CN111860087A - Information detection method, information detection device, computer device, and storage medium - Google Patents
Information detection method, information detection device, computer device, and storage medium Download PDFInfo
- Publication number
- CN111860087A CN111860087A CN201910981932.1A CN201910981932A CN111860087A CN 111860087 A CN111860087 A CN 111860087A CN 201910981932 A CN201910981932 A CN 201910981932A CN 111860087 A CN111860087 A CN 111860087A
- Authority
- CN
- China
- Prior art keywords
- information
- vehicle
- sampling
- license plate
- information 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 264
- 238000005070 sampling Methods 0.000 claims description 154
- 238000012549 training Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 8
- 238000000034 method Methods 0.000 description 18
- 238000004364 calculation method Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 7
- 238000012795 verification Methods 0.000 description 5
- 238000013136 deep learning model Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an information detection method, an information detection device, computer equipment and a storage medium. The information detection method comprises the following steps: acquiring a shot image; and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to obtain the license plate information. According to the technical scheme, the vehicle and the license plate are respectively detected, so that the detection speed is increased, and the detection precision is improved.
Description
Technical Field
The present invention relates to the field of information detection technologies, and in particular, to an information detection method, an information detection apparatus, a computer device, and a storage medium.
Background
In the related art, high-precision target detection is mainly deployed at a server end and has abundant high-performance computing hardware support such as a memory, a Graphics Processing Unit (GPU), a Field Programmable Gate Array (FPGA), and the like, but the memory of a mobile end is limited, and not only does not have a general GPU to support calculation of a target detection model, but also other application processes in a user terminal need to be considered to run in the process of using the CPU for calculation. Under the condition that all aspects on a terminal are limited, target detection is roughly divided into a first-order detection model framework and a second-order detection model framework, the second-order detection model is large, the detection precision is high, and the speed is low; although the speed of some classical first-order detection models (e.g., yolo, retinet) is fast, these miniaturized models are high-speed detection realized on the premise of sacrificing a certain precision, and cannot meet the high-precision detection task, and these two models cannot meet the business requirements in terms of speed and precision at the same time.
Disclosure of Invention
The present invention is directed to solving at least one of the problems of the prior art or the related art.
To this end, an aspect of the present invention is to provide an information detecting method.
Another aspect of the present invention is to provide an information detecting apparatus.
Yet another aspect of the invention is directed to a computer device.
Yet another aspect of the present invention is to provide a computer-readable storage medium.
In view of the above, according to an aspect of the present invention, an information detecting method is provided, including: acquiring a shot image; and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to obtain the license plate information.
The information detection method provided by the invention abandons the mode of simultaneously detecting all types of objects aiming at the contradiction between the speed and the precision of a classical general detection model, uses independent detection branches aiming at the vehicle (large object) and the license plate (small object), and respectively uses the characteristic layer of the information detection model to independently connect one exclusive detection branch with the other exclusive detection branch, namely, the vehicle detection branch of the information detection model is used for detecting the vehicle to obtain the vehicle information, and the license plate detection branch of the information detection model is used for detecting the license plate to obtain the license plate information. According to the technical scheme, the vehicle and the license plate are respectively detected, so that the detection speed is increased, and the detection precision is improved.
The information detection method according to the present invention may further include the following technical features:
in the above technical solution, the method further comprises: judging whether the shot image is effective or not according to the vehicle information and the license plate information; and when the shot image is invalid, sending out prompt information.
According to the technical scheme, when vehicle verification is needed, a driver detects a vehicle and a license plate in real time in the vehicle shooting process, if the detected vehicle information and the detected license plate information are wrong or incomplete, the shot image is determined to be invalid, prompt information is sent out in time to remind the driver, and it is ensured that a short video or picture of the vehicle submitted by the driver is valid video information containing the vehicle and the license plate.
In any of the above technical solutions, the vehicle information includes one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
According to the technical scheme, the vehicle door state recognition is integrated in the independent detection branch of the vehicle, two states of opening and closing of the vehicle door are output, the vehicle direction recognition is integrated in the independent detection branch of the vehicle, and the direction information of 0 degree, 90 degrees, 180 degrees and 270 degrees of a vehicle photo is output.
In any of the above technical solutions, the method further includes: acquiring a sampling image and acquiring sampling information in the sampling image; and establishing an information detection model according to the sampling information.
