channel gate video;
specifically, a camera is respectively arranged on the left side and the right side of the channel gate to ensure that the complete channel gate on the opposite side can be seen from respective visual angles, and then a channel gate video is obtained through the camera; taking fig. 2 as an example, the camera on the left channel gate can see the right complete channel gate from the viewing angle, and the camera on the right channel gate can also see the left complete channel gate from the viewing angle. And opening a detection channel gate after installation and fixation.
S102, selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set;
it should be noted that, in this step, based on the channel gate video recorded in the prepared rich scene obtained in step S101, a suitable picture is selected as a training sample, and the sample is manually labeled, and several types of targets mainly labeled may include an adult, a child, a bicycle, an electric vehicle, a trunk, a baby carriage, and the like inside the channel gate.
S103, training the centernet _ mobilenet _ v2 network by using a training sample set, and establishing a channel gate detection model;
it can be understood that, because the method of centernet is anchor _ free, it is not necessary to set a large number of candidate boxes like anchor _ based, which is time-consuming and labor-consuming, and the centernet directly performs nms in the network through a 3 × 3 max pool operation, which does not need nms based iou, and is fast and efficient. Therefore, the invention is based on the centernet target detection method, and in order to realize real-time detection, the mobile _ v2 is used as a backbone network to construct a lightweight centernet _ mobile _ v2 network, so that the detection rate and the detection performance can be improved;
s104, detecting a target to be detected in the channel gate according to the trained channel gate detection model;
it should be noted that the obtained final tunnel gate detection model can be used for testing the fabricated tunnel gate detection test data set under the GTX1080 platform. Through tests, the model can effectively detect various targets such as adults, children, bicycles, electric vehicles, baby carriages, traveling cases and the like, and has high detection speed.
In the method for detecting the channel gate provided by the embodiment of the invention, the established channel gate detection model takes a lightweight mobilent _ v2 network as a backbone network, has high real-time performance, can be deployed on an embedded platform with limited resources and the like, effectively detects various targets such as adults, children, bicycles, electric vehicles, baby carriages, traveling cases and the like in the channel gate, solves the problems of personnel clamping injury, collision and the like caused by a blind area of infrared correlation of the channel gate, and has excellent detection speed.
In specific implementation, in the method for detecting a tunnel barrier provided in the embodiment of the present invention, after the step S102 is executed to label the target in each picture in the training sample set, the method may further include: performing data cleaning on the training sample set to obtain a cleaned training sample set; and data enhancement is carried out during training, wherein the enhancement in the aspect of color is carried out on the pictures in the training sample set, namely, the random adjustment of corresponding proportion (such as 0.6 to 1.4) is respectively carried out on the brightness, the contrast, the hue and the saturation of the pictures on the original basis, the enhancement in the aspect of size is carried out on the pictures, the amplification and the reduction are randomly carried out in the corresponding proportion (such as 0.6 to 1.4), the mirroring operation is carried out on the pictures, and the random cutting is carried out on the pictures. This may generate a richer training sample set.
Further, in specific implementation, in the method for detecting a channel gate provided in the embodiment of the present invention, in order to improve robustness of a model, while performing step S103 to train a centernet _ mobilene _ v2 network, as shown in fig. 3 and 4, the method may further include: and accessing an RFB (perceptual Field enhanced Block) module for training after a feature extraction layer of the centernet _ mobilene _ v2 network. Therefore, the feature extraction capability of the model can be enhanced, and high discriminant features can be extracted.
Further, in a specific implementation, in the method for detecting a channel gate provided in the embodiment of the present invention, while the step S103 is executed to train the centernet _ mobilene _ v2 network, as shown in fig. 3 and 5, the method may further include: the feature extraction layer of the centernet _ mobilenet _ v2 network is followed by the SE (channel attention) module for training. By adding the attention mechanism, the useful features can be enhanced adaptively, and the invalid features can be suppressed.
In addition, while the step S103 is executed to train the centernet _ mobilenet _ v2 network, the method may further include: and constructing an elliptic Gaussian kernel generating thermodynamic diagram, and performing adaptive adjustment according to the size and the aspect ratio of the object. Therefore, the thermodynamic diagram is generated by constructing an appropriate elliptical Gaussian core according to the size and the aspect ratio of the object in a self-adaptive manner aiming at targets with different aspect ratios such as people, bicycles, luggage cases and the like.
