CN109766841A - Vehicle checking method, device and computer readable storage medium - Google Patents
Vehicle checking method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN109766841A CN109766841A CN201910029834.8A CN201910029834A CN109766841A CN 109766841 A CN109766841 A CN 109766841A CN 201910029834 A CN201910029834 A CN 201910029834A CN 109766841 A CN109766841 A CN 109766841A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- frame
- image frame
- detection
- described image
- 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.)
- Granted
Links
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of vehicle checking method, device and computer readable storage mediums, comprising the following steps: obtains the corresponding picture frame of current road conditions in video frame；It obtains in described image frame and the target area of vehicle occurs；The target area is input to deep learning model and carries out vehicle location calculating, obtains vehicle position information.Because the present invention can identify the minimum target region for occurring vehicle in picture frame by vehicle tracking technology, the minimum target region is input to the vehicle position information that deep learning model is calculated under current road conditions again, to solve the problems, such as to be difficult to take into account when DAS (Driver Assistant System) detects vehicle high-accuracy and efficient.
Description
Technical field
The present invention relates to field of vehicle detection more particularly to a kind of vehicle checking method, device and computer-readable deposit
Storage media.
Background technique
Automobile assistant driving system (ADAS) or automated driving system are the trend of following development and in recent years
Swift and violent development is obtained.Quickly, the vehicle in front of the detection of high-accuracy has very important significance, and is directly related to auxiliary
Help control loop that can effectively run.Vehicle-mounted DAS (Driver Assistant System) system is mounted in the embedded system of operation vehicle, by
Full in the calculating speed of its CPU, system power consumption is high, the factors such as low memory, and the computing resource that can be provided is very limited.
Traditional detection method speed of service is fast, and resource consumption is small, but the accuracy rate for detecting vehicle is not high；Another base
It is high in the vehicle checking method Detection accuracy of deep learning, but operating cost is high, and computing resource expends greatly, and detects speed
Slowly, detection effect is unsatisfactory.
Above content is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that above content is existing skill
Art.
Summary of the invention
The main purpose of the present invention is to provide a kind of vehicle checking method, device and computer readable storage medium,
It aims to solve the problem that and is difficult to take into account high-accuracy and efficient problem when DAS (Driver Assistant System) detects vehicle.
In order to achieve the above object, the present invention provides a kind of vehicle checking method, and the vehicle checking method includes following
Step:
Obtain the corresponding picture frame of current road conditions in video frame；
It obtains in described image frame and the target area of vehicle occurs；
The target area is input to deep learning model and carries out vehicle location calculating, obtains vehicle position information.
Preferably, the step of target area of vehicle occur in the acquisition described image frame include:
Judge whether described image frame has similar frame in history image frame；
When determining that described image frame has similar frame in history image frame, obtains in the similar frame and going through for vehicle occur
History region and zone shift amount；
The target area is obtained according to the history area and the zone shift amount.
Preferably, after there is the step of target area of vehicle in the acquisition described image frame, described in the judgement
Whether picture frame has the step of similar frame in history image frame, further includes:
When determining that described image frame does not have similar frame in the video frame, the road conditions letter in described image frame is obtained
Breath；
Corresponding detection window is obtained according to the traffic information；
Vehicle detection is carried out to described image frame according to the detection window；
When the detection window detects vehicle, the region that the vehicle is occurred is as the target area.
Preferably, described the step of carrying out vehicle detection to described image frame according to the detection window, includes:
Obtain starting detection position, default glide direction and default sliding distance；
The detection window is controlled to be slided in starting detection position with default glide direction and default sliding distance
It is dynamic；
The characteristics of image of the image in the detection window is extracted after the detection window slides every time；
It is determined in the detection window according to described image feature with the presence or absence of vehicle.
Preferably, described image feature includes local contrast feature and gradient orientation histogram feature.
Preferably, the size of the different detection windows of the traffic information is different.
Preferably, the different default sliding distances of the traffic information are different.
Preferably, described that the target area is input to the progress vehicle location calculating of deep learning model, obtain vehicle
The step of location information includes:
When getting multiple target areas, merges multiple target areas and obtain combined region；
The combined region is input to the deep learning model and carries out vehicle location calculating, obtains the vehicle location
Information.
