CN112784642B

CN112784642B - Vehicle detection method and device

Info

Publication number: CN112784642B
Application number: CN201911086705.9A
Authority: CN
Inventors: 梁云
Original assignee: Hangzhou Hikvision Digital Technology Co Ltd
Current assignee: Hangzhou Hikvision Digital Technology Co Ltd
Filing date: 2019-11-08
Publication date: 2024-06-21
Anticipated expiration: 2039-11-08

Abstract

The embodiment of the invention provides a vehicle detection method and device, wherein the method comprises the following steps: and acquiring a vehicle picture of the first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running. Inputting the vehicle picture into a preset model, and enabling the preset model to output a detection result corresponding to the vehicle picture, wherein the detection result comprises the region types of all regions in the vehicle picture and the confidence that all regions are of the corresponding region types. If the smoke area with the preset smoke type exists in the vehicle picture according to the detection result, determining whether the first vehicle generates preset smoke in the running process according to the first confidence coefficient corresponding to the smoke area. The region types of the regions in the vehicle picture are detected through the preset model, so that the regions with the region types similar to the preset smoke can be specifically removed, and meanwhile, whether the first vehicle generates the preset smoke or not is determined according to the confidence level, so that the accuracy of detecting the smoke vehicle is effectively improved.

Description

Vehicle detection method and device

Technical Field

The embodiment of the invention relates to a computer technology, in particular to a vehicle detection method and device.

Background

Vehicles generating smoke are typically representative of highly polluted vehicles, and the preset smoke generated by the vehicles during running causes pollution to the environment, so that detection of the smoke vehicles is very necessary.

At present, detection of a smoke vehicle mainly analyzes preset smoke based on an optical flow vector, specifically, the detection of the preset smoke is realized according to the characteristic that the speed distribution of the preset smoke has turbulent motion characteristics, however, when the road surface has rainwater, the vehicle passes through generated wheel vapor and raised dust generated when the wheels of the vehicle roll the road surface also has turbulent motion characteristics, and other areas similar to the preset smoke cannot be specifically removed by the current technology.

If other areas similar to the preset smoke cannot be eliminated, the accuracy of detecting the smoke vehicle is lower.

Disclosure of Invention

The embodiment of the invention provides a vehicle detection method and device, which are used for solving the problem of low accuracy of detecting a smoke vehicle.

In a first aspect, an embodiment of the present invention provides a vehicle detection method, including:

acquiring a vehicle picture of a first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running;

Inputting the vehicle picture into a preset model, so that the preset model outputs a detection result corresponding to the vehicle picture, wherein the detection result comprises the region type of each region in the vehicle picture and the confidence that each region is the corresponding region type;

If the fact that the smoke area with the preset smoke type exists in the vehicle picture is determined according to the detection result, whether the first vehicle generates preset smoke in the driving process is determined according to the first confidence coefficient corresponding to the smoke area.

In one possible design, the determining whether the first vehicle generates the preset smoke during the driving according to the first confidence coefficient corresponding to the smoke area includes:

judging whether the first confidence coefficient is larger than a preset confidence coefficient according to the first confidence coefficient corresponding to the preset smoke region;

if yes, determining that the first vehicle generates preset smoke in the running process.

In one possible design, before the inputting the vehicle picture into the preset model, the method further includes:

acquiring a training set, and classifying and calibrating each training vehicle picture in the training set to obtain a plurality of sub-training sets, wherein each sub-training set comprises a plurality of training vehicle pictures which are marked with areas corresponding to the same area type;

And training the preset model according to any one of the sub-training sets until the accuracy of the preset model for the current sub-training set is greater than or equal to the preset accuracy to obtain the trained preset model.

In one possible design, the region types include at least one of the following: preset smoke type, non-preset smoke type, water vapor type, shadow type and dust type;

Inputting the vehicle picture into a preset model, so that the preset model outputs a detection result corresponding to the vehicle picture, including:

Inputting the vehicle picture into a preset model, so that the preset model outputs the vehicle picture including at least one of the following areas: a first region of a preset smoke type, a second region of a non-preset smoke type, a third region of a water vapor type, a fourth region of a shadow type, and a fifth region of a dust type.

In one possible design, if the vehicle picture of the first vehicle includes a plurality of pictures, the method further includes:

Respectively inputting the plurality of vehicle pictures into a preset model, so that the preset model respectively outputs detection results corresponding to the vehicle pictures;

Respectively determining whether the first vehicle generates preset smoke in the running process according to the detection results corresponding to the vehicle pictures;

And if the number of the vehicle pictures for generating the preset smoke by the first vehicle is determined to be larger than the preset number, determining that the first vehicle generates the preset smoke in the driving process.

Performing image processing on the vehicle picture to obtain a processed vehicle picture, wherein the image processing comprises at least one of the following steps: image cropping, image scaling, image filtering.

In one possible design, the detection result further includes location information for indicating each region in the vehicle picture and/or identification information for characterizing the preset smoke level.

