CN109815856A - Status indication method, system and the computer readable storage medium of target vehicle - Google Patents
Status indication method, system and the computer readable storage medium of target vehicle Download PDFInfo
- Publication number
- CN109815856A CN109815856A CN201910017244.3A CN201910017244A CN109815856A CN 109815856 A CN109815856 A CN 109815856A CN 201910017244 A CN201910017244 A CN 201910017244A CN 109815856 A CN109815856 A CN 109815856A
- Authority
- CN
- China
- Prior art keywords
- target vehicle
- image information
- video
- detected
- loading
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000001514 detection method Methods 0.000 claims abstract description 86
- 238000012544 monitoring process Methods 0.000 claims abstract description 48
- 238000011897 real-time detection Methods 0.000 claims abstract description 19
- 238000011068 loading method Methods 0.000 claims description 97
- 238000012549 training Methods 0.000 claims description 60
- 230000011218 segmentation Effects 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 8
- 230000005856 abnormality Effects 0.000 abstract description 22
- 238000013527 convolutional neural network Methods 0.000 description 14
- 238000007689 inspection Methods 0.000 description 10
- 239000000284 extract Substances 0.000 description 9
- 230000002159 abnormal effect Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000000926 separation method Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000002547 anomalous effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention provides status indication method, system and the computer readable storage mediums of a kind of target vehicle.The status indication method of target vehicle includes: real-time detection video to be detected;When in video to be detected including target vehicle, the image information input state detection model of target vehicle is determined to the classification of image information;When the classification of image information is pre-set categories, marks target vehicle and reported to monitoring system.Apply technical solution provided by the invention, real-time detection video to be detected, when there is target vehicle in video to be detected, obtain the image information of target vehicle, it determines whether target vehicle is abnormality by state-detection model, mark the target vehicle of abnormality and is reported to monitoring system, and then can accurately detect the state of the target vehicles such as dump truck, and the target vehicle for being in abnormality is marked, more intelligent, the wider array of monitoring of environment scheme of application range is provided.
Description
Technical field
The present invention relates to vehicle recongnition technique field, in particular to a kind of status indication method of target vehicle, one
The status indication system of kind target vehicle and a kind of computer readable storage medium.
Background technique
Currently, in the related art, there are some special monitoring demands in traffic video monitoring, such as environmental protection aspect
Monitoring, it is desirable that monitor the delivery state of some particular vehicles to confirm whether vehicle meets environmental requirement.Wherein, dump truck due to
Its cargo delivered is usually to have article, such as coal, ore, dregs of higher pollution etc., and loading state is always ring
Protect the emphasis of monitoring.The identification of dump truck loading state also has practical significance and commercial value.
Application aspect technology of the depth convolutional neural networks in target identification is more and more mature, and especially in recent years one
(one kind is by ox by a little method such as AlexNet (a kind of convolutional neural networks structure designed by Alex Krizhevsky), VGG16
Saliva university provide convolutional neural networks structure), a kind of GoogLeNet (convolutional Neural proposed by Christian Szegedy
Network structure) etc. serial of methods significant effect and having begun be widely applied in actual items.But it is based on depth convolution
The identification project of neural network still has many problem needs and goes to solve.Firstly, being needed in the training process of model big
The mark sample of amount, however these samples are not easy to obtain in many cases.When especially carrying out anomalous identification, exceptional sample is
It needs to be identified.But the appearance of abnormal object is often small probability event, this also means that exceptional sample is even more to be difficult to
It obtains.The deficiency of sample makes model hardly result in sufficient study;Secondly, concentrating the target category to be identified in image data
Between usually have very high similarity, this is the main reason for causing erroneous detection.
In the prior art, although the common data sets obtained from internet such as COCO (image that team, Microsoft obtains
Data set), MIT-CBCL (Massachusetts Polytechnics obtain an image data set), KITTI (provide by www.cvilbs.net
Vehicle detection, tracking, semantic separation algorithms data set), UA-DETRAC (a kind of vehicle detection data set), ImageNet
Large-scale general data set as (a kind of large-scale visible database for visual object identification software research) is easy to obtain
, but since dump truck data compared to general vehicular traffic identification data are more difficult to obtain, this specific small field
Data set, the training requirement that sample size will meet deep learning need a large amount of data acquisition, pretreatment and data set mark
The investment of note;And need to do the covering state in the compartment of dump truck one identification during environment monitoring, it needs to identify
Be dump truck loading and the unusual condition that does not cover.Since camera position is in building site environment, installation site is different,
So that, usually there is dump truck abnormality mistake in these dump truck loadings and state not covered is difficult to capture in video image
The result dump truck loading exception and two class sample of normal condition of identification are obviously than the number of samples between the vehicle of different automobile types
More unbalanced therefore traditional target vehicle recognition methods is not suitable for dump truck state recognition, and traditional characteristic extracts
Algorithm ability to express is limited, largely reduces the robustness of algorithm, has very for complex environment in actual use
Big limitation is not used to the complex environment of the dump trucks such as building site, stockyard work.
Therefore, the effective method of one kind is needed at present to realize detection and label to dump truck is not covered.
Summary of the invention
The present invention is directed to solve at least one of the technical problems existing in the prior art or related technologies.
For this purpose, the first aspect of the present invention proposes a kind of status indication method of target vehicle.
The second aspect of the present invention proposes a kind of status indication system of target vehicle.
The third aspect of the present invention proposes a kind of computer readable storage medium.
In view of this, the first aspect of the present invention provides a kind of status indication method of target vehicle, comprising: inspection in real time
Survey video to be detected;When in video to be detected including target vehicle, the image information input state of target vehicle is detected into mould
Type is to determine the classification of image information;When the classification of image information is pre-set categories, target vehicle is marked and to monitoring system
Report.
In the technical scheme, by real-time detection video to be detected, specifically, video to be detected can be traffic monitoring
The video that the camera of the scene such as monitor video that camera obtains in real time or building site, stockyard setting is shot with video-corder, when detect to
When detecting in video comprising target vehicle, the image information of the target vehicle is input to state-detection model, is examined by state
Survey the classification of model inspection image information;When the classification that the result of state-detection model output is the image information is pre-set categories
When, target vehicle corresponding to the image information is labeled as abnormal vehicle, and report to monitoring system.The present invention is applied to mention
The technical solution of confession is confirmed whether target vehicle occur by real-time detection video to be detected, occurs in video to be detected
When target vehicle, the image information of target vehicle is obtained, the classification of image information is determined by state-detection model, and then determine
Whether target vehicle is abnormality, marks the target vehicle of abnormality and reports to monitoring system, and then can be accurately
The state of the target vehicles such as dump truck is detected, and marks the target vehicle for being in abnormality, more intelligent, application range is provided
Wider array of monitoring of environment scheme.