In addition, to achieve the above object, the present invention also provides a kind of vehicle detection apparatus, the vehicle detection apparatus includes
Processor, memory and it is stored in the vehicle detection program that can be run on the memory and on the processor, the vehicle
The step of detection program realizes vehicle checking method as described above when being executed by the processor.
In addition, to achieve the above object, it is described computer-readable the present invention also provides a kind of computer readable storage medium
Vehicle detection program is stored on storage medium, the vehicle detection program realizes vehicle as described above when being executed by processor
The step of detection method.
Vehicle checking method, device and computer readable storage medium provided by the invention, firstly, obtaining in video frame
Then the corresponding picture frame of current road conditions obtains in described image frame and the target area of vehicle occurs, finally, by the target
Region is input to deep learning model and carries out vehicle location calculating, obtains vehicle position information.Because the present invention can pass through vehicle
Tracking technique identifies the minimum target region for occurring vehicle in picture frame, then the minimum target region is input to depth
It practises model and the vehicle position information under current road conditions is calculated, be difficult to when to solving DAS (Driver Assistant System) detection vehicle simultaneous
Care for high-accuracy and efficient problem.
Detailed description of the invention
Detailed description of the invention is used to provide further understanding of the present invention, and constitutes part of specification, with the present invention
Embodiment be used to explain the present invention together, be not construed as limiting the invention.In the accompanying drawings:
Fig. 1 is the hardware structural diagram for the vehicle detection apparatus that the embodiment of the present invention is related to；
Fig. 2 is the flow diagram of vehicle checking method first embodiment of the present invention；
Fig. 3 is the flow diagram of vehicle checking method second embodiment of the present invention；
Fig. 4 is the flow diagram of vehicle checking method 3rd embodiment of the present invention；
Fig. 5 is the flow diagram of vehicle checking method fourth embodiment of the present invention；
Fig. 6 is the flow diagram of the 5th embodiment of vehicle checking method of the present invention；
Fig. 7 is the target area figure that vehicle checking method of the present invention detects；
Fig. 8 is vehicle checking method combined region figure of the present invention.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The primary solutions of the embodiment of the present invention are: obtaining the corresponding picture frame of current road conditions in video frame；Obtain institute
State the target area for occurring vehicle in picture frame；The target area is input to deep learning model and carries out vehicle location meter
It calculates, obtains vehicle position information.
Since vehicle-mounted DAS (Driver Assistant System) system is mounted in the embedded system of operation vehicle, the calculating speed of CPU
Full, system power consumption is high, the factors such as low memory, and the computing resource that can be provided is very limited.Traditional detection method runs speed
Degree is fast, and resource consumption is small, but the accuracy rate for detecting vehicle is not high；Another kind is detected based on the vehicle checking method of deep learning
Accuracy rate is high, but operating cost is high, and computing resource expends greatly, and detection speed is slow, and detection effect is unsatisfactory.
The present invention provides a solution, firstly, obtaining the corresponding picture frame of current road conditions in video frame；Then, it obtains
It takes and occurs the target area of vehicle in described image frame；Finally, the target area, which is input to deep learning model, carries out vehicle
Position calculates, and obtains vehicle position information.Because the present invention can identify occur vehicle in picture frame by vehicle tracking technology
Minimum target region, then the minimum target region is input to the vehicle that deep learning model is calculated under current road conditions
Location information, to solve the problems, such as to be difficult to take into account when DAS (Driver Assistant System) detects vehicle high-accuracy and efficient.
As shown in Figure 1, Fig. 1 is that the embodiment of the present invention is related to the hardware structural diagram of device.
Referring to Fig.1, the apparatus may include processor 1001, such as CPU, memory 1002, communication bus 1003, nets
Network interface 1004.Wherein, communication bus 1003 is for realizing the connection communication between each building block in the device.Network interface
1004 may include optionally standard wireline interface and wireless interface (such as WI-FI interface).Memory 1002 can be high speed
RAM memory is also possible to stable memory (non-volatile memory), such as magnetic disk storage.Memory 1002
It optionally can also be the storage device independently of aforementioned processor 1001.As shown in Figure 1, as a kind of computer storage medium
Memory 1002 in may include operating system, network communication module and vehicle detection program.
Optionally, described device can also include camera, RF (Radio Frequency, radio frequency) circuit, sensor,
Voicefrequency circuit, WiFi module etc..Wherein, sensor such as optical sensor, motion sensor and other sensors.Specifically
Ground, optical sensor may include ambient light sensor and proximity sensor, wherein ambient light sensor can be according to the bright of ambient light
The brightness of display screen secretly is adjusted, proximity sensor can close display screen and/or backlight when intelligent terminal is moved in one's ear.