In a second aspect, an embodiment of the present invention provides a vehicle detection apparatus including:

the vehicle image acquisition module is used for acquiring a vehicle image of a first vehicle, wherein the vehicle image is obtained by shooting the first vehicle in running;

The processing module is used for inputting the vehicle picture into a preset model so that the preset model outputs a detection result corresponding to the vehicle picture, wherein the detection result comprises the region type of each region in the vehicle picture and the confidence that each region is the corresponding region type;

the determining module is used for determining whether the first vehicle generates preset smoke in the driving process according to the first confidence coefficient corresponding to the smoke region if the smoke region with the preset smoke type exists in the vehicle picture according to the detection result.

In one possible design, the determining module is specifically configured to:

In one possible design, the acquisition module is further configured to:

The processing module is specifically configured to input the vehicle picture into a preset model, so that the preset model outputs the vehicle picture including at least one of the following areas: a first region of a preset smoke type, a second region of a non-preset smoke type, a third region of a water vapor type, a fourth region of a shadow type, and a fifth region of a dust type.

In one possible design, if the vehicle picture of the first vehicle includes a plurality of pictures, the processing module is further configured to:

the determining module is further configured to:

In one possible design, the processing module is further configured to:

Before the vehicle picture is input into a preset model, performing image processing on the vehicle picture to obtain a processed vehicle picture, wherein the image processing comprises at least one of the following steps: image cropping, image scaling, image filtering.

In a third aspect, an embodiment of the present invention provides a vehicle detection apparatus including:

a memory for storing a program;

A processor for executing the program stored by the memory, the processor being adapted to perform the method of the first aspect and any of the various possible designs of the first aspect as described above when the program is executed.

In a fourth aspect, embodiments of the present invention provide a computer readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of the first aspect above and any of the various possible designs of the first aspect.

The embodiment of the invention provides a vehicle detection method and device, wherein the method comprises the following steps: and acquiring a vehicle picture of the first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running. Inputting the vehicle picture into a preset model, and enabling the preset model to output a detection result corresponding to the vehicle picture, wherein the detection result comprises the region types of all regions in the vehicle picture and the confidence that all regions are of the corresponding region types. If the smoke area with the preset smoke type exists in the vehicle picture according to the detection result, determining whether the first vehicle generates preset smoke in the running process according to the first confidence coefficient corresponding to the smoke area. The region types of all regions in the vehicle picture are detected through the preset model, so that regions with region types similar to preset smoke can be specifically removed, and meanwhile, whether the first vehicle generates the preset smoke or not is determined through the confidence coefficient corresponding to the smoke region, so that the accuracy of detecting the smoke vehicle is effectively improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it will be obvious that the drawings in the following description are some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art.

FIG. 1 is a schematic diagram of a vehicle detection method according to an embodiment of the present invention;

Fig. 2 is a schematic diagram of an execution main body of a vehicle detection method according to an embodiment of the present invention;

FIG. 3 is a flowchart illustrating a vehicle detection method according to an embodiment of the present invention;

FIG. 4 is a second flowchart of a vehicle detection method according to the present invention;

FIG. 5 is a schematic diagram of a detection result of the vehicle detection method according to the present invention;

Fig. 6 is a schematic diagram of determining a preset smoke according to a plurality of vehicle pictures provided by the present invention;

Fig. 7 is a flowchart III of a vehicle detection method according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of training a preset model according to an embodiment of the present invention;

Fig. 9 is a schematic structural diagram of a vehicle detection device according to an embodiment of the present invention;

fig. 10 is a schematic hardware structure of a vehicle detection apparatus according to an embodiment of the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Fig. 1 is a schematic system diagram of a vehicle detection method according to an embodiment of the present invention, as shown in fig. 1: the photographing device 101 is disposed in the road, where the photographing device 101 is used for photographing a vehicle to obtain a vehicle picture, and specifically, the photographing device may be, for example, a monitoring camera of a cross bar disposed above the road, a monitoring camera of a pole set beside the road, or may also be a snapping camera specifically disposed in the road, which is not limited in this embodiment, so long as the photographing device 101 may achieve photographing of the vehicle picture.

After the photographing device 101 photographs a vehicle picture of the vehicle, the photographing device 101 sends the vehicle picture of the vehicle to the processing device 102, where the processing device 102 may analyze the vehicle picture to determine whether the vehicle corresponding to the current vehicle picture is a vehicle generating smoke.

The interaction of data exists between the photographing device 101 and the processing device 102, and the interaction manner may be, for example, a wired network, where the wired network may include, for example, a coaxial cable, a twisted pair wire, an optical fiber, etc., and the interaction manner may also be, for example, a wireless network, where the wireless network may be a 2G network, a 3G network, a 4G network, or a 5G network, a wireless fidelity (WIRELESS FIDELITY, abbreviated as WIFI) network, etc. The embodiment of the application does not limit the specific type or specific form of interaction, as long as the interaction function between the server and the terminal can be realized.