According to the technical scheme, sampling information in the sampling image is marked, the sampling information is input into the information detection model, parameters of the information detection model are trained through the sampling information and the marking, an optimal model is obtained, and therefore vehicle information and license plate information can be accurately and rapidly detected through the model.
In any of the above technical solutions, the information detection model includes five layers of backbone networks; the step of establishing an information detection model according to the sampling information specifically comprises the following steps: the vehicle sampling information training information detection model comprises a fifth layer trunk network and a vehicle detection branch of the vehicle sampling information training information detection model according to sampling information, and a first layer trunk network, a second layer trunk network, a third layer trunk network, a fourth layer trunk network and a license plate detection branch of the license plate sampling information training information detection model according to the sampling information, wherein the vehicle sampling information comprises sampling vehicle body central point coordinates, sampling vehicle body outer frame coordinates, sampling vehicle body width, sampling vehicle body height, sampling vehicle head or vehicle tail attributes, sampling vehicle door opening or closing information, sampling vehicle direction information, and license plate sampling information comprises sampling license plate central point coordinates, sampling license plate width, sampling license plate height, sampling license plate foreground and sampling license plate background.
In the technical scheme, the information detection model uses mobilent _ v3 (a deep learning model) as a main network, independent detection branches (a vehicle detection branch and a license plate detection branch) with separated features are used for respectively detecting a vehicle and a license plate, the detection branches customize different aspect ratios of anchors according to different shapes of the vehicle and the license plate, and are more attached to a detection object to remarkably improve detection precision, wherein the anchors are fixed reference frames which are mapped on an original image by default at each point in a detection feature map in a target detection task. The vehicle is easy to detect in a larger proportion in the figure, so the vehicle detection branch is connected behind the last 5 multiplied by 5 low-resolution feature layer of the main network, and simultaneously, a vehicle head or vehicle tail, a vehicle door state and a vehicle direction identification model are integrated on the layer, so that a large object (vehicle) can be detected with high precision, and extra attribute identification can be integrated under the condition of adding a small amount of calculation. The license plate occupies a small proportion in the picture, and is not easy to detect the number of anchorages required to ensure the detection precision, but the use of more main network feature layers to access extra detection branches to increase the anchorages coverage brings a large amount of calculation, so the technical scheme of the invention integrates the remaining 4 layers of the main network, and the scale level of the anchorages is increased under the feature layer with the high resolution of 20 multiplied by 20 to provide enough anchorages coverage, thereby not only meeting the anchorages coverage number detected with high precision, but also saving the calculation power.
According to another aspect of the present invention, there is provided an information detecting apparatus comprising: an image acquisition unit configured to acquire a captured image; and the information detection unit is used for detecting the vehicle in the shot image by using the vehicle detection branch of the information detection model to acquire the vehicle information, and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to acquire the license plate information.
The information detection device provided by the invention abandons the mode of simultaneously detecting all types of objects aiming at the contradiction between the speed and the precision of a classical general detection model, uses independent detection branches aiming at a vehicle (a large object) and a license plate (a small object), and respectively uses the characteristic layer of the information detection model to independently connect one exclusive detection branch with another, namely, the vehicle information is obtained by detecting the vehicle by using the vehicle detection branch of the information detection model, and the license plate information is obtained by detecting the license plate by using the license plate detection branch of the information detection model. According to the technical scheme, the vehicle and the license plate are respectively detected, so that the detection speed is increased, and the detection precision is improved.
The information detection apparatus according to the present invention may further include:
In the above technical solution, the method further comprises: the processing unit is used for judging whether the shot image is effective or not according to the vehicle information and the license plate information; and the prompting unit is used for sending out prompting information when the shot image is invalid.
According to the technical scheme, when vehicle verification is needed, the processing unit detects the vehicle and the license plate in real time in the process that a driver shoots the vehicle, if the detected vehicle information and the detected license plate information are wrong or incomplete, the shot image is determined to be invalid, and the prompting unit sends prompting information in time to remind the driver, so that the condition that the short video or picture of the vehicle submitted by the driver to be verified is effective video information containing the vehicle and the license plate is ensured.
In any of the above technical solutions, the vehicle information includes one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
According to the technical scheme, the vehicle door state recognition is integrated in the independent detection branch of the vehicle, two states of opening and closing of the vehicle door are output, the vehicle direction recognition is integrated in the independent detection branch of the vehicle, and the direction information of 0 degree, 90 degrees, 180 degrees and 270 degrees of a vehicle photo is output.