As a kind of motion sensor, gravity accelerometer can detect all directions on (generally three axis) acceleration it is big
It is small, it can detect that size and the direction of gravity when static, can be used to identify the application of intelligent terminal posture, (for example horizontal/vertical screen is cut
Change, dependent game, magnetometer pose calibrating), Vibration identification correlation function (such as pedometer, tap) etc.；Certainly, intelligent terminal
It can also configure the other sensors such as gyroscope, barometer, hygrometer, thermometer, infrared sensor, details are not described herein.
It will be understood by those skilled in the art that the restriction of the not structure twin installation of apparatus structure shown in Fig. 1, can wrap
It includes than illustrating more or fewer components, perhaps combines certain components or different component layouts.
In hardware involved in device shown in Fig. 1, network interface 1004 can be used for obtaining the real-time of current road segment
Road conditions and Weather information；And processor 1001 can be used for calling the vehicle detection program stored in memory 1002, and execute
It operates below:
Obtain the corresponding picture frame of current road conditions in video frame；
It obtains in described image frame and the target area of vehicle occurs；
The target area is input to deep learning model and carries out vehicle location calculating, obtains vehicle position information.
Further, processor 1001 can be used for calling the vehicle detection program stored in memory 1002, also execute
It operates below:
Judge whether described image frame has similar frame in history image frame；
When determining that described image frame has similar frame in history image frame, obtains in the similar frame and going through for vehicle occur
History region and zone shift amount；
The target area is obtained according to the history area and the zone shift amount.
Further, processor 1001 can be used for calling the vehicle detection program stored in memory 1002, also execute
It operates below:
When determining that described image frame does not have similar frame in the video frame, the road conditions letter in described image frame is obtained
Breath；
Corresponding detection window is obtained according to the traffic information；
Vehicle detection is carried out to described image frame according to the detection window；
When the detection window detects vehicle, the region that the vehicle is occurred is as the target area.
Further, processor 1001 can be used for calling the vehicle detection program stored in memory 1002, also execute
It operates below:
Obtain starting detection position, default glide direction and default sliding distance；
The detection window is controlled to be slided in starting detection position with default glide direction and default sliding distance
It is dynamic；
The characteristics of image of the image in the detection window is extracted after the detection window slides every time；
It is determined in the detection window according to described image feature with the presence or absence of vehicle.
Further, processor 1001 can be used for calling the vehicle detection program stored in memory 1002, also execute
It operates below:
When getting multiple target areas, merges multiple target areas and obtain combined region；
The combined region is input to the deep learning model and carries out vehicle location calculating, obtains the vehicle location
Information.
Referring to Fig. 2, Fig. 2 is the first embodiment of vehicle checking method of the present invention, and the vehicle checking method includes:
Step S10, the corresponding picture frame of current road conditions in video frame is obtained；
Vehicle checking method provided by the invention is mainly used for that DAS (Driver Assistant System) is allowed to take into account high-accuracy and high efficiency.This
The terminal that is related to of vehicle checking method that invention provides includes but is not limited to mobile phone, tablet computer and computer etc., in the terminal
It is pre-loaded with relevant application system.
Technical solution provided by the invention, on the way, control front camera obtains the corresponding view of current road conditions to vehicle driving
Then frequency frame obtains the corresponding picture frame of current road conditions in video frame.It is understood that including multiple go through in the video frame
History picture frame and current image frame, described image frame is the last frame of video frame, when described image frame is in the video frame
In when having similar history image frame, then determine that current road conditions do not have greatly changed, vehicle occur according in history image frame
History area and zone shift amount obtain vehicle position information；When described image frame is not similar in the video frame
History image frame when, then determine that apparent variation has occurred in current road conditions, pass through vehicle tracking technology obtain described image frame
In ROI region, and by the ROI region be input to deep learning frame calculate vehicle position information.