Specifically, target vehicle is set as dump truck, obtains the monitor video that traffic monitoring camera obtains in real time, detection
It whether include dump truck in monitor video;When in monitor video including dump truck, the corresponding image information of dump truck is inputted
Into state-detection model, so that state-detection model judges the classification of image information;The output knot of reception state detection model
Fruit, when state-detection model output the result shows that the classification of image information be pre-set categories when, assert the state of the dump truck
For exception, the specially state of the dump truck is to be loaded with cargo, but cargo is uncovered, at this time by the dump truck labeled as abnormal
Dump truck, while the information such as state, position and the license plate number of the dump truck are reported to monitoring system.
In addition, the status indication method of the target vehicle in above-mentioned technical proposal provided by the invention can also have it is as follows
Additional technical feature:
In the above-mentioned technical solutions, it is preferable that before the real-time detection video to be detected the step of, further includes: acquisition prison
Video is controlled, the history image information of target vehicle is obtained by monitor video;Sort command is received, according to sort command to history
Image information is classified, to form training dataset;By training dataset training pre-training model to obtain state-detection
Model.
In the technical scheme, by largely acquiring different location, Traffic Surveillance Video in different time periods, analysis monitoring
Video and therefrom obtain it includes target vehicle history image message sample, according to sort command to history image information sample
This is classified, and obtains the history image message sample collection of various different times, different location and different conditions, and form instruction
Practice data set;It chooses general using the convolutional neural networks model that the training of mass transportation vehicle data is got as pre-training
Model can be obtained the more accurate state with more high robust of discrimination by training dataset training pre-training model and examine
Model is surveyed, to realize to the accurate detection of target vehicle state and label.
In any of the above-described technical solution, it is preferable that the step of being classified according to sort command to history image information,
It specifically includes: when sort command instruction target vehicle is non-loading, image information being classified as first kind image information;When point
Class order instruction target vehicle is that when can not determine whether loading, image information is classified as the second class image information;Work as classification
Order instruction target vehicle is loading, and when cargo is capped, image information is classified as third class image information;Work as classification
Order instruction target vehicle is loading, and when cargo is uncovered, image information is classified as the 4th class image information;Its
In, the classification for being classified as the image information of the 4th class image information is pre-set categories.
In the technical scheme, according to target vehicle in history image information, i.e. the loading state of dump truck is by history figure
As information be classified as the non-loading of dump truck, dump truck loading and can not determine dump truck whether three kinds of loading, and further will
The case where dump truck loading, is specifically classified as dump truck loading, and cargo is capped and dump truck loading, and described
Cargo is uncovered, obtains specific 4 kinds of class categories, and specifically by the classification setting of history image information are as follows: First Kind Graph
As information, i.e. dump truck is loading;Second class image information, i.e., can not determine dump truck whether loading;Third class image information,
That is dump truck loading, and cargo is capped;And the 4th class image information, i.e. dump truck loading, and the cargo not by
Covering, and using sorted history image information as training dataset.Wherein, setting is classified as going through for the 4th class image information
The classification of history image information is pre-set categories.By by target vehicle, i.e., dump truck state according to loading, non-loading, can not be true
Determine loading state and be divided into 3 classes, whether cargo is capped by loading shape when then according to monitoring of environment demand further according to loading
The dump truck of state further segments, i.e., data and dump truck loading state are divided into 4 classes in data set manufacturing process, are increased
Harmony between data category can promote the accuracy rate that state-detection model identifies target vehicle state.
In any of the above-described technical solution, it is preferable that the step of real-time detection video to be detected, specifically include: to be checked
It surveys video to be pre-processed, to obtain the foreground picture of the image of each frame in video to be detected;Threshold segmentation is carried out to foreground picture,
And morphology opening operation and normalized are carried out to obtain binary image;Determine the profile of target area in binary image
Area;Judge whether contour area meets decision threshold range by rectangle frame detection method;When contour area meets decision threshold
When being worth range, determine in foreground picture comprising target vehicle.
In the technical scheme, after getting video to be detected, first using open source computer vision library OpenCV
(Open Source Computer Vision Library, computer vision library of increasing income) is assisted, to complete to be detected
The pretreatment of video, while the method for using mean filter and Gaussian Background to model establishes the image background model of video, with
By the prospect and background separation of the image of frame each in video to be detected, the prospect of the image of each frame in video to be detected is obtained
Figure;Also background subtraction can be used and obtain foreground picture;Then it is calculated using Ostu (Otsu algorithm) Threshold segmentation in OpenCV
Method carries out Threshold segmentation to foreground picture, while carrying out morphology opening operation to the image after segmentation, removes the noise in foreground picture
Point obtains binary image in further progress normalized;Binaryzation is determined by the contours extract function in OpenCV
The contour area of objective area in image;Judge whether contour area meets default decision threshold model by rectangle frame detection method
It encloses;When contour area meets decision threshold range, it can determine that, comprising target vehicle in foreground picture, this method is large-scale in detection
Vehicle, testing result is accurate when such as dump truck, and the calculation amount needed is small, while detection efficiency with higher.
In any of the above-described technical solution, it is preferable that when in foreground picture including target vehicle, further includes: obtain target
The coordinate information in region;The image information of target vehicle is intercepted in the image of each frame by coordinate information.
In the technical scheme, when determining in foreground picture includes target vehicle, the corresponding target area of target vehicle is obtained
The coordinate information in domain intercepts the image information of target vehicle by coordinate information in primitive frame image, and image information is defeated
Enter state-detection model, by the state of state-detection model inspection target vehicle, realizes to the target carriage for being in abnormality
Label and report.
Second aspect of the present invention provides a kind of status indication system of target vehicle, including memory and processor;It deposits
Reservoir is for storing computer program;Processor is for executing computer program with real-time detection video to be detected;When to be detected
When in video including target vehicle, the image information input state detection model of target vehicle is determined to the class of image information
Not;When the classification of image information is pre-set categories, marks target vehicle and reported to monitoring system.
In the technical scheme, by real-time detection video to be detected, specifically, video to be detected can be traffic monitoring
The video that the camera of the scene such as monitor video that camera obtains in real time or building site, stockyard setting is shot with video-corder, when detect to
When detecting in video comprising target vehicle, the image information of the target vehicle is input to state-detection model, is examined by state
Survey the classification of model inspection image information;When the classification that the result of state-detection model output is the image information is pre-set categories
When, target vehicle corresponding to the image information is labeled as abnormal vehicle, and report to monitoring system.The present invention is applied to mention
The technical solution of confession is confirmed whether target vehicle occur by real-time detection video to be detected, occurs in video to be detected
When target vehicle, the image information of target vehicle is obtained, the classification of image information is determined by state-detection model, and then determine
Whether target vehicle is abnormality, marks the target vehicle of abnormality and reports to monitoring system, and then can be accurately
The state of the target vehicles such as dump truck is detected, and marks the target vehicle for being in abnormality, more intelligent, application range is provided
Wider array of monitoring of environment scheme.