In a specific implementation, in the above method for detecting a tunnel gate provided in the embodiment of the present invention, a calculation formula of a multitask loss function of a tunnel gate detection model is as follows:
Ldet=Lk+λsizeLsize+λoffLoff (1)
wherein L isdetAs a multitask penalty function, LkAs a center point thermodynamic loss function, LsizeAs a function of the regression loss of the target dimension, λsizeWeight of the target size regression loss, LoffAs a function of center point offset loss, λoffThe weight of the regression loss of the target center point.
Specifically, the new feature layer features are fed into a multitasking loss function (from the centerpoint thermodynamic loss function L)kTarget size regression loss function LsizeAnd center point offset loss function LoffIn combination) LdetAnd (6) performing calculation.
In the invention, in order to pay attention to the central point thermodynamic diagram and the central point offset, lambda is usedsizeSet to 0.1, lambdaoffIs set to 1. L iskIs calculated as focal loss, as shown in formula (2):
wherein, YxycFor the label of the ground channel,and N is the number of central points of the label output by the network. x and y are coordinates of a target central point, and c is the category of the central point;andthe method is used for reducing the loss weight of the simple samples and increasing the loss weight of the difficult samples; (1-Y)xyc)βFor dealing with the problem of positive and negative sample imbalance, the loss weight near the center point is mainly reduced.
Labeling for each ground channelTransforming the true points according to the down-sampling multiple RThen it is transformed into a thermodynamic diagram by an elliptical Gaussian kernelWherein sigmaaAnd σbIs the standard deviation of the target size adaptation in x and y directions, as shown in equation (3):
Loffpredicting an offset by regression output for smoothL1 loss functionCalculating the loss between the predicted offset and the true offset, as shown in equation (4):
Lsizeas a function of smoothL1 loss, usingTo represent the coordinates of the upper left and lower right corners of the target bounding box, and k represents the class C of the targetkThe coordinates of the center point can be expressed asOutputting target size by regressionCalculating a predicted target size and a true target size skThe loss therebetween is shown in formula (5):
in particular implementations, the model parameters may be iteratively updated using a random gradient descent (SGD) method until the model converges.
Based on the same inventive concept, the embodiment of the present invention further provides a tunnel gate detection apparatus, and as the principle of the apparatus for solving the problem is similar to the foregoing tunnel gate detection method, the implementation of the apparatus can refer to the implementation of the tunnel gate detection method, and repeated details are not repeated.
In specific implementation, the channel gate detection apparatus provided in the embodiment of the present invention, as shown in fig. 6, may specifically include:
the video acquisition module 11 is used for acquiring a channel gate video;
the sample set generation module 12 is configured to select multiple pictures from the acquired channel gate video as a training sample set, and label a target in each picture in the training sample set;
a network training module 13, configured to train a centernet _ mobilenet _ v2 network by using a training sample set, and establish a channel gate detection model;
and the target detection module 14 is used for detecting a target to be detected in the channel gate according to the trained channel gate detection model.
In the gateway detection device provided by the embodiment of the invention, a gateway detection model can be established by the interaction of the four modules and taking a lightweight mobilenet _ v2 network as a backbone network, so that the gateway detection model has high real-time performance, can be deployed on an embedded platform with limited resources and the like, can effectively detect various targets such as adults, children, bicycles, electric vehicles, baby carriages, traveling cases and the like in a gateway, and has excellent detection speed.
Further, in a specific implementation, in the above channel gate detection apparatus provided in the embodiment of the present invention, the network training module 13 may be further configured to access the RFB module for training after the feature extraction layer of the centernet _ mobilene _ v2 network, and may be further configured to access the SE module for training after the feature extraction layer of the centernet _ mobilene _ v2 network.