Step S20, it obtains in described image frame and the target area of vehicle occurs；
In technical solution provided in this embodiment, in obtaining video frame after the corresponding picture frame of current road conditions, institute is judged
State whether picture frame has similar frame in history image frame；Each frame of history image frame is matched, calculates current figure
As the similarity between frame and history image frame is found out and gone through when determining that described image frame has similar frame in history image frame
Most like several frames in history result carry out estimation using optical flow method, obtain in the similar frame and the history area of vehicle occur
Domain and zone shift amount obtain the target area according to the history area and the zone shift amount.Determining institute
When stating picture frame does not have similar frame in the video frame, show that apparent variation has occurred in front vehicle condition or road conditions.Obtain institute
The traffic information in picture frame is stated, the vehicle in deep learning detection present frame is utilized.In order to accelerate the detection speed of deep learning
Degree, is slided in described image frame first with the detection window of more sizes.Institute is extracted after the detection window slides every time
Described image feature is input to support vector machines (SVM) model and obtains vehicle by the characteristics of image for stating the image in detection window
The target area of appearance.
It should be noted that the size of the different adjustment windows of traffic information is different, traffic information is different described default
Sliding distance is different；Characteristics of image includes local contrast feature and gradient orientation histogram feature.
Step S30, the target area is input to deep learning model and carries out vehicle location calculating, obtain vehicle location
Information.
It, will after there is the target area of vehicle in getting described image frame in technical solution provided in this embodiment
The target area is input to deep learning model together with described image frame and carries out vehicle location calculating, obtains vehicle location
Information.The present invention does not limit the deep learning model used, but this patent is calculated in picture frame using MXNET deep learning model
The location information that vehicle occurs.It is understood that multiple target areas are likely to occur in described image frame as shown in Figure 7,
A target area may only occur, when there are multiple target areas, as shown in figure 8, multiple target areas are merged
For a big combined region, and change the display color of the frame of combined region, the MXNET deep learning model passes through institute
Display color is stated to identify ROI region and carry out vehicle location calculating.
It should be noted that MXNET deep learning model only calculates the image in ROI region, ROI in picture frame
Image outside region will not participate in calculating, therefore greatly reduce time and consumption that deep learning model calculates vehicle location
Resource.
The present invention is according to above scheme, and on the way, control front camera obtains the corresponding video frame of preceding road conditions to vehicle driving,
Then the corresponding picture frame of current road conditions in video frame is obtained, and it is similar to judge whether described image frame has in history image frame
Frame；Each frame of history image frame is matched, similarity is calculated, has phase in history image frame in judgement described image frame
When like frame, several frames most like in historical results are found out, estimation is carried out using optical flow method, obtains and occurs in the similar frame
The history area and zone shift amount of vehicle obtain the target area according to the history area and the zone shift amount
Domain.When determining that described image frame does not have similar frame in the video frame, show that front vehicle condition or road conditions have occurred significantly
Variation.The traffic information in described image frame is obtained, the vehicle in deep learning detection present frame is utilized.In order to accelerate depth
The detection speed of habit is slided in described image frame first with the detection window of more sizes.Detection present frame possibility include
There is the region of vehicle.Wherein, the size difference of the different detection windows of the traffic information and the traffic information difference institute
It is different to state default sliding distance.The image that the image in the detection window is extracted after the detection window slides every time is special
Sign, whether it includes vehicle that described image feature is input in each window of support vector machines (SVM) model prediction.Its
In, described image feature includes local contrast feature and gradient orientation histogram feature followed by merges the area for predicting and
Domain forms a big region, i.e. ROI region (area-of-interest).The ROI region is finally input to the deep learning
Model carries out vehicle location calculating, obtains the vehicle position information.The present invention identifies picture frame by vehicle tracking technology
The middle minimum target region for vehicle occur, then the minimum target region is input to deep learning model, current road is calculated
Vehicle position information under condition improves use so that DAS (Driver Assistant System) be made to take into account high-accuracy and high efficiency when detecting vehicle
Family experience.
Further, referring to Fig. 3, Fig. 3 is the second embodiment of vehicle checking method of the present invention, above-mentioned shown in Fig. 2
On the basis of embodiment, the step S20, comprising:
Step S21, judge whether described image frame has similar frame in history image frame；
Step S22, it when determining that described image frame has similar frame in history image frame, obtains and occurs in the similar frame
The history area and zone shift amount of vehicle；
Step S23, the target area is obtained according to the history area and the zone shift amount.