In the above-mentioned technical solutions, it is preferable that processor is also used to: acquisition monitor video obtains mesh by monitor video
Mark the history image information of vehicle;Sort command is received, is classified according to sort command to history image information, to form instruction
Practice data set;By training dataset training pre-training model to obtain state-detection model.
In the technical scheme, by largely acquiring different location, Traffic Surveillance Video in different time periods, analysis monitoring
Video and therefrom obtain it includes target vehicle history image message sample, according to sort command to history image information sample
This is classified, and obtains the history image message sample collection of various different times, different location and different conditions, and form instruction
Practice data set;It chooses general using the convolutional neural networks model that the training of mass transportation vehicle data is got as pre-training
Model can be obtained the more accurate state with more high robust of discrimination by training dataset training pre-training model and examine
Model is surveyed, to realize to the accurate detection of target vehicle state and label.
In any of the above-described technical solution, it is preferable that processor is also used to: when sort command instruction target vehicle is not carry
When goods, image information is classified as first kind image information;When sort command instruction target vehicle is that can not determine whether loading
When, image information is classified as the second class image information;When sort command instruction target vehicle is loading, and cargo is capped
When, image information is classified as third class image information;When sort command instruction target vehicle is loading, and cargo is not coated
Image information is classified as the 4th class image information by Gai Shi;Wherein, it is classified as the classification of the image information of the 4th class image information
For pre-set categories.
In the technical scheme, according to target vehicle in history image information, i.e. the loading state of dump truck is by history figure
As information be classified as the non-loading of dump truck, dump truck loading and can not determine dump truck whether three kinds of loading, and further will
The case where dump truck loading, is specifically classified as dump truck loading, and cargo is capped and dump truck loading, and described
Cargo is uncovered, obtains specific 4 kinds of class categories, and specifically by the classification setting of history image information are as follows: First Kind Graph
As information, i.e. dump truck is loading;Second class image information, i.e., can not determine dump truck whether loading;Third class image information,
That is dump truck loading, and cargo is capped;And the 4th class image information, i.e. dump truck loading, and the cargo not by
Covering, and using sorted history image information as training dataset.Wherein, setting is classified as going through for the 4th class image information
The classification of history image information is pre-set categories.By by target vehicle, i.e., dump truck state according to loading, non-loading, can not be true
Determine loading state and be divided into 3 classes, whether cargo is capped by loading shape when then according to monitoring of environment demand further according to loading
The dump truck of state further segments, i.e., data and dump truck loading state are divided into 4 classes in data set manufacturing process, are increased
Harmony between data category can promote the accuracy rate that state-detection model identifies target vehicle state.
In any of the above-described technical solution, it is preferable that processor is also used to: video to be detected is pre-processed, with
The foreground picture of the image of each frame into video to be detected;To foreground picture carry out Threshold segmentation, and carry out morphology opening operation and
Normalized is to obtain binary image;Determine the contour area of target area in binary image;It is detected by rectangle frame
Method judges whether contour area meets decision threshold range;When contour area meets decision threshold range, foreground picture is determined
In include target vehicle.
In the technical scheme, after getting video to be detected, first using open source computer vision library OpenCV into
Row auxiliary to complete the pretreatment to video to be detected, while being established using the method that mean filter and Gaussian Background model
The prospect and background separation of the image of frame each in video to be detected are obtained view to be detected by the image background model of video
The foreground picture of the image of each frame in frequency;Also background subtraction can be used and obtain foreground picture;Then using in OpenCV
Ostu Threshold Segmentation Algorithm carries out Threshold segmentation to foreground picture, while carrying out morphology opening operation, removal to the image after segmentation
Noise spot in foreground picture obtains binary image in further progress normalized;Pass through the contours extract in OpenCV
Function determines the contour area of target area in binary image;Judge whether contour area meets by rectangle frame detection method
Default decision threshold range;When contour area meets decision threshold range, can determine comprising target vehicle in foreground picture, it should
Method is in detection oversize vehicle, and testing result is accurate when such as dump truck, and the calculation amount needed is small, while detection with higher is imitated
Rate.
In any of the above-described technical solution, it is preferable that processor is also used to: obtaining the coordinate information of target area;Pass through
Coordinate information intercepts the image information of target vehicle in the image of each frame.
In the technical scheme, when determining in foreground picture includes target vehicle, the corresponding target area of target vehicle is obtained
The coordinate information in domain intercepts the image information of target vehicle by coordinate information in primitive frame image, and image information is defeated
Enter state-detection model, by the state of state-detection model inspection target vehicle, realizes to the target carriage for being in abnormality
Label and report.
Third aspect present invention provides a kind of computer readable storage medium, is stored thereon with computer program, calculates
The status indication method that the target vehicle as described in any of the above-described technical solution is realized when machine program is executed by processor, because
This, which includes the complete of the status indication method of target vehicle described in any of the above-described technical solution
Portion's beneficial effect.
Additional aspect and advantage of the invention will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect of the invention and advantage will become from the description of the embodiment in conjunction with the following figures
Obviously and it is readily appreciated that, in which:
Fig. 1 shows the flow chart of the status indication method of target vehicle according to an embodiment of the invention;
Fig. 2 shows the flow charts of the status indication method of target vehicle according to another embodiment of the invention;
Fig. 3 shows the classification schematic diagram of target vehicle according to one embodiment of present invention;
Fig. 4 shows the flow chart of history image letter classification according to one embodiment of present invention;
Fig. 5 shows the flow chart of the status indication method of the target vehicle of still another embodiment in accordance with the present invention;
Fig. 6 shows the flow chart of the status indication method of target vehicle according to still another embodiment of the invention;
Fig. 7 shows the flow chart of the status indication method of target vehicle according to still another embodiment of the invention;
Fig. 8 shows the block diagram of the status indication system of target vehicle according to an embodiment of the invention.
Specific embodiment
To better understand the objects, features and advantages of the present invention, with reference to the accompanying drawing and specific real
Applying mode, the present invention is further described in detail.It should be noted that in the absence of conflict, the implementation of the application
Feature in example and embodiment can be combined with each other.
In the following description, numerous specific details are set forth in order to facilitate a full understanding of the present invention, still, the present invention may be used also
To be implemented using other than the one described here other modes, therefore, protection scope of the present invention is not by described below
Specific embodiment limitation.