In this embodiment, the processing device 102 may be a server or a camera (i.e. the photographing device 102), and the following description will be made with reference to fig. 2 for the case where the processing device is a server and a camera, and fig. 2 is a schematic diagram of an execution main body of the vehicle detection method according to the embodiment of the present invention, where the execution main body is shown in fig. 2:

After each shooting device shoots and obtains the vehicle picture, the vehicle picture can be sent to a server for vehicle detection processing; or the shooting device can be a processing device, specifically, a vehicle detection unit can be mounted in each shooting device, the shooting device shoots the current vehicle to obtain a vehicle picture, and the vehicle picture is processed according to the vehicle detection unit so as to judge whether the vehicle generates preset smoke or not.

The server and the photographing device described in fig. 2 may be used to implement the technical solution of the present invention, so the execution subject of the content described in the following embodiments may be the server or the photographing device, and will not be described in detail.

Based on the above-described system, there are two general implementations of the prior art for detecting whether a vehicle generates smoke during driving:

A target area of a suspected smoke area is detected through background modeling, and whether the detected target area is a preset smoke area or not is judged according to the gray level of an image, however, the method has larger misjudgment on gray pavement and shadow, so that the accuracy rate of smoke detection is lower.

The other is to analyze preset smog based on an optical flow vector, specifically, the detection of the preset smog is realized according to the characteristic that the speed distribution of the preset smog has turbulent motion characteristics, however, when the road surface has rainwater, the vehicle passes through the generated wheel water vapor, and the raised dust generated when the wheels of the vehicle roll the road surface also has turbulent motion characteristics, other areas similar to the preset smog cannot be specifically removed by the current technology, and the method also can cause lower accuracy of detecting the smog vehicle.

Based on the above-described problems, the present invention provides a vehicle detection method, so that other areas similar to a preset smoke area can be detected in a targeted manner for detecting smoke of a vehicle during driving, so as to exclude the areas, thereby improving accuracy of detecting smoke of the vehicle, and in the following, first, description is made with reference to fig. 3, and fig. 3 is a flowchart one of the vehicle detection methods provided in the embodiments of the present invention.

As shown in fig. 3, the method includes:

S301, acquiring a vehicle picture of a first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running.

Specifically, a photographing device is installed in the road, and related video parameters, image parameters, photographing parameters and the like can be configured before photographing, for example, a storage format of a vehicle picture, a frequency of the photographing device, a photographing number of the vehicle picture and the like can be set, and as can be understood by those skilled in the art, specific related parameters can be expanded and selected according to actual requirements, and the embodiment is not limited to this.

The vehicle picture of the first vehicle may be, for example, a picture in JPG format, where the JPG format is a compressed picture obtained by encoding a camera picture, or may be any possible picture format, where a specific picture format may be set by related parameters.

In one possible implementation manner, it may be determined whether the first vehicle is located within a shooting range of the shooting device, if so, the shooting device performs a snapshot operation, specifically, the shooting device may collect video frames in real time, and if the shooting device detects that the vehicle is located within the shooting range of the shooting device according to the moving target detection algorithm, the shooting device performs the snapshot operation to obtain a vehicle picture of the first vehicle.

In another possible implementation, for example, a triggering device may be provided in the road, and the triggering device may be, for example, a device with a certain height for capturing a frame composed of a trigger line, a lane line, and infrared rays, or the triggering device may also be a device provided on the road surface or under the road, or the like, and when a vehicle passes through the triggering device, the triggering device sends indication information to the photographing device, so that the photographing device photographs a first vehicle currently passing through the triggering device, for example, may photograph a vehicle picture, or may also photograph a preset number of vehicle pictures, or the like.

After the photographing device photographs the first vehicle in running to obtain the vehicle picture of the first vehicle, the photographing device can send the vehicle picture of the first vehicle to a server or the photographing device, specifically, the photographing device can perform Image signal processing (Image SignalProcessing, ISP) processing and Image compression coding processing on the vehicle picture of the first vehicle to obtain a code stream of the vehicle picture, so that the vehicle picture is sent to a client in a code stream mode, wherein the ISP processing can effectively realize functions of automatic exposure control, dead point removal, noise removal, automatic white balance, interpolation, color correction, nonlinear correction and the like, and details of a photographed scene can be well restored under different optical conditions through the ISP, so that the imaging quality of the vehicle picture can be effectively improved.

S302, inputting the vehicle picture into a preset model, and enabling the preset model to output a detection result corresponding to the vehicle picture, wherein the detection result comprises the region types of all regions in the vehicle picture and the confidence that all regions are corresponding region types.

In this embodiment, a preset model is provided to process a vehicle picture, where the preset model in this embodiment may determine a region type of each region in the vehicle picture, and a confidence that each region is a corresponding region type, where the region type includes, but is not limited to: the preset smoke type, the non-preset smoke type, the water vapor type, the shadow type and the dust type, wherein the preset smoke type can be black smoke, the non-preset smoke type can be white smoke, yellow smoke and the like, a person skilled in the art can understand that the specific area type can be selected according to actual requirements, and the embodiment is not limited to the specific area type.

It should be noted that, each region in the vehicle picture in the embodiment is a region that is automatically analyzed and divided by the preset model, for example, after the analysis of the preset model, it is determined that the first region exists in the vehicle picture as a black smoke region, then the preset model may output a detection result that the region type of the first region is a black smoke type, or after the analysis of the preset model, it is determined that the second region also exists in the same vehicle picture as a dust region, then the preset model may output a detection result that the region type of the second region is a dust type at the same time, then the vehicle picture in the present example includes two regions, which respectively correspond to the black smoke type region and the dust region.

Meanwhile, the detection result output by the preset model in this embodiment further includes a confidence coefficient of each region being a region type, where the confidence coefficient is used to indicate a confidence degree that the region is a region type indicated by the detection result, for example, 3 regions in a vehicle picture are detected by the current preset model, corresponding region identifiers are respectively a region 1, a region 2 and a region 3, where the region type corresponding to the region 1 is a black smoke type, the confidence coefficient is 0.9, the region type corresponding to the region 2 is a dust type, the confidence coefficient is 0.8, the region type corresponding to the region 3 is a shadow type, and the confidence coefficient is 0.3.

In this embodiment, the preset model may detect, in addition to the preset smoke area, a dust raising area, a shadow area, and the like that are relatively similar to the preset smoke area, so that other areas that are relatively similar to the preset smoke area may be specifically excluded.

And S303, if the smoke area with the preset smoke type exists in the vehicle picture according to the detection result, determining whether the first vehicle generates the preset smoke in the running process according to the first confidence coefficient corresponding to the smoke area.

In the above detection result, if it is determined that a smoke region of a preset smoke type exists in the vehicle picture, it may be initially determined that the first vehicle is suspected to generate the preset smoke, and then, according to a first confidence coefficient corresponding to the smoke region, it is determined whether the vehicle generates the preset smoke in the driving process.

In one possible implementation manner, whether the first confidence coefficient is greater than the preset confidence coefficient may be determined, if the first confidence coefficient is determined to be greater than the preset confidence coefficient, it may be determined that the first vehicle generates the preset smoke in the driving process, where the preset confidence coefficient may be selected according to actual requirements, and this embodiment is not limited to this, for example, in the above-described example, the confidence coefficient of the black smoke type in the area 1 is 0.9, and if the preset confidence coefficient is assumed to be 0.85, it may be determined that the first vehicle does generate the preset smoke in the driving process, and further determined by the confidence coefficient, so that accuracy of the preset smoke detection may be improved.

It should be noted that, in this embodiment, the setting of the preset confidence level may be adjusted according to an actual scene, for example, when judging whether the first vehicle generates the preset smoke in a scene where there is a dust situation, because the presence of the dust will have a certain influence on the detection Cao Cheng of the preset smoke, the first confidence level of the area of the preset smoke type will drop by 0.65, in this case, the first vehicle actually generates the preset smoke, if the preset confidence level is still set to 0.9, the condition of missed detection may occur, so the preset confidence level may be adjusted to 0.6 and then determined, for the deep learning algorithm, when the data in the training set is sufficient, the judgment will be better according to the preset confidence level directly, but for the special scene, and when the data in the training set of a certain type is insufficient, the setting of the preset confidence level should be adjusted according to the scene.

The vehicle detection method provided by the embodiment of the invention comprises the following steps: and acquiring a vehicle picture of the first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running. Inputting the vehicle picture into a preset model, and enabling the preset model to output a detection result corresponding to the vehicle picture, wherein the detection result comprises the region types of all regions in the vehicle picture and the confidence that all regions are of the corresponding region types. If the smoke area with the preset smoke type exists in the vehicle picture according to the detection result, determining whether the first vehicle generates preset smoke in the running process according to the first confidence coefficient corresponding to the smoke area. The region types of all regions in the vehicle picture are detected through the preset model, so that regions with region types similar to preset smoke can be specifically removed, and meanwhile, whether the first vehicle generates the preset smoke or not is determined through the confidence coefficient corresponding to the smoke region, so that the accuracy of detecting the smoke vehicle is effectively improved.

On the basis of the above embodiment, the above process may be understood as a detection performed according to a single vehicle picture, however, if the single vehicle picture is the vehicle picture with the tail approaching the lowest part of the image, the shadow formed by the imaging of the first vehicle in the imaging device may easily cause erroneous judgment on the detection result, and for this case, the vehicle picture provided by the present invention may be comprehensively judged according to multiple vehicle pictures to further improve the accuracy of the detection result, and the process of detecting multiple vehicle pictures is described with reference to fig. 4 to 6, fig. 4 is a flowchart two of the vehicle detection method provided by the present invention, fig. 5 is a schematic diagram of the detection result of the vehicle detection method provided by the present invention, and fig. 6 is a schematic diagram of determining the preset smoke according to multiple vehicle pictures provided by the present invention.

As shown in fig. 4, the method includes:

S401, acquiring a vehicle picture of a first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running.

In the embodiment, the first vehicle is a plurality of images, and the plurality of images in the embodiment are a plurality of images of the vehicle shot by the shooting device at one time, that is, when the first vehicle passes through the shooting area of the shooting device, the shooting area is shot for a plurality of times, so as to obtain a plurality of images.

S402, performing image processing on the vehicle picture to obtain a processed vehicle picture, wherein the image processing comprises at least one of the following steps: image cropping, image scaling, image filtering.

If a certain requirement exists on the input vehicle picture by the preset model, the embodiment finds the vehicle picture to be processed through image processing, and can obtain the processed vehicle picture meeting the requirement of the preset model.

For example, the resolution of the vehicle image is too large, and the vehicle image can be scaled to the target resolution through image scaling; or in order to avoid that the relevant detection area is also reduced in the image scaling process, so that the subsequent detection is not facilitated, the image can be cut firstly, and the size of the target area to be detected is kept; or the image can be more similar to a real smoke area through filtering, wherein the image filtering can be image signal Processing (IMAGE SIGNAL Processing, ISP) Processing, the ISP Processing can effectively realize the functions of automatic exposure control, dead point removal, noise removal, automatic white balance, interpolation, color correction, nonlinear correction and the like, and the ISP can better restore the details of a shooting scene under different optical conditions, so that the imaging quality of vehicle pictures can be effectively improved.

S402, respectively inputting a plurality of vehicle pictures into a preset model, so that the preset model respectively outputs detection results corresponding to the vehicle pictures.

The implementation of S402 is similar to that of S302, and will not be described here again.

The detection result output by the preset model for one vehicle picture is described in detail below with reference to fig. 5, and as shown in fig. 5, the detection result of the vehicle picture includes 3 regions, and the types corresponding to the 3 regions are respectively a black smoke type, a shadow type and a dust type, meanwhile, the confidence of the first region being the black smoke type is 0.9, the confidence of the second region being the dust type is 0,8, and the confidence of the third type being the shadow type is 0.3.

In this embodiment, the detection result is further used to indicate the region coordinates of each region in the vehicle picture, specifically, refer to fig. 5, where the first region, the second region, and the third region each correspond to a respective position in the vehicle picture, and the position information in this embodiment is used to indicate the position of the region in the vehicle picture, where in a possible implementation manner, when the shape of the region is rectangular, the position information may include, for example, the upper left corner coordinates of the region, and the length and the width of the region; or when the shape of the region is circular, the position information may include, for example, the coordinates of the center point of the region and the radius length of the region, and the display position information may also display the outline of each region in the vehicle picture, where the outline may be displayed in a solid line, a dotted line, or a translucent color filling different from the background color, and those skilled in the art can understand that the specific position information of the region, the shape of the region, and the display manner of the outline may all be selected according to the actual requirement, which is not limited in this embodiment.

And the detection result can also be used for representing the identification information of the preset smoke degree, in particular. The identification information of the preset smoke level may be, for example, a green man grade, which is a grade for representing the current black smoke, may be displayed to the user later to indicate what the specific black smoke grade is, for example, what the detected black smoke grade of the vehicle is to be displayed to the driver who has a black smoke vehicle to pay a fine.

By providing the position information of each region in the vehicle picture, the position of the region corresponding to each region type in the vehicle picture can be conveniently and rapidly determined in the subsequent processing process.

S403, respectively determining whether the first vehicle generates preset smoke in the running process according to the detection results corresponding to the vehicle pictures.

The implementation manner of S403 is similar to that of S303 described above, except that in step S303 in the above embodiment, it may be directly determined that the first vehicle generates the preset smoke in the driving process according to the first confidence coefficient corresponding to the smoke area of the single picture, however, in this embodiment, whether the first vehicle generates the preset smoke in the driving process is determined according to the detection results corresponding to the respective vehicle pictures.

And if the first vehicle picture, the second vehicle picture and the third vehicle picture exist currently, the three vehicle pictures are respectively processed so as to respectively determine detection results corresponding to the vehicle pictures, whether the first vehicle in the first vehicle picture generates preset smoke or not is determined according to the detection results of the first vehicle picture, and whether the first vehicle in the second vehicle picture and the third vehicle picture generates preset smoke or not is determined according to the detection results of the second vehicle picture and the third vehicle picture.

S404, judging and determining whether the number of vehicle pictures for generating preset smoke by the first vehicle is larger than the preset number, if so, executing S405, and if not, executing S406.

Next, the number of vehicle pictures that determine that the first vehicle generates the preset smoke is acquired, and whether the number is greater than the preset number is determined, and the process is described below with reference to fig. 6:

As shown in fig. 6, the first vehicle picture 601, the second vehicle picture 602, and the third vehicle picture 603 are a plurality of vehicle pictures taken for the first vehicle, where the second vehicle picture and the third vehicle picture correspondingly determine that the first vehicle generates the preset smoke, and the first vehicle picture correspondingly determines that the first vehicle does not generate the preset smoke, which indicates that the number of vehicle pictures currently determining that the first vehicle generates the preset smoke is 2, and if the preset number is 1, which indicates that the number of vehicle pictures determining that the first vehicle generates the preset smoke is greater than the preset number, where the preset number may be specifically set according to the number of vehicle pictures, the embodiment does not make any particular limitation on the preset number.

The determining, by the first vehicle picture, that the first vehicle picture does not generate the preset smoke may be that a smoke area of the preset smoke type does not exist in a detection result of the first vehicle picture, or may be that a smoke area of the preset smoke type exists in the detection result, where the first confidence coefficient corresponding to the smoke area is not greater than the preset confidence coefficient.

S405, determining that the first vehicle generates preset smoke in the driving process.

S406, determining that the first vehicle does not generate preset smoke in the driving process.

If the number of the vehicle pictures for generating the preset smoke by the first vehicle is determined to be larger than the preset number, the first vehicle can be determined to generate the preset smoke in the driving process; if the number is not greater than the preset number, for example, only one of the three vehicle pictures is determined to generate the preset smoke as a result of determining that the first vehicle generates the preset smoke, it may be determined that the first vehicle does not generate the preset smoke during the driving process.

In an alternative embodiment, after determining that the first vehicle generates the preset smoke, gray value calculation may be further performed according to a smoke area in the vehicle image, so as to obtain a ringeman coefficient commonly used in the industry, and determining a black smoke level of the smoke area according to the ringeman coefficient.

The vehicle detection method provided by the embodiment of the invention comprises the following steps: and acquiring a vehicle picture of the first vehicle, wherein the vehicle picture is obtained by shooting the first vehicle in running. And respectively inputting the plurality of vehicle pictures into a preset model, so that the preset model respectively outputs detection results corresponding to the vehicle pictures. And respectively determining whether the first vehicle generates preset smoke in the running process according to the detection results corresponding to the vehicle pictures. Judging whether the number of the vehicle pictures for determining that the first vehicle generates the preset smoke is larger than the preset number, and if so, determining that the first vehicle generates the preset smoke in the driving process. If not, determining that the first vehicle does not generate preset smoke in the driving process. Whether the first vehicle generates preset smoke or not is comprehensively determined through a plurality of vehicle pictures, so that false detection caused by imaging of a single vehicle picture due to erection of a shooting device is avoided, and accuracy of a detection result is effectively guaranteed.

On the basis of the above embodiment, in order to enable the preset model to detect the region types of each region, before inputting the vehicle image into the preset model, the preset model needs to be pre-trained according to the vehicle image identifying the regions corresponding to the different types of regions, the model training in the present application is described below with reference to fig. 7, fig. 7 is a flowchart three of the vehicle detection method provided by the embodiment of the present application, and fig. 8 is a schematic diagram of the preset model training provided by the embodiment of the present application.

As shown in fig. 7, the method includes:

S701, acquiring training sets, and classifying and calibrating each training vehicle picture in the training sets to obtain a plurality of sub-training sets, wherein each sub-training set comprises a plurality of training vehicle pictures marked with areas corresponding to the same area type.

Specifically, the training set includes a plurality of training vehicle pictures, and in this embodiment, in order to enable each region corresponding to a different region type in the vehicle pictures to be identified for the preset model, the plurality of training vehicle pictures included in the training set are vehicle pictures for each different region type.

Classifying and calibrating each training vehicle picture to obtain a plurality of sub-training sets, wherein the assumed region types comprise: the black smoke type, the water vapor type, the shadow type and the dust type can be identified as the vehicle image of the black smoke type, and specifically identify the region corresponding to the black smoke type, or the vehicle image of the water vapor type, the vehicle image of the shadow type and the vehicle image of the dust type, and specifically identify the type of one of the vehicle image, then a plurality of training vehicle images corresponding to the same region type can form a sub-training set, for example, a first sub-training set of the black smoke type can be obtained, the first sub-training set comprises the vehicle image identifying the region corresponding to the black smoke type, a second sub-training set of the water vapor type, a third sub-training set of the shadow type and a fourth sub-training set of the dust type, and the specific region type can be selected according to actual requirements.

It can be appreciated that since the preset model is required to perform effective and correct learning, the correctness of the identification needs to be ensured, and the identification of the region type of the region of the vehicle picture in the training set is manually operated.

S702, training a preset model according to any one of the sub-training sets until the accuracy of the preset model for the current sub-training set is greater than or equal to the preset accuracy to obtain a trained preset model.

Specifically, because the preset model in the embodiment needs to identify the vehicle pictures of different region types, for any one of the sub-training sets, the preset model needs to be trained according to the sub-training set.

The process of training the preset model may refer to fig. 8, and it is assumed that the current first sub-training set includes 1000 vehicle pictures of the area marked with the black smoke type, the second sub-training set includes 1000 vehicle pictures of the area marked with the water vapor type, the third sub-training set includes 1000 vehicle pictures of the area marked with the shadow type, the fourth sub-training set includes 1000 vehicle pictures of the area marked with the dust type, and the sub-training sets corresponding to these pictures together form a training set corresponding to each area type.

In one possible implementation manner, the deep learning framework CAFFE (Convolutional Architecture for Fast Feature Embedding) deep learning network YOLO may be used to perform training of the preset model, where CAFFE is a deep learning framework with expressive, speed and thinking modularization, CAFFE supports multiple types of deep learning frameworks, faces image classification and image segmentation, supports CNN network design, or may also use any feasible learning framework and learning network as long as training of the model can be achieved, which is not limited in this embodiment.

Training the preset model aiming at any one of the sub-training sets until the accuracy rate of the preset model for the region type identification in the current sub-training set is greater than or equal to the preset accuracy rate, and considering that the training of the sub-training set model aiming at the current region type is finished, and obtaining the trained preset model after the training of all the sub-training sets is finished.

The training vehicle pictures in the training set are classified and calibrated to perform model training, rather than model training for single black smoke detection, and not only calibration training for target pictures containing black smoke tail gas, so that models for classification judgment of various different conditions, such as black smoke emission under dust conditions or dust imaging characteristics, can be better trained, and recognition accuracy of preset models is effectively improved.

It should be noted that, in this embodiment, the preset model may identify the areas corresponding to the multiple different area types at the same time, and then the detection result output by the preset model in this embodiment may include a first area of the preset smoke type, and meanwhile, when the vehicle picture includes an area of the non-preset smoke type, a second area of the non-preset smoke type may be output, and a confidence level that the second area is of the non-preset smoke type may be output, and further, on the premise that the vehicle picture includes the current area type, a third area of the water vapor type, a fourth area of the shadow type, and a fifth area of the dust type may be output, so that the comprehensiveness of the detection result of the preset model may be improved.

The vehicle detection method provided by the embodiment of the invention comprises the following steps: and obtaining training sets, and classifying and calibrating each training vehicle picture in the training sets to obtain a plurality of sub-training sets, wherein each sub-training set comprises a plurality of training vehicle pictures with the areas corresponding to the same area types. And training the preset model according to any one of the sub-training sets until the accuracy of the preset model for the current sub-training set is greater than or equal to the preset accuracy to obtain the trained preset model. Through training the model of predetermineeing respectively according to the sub-training set of each regional type to make the model of predetermineeing can be pointed detect with predetermineeing other regions that smog region is comparatively similar, avoided predetermineeing the erroneous detection in smog region, can train better simultaneously to the model of classifying judgement of various different conditions, in order to effectively promote the discernment rate of accuracy of predetermineeing the model. .

Fig. 9 is a schematic structural diagram of a vehicle detection device according to an embodiment of the present invention. As shown in fig. 9, the apparatus 90 includes: an acquisition module 901, a processing module 902, and a determination module 903.

An obtaining module 901, configured to obtain a vehicle picture of a first vehicle, where the vehicle picture is obtained by shooting the first vehicle in running;

the processing module 902 is configured to input the vehicle picture to a preset model, so that the preset model outputs a detection result corresponding to the vehicle picture, where the detection result includes a region type of each region in the vehicle picture and a confidence level that each region is the corresponding region type;

The determining module 903 is configured to determine, if a smoke area of a preset smoke type exists in the vehicle picture according to the detection result, whether the first vehicle generates preset smoke during running according to a first confidence level corresponding to the smoke area.

In one possible design, the determining module 903 is specifically configured to:

In one possible design, the acquisition module 901 is further configured to:

before said inputting of said vehicle picture into a preset model,

The processing module 902 is specifically configured to input the vehicle picture into a preset model, so that the preset model outputs the vehicle picture including at least one of the following areas: a first region of a preset smoke type, a second region of a non-preset smoke type, a third region of a water vapor type, a fourth region of a shadow type, and a fifth region of a dust type.

In one possible design, if the vehicle picture of the first vehicle includes a plurality of pictures, the processing module 902 is further configured to:

The determining module 903 is further configured to:

In one possible design, the processing module 902 is further configured to:

The device provided in this embodiment may be used to implement the technical solution of the foregoing method embodiment, and its implementation principle and technical effects are similar, and this embodiment will not be described herein again.

Fig. 10 is a schematic hardware structure of a vehicle detection device according to an embodiment of the present invention, as shown in fig. 10, an apparatus 100 of the present embodiment includes: a processor 1001 and a memory 1002; wherein the method comprises the steps of

Memory 1002 for storing computer-executable instructions;

The processor 1001 is configured to execute computer-executable instructions stored in the memory to implement the steps executed by the vehicle detection method in the above embodiment. Reference may be made in particular to the relevant description of the embodiments of the method described above.

Alternatively, the memory 1002 may be separate or integrated with the processor 1001.

When the memory 1002 is provided separately, the device further comprises a bus 1003 for connecting said memory 1002 and the processor 1001.

The embodiment of the invention also provides a computer readable storage medium, wherein computer execution instructions are stored in the computer readable storage medium, and when a processor executes the computer execution instructions, the vehicle detection method executed by the vehicle detection device is realized.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be additional divisions when actually implemented, for example, multiple modules may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.

The integrated modules, which are implemented in the form of software functional modules, may be stored in a computer readable storage medium. The software functional module is stored in a storage medium, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (english: processor) to perform some of the steps of the methods according to the embodiments of the application.

It should be understood that the above Processor may be a central processing unit (english: central Processing Unit, abbreviated as CPU), or may be other general purpose processors, a digital signal Processor (english: DIGITAL SIGNAL Processor, abbreviated as DSP), an Application-specific integrated Circuit (english: application SPECIFIC INTEGRATED Circuit, abbreviated as ASIC), or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present invention may be embodied directly in a hardware processor for execution, or in a combination of hardware and software modules in a processor for execution.

The memory may comprise a high-speed RAM memory, and may further comprise a non-volatile memory NVM, such as at least one magnetic disk memory, and may also be a U-disk, a removable hard disk, a read-only memory, a magnetic disk or optical disk, etc.

The bus may be an industry standard architecture (Industry Standard Architecture, ISA) bus, an external device interconnect (PERIPHERAL COMPONENT, PCI) bus, or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, among others. The buses may be divided into address buses, data buses, control buses, etc. For ease of illustration, the buses in the drawings of the present application are not limited to only one bus or to one type of bus.

The storage medium may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.

Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the method embodiments described above may be performed by hardware associated with program instructions. The foregoing program may be stored in a computer readable storage medium. The program, when executed, performs steps including the method embodiments described above; and the aforementioned storage medium includes: various media that can store program code, such as ROM, RAM, magnetic or optical disks.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.

Claims

1. A vehicle detection method, characterized by comprising:

If a smoke region with a preset smoke type exists in the vehicle picture according to the detection result, determining whether preset smoke is generated by the first vehicle in the running process according to a first confidence coefficient corresponding to the smoke region;

if the vehicle picture of the first vehicle includes a plurality of pictures, the images of the first vehicle in the plurality of pictures are different, the method further includes:

If the number of the vehicle pictures for generating the preset smoke by the first vehicle is determined to be larger than the preset number, determining that the first vehicle generates the preset smoke in the driving process;

before the vehicle picture is input to the preset model, the method further comprises:

Training the preset model according to any one of the sub-training sets until the accuracy of the preset model for the current sub-training set is greater than or equal to the preset accuracy to obtain the trained preset model;

after the first vehicle is determined to generate preset smoke in the driving process, gray value calculation is performed according to a smoke area of a preset smoke type in the vehicle picture so as to determine the black smoke level of the smoke area;

The region type includes at least two of: preset smoke type, non-preset smoke type, water vapor type, shadow type and dust type;

Inputting the vehicle picture into a preset model, so that the preset model outputs the vehicle picture including at least two areas: a first region of a preset smoke type, a second region of a non-preset smoke type, a third region of a water vapor type, a fourth region of a shadow type and a fifth region of a dust type.

2. The method of claim 1, wherein determining whether the first vehicle generates the preset smoke during the driving according to the first confidence level corresponding to the smoke area comprises:

3. The method of claim 1, wherein prior to said inputting the vehicle picture into a preset model, the method further comprises:

4. The method according to claim 1, wherein the detection result further comprises position information for indicating each region in the vehicle picture and/or identification information characterizing the preset smoke level.

5. A vehicle detection apparatus, characterized by comprising:

the determining module is used for determining whether the first vehicle generates preset smoke in the driving process according to the first confidence coefficient corresponding to the smoke region if the smoke region with the preset smoke type exists in the vehicle picture according to the detection result;

If the vehicle picture of the first vehicle includes a plurality of pictures, the images of the first vehicle in the plurality of pictures are different, the processing module is further configured to:

the determining module is further configured to:

The acquisition module is further configured to:

The determining module is further configured to: carrying out gray value calculation according to a smoke area of a preset smoke type in the vehicle picture so as to determine the black smoke level of the smoke area;

The processing module is specifically configured to input the vehicle picture to a preset model, so that the preset model outputs the vehicle picture including at least two areas: a first region of a preset smoke type, a second region of a non-preset smoke type, a third region of a water vapor type, a fourth region of a shadow type, and a fifth region of a dust type.

6. The apparatus of claim 5, wherein the determining module is specifically configured to:

7. The apparatus of claim 5, wherein the processing module is further configured to:

8. The apparatus of claim 5, wherein the detection result further comprises location information for indicating each region in the vehicle picture and/or identification information characterizing the preset smoke level.

9. A vehicle detection apparatus, characterized by comprising:

a memory for storing a program;

A processor for executing the program stored by the memory, the processor being for performing the method of any one of claims 1 to 4 when the program is executed.

10. A computer readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any one of claims 1 to 4.