In technical solution provided in this embodiment, in obtaining video frame after the corresponding picture frame of current road conditions, institute is judged
State whether picture frame has similar frame in history image frame；Each frame of history image frame is matched, calculates current figure
As the similarity between frame and history image frame is found out and gone through when determining that described image frame has similar frame in history image frame
Most like several frames in history result carry out estimation using optical flow method, obtain in the similar frame and the history area of vehicle occur
Domain and zone shift amount obtain the target area according to the history area and the zone shift amount.
Further, when obtaining the target area according to the history area and the zone shift amount, in order to
The more accurate location information for obtaining current vehicle location, can also be weighted multiple history image frames, calculation formula
It is as follows:
Wherein, BkIt is the boundary rectangle frame of k-th of vehicle, wi, δiAnd HBiIt is the weight of i-th of history image frame respectively,
The boundary rectangle of zone shift amount and corresponding vehicle.
When described image frame has similar history image frame, by history image frame obtain the location information of vehicle without
Vehicle location calculating is carried out by deep learning model, to efficiently get the vehicle location letter of the picture frame of current road conditions
Breath.
Further, referring to Fig. 4, Fig. 4 is the 3rd embodiment of vehicle checking method of the present invention, above-mentioned shown in Fig. 3
On the basis of embodiment, the front and back of the step S23, further includes:
Step S24, it when determining that described image frame does not have similar frame in the video frame, obtains in described image frame
Traffic information；
Step S25, corresponding detection window is obtained according to the traffic information；
Step S26, vehicle detection is carried out to described image frame according to the detection window；
Step S27, when the detection window detects vehicle, the region that the vehicle is occurred is as the target area
Domain.
In technical solution provided in this embodiment, in obtaining video frame after the corresponding picture frame of current road conditions, institute is judged
State whether picture frame has similar frame in history image frame；Each frame of history image frame is matched, calculates current figure
As the similarity between frame and history image frame.When determining that described image frame does not have similar frame in the video frame, show
Apparent variation has occurred in front vehicle condition or road conditions.The traffic information in described image frame is obtained, is worked as using deep learning detection
Vehicle in previous frame.In order to accelerate the detection speed of deep learning, first with the detection window of more sizes in described image frame
Middle sliding.The LBP (local contrast) and HOG of the image in the detection window are extracted after the detection window slides every time
LBP the and HOG feature is input to support vector machines (SVM) model, to judge by (gradient orientation histogram feature) feature
The location of current detection window whether there is vehicle.
ROI region, which is obtained, by vehicle tracking technology improves acquisition figure to reduce the calculation amount of deep learning model
As the efficiency of frame vehicle position information.
Further, referring to Fig. 5, Fig. 5 is the fourth embodiment of vehicle checking method of the present invention, above-mentioned shown in Fig. 4
On the basis of embodiment, the step S26, comprising:
Step S261, starting detection position, default glide direction and default sliding distance are obtained；
Step S262, control the detection window the starting detect position with default glide direction and it is default slide away from
From being slided；
Step S263, the characteristics of image of the image in the detection window is extracted after the detection window slides every time；
Step S264, it is determined in the detection window according to described image feature with the presence or absence of vehicle.
In technical solution provided in this embodiment, in order to accelerate the detection speed of deep learning, first with more sizes
Detection window slides in described image frame, and terminal obtains the starting detection position of sliding, presets glide direction and preset and slide
Then dynamic distance controls the detection window in the starting and detects position with default glide direction and the progress of default sliding distance
Sliding, extracts the characteristics of image of the image in the detection window, according to described image after the detection window slides every time
Feature determines in the detection window with the presence or absence of vehicle.
It should be noted that terminal can be slided according to the size and adjustment of traffic information dynamic adjustment detection window
Distance, for example, the vehicle in picture frame generally only occupies the position of image very little when the traffic information is expressway, because
This will test the size reduction of window.The number of vehicle is not detected in terminal record detection window, reaches preset quantity in number
When dynamic increase detection window sliding distance, the preset quantity be administrative staff be arranged, can be 3 times or 3 times or more.
Terminal completely detects picture frame by way of sliding, to improve the accuracy rate of vehicle detection.
Further, referring to Fig. 6, Fig. 6 is the 5th embodiment of vehicle checking method of the present invention, above-mentioned shown in Fig. 2
On the basis of embodiment, the step S30, comprising:
Step S31, when getting multiple target areas, merge multiple target areas and obtain combined region；
Step S32, the combined region is input to the deep learning model and carries out vehicle location calculating, obtained described
Vehicle position information.