Status indication method, the mesh of the target vehicle described according to some embodiments of the invention are described referring to Fig. 1 to Fig. 8
Mark the status indication system and computer readable storage medium of vehicle.
As shown in Figure 1, providing a kind of status indication side of target vehicle in the embodiment of first aspect present invention
Method, comprising:
S102, real-time detection video to be detected;
The image information input state of target vehicle is detected mould when in video to be detected including target vehicle by S104
Type is to determine the classification of image information;
S106 marks target vehicle and reports to monitoring system when the classification of image information is pre-set categories.
In this embodiment, by real-time detection video to be detected, specifically, video to be detected can be taken the photograph for traffic monitoring
The video that the camera of the scene such as the monitor video obtained in real time as head or building site, stockyard setting is shot with video-corder, it is to be checked when detecting
When surveying in video comprising target vehicle, the image information of the target vehicle is input to state-detection model, passes through state-detection
The classification of model inspection image information;When the classification that the result of state-detection model output is the image information is pre-set categories
When, target vehicle corresponding to the image information is labeled as abnormal vehicle, and report to monitoring system.The present invention is applied to mention
The technical solution of confession is confirmed whether target vehicle occur by real-time detection video to be detected, occurs in video to be detected
When target vehicle, the image information of target vehicle is obtained, the classification of image information is determined by state-detection model, and then determine
Whether target vehicle is abnormality, marks the target vehicle of abnormality and reports to monitoring system, and then can be accurately
The state of the target vehicles such as dump truck is detected, and marks the target vehicle for being in abnormality, more intelligent, application range is provided
Wider array of monitoring of environment scheme.
Specifically, target vehicle is set as dump truck, obtains the monitor video that traffic monitoring camera obtains in real time, detection
It whether include dump truck in monitor video;When in monitor video including dump truck, the corresponding image information of dump truck is inputted
Into state-detection model, so that state-detection model judges the classification of image information;The output knot of reception state detection model
Fruit, when state-detection model output the result shows that the classification of image information be pre-set categories when, assert the state of the dump truck
For exception, the specially state of the dump truck is to be loaded with cargo, but cargo is uncovered, at this time by the dump truck labeled as abnormal
Dump truck, while the information such as state, position and the license plate number of the dump truck are reported to monitoring system.
In one embodiment of the invention, it is preferable that as shown in Fig. 2, the status indication method of target vehicle includes:
S202 acquires monitor video, and the history image information of target vehicle is obtained by monitor video;
S204 receives sort command, is classified according to sort command to history image information, to form training data
Collection;
S206, by training dataset training pre-training model to obtain state-detection model;
S208, real-time detection video to be detected;
The image information input state of target vehicle is detected mould when in video to be detected including target vehicle by S210
Type is to determine the classification of image information;
S212 marks target vehicle and reports to monitoring system when the classification of image information is pre-set categories.
In this embodiment, by largely acquiring different location, Traffic Surveillance Video in different time periods, analysis monitoring view
Frequency and therefrom obtain it includes target vehicle history image message sample, according to sort command to history image message sample
Classify, obtains the history image message sample collection of various different times, different location and different conditions, and form training
Data set;It chooses general using the convolutional neural networks model that the training of mass transportation vehicle data is got as pre-training mould
Type, it is preferable that select a kind of VGGNet-16 (convolutional neural networks structure provided by Oxford University) model as pre-training mould
Type;By training dataset training pre-training model can be obtained discrimination more accurately, the state-detection with more high robust
Model, to realize to the accurate detection of target vehicle state and label.
Specifically, as the convolutional neural networks model of pre-training model there are many selection, such as if favor speed more
If fast convolutional neural networks model, it can choose that structure level number is few, fireballing AlexNet model;If it is preferred that precision
If higher convolutional neural networks model, a kind of VGGNet (convolutional neural networks knot provided by Oxford University can choose
Structure) model, NIN (Network in Network, Webweb convolutional neural networks structure) model, GoogLeNet model and
A kind of numbers of plies such as ResNet (Residual Neural Network, convolutional neural networks structure that Microsoft proposes) model are more
Network model.
In one embodiment of the invention, it is preferable that the step classified according to sort command to history image information
Suddenly, it specifically includes: when sort command instruction target vehicle is non-loading, image information being classified as first kind image information;
When sort command instruction target vehicle is that can not determine whether loading, image information is classified as the second class image information;When
Sort command indicates that target vehicle is loading, and when cargo is capped, image information is classified as third class image information;When
Sort command indicates that target vehicle is loading, and when cargo is uncovered, image information is classified as the 4th class image information;
Wherein, the classification for being classified as the image information of the 4th class image information is pre-set categories.
In this embodiment, specifically, as shown in figure 3, according to target vehicle in history image information, i.e. the load of dump truck
Goods state history image information is classified as the non-loading of dump truck, dump truck loading and can not determine dump truck whether loading three
Kind, and the case where dump truck loading, is specifically further classified as dump truck loading, and cargo is capped and dump truck
Loading, and the cargo is uncovered, obtains specific 4 kinds of class categories, and specifically set the classification of history image information
Be set to: first kind image information, i.e. dump truck are loading;Second class image information, i.e., can not determine dump truck whether loading;The
The loading of three classes image information, i.e. dump truck, and cargo is capped;And the 4th class image information, i.e. dump truck loading, and
The cargo is uncovered, and using sorted history image information as training dataset.Wherein, setting is classified as the 4th class
The classification of the history image information of image information is pre-set categories.By by target vehicle, i.e., dump truck state according to loading, not
Loading can not determine that loading state is divided into 3 classes, and whether cargo is coated when then according to monitoring of environment demand further according to loading
Lid further segments the dump truck of loading state, i.e., is divided into data and dump truck loading state in data set manufacturing process
4 classes increase the harmony between data category, can be promoted state-detection model target vehicle state is identified it is accurate
Rate.
Wherein, the detailed process classified to history image information is as shown in Figure 4:
S402 acquires monitor video data;
S404 disassembles monitor video data to obtain the frame image containing dump truck;
S406 extracts the history image information of dump truck region in frame image;
History image information is divided into 4 classes according to the state of dump truck by S408.
When acquisition is used for the training dataset of physical training condition detection model, splitted data into according to dump truck loading state
4 classes, classification in this way can increase the harmony between data category and be conducive to promote recognition accuracy, pre- having chosen
After training pattern, the other parameter of the last layer output class of this pre-training model is revised as 4, and initialize this with random number
The weight parameter of layer, and keep other layers of structure and parameter constant, it is then modified pre- with obtained training dataset training
Training pattern, even if the available state-detection mould for still ensuring that higher recognition accuracy when dump truck sample size is less
Type.
In one embodiment of the invention, it is preferable that as shown in figure 5, the status indication method of target vehicle includes:
S502 pre-processes video to be detected, to obtain the foreground picture of the image of each frame in video to be detected;
S504 carries out Threshold segmentation to foreground picture, and carries out morphology opening operation and normalized to obtain binaryzation
Image;
S506 determines the contour area of target area in binary image;
S508 judges whether contour area meets decision threshold range by rectangle frame detection method;
S510 is determined in foreground picture when contour area meets decision threshold range comprising target vehicle;
The image information input state detection model of target vehicle is determined the classification of image information by S512;
S514 marks target vehicle and reports to monitoring system when the classification of image information is pre-set categories.
In this embodiment, it after getting video to be detected, is carried out first using open source computer vision library OpenCV
Auxiliary to complete the pretreatment to video to be detected, while establishing view using the method that mean filter and Gaussian Background model
The prospect and background separation of the image of frame each in video to be detected are obtained video to be detected by the image background model of frequency
In each frame image foreground picture;Also background subtraction can be used and obtain foreground picture;Then using the Ostu in OpenCV
Threshold Segmentation Algorithm carries out Threshold segmentation to foreground picture, while carrying out morphology opening operation to the image after segmentation, removes prospect
Noise spot in figure obtains binary image in further progress normalized;Pass through the contours extract function in OpenCV
Determine the contour area of target area in binary image;It is default to judge whether contour area meets by rectangle frame detection method
Decision threshold range;When contour area meets decision threshold range, can determine in foreground picture comprising target vehicle, this method
In detection oversize vehicle, testing result is accurate when such as dump truck, and the calculation amount needed is small, while detection efficiency with higher.
In one embodiment of the invention, it is preferable that as shown in fig. 6, the status indication method of target vehicle includes:
S602 pre-processes video to be detected, to obtain the foreground picture of the image of each frame in video to be detected;
S604 carries out Threshold segmentation to foreground picture, and carries out morphology opening operation and normalized to obtain binaryzation
Image;
S606 determines the contour area of target area in binary image;
S608 judges whether contour area meets decision threshold range by rectangle frame detection method;
S610 is determined in foreground picture when contour area meets decision threshold range comprising target vehicle;
S612 obtains the coordinate information of target area;
S614 intercepts the image information of target vehicle by coordinate information in the image of each frame;
The image information input state detection model of target vehicle is determined the classification of image information by S616;
S618 marks target vehicle and reports to monitoring system when the classification of image information is pre-set categories.
In this embodiment, when determining in foreground picture includes target vehicle, the corresponding target area of target vehicle is obtained
Coordinate information, the image information of target vehicle is intercepted in primitive frame image by coordinate information, image information is inputted
State-detection model is realized by the state of state-detection model inspection target vehicle to the target vehicle for being in abnormality
Label and report.
In one embodiment of the invention, it is preferable that as shown in fig. 7, the reality of the status indication method of target vehicle
Workflow includes:
S702 obtains video to be detected;
S704 models the background of video to be detected;
S706 carries out mean filter to kth frame image in video to be detected;
S708 carries out adaptive threshold fuzziness and binary conversion treatment to filtered image;
S710, is normalized binary image and morphology opening operation is handled;
S712, the profile of target area is to judge whether target area meets judgement in the binary image after extraction process
Condition is then to enter S716, otherwise enters S714;
S714 carries out mean filter to+1 frame image of kth;
S716 returns to the coordinate information of target area;
S718 extracts image information and input state detection model by coordinate information from kth frame image;
S720 judges whether image information is pre-set categories, is then to enter S722, otherwise enters S714;
S722 marks target vehicle and reports to monitoring system;
S724, judges whether video to be detected terminates, and is, process terminates, and otherwise returns to S714.
In this embodiment, obtain video to be detected, to the background of video to be detected modeled with when video prospect
With background separation;It extracts kth frame image in video and mean filter is carried out to kth frame image, wherein k=1,2,3 ...;It is first
Adaptive threshold fuzziness and binary conversion treatment first are carried out to filtered kth frame image, then to filtered kth frame image
Adaptive threshold fuzziness and binary conversion treatment are carried out, then binary image is normalized and shape to treated
State opening operation is handled, and the profile of target area is in the binary image after last extraction process to judge whether target area is full
Sufficient decision condition continues to extract+1 frame image of kth and be filtered if being unsatisfactory for decision condition;If satisfaction is sentenced
Fixed condition then returns to the coordinate information of target area, extracts image information from kth frame image according to coordinate information and inputs shape
State detection model judges whether image information is pre-set categories by state-detection model, that is, judge target dump truck whether be
Loading and the unlapped state of cargo;If image information is not pre-set categories, continues to extract+1 frame image of kth and be filtered
Wave processing;If image information is pre-set categories, target dump truck is marked to be abnormality and report to monitoring system;Continue
The frame image in video is extracted to judge whether video terminates, if video is not finished, continues extraction next frame image and goes forward side by side
Row detection, if video is over, then terminates testing process.
As shown in figure 8, providing a kind of status indication system of target vehicle in the embodiment of second aspect of the present invention
800, including memory 802 and processor 804;Memory 802 is for storing computer program;Processor 804 is based on executing
Calculation machine program is with real-time detection video to be detected;When in video to be detected including target vehicle, the image of target vehicle is believed
Input state detection model is ceased to determine the classification of image information;When the classification of image information is pre-set categories, target is marked
Vehicle is simultaneously reported to monitoring system.
In this embodiment, by real-time detection video to be detected, specifically, video to be detected can be taken the photograph for traffic monitoring
The video that the camera of the scene such as the monitor video obtained in real time as head or building site, stockyard setting is shot with video-corder, it is to be checked when detecting
When surveying in video comprising target vehicle, the image information of the target vehicle is input to state-detection model, passes through state-detection
The classification of model inspection image information;When the classification that the result of state-detection model output is the image information is pre-set categories
When, target vehicle corresponding to the image information is labeled as abnormal vehicle, and report to monitoring system.The present invention is applied to mention
The technical solution of confession is confirmed whether target vehicle occur by real-time detection video to be detected, occurs in video to be detected
When target vehicle, the image information of target vehicle is obtained, the classification of image information is determined by state-detection model, and then determine
Whether target vehicle is abnormality, marks the target vehicle of abnormality and reports to monitoring system, and then can be accurately
The state of the target vehicles such as dump truck is detected, and marks the target vehicle for being in abnormality, more intelligent, application range is provided
Wider array of monitoring of environment scheme.
In one embodiment of the invention, it is preferable that processor 804 is also used to: acquisition monitor video is regarded by monitoring
Frequency obtains the history image information of target vehicle;Sort command is received, is classified according to sort command to history image information,
To form training dataset;By training dataset training pre-training model to obtain state-detection model.
In this embodiment, by largely acquiring different location, Traffic Surveillance Video in different time periods, analysis monitoring view
Frequency and therefrom obtain it includes target vehicle history image message sample, according to sort command to history image message sample
Classify, obtains the history image message sample collection of various different times, different location and different conditions, and form training
Data set;It chooses general using the convolutional neural networks model that the training of mass transportation vehicle data is got as pre-training mould
Type, by training dataset training pre-training model can be obtained discrimination more accurately, the state-detection with more high robust
Model, to realize to the accurate detection of target vehicle state and label.
In one embodiment of the invention, it is preferable that processor 804 is also used to: when sort command indicates target vehicle
When for non-loading, image information is classified as first kind image information;When sort command instruction target vehicle is that can not determine to be
When no loading, image information is classified as the second class image information;When sort command instruction target vehicle is loading, and cargo
When capped, image information is classified as third class image information;When sort command instruction target vehicle is loading, and cargo
When uncovered, image information is classified as the 4th class image information;Wherein, it is classified as the image information of the 4th class image information
Classification be pre-set categories.
In this embodiment, according to target vehicle in history image information, i.e. the loading state of dump truck is by history image
Information be classified as the non-loading of dump truck, dump truck loading and can not determine dump truck whether three kinds of loading, and further by mud
The case where head vehicle loading, is specifically classified as dump truck loading, and cargo is capped and dump truck loading, and the goods
Object is uncovered, obtains specific 4 kinds of class categories, and specifically by the classification setting of history image information are as follows: First Kind Graph picture
Information, i.e. dump truck are loading;Second class image information, i.e., can not determine dump truck whether loading;Third class image information, i.e.,
Dump truck loading, and cargo is capped;And the 4th class image information, i.e. dump truck loading, and the cargo is not coated
Lid, and using sorted history image information as training dataset.Wherein, setting is classified as the history of the 4th class image information
The classification of image information is pre-set categories.By by target vehicle, i.e., dump truck state according to loading, non-loading, can not determine
Loading state is divided into 3 classes, and whether cargo is capped by loading state when then according to monitoring of environment demand further according to loading
Dump truck further segment, i.e., data and dump truck loading state are divided into 4 classes in data set manufacturing process, increase number
According to the harmony between classification, the accuracy rate that state-detection model identifies target vehicle state can be promoted.
In one embodiment of the invention, it is preferable that processor 804 is also used to: video to be detected is pre-processed,
To obtain the foreground picture of the image of each frame in video to be detected;Threshold segmentation is carried out to foreground picture, and carries out morphology and opens fortune
It calculates and normalized is to obtain binary image;Determine the contour area of target area in binary image;Pass through rectangle frame
Detection method judges whether contour area meets decision threshold range;When contour area meets decision threshold range, before determining
It include target vehicle in scape figure.
In this embodiment, it after getting video to be detected, is carried out first using open source computer vision library OpenCV
Auxiliary to complete the pretreatment to video to be detected, while establishing view using the method that mean filter and Gaussian Background model
The prospect and background separation of the image of frame each in video to be detected are obtained video to be detected by the image background model of frequency
In each frame image foreground picture;Also background subtraction can be used and obtain foreground picture;Then using the Ostu in OpenCV
Threshold Segmentation Algorithm carries out Threshold segmentation to foreground picture, while carrying out morphology opening operation to the image after segmentation, removes prospect
Noise spot in figure obtains binary image in further progress normalized;Pass through the contours extract function in OpenCV
Determine the contour area of target area in binary image;It is default to judge whether contour area meets by rectangle frame detection method
Decision threshold range;When contour area meets decision threshold range, can determine in foreground picture comprising target vehicle, this method
In detection oversize vehicle, testing result is accurate when such as dump truck, and the calculation amount needed is small, while detection efficiency with higher.
In one embodiment of the invention, it is preferable that processor 804 is also used to: obtaining the coordinate information of target area;
The image information of target vehicle is intercepted in the image of each frame by coordinate information.
In this embodiment, when determining in foreground picture includes target vehicle, the corresponding target area of target vehicle is obtained
Coordinate information, the image information of target vehicle is intercepted in primitive frame image by coordinate information, image information is inputted
State-detection model is realized by the state of state-detection model inspection target vehicle to the target vehicle for being in abnormality
Label and report.
In the embodiment of third aspect present invention, provides when determining in foreground picture comprising target vehicle, obtain mesh
The coordinate information for marking the corresponding target area of vehicle intercepts the image letter of target vehicle by coordinate information in primitive frame image
Breath by the state of state-detection model inspection target vehicle, realizes image information input state detection model to being in
The label of the target vehicle of abnormality and report.
In the description of the present invention, term " multiple " then refers to two or more, unless otherwise restricted clearly, term
The orientation or positional relationship of the instructions such as "upper", "lower" is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of retouching
It states the present invention and simplifies description, rather than the device or element of indication or suggestion meaning must have a particular orientation, with specific
Orientation construction and operation, therefore be not considered as limiting the invention;Term " connection ", " installation ", " fixation " etc. should all
It is interpreted broadly, for example, " connection " may be fixed connection or may be dismantle connection, or integral connection;It can be straight
Connect it is connected, can also be indirectly connected through an intermediary.It for the ordinary skill in the art, can be according to specific feelings
Condition understands the concrete meaning of above-mentioned term in the present invention.
In the description of the present invention, the description meaning of term " one embodiment ", " some embodiments ", " specific embodiment " etc.
Refer to that particular features, structures, materials, or characteristics described in conjunction with this embodiment or example are contained at least one implementation of the invention
In example or example.In the present invention, schematic expression of the above terms are not necessarily referring to identical embodiment or example.And
And the particular features, structures, materials, or characteristics of description can be in any one or more of the embodiments or examples with suitable
Mode combines.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (11)
1. a kind of status indication method of target vehicle characterized by comprising
Real-time detection video to be detected;
When in video to be detected including target vehicle, by the image information input state detection model of the target vehicle with true
Determine the classification of described image information;
When the classification of described image information is pre-set categories, marks the target vehicle and reported to monitoring system.
2. the status indication method of target vehicle according to claim 1, which is characterized in that be checked in the real-time detection
Before the step of surveying video, further includes:
Monitor video is acquired, the history image information of the target vehicle is obtained by the monitor video;
Sort command is received, is classified according to the sort command to the history image information, to form training dataset;
By training dataset training pre-training model to obtain the state-detection model.