For more specific working processes of the modules, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Correspondingly, the embodiment of the invention also discloses a channel gate detection device, which comprises a processor and a memory; the processor implements the channel gate detection method disclosed in the foregoing embodiments when executing the computer program stored in the memory.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Further, the present invention also discloses a computer readable storage medium for storing a computer program; the computer program, when executed by a processor, implements the channel gate detection method disclosed above.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other. The device, the equipment and the storage medium disclosed by the embodiment correspond to the method disclosed by the embodiment, so that the description is relatively simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The embodiment of the invention provides a channel gate detection method, which comprises the following steps: acquiring a channel gate video; selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set; training the centernet _ mobilenet _ v2 network by using a training sample set, and establishing a channel gate detection model; and detecting the target to be detected in the channel gate according to the trained channel gate detection model. The channel gate detection model established by the invention takes a lightweight mobilent _ v2 network as a backbone network, has higher real-time performance, can be deployed on an embedded platform with limited resources and the like, effectively detects various targets such as adults, children, bicycles, electric vehicles, strollers, traveling cases and the like in the channel gate, solves the problems of personnel injury, gate collision and the like caused by a blind area of infrared opposite emission of the channel gate, and has superior detection speed. In addition, the invention also provides a corresponding device, equipment and a computer readable storage medium aiming at the channel gate detection method, so that the method has higher practicability, and the device, the equipment and the computer readable storage medium have corresponding advantages.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The channel gate detection method, device, apparatus and storage medium provided by the present invention are described in detail above, and the principle and implementation of the present invention are explained herein by applying specific examples, and the description of the above examples is only used to help understanding the method and core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A channel gate detection method, comprising:
acquiring a channel gate video;
selecting a plurality of pictures from the acquired channel gate video as a training sample set, and labeling targets in the pictures in the training sample set;
training the centernet _ mobilenet _ v2 network by using the training sample set, and establishing a channel gate detection model;
and detecting a target to be detected in the channel gate according to the trained channel gate detection model.
2. The channel gate detection method of claim 1, wherein while training the centernet _ mobilene _ v2 network, further comprising:
and accessing an RFB module for training after a feature extraction layer of the centernet _ mobilene _ v2 network.
3. The channel gate detection method of claim 2, wherein while training the centernet _ mobilene _ v2 network, further comprising:
and connecting an SE module to train after the feature extraction layer of the centrenetet _ mobilenet _ v2 network.
4. The method according to claim 3, wherein the calculation formula of the multitask loss function of the tunnel gate detection model is as follows:
Ldet=Lk+λsizeLsize+λoffLoff
wherein L isdetAs a multitask penalty function, LkAs a center point thermodynamic loss function, LsizeAs a function of the regression loss of the target dimension, λsizeWeight of the target size regression loss, LoffAs a function of center point offset loss, λoffThe weight of the regression loss of the target center point.
5. The method of claim 4, further comprising, after labeling the targets in the pictures in the training sample set:
performing data cleaning on the training sample set;
randomly adjusting the corresponding proportion of brightness, contrast, hue, saturation and size of the pictures in the training sample set;
and carrying out mirror image operation and random clipping on the pictures in the training sample set.
6. The method according to claim 5, wherein the acquiring a gateway video specifically comprises:
and acquiring channel gate videos through cameras arranged on the left side and the right side of the channel gate.
7. A tunnel gate detection device, comprising:
the video acquisition module is used for acquiring a channel gate video;
the sample set generation module is used for selecting a plurality of pictures from the acquired channel gate video as a training sample set and marking targets in the pictures in the training sample set;
the network training module is used for training the centernet _ mobilenet _ v2 network by using the training sample set to establish a channel gate detection model;
and the target detection module is used for detecting a target to be detected in the channel gate according to the trained channel gate detection model.
8. The apparatus of claim 7, wherein the network training module is further configured to access an RFB module for training after a feature extraction layer of the centernet _ mobilene _ v2 network, and further configured to access an SE module for training after the feature extraction layer of the centernet _ mobilene _ v2 network.
9. A tunnel gate detection apparatus comprising a processor and a memory, wherein the processor implements the tunnel gate detection method according to any one of claims 1 to 6 when executing a computer program stored in the memory.