It, will after there is the target area of vehicle in getting described image frame in technical solution provided in this embodiment
The target area is input to deep learning model together with described image frame and carries out vehicle location calculating, obtains vehicle location
Information.Multiple target areas are likely to occur in described image frame as shown in Figure 7, it is also possible to a target area are only occurred, gone out
When existing multiple target areas, as shown in figure 8, a big combined region is merged into multiple target areas, and change conjunction
And the display color of the frame in region, the MXNET deep learning model identify that ROI region is gone forward side by side by the display color
Row vehicle location calculates.
A combined region is merged into multiple target areas, to avoid the deep learning frame when there are multiple target areas
Frame, which is omitted, calculates target area.
To achieve the above object, the present invention also provides a kind of vehicle detection apparatus, the vehicle detection apparatus includes processing
Device, memory and it is stored in the vehicle detection program that can be run on the memory and on the processor, the vehicle inspection
The step of ranging sequence realizes vehicle checking method as described above when being executed by the processor.
To achieve the above object, the present invention also provides a kind of computer readable storage medium, the computer-readable storages
Vehicle detection program is stored on medium, the vehicle detection program realizes vehicle detection as described above when being executed by processor
The step of method.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in one as described above
In storage medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be TV
Machine, mobile phone, computer, device, air conditioner or network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of vehicle checking method, which is characterized in that the vehicle checking method the following steps are included:
Obtain the corresponding picture frame of current road conditions in video frame；
It obtains in described image frame and the target area of vehicle occurs；
The target area is input to deep learning model and carries out vehicle location calculating, obtains vehicle position information.
2. vehicle checking method as described in claim 1, which is characterized in that vehicle occur in the acquisition described image frame
The step of target area includes:
Judge whether described image frame has similar frame in history image frame；
When determining that described image frame has similar frame in history image frame, obtains in the similar frame and the history area of vehicle occur
Domain and zone shift amount；
The target area is obtained according to the history area and the zone shift amount.
3. vehicle checking method as claimed in claim 2, which is characterized in that vehicle occur in the acquisition described image frame
It is described to judge whether described image frame has the step of similar frame in history image frame after the step of target area, further includes:
When determining that described image frame does not have similar frame in the video frame, the traffic information in described image frame is obtained；
Corresponding detection window is obtained according to the traffic information；
Vehicle detection is carried out to described image frame according to the detection window；
When the detection window detects vehicle, the region that the vehicle is occurred is as the target area.
4. vehicle checking method as claimed in claim 3, which is characterized in that it is described according to the detection window to described image
Frame carry out vehicle detection the step of include:
Obtain starting detection position, default glide direction and default sliding distance；
The detection window is controlled to be slided in starting detection position with default glide direction and default sliding distance；
The characteristics of image of the image in the detection window is extracted after the detection window slides every time；
It is determined in the detection window according to described image feature with the presence or absence of vehicle.
5. vehicle checking method as claimed in claim 4, which is characterized in that described image feature includes local contrast feature
And gradient orientation histogram feature.
6. vehicle checking method as claimed in claim 3, which is characterized in that the different detection windows of the traffic information
Size is different.
7. vehicle checking method as claimed in claim 4, which is characterized in that the different default slidings of the traffic information away from
From difference.
8. vehicle checking method as described in claim 1, which is characterized in that described that the target area is input to depth
Practising the step of model carries out vehicle location calculating, obtains vehicle position information includes:
When getting multiple target areas, merges multiple target areas and obtain combined region；
The combined region is input to the deep learning model and carries out vehicle location calculating, obtains the vehicle location letter
Breath.
9. a kind of vehicle detection apparatus, which is characterized in that the vehicle detection apparatus includes processor, memory and is stored in institute
The vehicle detection program that can be run on memory and on the processor is stated, the vehicle detection program is held by the processor
It realizes when row such as the step of vehicle checking method described in any item of the claim 1 to 8.