3. the status indication method of target vehicle according to claim 2, which is characterized in that described to be ordered according to the classification
It the step of order classifies to history image information, specifically includes:
When the sort command indicates that the target vehicle is non-loading, the history image information is classified as First Kind Graph
As information;
When the sort command indicates that the target vehicle is that can not determine whether loading, the history image information is classified
For the second class image information;
When the sort command indicates that the target vehicle is loading, and cargo is capped, by the history image information
It is classified as third class image information;
When the sort command indicates that the target vehicle is loading, and the cargo is uncovered, by the history figure
As information is classified as the 4th class image information;
Wherein, the classification for being classified as the history image information of the 4th class image information is the pre-set categories.
4. the status indication method of target vehicle according to any one of claim 1 to 3, which is characterized in that the reality
When the step of detecting video to be detected, specifically include:
The video to be detected is pre-processed, to obtain the foreground picture of the image of each frame in the video to be detected;
Threshold segmentation is carried out to the foreground picture, and carries out morphology opening operation and normalized to obtain binary image;
Determine the contour area of target area in the binary image;
Judge whether the contour area meets decision threshold range by rectangle frame detection method;
When the contour area meets the decision threshold range, determine in the foreground picture comprising the target vehicle.
5. the status indication method of target vehicle according to claim 4, which is characterized in that when including in the foreground picture
When the target vehicle, the method also includes:
Obtain the coordinate information of the target area;
The described image information of the target vehicle is intercepted in the image of each frame by the coordinate information.
6. a kind of status indication system of target vehicle characterized by comprising
Memory, for storing computer program;
Processor, for execute the computer program with:
Real-time detection video to be detected;
When in video to be detected including target vehicle, by the image information input state detection model of the target vehicle with true
Determine the classification of described image information;
When the classification of described image information is pre-set categories, marks the target vehicle and reported to monitoring system.
7. the status indication system of target vehicle according to claim 6, which is characterized in that the processor is also used to:
Monitor video is acquired, the history image information of the target vehicle is obtained by the monitor video;
Sort command is received, is classified according to the sort command to the history image information, to form training dataset;
By training dataset training pre-training model to obtain the state-detection model.
8. the status indication system of target vehicle according to claim 7, which is characterized in that the processor is also used to:
When the sort command indicates that the target vehicle is non-loading, the history image information is classified as First Kind Graph
As information;
When the sort command indicates that the target vehicle is that can not determine whether loading, the history image information is classified
For the second class image information;
When the sort command indicates that the target vehicle is loading, and cargo is capped, by the history image information
It is classified as third class image information;
When the sort command indicates that the target vehicle is loading, and the cargo is uncovered, by the history figure
As information is classified as the 4th class image information;
Wherein, the classification for being classified as the history image information of the 4th class image information is the pre-set categories.
9. the status indication system of the target vehicle according to any one of claim 6 to 8, which is characterized in that the place
Reason device is also used to:
The video to be detected is pre-processed, to obtain the foreground picture of the image of each frame in the video to be detected;
Threshold segmentation is carried out to the foreground picture, and carries out morphology opening operation and normalized to obtain binary image;
Determine the contour area of target area in the binary image;
Judge whether the contour area meets decision threshold range by rectangle frame detection method;
When the contour area meets the decision threshold range, determine in the foreground picture comprising the target vehicle.
10. the status indication system of target vehicle according to claim 9, which is characterized in that the processor is also used to:
Obtain the coordinate information of the target area;
The described image information of the target vehicle is intercepted in the image of each frame by the coordinate information.
11. a kind of computer readable storage medium, is stored thereon with computer program, it is characterised in that: the computer program
The status indication method of the target vehicle as described in any one of claims 1 to 5 is realized when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910017244.3A CN109815856A (en) | 2019-01-08 | 2019-01-08 | Status indication method, system and the computer readable storage medium of target vehicle |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910017244.3A CN109815856A (en) | 2019-01-08 | 2019-01-08 | Status indication method, system and the computer readable storage medium of target vehicle |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109815856A true CN109815856A (en) | 2019-05-28 |
Family
ID=66604205
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910017244.3A Pending CN109815856A (en) | 2019-01-08 | 2019-01-08 | Status indication method, system and the computer readable storage medium of target vehicle |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109815856A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110324583A (en) * | 2019-07-15 | 2019-10-11 | 深圳中兴网信科技有限公司 | A kind of video monitoring method, video monitoring apparatus and computer readable storage medium |
CN110378258A (en) * | 2019-07-04 | 2019-10-25 | 上海眼控科技股份有限公司 | A kind of vehicle seat information detecting method and equipment based on image |
CN110929589A (en) * | 2019-10-31 | 2020-03-27 | 浙江大华技术股份有限公司 | Method, device, computer device and storage medium for vehicle feature recognition |
CN111160270A (en) * | 2019-12-31 | 2020-05-15 | 中铁大桥科学研究院有限公司 | Bridge monitoring method based on intelligent video identification |
CN111209923A (en) * | 2020-04-23 | 2020-05-29 | 北京慧智数据科技有限公司 | Deep learning technology-based muck truck cover or uncover identification method |
CN112604841A (en) * | 2020-12-10 | 2021-04-06 | 中国联合网络通信集团有限公司 | Vehicle spraying detection method, device, equipment and storage medium |
CN112822458A (en) * | 2021-01-12 | 2021-05-18 | 南京易自助网络科技有限公司 | Video monitoring method and terminal for car washing process |