10. A computer-readable storage medium for storing a computer program, wherein the computer program when executed by a processor implements the channel gate detection method of any of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011584043.0A CN112733630A (en) | 2020-12-28 | 2020-12-28 | Channel gate detection method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011584043.0A CN112733630A (en) | 2020-12-28 | 2020-12-28 | Channel gate detection method, device, equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112733630A true CN112733630A (en) | 2021-04-30 |
Family
ID=75606698
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011584043.0A Pending CN112733630A (en) | 2020-12-28 | 2020-12-28 | Channel gate detection method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112733630A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113627346A (en) * | 2021-08-11 | 2021-11-09 | 深圳市捷顺科技实业股份有限公司 | Channel gate detection method, device, equipment and readable storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110852257A (en) * | 2019-11-08 | 2020-02-28 | 深圳和而泰家居在线网络科技有限公司 | Method and device for detecting key points of human face and storage medium |
CN111914937A (en) * | 2020-08-05 | 2020-11-10 | 湖北工业大学 | Lightweight improved target detection method and detection system |
CN111985299A (en) * | 2020-06-29 | 2020-11-24 | 济南浪潮高新科技投资发展有限公司 | Mask identification method and device based on improved convolutional neural network |
CN112101113A (en) * | 2020-08-14 | 2020-12-18 | 北京航空航天大学 | Lightweight unmanned aerial vehicle image small target detection method |
-
2020
- 2020-12-28 CN CN202011584043.0A patent/CN112733630A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110852257A (en) * | 2019-11-08 | 2020-02-28 | 深圳和而泰家居在线网络科技有限公司 | Method and device for detecting key points of human face and storage medium |
CN111985299A (en) * | 2020-06-29 | 2020-11-24 | 济南浪潮高新科技投资发展有限公司 | Mask identification method and device based on improved convolutional neural network |
CN111914937A (en) * | 2020-08-05 | 2020-11-10 | 湖北工业大学 | Lightweight improved target detection method and detection system |
CN112101113A (en) * | 2020-08-14 | 2020-12-18 | 北京航空航天大学 | Lightweight unmanned aerial vehicle image small target detection method |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113627346A (en) * | 2021-08-11 | 2021-11-09 | 深圳市捷顺科技实业股份有限公司 | Channel gate detection method, device, equipment and readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP3869459B1 (en) | Target object identification method and apparatus, storage medium and electronic apparatus | |
US10102421B2 (en) | Method and device for face recognition in video | |
CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
CN109784290B (en) | Target detection method, device, equipment and readable storage medium | |
US9159137B2 (en) | Probabilistic neural network based moving object detection method and an apparatus using the same | |
CN111126399A (en) | Image detection method, device and equipment and readable storage medium | |
CN107169993A (en) | Detection recognition method is carried out to object using public security video monitoring blurred picture | |
CN107194396A (en) | Method for early warning is recognized based on the specific architecture against regulations in land resources video monitoring system | |
CN107220603A (en) | Vehicle checking method and device based on deep learning | |
Uy et al. | Automated traffic violation apprehension system using genetic algorithm and artificial neural network | |
CN110543877A (en) | Identification recognition method, training method and device of model thereof and electronic system | |
CN112434566B (en) | Passenger flow statistics method and device, electronic equipment and storage medium | |
JP4975801B2 (en) | Monitoring method and monitoring apparatus using hierarchical appearance model | |
CN112733630A (en) | Channel gate detection method, device, equipment and storage medium | |
CN112465854A (en) | Unmanned aerial vehicle tracking method based on anchor-free detection algorithm | |
Bedruz et al. | Philippine vehicle plate localization using image thresholding and genetic algorithm | |
CN109684982B (en) | Flame detection method based on video analysis and combined with miscible target elimination | |
CN111767881A (en) | Self-adaptive crowd density estimation device based on AI technology | |
CN107635188A (en) | A kind of video frequency vehicle trace analysis method based on Docker platforms | |
CN110503049A (en) | Based on the satellite video number of vehicles estimation method for generating confrontation network | |
CN109241952B (en) | Figure counting method and device in crowded scene | |
CN114092885A (en) | High-speed service area vehicle protection method and system based on video analysis | |
CN113920585A (en) | Behavior recognition method and device, equipment and storage medium | |
CN103325121A (en) | Method and system for estimating network topological relations of cameras in monitoring scenes | |
CN105959707A (en) | Static state background video compression algorithm based on motion perception |
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 |