In any of the above technical solutions, the method further includes: the model establishing unit is used for acquiring a sampling image and acquiring sampling information in the sampling image; and establishing an information detection model according to the sampling information.
According to the technical scheme, sampling information in the sampling image is marked, the sampling information is input into the information detection model, parameters of the information detection model are trained through the sampling information and the marking, an optimal model is obtained, and therefore vehicle information and license plate information can be accurately and rapidly detected through the model.
In any of the above technical solutions, the information detection model includes five layers of backbone networks; the model establishing unit establishes an information detection model according to the sampling information, and specifically comprises the following steps: the vehicle sampling information training information detection model comprises a fifth layer trunk network and a vehicle detection branch of the vehicle sampling information training information detection model according to sampling information, and a first layer trunk network, a second layer trunk network, a third layer trunk network, a fourth layer trunk network and a license plate detection branch of the license plate sampling information training information detection model according to the sampling information, wherein the vehicle sampling information comprises sampling vehicle body central point coordinates, sampling vehicle body outer frame coordinates, sampling vehicle body width, sampling vehicle body height, sampling vehicle head or vehicle tail attributes, sampling vehicle door opening or closing information, sampling vehicle direction information, and license plate sampling information comprises sampling license plate central point coordinates, sampling license plate width, sampling license plate height, sampling license plate foreground and sampling license plate background.
In the technical scheme, the information detection model uses mobilent _ v3 (a deep learning model) as a main network, independent detection branches (a vehicle detection branch and a license plate detection branch) with separated features are used for respectively detecting a vehicle and a license plate, the detection branches customize different aspect ratios of anchors according to different shapes of the vehicle and the license plate, and are more attached to a detection object to remarkably improve detection precision, wherein the anchors are fixed reference frames which are mapped on an original image by default at each point in a detection feature map in a target detection task. The vehicle is easy to detect in a larger proportion in the figure, so the vehicle detection branch is connected behind the last 5 multiplied by 5 low-resolution feature layer of the main network, and simultaneously, a vehicle head or vehicle tail, a vehicle door state and a vehicle direction identification model are integrated on the layer, so that a large object (vehicle) can be detected with high precision, and extra attribute identification can be integrated under the condition of adding a small amount of calculation. The license plate occupies a small proportion in the picture, and is not easy to detect the number of anchorages required to ensure the detection precision, but the use of more main network feature layers to access extra detection branches to increase the anchorages coverage brings a large amount of calculation, so the technical scheme of the invention integrates the remaining 4 layers of the main network, and the scale level of the anchorages is increased under the feature layer with the high resolution of 20 multiplied by 20 to provide enough anchorages coverage, thereby not only meeting the anchorages coverage number detected with high precision, but also saving the calculation power.
According to another aspect of the present invention, a computer device is provided, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the information detection method according to any one of the above-mentioned technical solutions.
According to the computer device provided by the invention, when the processor executes the computer program, the steps of the information detection method according to any one of the above technical schemes are realized, so that the computer device has all the beneficial effects of the information detection method according to any one of the above technical schemes.
According to yet another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the information detection method according to any one of the above-mentioned technical solutions.
The computer-readable storage medium provided by the present invention, when being executed by a processor, implements the steps of the information detection method according to any one of the above-mentioned technical solutions, and therefore, the computer-readable storage medium includes all the advantageous effects of the information detection method according to any one of the above-mentioned technical solutions.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 shows a schematic flow chart of an information detection method according to a first embodiment of the present invention;
FIG. 2 is a flow chart illustrating an information detection method according to a second embodiment of the present invention;
fig. 3 is a flowchart illustrating an information detection method according to a third embodiment of the present invention;
FIG. 4 is a schematic structural diagram of an information detection model according to a fourth embodiment of the present invention;
fig. 5 shows a schematic block diagram of an information detection apparatus of a first embodiment of the present invention;
fig. 6 shows a schematic block diagram of an information detecting apparatus of a second embodiment of the present invention;
fig. 7 shows a schematic block diagram of an information detection apparatus of a third embodiment of the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments of the present invention and features of the embodiments may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and therefore the scope of the present invention is not limited to the specific embodiments disclosed below.
Embodiments of the first aspect of the present invention provide an information detection method, which is described in detail below with reference to a plurality of embodiments.
First embodiment, fig. 1 shows a schematic flow chart of an information detection method according to a first embodiment of the present invention. Wherein, the method comprises the following steps:
and step 104, detecting the vehicle in the shot image by using the vehicle detection branch of the information detection model to obtain the vehicle information, and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to obtain the license plate information.