CN113378629A (en) * | 2021-04-27 | 2021-09-10 | 阿里云计算有限公司 | Method and device for detecting abnormal vehicle in smoke discharge |
CN113688750A (en) * | 2021-08-27 | 2021-11-23 | 上海高德威智能交通系统有限公司 | Carriage state detection method and device |
CN116030418A (en) * | 2023-02-14 | 2023-04-28 | 北京建工集团有限责任公司 | Automobile lifting line state monitoring system and method |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN202563526U (en) * | 2012-03-22 | 2012-11-28 | 北京尚易德科技有限公司 | Transportation vehicle detection and recognition system based on video |
CN104915650A (en) * | 2015-06-05 | 2015-09-16 | 福建天晴数码有限公司 | Vehicle cargo-carrying state identification method and device |
CN105930833A (en) * | 2016-05-19 | 2016-09-07 | 重庆邮电大学 | Vehicle tracking and segmenting method based on video monitoring |
CN206162709U (en) * | 2016-08-23 | 2017-05-10 | 吴晓栋 | Fork truck monitored control system |
CN106875424A (en) * | 2017-01-16 | 2017-06-20 | 西北工业大学 | A kind of urban environment driving vehicle Activity recognition method based on machine vision |
CN107767399A (en) * | 2016-08-23 | 2018-03-06 | 吴晓栋 | Fork truck monitoring method and monitoring system |
CN108765943A (en) * | 2018-05-30 | 2018-11-06 | 深圳市城市公共安全技术研究院有限公司 | Intelligent vehicle monitoring method, monitoring system and server |
-
2019
- 2019-01-08 CN CN201910017244.3A patent/CN109815856A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN202563526U (en) * | 2012-03-22 | 2012-11-28 | 北京尚易德科技有限公司 | Transportation vehicle detection and recognition system based on video |
CN104915650A (en) * | 2015-06-05 | 2015-09-16 | 福建天晴数码有限公司 | Vehicle cargo-carrying state identification method and device |
CN105930833A (en) * | 2016-05-19 | 2016-09-07 | 重庆邮电大学 | Vehicle tracking and segmenting method based on video monitoring |
CN206162709U (en) * | 2016-08-23 | 2017-05-10 | 吴晓栋 | Fork truck monitored control system |
CN107767399A (en) * | 2016-08-23 | 2018-03-06 | 吴晓栋 | Fork truck monitoring method and monitoring system |
CN106875424A (en) * | 2017-01-16 | 2017-06-20 | 西北工业大学 | A kind of urban environment driving vehicle Activity recognition method based on machine vision |
CN108765943A (en) * | 2018-05-30 | 2018-11-06 | 深圳市城市公共安全技术研究院有限公司 | Intelligent vehicle monitoring method, monitoring system and server |
Non-Patent Citations (2)
Title |
---|
吴亚斌;陈淑荣;: "一种基于监控视频的车型识别和车流量检测算法" * |
隋修武;田松;余保付;葛辉;: "交通场景中的实时多目标自动跟踪算法" * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110378258A (en) * | 2019-07-04 | 2019-10-25 | 上海眼控科技股份有限公司 | A kind of vehicle seat information detecting method and equipment based on image |
CN110378258B (en) * | 2019-07-04 | 2021-02-09 | 上海眼控科技股份有限公司 | Image-based vehicle seat information detection method and device |
CN110324583A (en) * | 2019-07-15 | 2019-10-11 | 深圳中兴网信科技有限公司 | A kind of video monitoring method, video monitoring apparatus and computer readable storage medium |
CN110929589B (en) * | 2019-10-31 | 2023-07-07 | 浙江大华技术股份有限公司 | Method, apparatus, computer apparatus and storage medium for identifying vehicle characteristics |
CN110929589A (en) * | 2019-10-31 | 2020-03-27 | 浙江大华技术股份有限公司 | Method, device, computer device and storage medium for vehicle feature recognition |
CN111160270A (en) * | 2019-12-31 | 2020-05-15 | 中铁大桥科学研究院有限公司 | Bridge monitoring method based on intelligent video identification |
CN111160270B (en) * | 2019-12-31 | 2023-09-01 | 中铁大桥科学研究院有限公司 | Bridge monitoring method based on intelligent video recognition |
CN111209923A (en) * | 2020-04-23 | 2020-05-29 | 北京慧智数据科技有限公司 | Deep learning technology-based muck truck cover or uncover identification method |
CN112604841A (en) * | 2020-12-10 | 2021-04-06 | 中国联合网络通信集团有限公司 | Vehicle spraying detection method, device, equipment and storage medium |
CN112604841B (en) * | 2020-12-10 | 2022-02-15 | 中国联合网络通信集团有限公司 | Vehicle spraying detection method, device, equipment and storage medium |
CN112822458A (en) * | 2021-01-12 | 2021-05-18 | 南京易自助网络科技有限公司 | Video monitoring method and terminal for car washing process |
CN112822458B (en) * | 2021-01-12 | 2022-05-13 | 南京易自助网络科技有限公司 | Video monitoring method and terminal for car washing process |
CN113378629A (en) * | 2021-04-27 | 2021-09-10 | 阿里云计算有限公司 | Method and device for detecting abnormal vehicle in smoke discharge |
CN113688750A (en) * | 2021-08-27 | 2021-11-23 | 上海高德威智能交通系统有限公司 | Carriage state detection method and device |
CN116030418A (en) * | 2023-02-14 | 2023-04-28 | 北京建工集团有限责任公司 | Automobile lifting line state monitoring system and method |
CN116030418B (en) * | 2023-02-14 | 2023-09-12 | 北京建工集团有限责任公司 | Automobile lifting line state monitoring system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109815856A (en) | Status indication method, system and the computer readable storage medium of target vehicle | |
CN111292305B (en) | Improved YOLO-V3 metal processing surface defect detection method | |
CN109816024B (en) | Real-time vehicle logo detection method based on multi-scale feature fusion and DCNN | |
Akagic et al. | Pavement crack detection using Otsu thresholding for image segmentation | |
Hsieh et al. | Automatic traffic surveillance system for vehicle tracking and classification | |
CN103593670B (en) | A kind of copper plate/strip detection method of surface flaw based on online limit of sequence learning machine | |
CN111325713A (en) | Wood defect detection method, system and storage medium based on neural network | |
CN109977808A (en) | A kind of wafer surface defects mode detection and analysis method | |
Piccioli et al. | Robust road sign detection and recognition from image sequences | |
CN107729818A (en) | A kind of multiple features fusion vehicle recognition methods again based on deep learning | |
CN112308826B (en) | Bridge structure surface defect detection method based on convolutional neural network | |
CN105574550A (en) | Vehicle identification method and device | |
CN102998316B (en) | Transparent liquid impurity detection system and detection method thereof | |
Zhou et al. | Exploring faster RCNN for fabric defect detection | |
CN112465706A (en) | Automatic gate container residual inspection method | |
CN107194393A (en) | A kind of method and device for detecting Provisional Number Plate | |
CN111126393A (en) | Vehicle appearance refitting judgment method and device, computer equipment and storage medium | |
Ying et al. | A novel shadow-free feature extractor for real-time road detection | |
CN110516691A (en) | A kind of Vehicular exhaust detection method and device | |
CN114049624A (en) | Intelligent detection method and system for ship cabin based on machine vision | |
CN115830359A (en) | Workpiece identification and counting method based on target detection and template matching in complex scene | |
CN111369526A (en) | Multi-type old bridge crack identification method based on semi-supervised deep learning | |
CN113111875A (en) | Seamless steel rail weld defect identification device and method based on deep learning | |
CN112881412A (en) | Method for detecting non-metal foreign bodies in scrap steel products | |
Ghahremannezhad et al. | Automatic road detection in traffic videos |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190528 |
|
WD01 | Invention patent application deemed withdrawn after publication |