10. a kind of computer readable storage medium, which is characterized in that be stored with vehicle inspection on the computer readable storage medium
Ranging sequence realizes such as vehicle detection described in any item of the claim 1 to 8 when the vehicle detection program is executed by processor
The step of method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910029834.8A CN109766841B (en) | 2019-01-10 | 2019-01-10 | Vehicle detection method, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910029834.8A CN109766841B (en) | 2019-01-10 | 2019-01-10 | Vehicle detection method, device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109766841A true CN109766841A (en) | 2019-05-17 |
| CN109766841B CN109766841B (en) | 2022-03-29 |
Family
ID=66453855
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910029834.8A Active CN109766841B (en) | 2019-01-10 | 2019-01-10 | Vehicle detection method, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109766841B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188645A (en) * | 2019-05-22 | 2019-08-30 | 北京百度网讯科技有限公司 | For the method for detecting human face of vehicle-mounted scene, device, vehicle and storage medium |
| CN112084833A (en) * | 2019-06-14 | 2020-12-15 | 丰田自动车株式会社 | Image recognition device |
| CN112348035A (en) * | 2020-11-11 | 2021-02-09 | 东软睿驰汽车技术（沈阳）有限公司 | Vehicle key point detection method and device and electronic equipment |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103021177A (en) * | 2012-11-05 | 2013-04-03 | 北京理工大学 | Method and system for processing traffic monitoring video image in foggy day |
| CN103605953A (en) * | 2013-10-31 | 2014-02-26 | 电子科技大学 | Vehicle interest target detection method based on sliding window search |
| CN103871079A (en) * | 2014-03-18 | 2014-06-18 | 南京金智视讯技术有限公司 | Vehicle tracking method based on machine learning and optical flow |
| CN106611147A (en) * | 2015-10-15 | 2017-05-03 | 腾讯科技（深圳）有限公司 | Vehicle tracking method and device |
| CN106919916A (en) * | 2017-02-23 | 2017-07-04 | 上海蔚来汽车有限公司 | For the face front attitude parameter method of estimation and device of driver status detection |
| CN106919902A (en) * | 2017-01-19 | 2017-07-04 | 南京邮电大学盐城大数据研究院有限公司 | A kind of vehicle identification and trajectory track method based on CNN |
| CN106981175A (en) * | 2017-05-04 | 2017-07-25 | 广东轻工职业技术学院 | It is a kind of to prevent the method and system of fatigue driving |
| CN107328410A (en) * | 2017-06-30 | 2017-11-07 | 百度在线网络技术（北京）有限公司 | Method and automobile computer for positioning automatic driving vehicle |
| CN107463890A (en) * | 2017-07-20 | 2017-12-12 | 浙江零跑科技有限公司 | A kind of Foregut fermenters and tracking based on monocular forward sight camera |
| US20180101178A1 (en) * | 2016-10-12 | 2018-04-12 | Hyundai Motor Company | Autonomous driving control apparatus, vehicle having the same, and method for controlling the same |
| CN108229475A (en) * | 2018-01-03 | 2018-06-29 | 深圳中兴网信科技有限公司 | Wireless vehicle tracking, system, computer equipment and readable storage medium storing program for executing |
| CN108446622A (en) * | 2018-03-14 | 2018-08-24 | 海信集团有限公司 | Detecting and tracking method and device, the terminal of target object |
| CN108830188A (en) * | 2018-05-30 | 2018-11-16 | 西安理工大学 | Vehicle checking method based on deep learning |
-
2019
- 2019-01-10 CN CN201910029834.8A patent/CN109766841B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103021177A (en) * | 2012-11-05 | 2013-04-03 | 北京理工大学 | Method and system for processing traffic monitoring video image in foggy day |
| CN103605953A (en) * | 2013-10-31 | 2014-02-26 | 电子科技大学 | Vehicle interest target detection method based on sliding window search |
| CN103871079A (en) * | 2014-03-18 | 2014-06-18 | 南京金智视讯技术有限公司 | Vehicle tracking method based on machine learning and optical flow |
| CN106611147A (en) * | 2015-10-15 | 2017-05-03 | 腾讯科技（深圳）有限公司 | Vehicle tracking method and device |
| US20180101178A1 (en) * | 2016-10-12 | 2018-04-12 | Hyundai Motor Company | Autonomous driving control apparatus, vehicle having the same, and method for controlling the same |
| CN106919902A (en) * | 2017-01-19 | 2017-07-04 | 南京邮电大学盐城大数据研究院有限公司 | A kind of vehicle identification and trajectory track method based on CNN |
| CN106919916A (en) * | 2017-02-23 | 2017-07-04 | 上海蔚来汽车有限公司 | For the face front attitude parameter method of estimation and device of driver status detection |