The information detection method provided by the invention abandons the mode of simultaneously detecting all types of objects aiming at the contradiction between the speed and the precision of a classical general detection model, uses independent detection branches aiming at the vehicle (large object) and the license plate (small object), and respectively uses the characteristic layer of the information detection model to independently connect one exclusive detection branch with the other exclusive detection branch, namely, the vehicle detection branch of the information detection model is used for detecting the vehicle to obtain the vehicle information, and the license plate detection branch of the information detection model is used for detecting the license plate to obtain the license plate information. According to the technical scheme, the vehicle and the license plate are respectively detected, so that the detection speed is increased, and the detection precision is improved.
In this embodiment, the vehicle information includes one or a combination of: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background. The method integrates the vehicle door state recognition in the independent detection branch of the vehicle, outputs two states of vehicle door opening and closing, integrates the vehicle direction recognition in the independent detection branch of the vehicle, and outputs the direction information of the vehicle photos of 0 degree, 90 degrees, 180 degrees and 270 degrees.
Second embodiment, fig. 2 is a schematic flow chart illustrating an information detection method according to a second embodiment of the present invention. Wherein, the method comprises the following steps:
and step 208, sending out prompt information when the shot image is invalid.
In the embodiment, when vehicle verification is required, the vehicle and the license plate are detected in real time in the process of shooting the vehicle by the driver, if the detected vehicle information and the license plate information are wrong or incomplete, the shot image is determined to be invalid, prompt information is sent out in time to remind the driver, and the condition that the short video or the picture of the vehicle submitted by the driver is valid video information containing the vehicle and the license plate is ensured.
Wherein the vehicle information comprises one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
Third embodiment, fig. 3 is a flowchart illustrating an information detection method according to a third embodiment of the present invention. Wherein, the method comprises the following steps:
and step 310, sending out prompt information when the shot image is invalid.
In the embodiment, the sampling information in the sampling image is marked, the sampling information is input into the information detection model, and the parameters of the information detection model are trained through the sampling information and the marking to obtain the optimal model, so that the vehicle information and the license plate information can be accurately and quickly detected through the model.
The information detection model comprises five layers of backbone networks; in step 302, a fifth layer backbone network and a vehicle detection branch of the vehicle sampling information training information detection model according to the sampling information, and a first layer backbone network, a second layer backbone network, a third layer backbone network, a fourth layer backbone network and a license plate detection branch of the license plate sampling information training information detection model according to the sampling information, wherein the vehicle sampling information comprises a sampling vehicle body center point coordinate, a sampling vehicle body outer frame coordinate, a sampling vehicle body width, a sampling vehicle body height, a sampling vehicle head or vehicle tail attribute, sampling vehicle door opening or closing information, and sampling vehicle direction information, and the license plate sampling information comprises a license plate sampling center point coordinate, a sampling license plate width, a sampling license plate height, a sampling license plate foreground, and a sampling license plate background.
The embodiment of the invention provides a method for detecting a license plate of a vehicle at a mobile terminal, which comprises the following steps:
1. the model is designed, the model structure is as shown in fig. 4, the information detection model uses mobilent _ v3 (deep learning model) as a main network, independent detection branches (vehicle detection branch and license plate detection branch) with separated features are used for respectively detecting a vehicle and a license plate, the detection branches customize different aspect ratios of anchors according to different shapes of the vehicle and the license plate, and are more attached to a detection object to remarkably improve detection precision, wherein the anchors are fixed reference frames which are mapped on an original image by default at each point in a detection feature map in a target detection task. The vehicle is easy to detect in a larger proportion in the figure, so the vehicle detection branch is connected behind the last 5 multiplied by 5 low-resolution feature layer of the main network, and simultaneously, a vehicle head or vehicle tail, a vehicle door state and a vehicle direction identification model are integrated on the layer, so that a large object (vehicle) can be detected with high precision, and extra attribute identification can be integrated under the condition of adding a small amount of calculation. Specifically, 2 layers of door state recognition layers can be integrated in the vehicle detection branch, two states of opening and closing of the door can be output, 4 layers of vehicle photo direction recognition layers can be integrated in the vehicle detection branch, and direction information of 0 degree, 90 degrees counterclockwise, 180 degrees counterclockwise and 270 degrees counterclockwise of the vehicle photo can be output. The license plate occupies a small proportion in the picture, and is not easy to detect the number of anchorages required to ensure the detection precision, but the use of more main network feature layers to access extra detection branches to increase the anchorages coverage brings a large amount of calculation, so the technical scheme of the invention integrates the remaining 4 layers of the main network, and the scale level of the anchorages is increased under the feature layer with the high resolution of 20 multiplied by 20 to provide enough anchorages coverage, thereby not only meeting the anchorages coverage number detected with high precision, but also saving the calculation power.
2. And marking the frame coordinate data of the vehicle and the license plate.
3. And marking the attribute of the head or the tail of the vehicle.
4. The vehicle picture door status (closed or open) is noted.
5. And adjusting the directions of all the vehicle pictures to be positive directions, randomly rotating by 0 degree, 90 degrees, 180 degrees and 270 degrees anticlockwise during training, and outputting labels.
6. During training, an RGB (Red Green blue) image is input, and a 5-layer feature layer is extracted after the RGB (Red Green blue) image is output through a backbone network mobilenet _ v 3. Wherein the last layer is used for training the branch of vehicle detection, the coordinates of the outer frame of the vehicle, the direction of the vehicle picture, the state of the vehicle door and the labels of the head/tail of the vehicle; and the rest 4 layers are fused into a layer serving as a branch of license plate detection through pyramid up-sampling, and are trained together with license plate outer frame coordinates and license plate foreground/background labels.
7. After training, the information detection model is used for detecting the positions of the vehicle and the license plate and outputting the state of the vehicle door, the picture sharing of the vehicle and the attribute of the vehicle head or the vehicle tail.
The second aspect of the present invention provides an information detecting apparatus, and the following describes in detail a plurality of embodiments.
First embodiment, fig. 5 shows a schematic block diagram of an information detection apparatus 50 according to a first embodiment of the present invention. Wherein, this information detection device 50 includes:
An image acquisition unit 502 for acquiring a captured image;
an information detection unit 504, configured to detect a vehicle in the captured image using the vehicle detection branch of the information detection model, obtain vehicle information, and detect a license plate in the captured image using the license plate detection branch of the information detection model, obtain license plate information.
The information detection device 50 provided by the invention abandons the mode of simultaneously detecting all types of objects aiming at the contradiction between the speed and the precision of a classical general detection model, uses independent detection branches aiming at a vehicle (a large object) and a license plate (a small object), and respectively uses the characteristic layer of the information detection model to independently connect one exclusive detection branch with another, namely, the vehicle information is obtained by detecting the vehicle by using the vehicle detection branch of the information detection model, and the license plate information is obtained by detecting the license plate by using the license plate detection branch of the information detection model. According to the technical scheme, the vehicle and the license plate are respectively detected, so that the detection speed is increased, and the detection precision is improved.
In the above embodiment, the vehicle information includes one or a combination of: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
In the embodiment, the vehicle door state recognition is integrated in the independent detection branch of the vehicle, two states of opening and closing of the vehicle door are output, the vehicle direction recognition is also integrated in the independent detection branch of the vehicle, and the direction information of 0 degree, 90 degrees, 180 degrees and 270 degrees of a vehicle photo is output.
Second embodiment, fig. 6 is a schematic block diagram showing an information detecting apparatus 60 according to a second embodiment of the present invention. Wherein, this information detection device 60 includes:
an image acquisition unit 602 for acquiring a captured image;
an information detection unit 604, configured to detect a vehicle in the captured image by using the vehicle detection branch of the information detection model, obtain vehicle information, and detect a license plate in the captured image by using the license plate detection branch of the information detection model, so as to obtain license plate information;
the processing unit 606 is configured to determine whether the captured image is valid according to the vehicle information and the license plate information;
and a prompting unit 608 for sending out a prompting message when the captured image is invalid.
In this embodiment, when vehicle verification is required, the processing unit 606 judges the vehicle and the license plate in real time during the process of shooting the vehicle by the driver, if the detected vehicle information and the license plate information are wrong or incomplete, it is determined that the shot image is invalid, and the prompting unit 608 sends prompt information to prompt the driver in time, so as to ensure that the short video or the picture of the vehicle submitted by the driver is valid video information including the vehicle and the license plate.
Wherein the vehicle information comprises one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
Third embodiment, fig. 7 shows a schematic block diagram of an information detecting apparatus 70 according to a third embodiment of the present invention. Wherein, this information detection device 70 includes:
a model establishing unit 702, configured to obtain a sampling image and obtain sampling information in the sampling image; establishing an information detection model according to the sampling information;
an image acquisition unit 704 for acquiring a captured image;
an information detection unit 706, configured to detect a vehicle in the captured image by using the vehicle detection branch of the information detection model, obtain vehicle information, and detect a license plate in the captured image by using the license plate detection branch of the information detection model, so as to obtain license plate information;
the processing unit 708 is used for judging whether the shot image is effective according to the vehicle information and the license plate information;
and the prompting unit 710 is used for sending out prompting information when the shot image is invalid.
In the embodiment, the sampling information in the sampling image is marked, the sampling information is input into the information detection model, and the parameters of the information detection model are trained through the sampling information and the marking to obtain the optimal model, so that the vehicle information and the license plate information can be accurately and quickly detected through the model.
In the above embodiment, the information detection model includes five layers of backbone networks; the model establishing unit establishes an information detection model according to the sampling information, and specifically comprises the following steps: the vehicle sampling information training information detection model comprises a fifth layer trunk network and a vehicle detection branch of the vehicle sampling information training information detection model according to sampling information, and a first layer trunk network, a second layer trunk network, a third layer trunk network, a fourth layer trunk network and a license plate detection branch of the license plate sampling information training information detection model according to the sampling information, wherein the vehicle sampling information comprises sampling vehicle body central point coordinates, sampling vehicle body outer frame coordinates, sampling vehicle body width, sampling vehicle body height, sampling vehicle head or vehicle tail attributes, sampling vehicle door opening or closing information, sampling vehicle direction information, and license plate sampling information comprises sampling license plate central point coordinates, sampling license plate width, sampling license plate height, sampling license plate foreground and sampling license plate background.
In this embodiment, the information detection model uses mobilent _ v3 (deep learning model) as a backbone network, and uses independent detection branches (vehicle detection branch and license plate detection branch) with separated features to respectively detect a vehicle and a license plate, and the detection branches customize different aspect ratios of anchors according to different shapes of the vehicle and the license plate, and are more attached to a detection object to significantly improve detection accuracy, wherein the anchors are fixed reference frames in which each point in a detection feature map is mapped onto an original image by default in a target detection task. The vehicle is easy to detect in a larger proportion in the figure, so the vehicle detection branch is connected behind the last 5 multiplied by 5 low-resolution feature layer of the main network, and simultaneously, a vehicle head or vehicle tail, a vehicle door state and a vehicle direction identification model are integrated on the layer, so that a large object (vehicle) can be detected with high precision, and extra attribute identification can be integrated under the condition of adding a small amount of calculation. The license plate occupies a small proportion in the picture, and is not easy to detect the number of anchorages required to ensure the detection precision, but the use of more main network feature layers to access extra detection branches to increase the anchorages coverage brings a large amount of calculation, so the technical scheme of the invention integrates the remaining 4 layers of the main network, and the scale level of the anchorages is increased under the feature layer with the high resolution of 20 multiplied by 20 to provide enough anchorages coverage, thereby not only meeting the anchorages coverage number detected with high precision, but also saving the calculation power.
In this embodiment, when vehicle verification is required, the processing unit 708 determines the vehicle and the license plate in real time during the process of shooting the vehicle by the driver, and if the detected vehicle information and the license plate information are wrong or incomplete, it is determined that the shot image is invalid, and the prompting unit 710 sends prompt information to prompt the driver in time, so as to ensure that the short video or the picture of the vehicle submitted by the driver is valid video information including the vehicle and the license plate.
Wherein the vehicle information comprises one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information; the license plate information comprises one or the combination of the following components: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
In an embodiment of the third aspect of the present invention, a computer device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the steps of the information detection method according to any of the above embodiments are implemented.
The computer device provided by the present invention implements the steps of the information detection method according to any of the above embodiments when the processor executes the computer program, and therefore, the computer device includes all the advantages of the information detection method according to any of the above embodiments.
Embodiments of the fourth aspect of the present invention provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the information detection method according to any one of the above embodiments.
The present invention provides a computer-readable storage medium, wherein a computer program is executed by a processor to implement the steps of the information detection method according to any of the above embodiments, and therefore the computer-readable storage medium includes all the advantages of the information detection method according to any of the above embodiments.
In the description herein, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance unless explicitly stated or limited otherwise; the terms "connected," "mounted," "secured," and the like are to be construed broadly and include, for example, fixed connections, removable connections, or integral connections; may be directly connected or indirectly connected through an intermediate. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the description herein, the description of the terms "one embodiment," "some embodiments," "specific embodiments," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (12)
1. An information detection method, comprising:
acquiring a shot image;
and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to obtain the license plate information.
2. The information detection method according to claim 1, further comprising:
judging whether the shot image is effective or not according to the vehicle information and the license plate information;
and sending out prompt information when the shot image is invalid.
3. The information detection method according to claim 1,
the vehicle information includes one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information;
The license plate information comprises one or the combination of the following: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
4. The information detection method according to any one of claims 1 to 3, characterized by further comprising:
acquiring a sampling image and acquiring sampling information in the sampling image;
and establishing the information detection model according to the sampling information.
5. The information detection method according to claim 4, wherein the information detection model comprises a five-layer backbone network;
establishing the information detection model according to the sampling information, which specifically comprises the following steps:
the vehicle sampling information training of the information detection model is trained according to the fifth layer trunk network of the information detection model and the vehicle detection branch, and the license plate sampling information training of the sampling information is trained according to the first layer trunk network, the second layer trunk network, the third layer trunk network, the fourth layer trunk network of the information detection model and the license plate detection branch, wherein the vehicle sampling information comprises sampling vehicle body central point coordinates, sampling vehicle body outer frame coordinates, sampling vehicle body width, sampling vehicle body height, sampling vehicle head or vehicle tail attributes, sampling vehicle door opening or closing information and sampling vehicle direction information, and the license plate sampling information comprises sampling vehicle body central point coordinates, sampling vehicle plate width, sampling vehicle plate height, sampling vehicle plate foreground and sampling vehicle plate background.
6. An information detecting apparatus, characterized by comprising:
an image acquisition unit configured to acquire a captured image;
and the information detection unit is used for detecting the vehicle in the shot image by using the vehicle detection branch of the information detection model to acquire vehicle information, and detecting the license plate in the shot image by using the license plate detection branch of the information detection model to acquire the license plate information.
7. The information detection apparatus according to claim 6, characterized by further comprising:
the processing unit is used for judging whether the shot image is effective or not according to the vehicle information and the license plate information;
and the prompting unit is used for sending out prompting information when the shot image is invalid.
8. The information detecting apparatus according to claim 6,
the vehicle information includes one or a combination of the following: the system comprises a vehicle body center point coordinate, a vehicle body outer frame coordinate, a vehicle body width, a vehicle body height, a vehicle head or vehicle tail attribute, vehicle door opening or closing information and vehicle direction information;
the license plate information comprises one or the combination of the following: the license plate center point coordinates, the license plate width, the license plate height, the license plate foreground and the license plate background.
9. The information detection apparatus according to any one of claims 6 to 8, characterized by further comprising:
the model establishing unit is used for acquiring a sampling image and acquiring sampling information in the sampling image; and establishing the information detection model according to the sampling information.
10. The information detection apparatus according to claim 9, wherein the information detection model includes a five-layer backbone network;
the model establishing unit establishes the information detection model according to the sampling information, and specifically includes:
the vehicle sampling information training of the information detection model is trained according to the fifth layer trunk network of the information detection model and the vehicle detection branch, and the license plate sampling information training of the sampling information is trained according to the first layer trunk network, the second layer trunk network, the third layer trunk network, the fourth layer trunk network of the information detection model and the license plate detection branch, wherein the vehicle sampling information comprises sampling vehicle body central point coordinates, sampling vehicle body outer frame coordinates, sampling vehicle body width, sampling vehicle body height, sampling vehicle head or vehicle tail attributes, sampling vehicle door opening or closing information and sampling vehicle direction information, and the license plate sampling information comprises sampling vehicle body central point coordinates, sampling vehicle plate width, sampling vehicle plate height, sampling vehicle plate foreground and sampling vehicle plate background.
11. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the information detection method according to any one of claims 1 to 5 when executing the computer program.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the information detection method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910981932.1A CN111860087A (en) | 2019-10-16 | 2019-10-16 | Information detection method, information detection device, computer device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910981932.1A CN111860087A (en) | 2019-10-16 | 2019-10-16 | Information detection method, information detection device, computer device, and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111860087A true CN111860087A (en) | 2020-10-30 |
Family
ID=72970669
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910981932.1A Pending CN111860087A (en) | 2019-10-16 | 2019-10-16 | Information detection method, information detection device, computer device, and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111860087A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830192A (en) * | 2018-05-31 | 2018-11-16 | 珠海亿智电子科技有限公司 | Vehicle and detection method of license plate under vehicle environment based on deep learning |
| CN109635656A (en) * | 2018-11-12 | 2019-04-16 | 平安科技（深圳）有限公司 | Vehicle attribute recognition methods, device, equipment and medium neural network based |
| CN109635825A (en) * | 2018-12-19 | 2019-04-16 | 苏州市科远软件技术开发有限公司 | Vehicle attribute detection method, device and storage medium |
| WO2019169816A1 (en) * | 2018-03-09 | 2019-09-12 | 中山大学 | Deep neural network for fine recognition of vehicle attributes, and training method thereof |
| KR102030628B1 (en) * | 2019-04-04 | 2019-10-10 | (주)아이엠시티 | Recognizing method and system of vehicle license plate based convolutional neural network |
-
2019
- 2019-10-16 CN CN201910981932.1A patent/CN111860087A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019169816A1 (en) * | 2018-03-09 | 2019-09-12 | 中山大学 | Deep neural network for fine recognition of vehicle attributes, and training method thereof |
| CN108830192A (en) * | 2018-05-31 | 2018-11-16 | 珠海亿智电子科技有限公司 | Vehicle and detection method of license plate under vehicle environment based on deep learning |
| CN109635656A (en) * | 2018-11-12 | 2019-04-16 | 平安科技（深圳）有限公司 | Vehicle attribute recognition methods, device, equipment and medium neural network based |
| CN109635825A (en) * | 2018-12-19 | 2019-04-16 | 苏州市科远软件技术开发有限公司 | Vehicle attribute detection method, device and storage medium |
| KR102030628B1 (en) * | 2019-04-04 | 2019-10-10 | (주)아이엠시티 | Recognizing method and system of vehicle license plate based convolutional neural network |
Non-Patent Citations (2)
| Title |
|---|
| 柳思健;: "基于卷积网络的车辆定位与细粒度分类算法", 自动化与仪表, no. 07 * |
| 赵田;安婧;韩彩霞;贾学科;: "面向自然环境的车牌检测算法设计", 兰州文理学院学报(自然科学版), no. 04 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11830230B2 (en) | Living body detection method based on facial recognition, and electronic device and storage medium | |
| CN109671023B (en) | Face image super-resolution secondary reconstruction method | |
| US20210343041A1 (en) | Method and apparatus for obtaining position of target, computer device, and storage medium | |
| CN110866871A (en) | Text image correction method and device, computer equipment and storage medium | |
| CN109711407B (en) | License plate recognition method and related device | |
| CN111985281B (en) | Image generation model generation method and device and image generation method and device | |
| CN111259968A (en) | Illegal image recognition method, device, equipment and computer readable storage medium | |
| CN112307853A (en) | Detection method of aerial image, storage medium and electronic device | |
| CN110796580B (en) | Intelligent traffic system management method and related products | |
| CN110674759A (en) | Monocular face in-vivo detection method, device and equipment based on depth map | |
| CN110298281A (en) | Video structural method, apparatus, electronic equipment and storage medium | |
| WO2023138537A1 (en) | Image processing method and apparatus, terminal device and storage medium | |
| CN112101205A (en) | Training method and device based on multi-task network | |
| CN110554356A (en) | Equipment positioning method and system in visible light communication | |
| US20220357159A1 (en) | Navigation Method, Navigation Apparatus, Electronic Device, and Storage Medium | |
| US20240153041A1 (en) | Image processing method and apparatus, computer, readable storage medium, and program product | |
| CN112070703B (en) | Method and system for enhancing underwater visual image of bionic robot fish | |
| CN109698908A (en) | Intelligence calls method, terminal and the storage medium of front camera and rear camera imaging | |
| CN111724437B (en) | Visual positioning method and related device, equipment and storage medium | |
| WO2024001617A1 (en) | Method and apparatus for identifying behavior of playing with mobile phone | |
| CN112446845A (en) | Map construction method, map construction device, SLAM system, and storage medium | |
| CN111860087A (en) | Information detection method, information detection device, computer device, and storage medium | |
| CN113112404A (en) | Image splicing method and device based on sliding window | |
| WO2021022989A1 (en) | Calibration parameter obtaining method and apparatus, processor, and electronic device | |
| CN116777914B (en) | Data processing method, device, equipment and computer readable storage medium |
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 |