| CN106981175A (en) * | 2017-05-04 | 2017-07-25 | 广东轻工职业技术学院 | It is a kind of to prevent the method and system of fatigue driving |
| CN107328410A (en) * | 2017-06-30 | 2017-11-07 | 百度在线网络技术（北京）有限公司 | Method and automobile computer for positioning automatic driving vehicle |
| CN107463890A (en) * | 2017-07-20 | 2017-12-12 | 浙江零跑科技有限公司 | A kind of Foregut fermenters and tracking based on monocular forward sight camera |
| CN108229475A (en) * | 2018-01-03 | 2018-06-29 | 深圳中兴网信科技有限公司 | Wireless vehicle tracking, system, computer equipment and readable storage medium storing program for executing |
| CN108446622A (en) * | 2018-03-14 | 2018-08-24 | 海信集团有限公司 | Detecting and tracking method and device, the terminal of target object |
| CN108830188A (en) * | 2018-05-30 | 2018-11-16 | 西安理工大学 | Vehicle checking method based on deep learning |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188645A (en) * | 2019-05-22 | 2019-08-30 | 北京百度网讯科技有限公司 | For the method for detecting human face of vehicle-mounted scene, device, vehicle and storage medium |
| CN110188645B (en) * | 2019-05-22 | 2022-10-21 | 阿波罗智联(北京)科技有限公司 | Face detection method and device for vehicle-mounted scene, vehicle and storage medium |
| CN112084833A (en) * | 2019-06-14 | 2020-12-15 | 丰田自动车株式会社 | Image recognition device |
| CN112084833B (en) * | 2019-06-14 | 2024-04-30 | 丰田自动车株式会社 | Image recognition apparatus |
| CN112348035A (en) * | 2020-11-11 | 2021-02-09 | 东软睿驰汽车技术（沈阳）有限公司 | Vehicle key point detection method and device and electronic equipment |
| CN112348035B (en) * | 2020-11-11 | 2024-05-24 | 东软睿驰汽车技术（沈阳）有限公司 | Vehicle key point detection method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109766841B (en) | 2022-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11341375B2 (en) | Image processing apparatus and method based on deep learning and neural network learning | |
| US11183051B2 (en) | Deep learning methods for estimating density and/or flow of objects, and related methods and software | |
| WO2020253657A1 (en) | Video clip positioning method and apparatus, computer device, and storage medium | |
| EP2695143B1 (en) | Image processing apparatus, display control method and program | |
| US20200271450A1 (en) | Systems and methods for providing immersive extended reality experiences on moving platforms | |
| JP4467838B2 (en) | Image recognition apparatus and image recognition method | |
| US20180124319A1 (en) | Method and apparatus for real-time traffic information provision | |
| CN104103030B (en) | Image analysis method, camera apparatus, control apparatus and control method | |
| CN107679455A (en) | Target tracker, method and computer-readable recording medium | |
| CN109664820A (en) | Driving reminding method, device, equipment and storage medium based on automobile data recorder | |
| CN109766841A (en) | Vehicle checking method, device and computer readable storage medium | |
| CN103514432A (en) | Method, device and computer program product for extracting facial features | |
| KR20160084706A (en) | Device for Providing Description Information Regarding Workout Record and Method Thereof | |
| CN113673305A (en) | Image marking using geodesic features | |
| CN103871046A (en) | Setting apparatus, control method thereof, image processing apparatus, and control method thereof | |
| CN111989915B (en) | Methods, media, and systems for automatic visual inference of environment in an image | |
| CN108683744A (en) | Information-pushing method, device, computer equipment and storage medium | |
| CN112052778B (en) | Traffic sign identification method and related device | |
| JP6462528B2 (en) | MOBILE BODY TRACKING DEVICE, MOBILE BODY TRACKING METHOD, AND MOBILE BODY TRACKING PROGRAM | |
| CN107194948B (en) | Video significance detection method based on integrated prediction and time-space domain propagation | |
| US20120128255A1 (en) | Part detection apparatus, part detection method, and program | |
| JP6320942B2 (en) | Weight matrix learning device, gaze direction prediction system, warning system, and weight matrix learning method | |
| JP6794895B2 (en) | Electronic devices, location identification systems, location identification methods and programs | |
| US20230091536A1 (en) | Camera Placement Guidance | |
| WO2022179016A1 (en) | Lane detection method and apparatus, device, and 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |