CN107766835A - traffic safety detection method and device - Google Patents
traffic safety detection method and device Download PDFInfo
- Publication number
- CN107766835A CN107766835A CN201711080164.XA CN201711080164A CN107766835A CN 107766835 A CN107766835 A CN 107766835A CN 201711080164 A CN201711080164 A CN 201711080164A CN 107766835 A CN107766835 A CN 107766835A
- Authority
- CN
- China
- Prior art keywords
- automobile
- image
- driver
- facial image
- camera
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60Q—ARRANGEMENT OF SIGNALLING OR LIGHTING DEVICES, THE MOUNTING OR SUPPORTING THEREOF OR CIRCUITS THEREFOR, FOR VEHICLES IN GENERAL
- B60Q9/00—Arrangement or adaptation of signal devices not provided for in one of main groups B60Q1/00 - B60Q7/00, e.g. haptic signalling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
Abstract
The present invention provides a kind of traffic safety detection method and device, and applied to the automobile for being provided with camera, the camera is used to gather the image information in the automobile.Method includes：When electric on detecting automobile, camera is opened；The video information of camera collection is obtained, the every two field picture included according to the first algorithm to video information is analyzed, and identifies the facial image in every two field picture；Quantity of the statistics per the facial image in two field picture, and when the face quantity in any one two field picture is more than the first preset value, the two field picture is labeled as target image；The quantity for the target image that statistics video information includes, and when the quantity of target image reaches the second preset value, every frame target image is analyzed according to the second algorithm, re-recognizes the facial image in every frame target image；When the quantity for the facial image being identified in any one frame target image is more than the first preset value, automobile overload is determined, No starting automobile is simultaneously alarmed.
Description
Technical field
The present invention relates to safety monitoring technology field, in particular to a kind of traffic safety detection method and device.
Background technology
With the continuous development of scientific technology, thus people also begin to concern while the facility of automobile belt is enjoyed
Caused driving safety problem.Bad habit, which is driven, according to statistics of traffic accidents data display overload etc. occupies phase in traffic safety hidden danger
When big ratio.Sometimes know that overload has risk perfectly well, but some users still have idea of leaving things to chance, thus how applicating technology hand
Section carries out timely early warning to violations of rules and regulations such as overload, and try to forestall traffic accidents generation, it appears particularly important.
The content of the invention
In view of this, it is an object of the invention to provide a kind of traffic safety detection method, applied to being provided with camera
Automobile, the camera is used to gather image information in the automobile；Methods described includes：
When electric on detecting the automobile, the camera is opened；
The video information that the camera collects is obtained, the every frame figure included according to the first algorithm to the video information
As being analyzed, the facial image in every two field picture is identified；
The quantity of facial image in the every two field picture of statistics, and the face quantity in any one two field picture is pre- more than first
If during value, the two field picture is labeled as target image；
Count the quantity for the target image that the video information includes, and to reach second default in the quantity of target image
During value, every frame target image is analyzed according to the second algorithm, re-recognizes the facial image in every frame target image；
When the quantity for the facial image being identified in any one frame target image is more than first preset value, it is determined that
The automobile overload, automobile described in No starting are simultaneously alarmed.
Another object of the present invention is to provide a kind of traffic safety detection method, applied to the vapour for being provided with camera
Car, the camera are used to gather the image information in the automobile；Methods described includes：
When electric on detecting automobile, the camera is opened, and obtains an at least two field picture for the camera collection,
The facial image in an at least two field picture is identified according to the first algorithm and/or the second algorithm；
The relative position relation of each facial image to being identified in an at least two field picture is analyzed, and obtains institute
State the facial image of the front passenger of automobile；
The first age grader that the facial image input of the front passenger is trained, and according to first age
The classification results of grader obtain the first age group belonging to the front passenger；
Judge whether include first age group in default first object age group；
If including first age group, Gabor characteristic and office are extracted from the facial image of the front passenger
Portion's binary pattern feature, and the second age point that the Gabor characteristic extracted and the input of local binary patterns feature are trained
Class device, the second age group according to belonging to the classification results of the second age grader obtain the front passenger, described
One target age group includes second age group；
Judge whether include second age group in default second target age group, if so, then described in No starting
Automobile is simultaneously alarmed.
Another object of the present invention is to provide a kind of traffic safety detection method, applied to the vapour for being provided with camera
Car, the camera are used to gather the image information in the automobile；Methods described includes：
After automobile starting, the picture frame that the camera currently collects is obtained, and identify in described image frame
Facial image；
The relative position relation of each facial image in described image frame is analyzed, obtains the driver of the automobile
Facial image；
The gesture recognition model that the facial image input of the driver is trained, obtains the head court of the driver
To；
Judge whether the driver is in fatigue driving state according to the expression posture of the driver and head direction,
If being in fatigue driving state, alarmed.
Another object of the present invention is to provide a kind of traffic safety detection means, applied to the vapour for being provided with camera
Car, the camera are used to gather the image information in the automobile；Described device includes：
Opening module, for when electric on detecting the automobile, opening the camera；
Identification module, the video information collected for obtaining the camera, the video is believed according to the first algorithm
Every two field picture that ceasing includes is analyzed, and identifies the facial image in every two field picture；
First statistical module, for counting the quantity of the facial image in every two field picture, and in any one two field picture
When face quantity is more than the first preset value, the two field picture is labeled as target image；
Second statistical module, the quantity of the target image included for counting the video information, and in target image
Quantity when reaching the second preset value, every frame target image is analyzed according to the second algorithm, re-recognizes every frame target figure
Facial image as in；
Alarm module, it is more than described first for the quantity when the facial image being identified in any one frame target image
During preset value, the automobile overload is determined, automobile described in No starting is simultaneously alarmed.
The embodiment of the present invention provides a kind of traffic safety detection method and device, opens camera to vapour when electric on automobile
In-car image information is acquired, and identifies facial image from the every two field picture collected by the first algorithm, by institute
The picture frame for being more than the first preset value including face quantity is labeled as target image, and to reach second pre- in the quantity of target image
If during value, the facial image in each target image is re-recognized using the second algorithm, the face in any one frame target image
When image is more than the first preset value, the automobile overload is determined, the No starting automobile is simultaneously alarmed.In this way, on the one hand it may insure
To whether the accuracy of the judged result to overload, on the other hand, early warning can be carried out to driver and passenger in time, eliminated lucky
Good fortune psychology, reaching prevents because overload produces the effect of traffic accident.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by embodiment it is required use it is attached
Figure is briefly described, it will be appreciated that the following drawings illustrate only certain embodiments of the present invention, therefore be not construed as pair
The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to this
A little accompanying drawings obtain other related accompanying drawings.
Fig. 1 is a kind of block diagram for automobile that first embodiment of the invention provides；
Fig. 2 is a kind of schematic flow sheet for traffic safety detection method that first embodiment of the invention provides；
Fig. 3 is the another schematic flow sheet for the traffic safety detection method that first embodiment of the invention provides；
Fig. 4 is the another schematic flow sheet for the traffic safety detection method that first embodiment of the invention provides；
Fig. 5 is the another schematic flow sheet for the traffic safety detection method that first embodiment of the invention provides；
Fig. 6 is the another schematic flow sheet for the traffic safety detection method that first embodiment of the invention provides；
Fig. 7 is the another schematic flow sheet for the traffic safety detection method that first embodiment of the invention provides；
Fig. 8 is a kind of functional block diagram for traffic safety detection means that first embodiment of the invention provides.
Icon：100- automobiles；110- memories；120- processors；130- detection units；140- alarm units；200- rows
Car safety detection device；210- opening modules；220- identification modules；The statistical modules of 230- first；The statistical modules of 240- second；
250- alarm modules；260- violation passenger detection modules；270- fatigue driving detection modules；280- dangerous drivings detect mould
Block.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, rather than whole embodiments.The present invention implementation being generally described and illustrated herein in the accompanying drawings
The component of example can be configured to arrange and design with a variety of.
Therefore, below the detailed description of the embodiments of the invention to providing in the accompanying drawings be not intended to limit it is claimed
The scope of the present invention, but be merely representative of the present invention selected embodiment.It is common based on the embodiment in the present invention, this area
The every other embodiment that technical staff is obtained under the premise of creative work is not made, belong to the model that the present invention protects
Enclose.
It should be noted that：Similar label and letter represents similar terms in following accompanying drawing, therefore, once a certain Xiang Yi
It is defined, then it further need not be defined and explained in subsequent accompanying drawing in individual accompanying drawing.
As shown in figure 1, being a kind of block diagram of automobile 100 provided in an embodiment of the present invention, the automobile 100 includes
Traffic safety detection means 200, memory 110, processor 120, detection unit 130 and alarm unit 140.
The memory 110, processor 120, detection unit 130 and 140 each element of alarm unit between each other directly or
It is electrically connected with indirectly, to realize the interaction of data.For example, these elements can pass through one or more communication bus between each other
Or signal wire is realized and is electrically connected with.The traffic safety detection means 200 include it is at least one can be with software or firmware
(firmware) form is stored in the software function module in the memory 110 of the automobile 100.The processor 120 is used
When execute instruction is being received, the executable module stored in the memory 110 is performed.
In the present embodiment, the memory 110 may be, but not limited to, random access memory (Random
Access Memory, RAM), read-only storage (Read Only Memory, ROM), programmable read only memory
(Programmable Read-Only Memory, PROM), erasable read-only memory (Erasable Programmable
Read-Only Memory, EPROM), electricallyerasable ROM (EEROM) (Electric Erasable Programmable
Read-Only Memory, EEPROM) etc..
The processor 120 can be a kind of IC chip, have the disposal ability of signal.Above-mentioned processor can
To be general processor, including central processing unit (Central Processing Unit, CPU), network processing unit (Network
Processor, NP) etc.；Can also be digital signal processor (DSP)), application specific integrated circuit (ASIC), field programmable gate
Array (FPGA) either other PLDs, discrete gate or transistor logic, discrete hardware components.Can be real
Now or perform the embodiment of the present invention in disclosed each method, step and logic diagram.General processor can be microprocessor
Device or the processor can also be any conventional processors etc..
The detection unit 130 includes imaging sensor, such as camera.The camera is installed on the automobile 100, uses
Image information in the collection automobile 100.In the present embodiment, in order to ensure the camera can be adopted more precisely
Collect the facial image in automobile 100, the camera can be installed on to the front of the automobile 100, make the camera
Camera lens deviates from the windshield of the automobile 100.
The alarm unit 140 can be voice announcer, buzzer, flash lamp or audible-visual annunciator etc., the present embodiment
It is without limitation.
It should be appreciated that being only the structural representation of the automobile 100 shown in Fig. 1, the automobile 100 can also have than Fig. 1
Shown more or less components, or have and configuration entirely different shown in Fig. 1.It should be noted that group shown in Fig. 1
Part can use hardware, software or its be implemented in combination with, the present embodiment is without limitation.
As shown in Fig. 2 being a kind of schematic flow sheet of traffic safety detection method provided in an embodiment of the present invention, it is applied to
Automobile 100 shown in Fig. 1.Methods described is elaborated with reference to Fig. 2.
Step S110, when electric on detecting the automobile 100, open the camera.
In detail, when the ignition system for detecting the automobile 100 starts, the camera is opened, with to the vapour
Image information in car 100 is acquired.
Step S120, the video information that the camera collects is obtained, according to the first algorithm to the video information bag
The every two field picture included is analyzed, and identifies the facial image in every two field picture.
Wherein, first algorithm can be the method for detecting human face based on AdaBoost.In detail, first algorithm
Can be by calling cvHaarDetectObjects () function in OpenCV to realize.
Step S130, the quantity of the facial image in counting per two field picture, and the face quantity in any one two field picture
During more than the first preset value, the two field picture is labeled as target image.
Wherein, first preset value can specifically be set according to the passengers quantity that the automobile 100 can carry, for example, one
As car first preset value can be arranged to 5.In this situation, when the face quantity in any one two field picture is more than 5
When, the two field picture is labeled as target image.
In the present embodiment, target image refers to that the quantity of facial image exceedes the picture frame that can carry passengers quantity.
Step S140, the quantity for the target image that the video information includes is counted, and reached in the quantity of target image
During to the second preset value, every frame target image is analyzed according to the second algorithm, re-recognizes the people in every frame target image
Face image.
In the present embodiment, second preset value can flexibly be set according to actual conditions, for example, described second
Preset value can be 2.In this situation, when 2 frame target image in the video information be present, then second is further used
Algorithm is analyzed every frame target image.
In other words, when the image that more than two frames or two frames be present in the video information has overload condition, using second
Algorithm carries out second to the picture frame with overload condition and verified.
Alternatively, in the present embodiment, second algorithm can be the Faster-RCNN algorithms based on deep learning.
In detail, Faster-RCNN network models can be carried out in the following way first to be trained：
(1) using model initialization RPN (Region Proposal Networks) net of the pre-training on ImageNet
Network parameter, RPN networks are finely tuned using existing face database；
(2) candidate region is extracted from the RPN networks, and Fast- is trained using the candidate region extracted
RCNN parameters；
(3) RPN networks are trained again using Fast RCNN on the basis of (2), it is micro- in the case of fixed convolutional layer
RPN networks are adjusted, while keep inclusion layer not train；
(4) in the case of the convolutional layer of the Fast RCNN in fixation (2), time is extracted in the RPN networks in (3)
Favored area, and Fast RCNN networks are finely adjusted based on the candidate region extracted, so as to obtain the Faster-RCNN
Network model.
During implementation, target image to be identified is inputted into the Faster-RCNN network models, the target figure can be exported
All people's face image and the positional information of each facial image as in.
Step S150, preset when the quantity for the facial image being identified in any one frame target image is more than described first
During value, determine that the automobile 100 overloads, automobile 100 described in No starting is simultaneously alarmed.
During implementation, the quantity of the facial image to being identified in each target image based on the second algorithm is counted, when appoint
When the quantity of the facial image being identified in a frame target image of anticipating is more than first preset value, the automobile 100 is determined
Overload.In the case of it is determined that the automobile 100 overloads, automobile 100 described in No starting is simultaneously alarmed.
Wherein, the mode alarmed can include exporting warning message by the alarm unit 140 of the automobile 100,
For example, when the alarm unit 140 is phonetic alarm, user's automobile 100 can be prompted to have overloaded, may be caused " XXX "
Accident, to eliminate the idea of leaving things to chance of passenger and driver, wherein " XXX " accident can determine according to actual count data.
Study and find through inventor, in actual applications, many times passenger and driver are the danger for knowing overload
, but can always have idea of leaving things to chance, fright and effect of contraction can be played to passenger and driver by above-mentioned alarm, so as to
The generation of overloading is avoided, and then is prevented because overload produces traffic accident.
In addition, the mode alarmed can also include sending warning message, the report to the communication apparatus of traffic control department
Alert information can include the identification information of the target image and automobile 100 detected, to take enforceable means to prevent
Overload.
Alternatively, in the present embodiment, methods described can also comprise the following steps：
When the camera opens failure, automobile 100 described in No starting；Or
When the camera is opened successfully, judge whether the camera is blocked, if the camera is blocked,
Automobile 100 described in No starting.
Discovery is carefully studied through inventor, sometimes driver or passenger, may in order to avoid being collected loaded information
It can select to block camera or manual-lock camera.By above-mentioned steps, can avoid the occurrence of this kind of.
Alternatively, in the present embodiment, judge that the step of whether camera is blocked can be real in the following way
It is existing：
Obtain at least two field picture that the camera collects, and an at least two field picture described in calculating and default automobile
The similarity of scene image in 100；
Judge whether the similarity reaches default similarity threshold, if not up to described similarity threshold, it is determined that
The camera is blocked.
During implementation, an at least frame is chosen in the image that the camera has collected and is used for whether differentiating the camera
It is blocked.By scene in an at least two field picture and the automobile 100 being stored in advance in the memory 110 of the automobile 100
Image is analyzed, the similarity of an at least two field picture and scene image in the automobile 100 described in calculating, when the phase
Similarity threshold set in advance is not up to like degree, shows that camera is blocked.Because in the situation that camera is not blocked
Under, image that camera collects should be substantially similar with the image that collects in advance.
Further, in order to ensure the accuracy of judgement, at least predeterminable area of a two field picture and institute can be detected
State the similarity of the predeterminable area of scene image in automobile 100.The predeterminable area refers to its in addition to seat location
His region.Because the passenger seated image with when not sitting passenger inherently has certain difference on seat, this region is excluded
The differentiation for carrying out image similarity after opening again can be more accurate.
Study and find through inventor, when the children of below 12 one full year of life (or less than 1.4 meters), or adult bosom are taken in copilot
Embrace baby and be sitting in dress circle etc., safety problem extremely easily occur.
For this problem, alternatively, as shown in figure 3, the facial image in an at least two field picture is identified (passes through
One algorithm or the second algorithm) after, methods described can also include step S160, step S170, step S180, step S190
And step S1100.
Step S160, the relative position relation of each facial image in an at least two field picture is analyzed, obtained
The facial image of the front passenger of the automobile 100.
Because the front passenger of the automobile 100 is different with the relative distance of the camera from rear passenger, cause institute
State in the image that camera collects, there can be relative position relation between front passenger and the facial image of rear passenger,
The front passenger of the automobile 100 can be extracted on the basis of conventional images treatment technology using this relative position relation
Facial image.
Step S170, the first age grader that the facial image input of the front passenger is trained, and according to institute
The classification results for stating the first age grader obtain the first age group belonging to the front passenger.
In the present embodiment, normalized can be done to the facial image of the front passenger, after normalized
Image be input to again in the first age grader trained.
Wherein, the first age grader can be built in the following way：
The facial image sample of teenager's (0-24 year), middle aged (25-54 year) and old (50-90 year) are collected, and is marked
Its age group information, AAM (the Active Appearance Model) method of recycling extract the AAM features of each face sample；
The AAM features of face sample age group information corresponding with its is input in SVMs (SVM) and carries out classifier training,
It can obtain the first age grader.
During implementation, AAM features can be extracted from the facial image of the front passenger, then the AAM extracted is special
Sign inputs the first age grader, to judge the age group belonging to the front passenger.
In the present embodiment, first age group include teenager's group, middle aged group and old group, the age bracket each organized
The grouping information of face sample with building the first age grader matches.
Step S180, judge whether include first age group in default first object age group.
In the present embodiment, the default first object age group can include teenager's group and old group.
Step S190, if including first age group, Gabor is extracted from the facial image of the front passenger
Feature and local binary patterns feature, and that the Gabor characteristic extracted and the input of local binary patterns feature are trained
Two character classification by age devices, the second age according to belonging to the classification results of the second age grader obtain the front passenger
Group, the first object age group include second age group.
In the present embodiment, when the first object age group includes first age group, in other words, before described
, it is necessary to further analyze the facial image of the front passenger when row passenger belongs to teenager's group or belongs to old group.
In the present embodiment, teenager can be organized and old group builds a character classification by age device respectively, to be classified.Its
In, by taking teenager's group as an example, the face sample of teenager's (0-25 year) is collected, then the sample being collected into is subdivided into children (0-
12 years old) and juvenile (13-25 year) two class, and mark the age group information of each subclass.
Then, extracted from each face sample Gabor characteristic and local binary patterns (Local Binary Patterns,
LBP) feature and PCA dimension-reduction treatment is carried out, and the Gabor characteristic after dimensionality reduction, LBP features and its corresponding age group information is defeated
Enter to supporting vector machine model to be trained to obtain and the second of children's group or juvenile group is divided into for the passenger for organizing teenager
Character classification by age device.
For old group, it also can in a comparable manner construct and the passenger of old age group is divided into person in middle and old age (50-70
Year) (70-90 year) the second age grader organized of group or Elder.
In the present embodiment, the second age grader can refer to the old group of grader classified,
It can refer to the grader classified to teenager's group.
Step S1100, judge whether include second age group in default second target age group, if so, then prohibiting
Only start the automobile 100 and alarmed.
In the present embodiment, default second target age group includes children's group and Elder group.During implementation, when
When front passenger belongs to children's group or Elder group, then judge that front-seat take of automobile 100 has offender, No starting institute
State automobile 100 and alarm.Wherein, the type of alarm mode alarmed in step S150 with overloading herein when is similar, herein
Repeat no more.
Alternatively, as shown in figure 4, in the present embodiment, in step S190 is performed, by the Gabor characteristic extracted and
After the second age grader for training of LBP features input, methods described can also include step S1110, step S1120 and
Step S1130.
Step S1110, according to the classification results of the second age grader judge the front passenger age whether
For the boundary value of two neighboring second age group.
Wherein, the boundary value of two neighboring second age group can be able to be set in advance how discrete with more than one
The set at age, the age in the set are the ages of the close boundary of two neighboring second age group.For example, children's group
Boundary value with teenager's group can be [11,12,13,14], the boundary value that person in middle and old age combine Elder group can be [69,70,
71,72]。
In the present embodiment, in order to ensure the accuracy of classification, judging that age of the front passenger is adjacent two
During the boundary value of individual second age group, the accurate age of the passenger can be further confirmed using the height of passenger.
Step S1120, if the boundary value of two neighboring second age group, then the face of the front passenger is obtained in institute
With the relative position relation of seat image in picture frame, and the body of the front passenger is calculated according to the relative position relation
It is high.
Because the people of age groups has different height, and the people of different heights is sitting in identical seat (front-row seats)
When upper and the relative position relation of seat also can be distinct, and thus, the face based on the front passenger is in place picture frame
In can analyze to obtain the height of the front passenger with the relative position relation of seat image.
Step S1130, when the height of the front passenger is less than predetermined threshold value, automobile 100 is gone forward side by side described in No starting
Row alarm.
In the present embodiment, the predetermined threshold value can be set according to statistics, and detected data according to history and entered
Row renewal, to lift the accuracy of judged result.
Wherein, type of alarm herein is similar with type of alarm during foregoing generation overloading, will not be repeated here.
Statistics of traffic accidents data are shown, during the traveling of automobile 100, fatigue driving or the use of mobile phone etc. is to make
Into the main reason for traffic accident.If being detected with regard to such phenomenon, and corresponding early warning is carried out, then can avoided traffic accident
Generation, so that it is guaranteed that traffic safety.
Alternatively, as shown in figure 5, after the automobile 100 startup, methods described can also include step S1140, step
S1150, step S1160, step S1170 and step S1180.
Step S1140, the picture frame that the camera currently collects is obtained, and identify the face figure of described image frame
Picture.
Alternatively, in the driving procedure of automobile 100, the image that the camera currently collects can periodically be obtained
Frame is detected, and the specific cycle can flexibly be set according to actual conditions.
Step S1150, the relative position relation of each facial image in described image frame is analyzed, obtains the vapour
The facial image of the driver of car 100.
Step S1160, the facial image of the driver is inputted into the Expression Recognition model based on GoogLeNet, obtained
The emotional state of the driver.
Wherein, the Expression Recognition model based on GoogLeNet can train acquisition in the following way：
Collect respectively with default emotional state (e.g., it is normal, speak, be absorbed in and tiredness etc.) face sample, and mark
The emotional state of each face sample；Above-mentioned face sample is trained using GoogLeNet deep neural networks, obtains face
Human facial feature extraction model.
During implementation, facial image (that is, the facial image of driver) to be identified is input to the GoogLeNet trained
Model, to extract the human face expression characteristic vector in the facial image, then the human face expression characteristic vector extracted inputted
Classified to SVMs according to the default emotional state, you can obtain corresponding to the facial image to be identified
Emotional state.
Step S1170, the gesture recognition model that the facial image input of the driver is trained, obtains the driving
The head direction of member.
Alternatively, in the present embodiment, human face posture can be divided into left avertence turns, right avertence turns, positive and bow four kinds,
The training of gesture recognition model is carried out according still further to four kinds of human face postures.Wherein, the training method of the gesture recognition model with
The training method of the Expression Recognition model is similar.
Step S1180, judge the driver whether in tired according to the emotional state of the driver and head direction
Labor driving condition, if being in fatigue driving state, alarmed.
Alternatively, in the present embodiment, in the step S1180, according to the emotional state of the driver and head court
To judging whether the driver may include steps of in fatigue driving state：
When the emotional state of the driver is tired state, and the head of the driver is oriented when bowing, it is determined that
The driver is in fatigue driving state.
Alternatively, as shown in fig. 6, after the facial image of driver of the automobile 100 is obtained, methods described may be used also
With including step S1190 and step S1200.
Step S1190, Gabor characteristic and histograms of oriented gradients feature are extracted from the facial image of the driver,
And the Image Classifier for training the Gabor characteristic extracted and the input of histograms of oriented gradients feature, to be driven described in identification
Whether the person's of sailing has mobile phone in one's ear.
Alternatively, in the present embodiment, described image grader can train acquisition in the following way：
Collect face has mobile phone and without the class facial image sample of mobile phone two in one's ear, and marks the class of each facial image sample
Other information；Gabor characteristic and histograms of oriented gradients (Histogram of are extracted from each facial image sample
Oriented Gridients, HOG) feature, and the feature extracted and its classification information are input in SVMs
Row training, you can obtain described image grader.
During implementation, Gabor characteristic and HOG features are extracted from the facial image of driver to be identified, and will propose
Gabor characteristic and HOG features be input to described image grader, you can whether that identifies the driver has hand in one's ear
Machine.
Step S1200, if the driver's has a mobile phone in one's ear, and the current emotional state of the driver is speaks,
Then determine that the driver is in dangerous driving state, then alarmed.
The driver have mobile phone in one's ear in the case of, if identifying the current emotional state of the driver to say
Words, it is determined that the driver makes a phone call when driving, in the hole, then can be alarmed, the mode of alarm is specific
Referring to the type of alarm in the above during generation overloading.
Study and find through inventor, when driver's long-time is not towards front, the probability that traffic accident occurs is larger.Cause
And the multiple image in preset duration is identified for the present embodiment selection, to judge driver whether in dangerous driving.
Alternatively, in order to ensure the accuracy of detection, as shown in fig. 7, after the automobile 100 startup, methods described is also
Step S1210, step S1220, step S1230 and step S1240 can be included.
Step S1210, the multiple image that the camera collects in preset duration is obtained, and identify every two field picture
In facial image.
In the present embodiment, the preset duration can flexibly be set according to actual conditions, for example, 1 minute~3 points
Clock.
Step S1220, the relative position relation of the facial image in every two field picture is analyzed, extract the automobile
Facial image of 100 driver in the two field picture, obtain multiple facial images of the driver.
Wherein, multiple facial images of the driver refer to, the driver is collected in the preset duration
Multiple facial images.
Step S1230, the gesture recognition model for respectively training the input of the multiple facial image, obtains the driving
Multiple head directions of member.
During implementation, the head direction of driver described in the multiple facial image is judged respectively, obtains the multiple head
Portion's direction.
Step S1240, if the multiple head is towards immediately ahead of being not, it is determined that the driver is in fatigue driving
State, and alarmed.
If it was found that the multiple head direction is not front, can determine in the preset duration, it is described to drive
The person of sailing may thereby determine that the driver is not driving attentively, so as to judge that it is in dangerous driving not towards front
State, then it can be alarmed in the manner previously described.
As shown in figure 8, it is a kind of functional block diagram of traffic safety detection means 200 provided in an embodiment of the present invention.
Described device includes opening module 210, identification module 220, the first statistical module 230, the second statistical module 240 and alarm module
250。
Wherein, the opening module 210 is used for when electric on detecting the automobile 100, opens the camera.
In the present embodiment, the description as described in the opening module 210 is specifically referred to the detailed of step S110 shown in Fig. 2
Thin description, that is, the step S110 can be performed by the opening module 210.
Alternatively, in the present embodiment, the opening module 210 can be also used for when the camera opens failure,
Automobile 100 described in No starting, and when the camera is opened successfully, judge whether the camera is blocked, if institute
State camera to be blocked, then automobile 100 described in No starting.
Alternatively, the opening module 210 judges the mode whether camera is blocked, and can include：
Obtain at least two field picture that the camera collects, and an at least two field picture described in calculating and default automobile
The similarity of scene image in 100；
Judge whether the similarity reaches default similarity threshold, if not up to described similarity threshold, it is determined that
The camera is blocked.
The identification module 220 is used to obtain the video information that the camera collects, according to the first algorithm to described
Every two field picture that video information includes is analyzed, and identifies the facial image in every two field picture.
In the present embodiment, the description as described in the identification module 220 is specifically referred to the detailed of step S120 shown in Fig. 2
Thin description, that is, the step S120 can be performed by the identification module 220.
First statistical module 230 is used for the quantity for counting the facial image in every two field picture, and in any one frame figure
When face quantity as in is more than the first preset value, the two field picture is labeled as target image.
In the present embodiment, the description as described in first statistical module 230 is specifically referred to step S130 shown in Fig. 2
Detailed description, that is, the step S130 can be performed by first statistical module 230.
Second statistical module 240 is used for the quantity for counting the target image that the video information includes, and in mesh
When the quantity of logo image reaches the second preset value, every frame target image is analyzed according to the second algorithm, re-recognizes every frame
Facial image in target image.
In the present embodiment, the description as described in second statistical module 240 is specifically referred to step S140 shown in Fig. 2
Detailed description, that is, the step S140 can be performed by second statistical module 240.
The quantity that the alarm module 250 is used to work as the facial image being identified in any one frame target image is more than institute
When stating the first preset value, determine that the automobile 100 overloads, automobile 100 described in No starting is simultaneously alarmed.
In the present embodiment, the description as described in the alarm module 250 is specifically referred to the detailed of step S150 shown in Fig. 2
Thin description, that is, the step S150 can be performed by the alarm module 250.
Alternatively, described device can also include violation passenger detection module 260.
The violation is taken detection module and is used for after the facial image in identifying an at least two field picture, to it is described at least
The relative position relation of each facial image in one two field picture is analyzed, and obtains the face of the front passenger of the automobile 100
Image；The first age grader that the facial image input of the front passenger is trained, and according to first age point
The classification results of class device obtain the first age group belonging to the front passenger；Judge in default first object age group whether
Including first age group；If including first age group, extracted from the facial image of the front passenger
Gabor characteristic and local binary patterns (LBP) feature, and that the Gabor characteristic extracted and the input of LBP features are trained
Two character classification by age devices, the second age point according to belonging to the classification results of the second age grader obtain the front passenger
Group, the first object age group include second age group；Judge whether include institute in default second target age group
The second age group is stated, if so, then automobile 100 described in No starting and being alarmed.
Alternatively, in the present embodiment, the violation passenger detection module 260 can be also used for extracting
After the second age grader that Gabor characteristic and the input of local binary patterns feature train, according to second age point
The classification results of class device judge the front passenger age whether be two neighboring second age group boundary value；If adjacent
The boundary value of two the second age groups, the then face for obtaining the front passenger are relative with seat image in the picture frame of place
Position relationship, and according to the height of the relative position relation calculating front passenger；It is less than in the height of the front passenger
During predetermined threshold value, automobile 100 described in No starting is simultaneously alarmed.
Alternatively, described device can also include fatigue driving detection module 270.
The fatigue detecting drives module and is used for after the startup of automobile 100, obtains the figure that the camera currently collects
As frame, and identify the facial image of described image frame；The relative position relation of each facial image in described image frame is entered
Row analysis, obtains the facial image of the driver of the automobile 100；The facial image input of the driver is based on
GoogLeNet Expression Recognition model, obtains the emotional state of the driver；The facial image of the driver is inputted into instruction
The gesture recognition model perfected, obtain the head direction of the driver；According to the emotional state of the driver and head court
To judging whether the driver is in fatigue driving state, if being in fatigue driving state, alarmed.
Alternatively, the fatigue driving module judges the driving according to the emotional state of the driver and head direction
Whether member is in the mode of fatigue driving state, can include：
When the emotional state of the driver is tired state, and the head of the driver is oriented when bowing, it is determined that
The driver is in fatigue driving state.
Alternatively, described device can also include dangerous driving detection module 280.
The dangerous driving detection module 280 is used for after the facial image of driver of the automobile 100 is obtained, from
Gabor characteristic and histograms of oriented gradients feature, and the Gabor characteristic that will be extracted are extracted in the facial image of the driver
And the Image Classifier that the input of histograms of oriented gradients feature trains, to identify whether the driver's has mobile phone in one's ear；
If the driver's has a mobile phone in one's ear, and the current emotional state of the driver is speaks, it is determined that at the driver
In dangerous driving state, then alarmed.
Alternatively, in the present embodiment, the dangerous driving detection module 280 is used for after the automobile 100 startup, obtains
The multiple image for taking the camera to be collected in preset duration, and identify the facial image in every two field picture；To every frame
The relative position relation of facial image in image is analyzed, and extracts the driver of the automobile 100 in the two field picture
Facial image, obtain multiple facial images of the driver；The appearance that the input of the multiple facial image is trained respectively
State identification model, obtain multiple head directions of the driver；If the multiple head is towards immediately ahead of being not, it is determined that
The driver is in dangerous driving state, and is alarmed.
Second embodiment
Second embodiment of the invention provides a kind of traffic safety detection method, applied to the automobile 100 shown in Fig. 1.It is described
Method comprise the following steps step S160 in S210 and first embodiment, step S170, step S180, step S190 and
Step S1100.
Step S210, when electric on detecting automobile 100, the camera is opened, and obtain the camera collection
An at least two field picture, the facial image in an at least two field picture is identified according to the first algorithm and/or the second algorithm.
Wherein, the first algorithm and the second algorithm are identical with the first algorithm referred in first embodiment and the second algorithm.
Alternatively, other steps of second embodiment of the invention can also include other steps in first embodiment of the invention
Suddenly.
By above-mentioned design, traffic safety can also detect and and alarm, to constrain driver and passenger.
3rd embodiment
Second embodiment of the invention provides a kind of traffic safety detection method, applied to the automobile 100 shown in Fig. 1.It is described
Method includes step S1140, step S1150, step S1160, step S1170 and the step of first embodiment of the invention
S1180。
Alternatively, methods described can also include the step S1190 and step S1200 in first embodiment of the invention.
, can also be by analyzing the facial image that camera collects, with to traffic safety by above-mentioned design
Detected, so as to avoid causing traffic accident.
In summary, traffic safety detection method and device provided in an embodiment of the present invention, open when electric on automobile 100
Camera is acquired to the image information in automobile 100, and is identified by the first algorithm from the every two field picture collected
Facial image, the picture frame that included face quantity is more than to the first preset value are labeled as target image, and in target image
When quantity reaches the second preset value, the facial image in each target image is re-recognized using the second algorithm, when any one frame mesh
When facial image in logo image is more than the first preset value, determine that the automobile 100 overloads, the No starting automobile 100 is simultaneously alarmed.
In this way, on the one hand may insure to whether the accuracy of the judged result to overload, on the other hand, can in time to driver and
Passenger carries out early warning, eliminates idea of leaving things to chance, and reaching prevents because overload produces the effect of traffic accident.
In addition, the above method and device can also detect to whether the front row of automobile 100 has taken offender,
Detect that front-seat take is alarmed when having children or man at an advanced age, can avoid occurring to cause serious injury during traffic accident.
In addition, the above method and device can also whether fatigue driving and dangerous driving be carried out to the driver of automobile 100
Detection, and alarmed when detecting driver tired driving or dangerous driving, avoid causing severe traffic accidents.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.Within the spirit and principles of the invention, that is made any repaiies
Change, equivalent substitution, improvement etc., should be included in the scope of the protection.
Claims (13)
1. a kind of traffic safety detection method, it is characterised in that applied to the automobile for being provided with camera, the camera is used for
Gather the image information in the automobile；Methods described includes：
When electric on detecting the automobile, the camera is opened；
The video information that the camera collects is obtained, every two field picture that the video information includes is entered according to the first algorithm
Row analysis, identifies the facial image in every two field picture；
The quantity of facial image in the every two field picture of statistics, and the face quantity in any one two field picture is more than the first preset value
When, the two field picture is labeled as target image；
The quantity for the target image that the video information includes is counted, and reaches the second preset value in the quantity of target image
When, every frame target image is analyzed according to the second algorithm, re-recognizes the facial image in every frame target image；
When the quantity for the facial image being identified in any one frame target image is more than first preset value, it is determined that described
Automobile overload, automobile described in No starting are simultaneously alarmed.
2. according to the method for claim 1, it is characterised in that methods described also includes：
When the camera opens failure, automobile described in No starting；Or
When the camera is opened successfully, judge whether the camera is blocked, if the camera is blocked, forbid
Start the automobile.
3. according to the method for claim 2, it is characterised in that judge the step of whether camera is blocked, including：
Obtain at least two field picture that the camera collects, and an at least two field picture described in calculating and default automobile internal field
The similarity of scape image；
Judge whether the similarity reaches default similarity threshold, if not up to described similarity threshold, it is determined that described
Camera is blocked.
4. according to the method described in any one of claims 1 to 3, it is characterised in that the people in an at least two field picture is identified
After face image, methods described also includes：
The relative position relation of each facial image in an at least two field picture is analyzed, obtains the front row of the automobile
The facial image of passenger；
The first age grader that the facial image input of the front passenger is trained, and according to first character classification by age
The classification results of device obtain the first age group belonging to the front passenger；
Judge whether include first age group in default first object age group；
If including first age group, Gabor characteristic and local two are extracted from the facial image of the front passenger
It is worth pattern feature, and the second age grader that the Gabor characteristic extracted and the input of local binary patterns feature are trained,
The second age group according to belonging to the classification results of the second age grader obtain the front passenger, the first object
Age group includes second age group；
Judge whether include second age group in default second target age group, if so, then automobile described in No starting
And alarmed.
5. according to the method for claim 4, it is characterised in that the Gabor characteristic extracted and local binary patterns is special
After the second age grader that sign input trains, methods described also includes：
Whether the age that the front passenger is judged according to the classification results of the second age grader is two neighboring second
The boundary value of age group；
If the boundary value of two neighboring second age group, then obtain the face of the front passenger in the picture frame of place with seat
The relative position relation of chair image, and according to the height of the relative position relation calculating front passenger；
When the height of the front passenger is less than predetermined threshold value, automobile described in No starting is simultaneously alarmed.
6. according to the method described in any one of claims 1 to 3, it is characterised in that after automobile starting, methods described is also wrapped
Include：
The picture frame that the camera currently collects is obtained, and identifies the facial image of described image frame；
The relative position relation of each facial image in described image frame is analyzed, obtains the people of the driver of the automobile
Face image；
The facial image of the driver is inputted into the Expression Recognition model based on GoogLeNet, obtains the table of the driver
Situation state；
The gesture recognition model that the facial image input of the driver is trained, obtains the head direction of the driver；
According to the emotional state of the driver and head towards judging whether the driver is in fatigue driving state, if place
In fatigue driving state, then alarmed.
7. according to the method for claim 6, it is characterised in that sentenced according to the emotional state of the driver and head direction
The step of whether driver of breaking is in fatigue driving state, including：
When the emotional state of the driver is tired state, and the head of the driver is oriented when bowing, it is determined that described
Driver is in fatigue driving state.
8. according to the method for claim 6, it is characterised in that obtain the driver of the automobile facial image it
Afterwards, methods described also includes：
Extract Gabor characteristic and histograms of oriented gradients feature from the facial image of the driver, and will extract
The Image Classifier that Gabor characteristic and the input of histograms of oriented gradients feature train, it is in one's ear with identify the driver
It is no to have mobile phone；
If the driver's has a mobile phone in one's ear, and the current emotional state of the driver is speaks, it is determined that the driving
Member is in dangerous driving state, then is alarmed.
9. according to the method described in any one of claims 1 to 3, it is characterised in that after the automobile starting, methods described is also
Including：
The multiple image that the camera collects in preset duration is obtained, and identifies the facial image in every two field picture；
The relative position relation of facial image in every two field picture is analyzed, extracts the driver of the automobile in the frame
Facial image in image, obtain multiple facial images of the driver；
The gesture recognition model that the input of the multiple facial image is trained respectively, obtains multiple head courts of the driver
To；
If the multiple head is towards immediately ahead of being not, it is determined that the driver is in dangerous driving state, and is reported
It is alert.
10. a kind of traffic safety detection method, it is characterised in that applied to the automobile for being provided with camera, the camera is used
Image information in the collection automobile；Methods described includes：
When electric on detecting automobile, the camera is opened, and obtains an at least two field picture for the camera collection, according to
The facial image in an at least two field picture is identified for first algorithm and/or the second algorithm；
The relative position relation of each facial image to being identified in an at least two field picture is analyzed, and obtains the vapour
The facial image of the front passenger of car；
The first age grader that the facial image input of the front passenger is trained, and according to first character classification by age
The classification results of device obtain the first age group belonging to the front passenger；
Judge whether include first age group in default first object age group；
If including first age group, Gabor characteristic and local two are extracted from the facial image of the front passenger
It is worth pattern feature, and the second age grader that the Gabor characteristic extracted and the input of local binary patterns feature are trained,
The second age group according to belonging to the classification results of the second age grader obtain the front passenger, the first object
Age group includes second age group；
Judge whether include second age group in default second target age group, if so, then automobile described in No starting
And alarmed.
11. a kind of traffic safety detection method, it is characterised in that applied to the automobile for being provided with camera, the camera is used
Image information in the collection automobile；Methods described includes：
After automobile starting, the picture frame that the camera currently collects is obtained, and identify the face in described image frame
Image；
The relative position relation of each facial image in described image frame is analyzed, obtains the people of the driver of the automobile
Face image；
The gesture recognition model that the facial image input of the driver is trained, obtains the head direction of the driver；
According to the expression posture of the driver and head towards judging whether the driver is in fatigue driving state, if place
In fatigue driving state, then alarmed.
12. according to the method for claim 11, it is characterised in that obtain the driver of the automobile facial image it
Afterwards, methods described also includes：
Extract Gabor characteristic and histograms of oriented gradients feature from the facial image of the driver, and will extract
The Image Classifier that Gabor characteristic and the input of histograms of oriented gradients feature train, it is in one's ear with identify the driver
It is no to have mobile phone；
If the driver's has mobile phone in one's ear, and the emotional state of the driver is to speak, it is determined that the driving
Member is in dangerous driving state, then is alarmed.
13. a kind of traffic safety detection means, it is characterised in that applied to the automobile for being provided with camera, the camera is used
Image information in the collection automobile；Described device includes：
Opening module, for when electric on detecting the automobile, opening the camera；
Identification module, the video information collected for obtaining the camera, according to the first algorithm to the video information bag
The every two field picture included is analyzed, and identifies the facial image in every two field picture；
First statistical module, for counting the quantity of the facial image in every two field picture, and the face in any one two field picture
When quantity is more than the first preset value, the two field picture is labeled as target image；
Second statistical module, the quantity of the target image included for counting the video information, and in the number of target image
When amount reaches the second preset value, every frame target image is analyzed according to the second algorithm, re-recognized in every frame target image
Facial image；
Alarm module, for being preset when the quantity for the facial image being identified in any one frame target image is more than described first
During value, the automobile overload is determined, automobile described in No starting is simultaneously alarmed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711080164.XA CN107766835A (en) | 2017-11-06 | 2017-11-06 | traffic safety detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711080164.XA CN107766835A (en) | 2017-11-06 | 2017-11-06 | traffic safety detection method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107766835A true CN107766835A (en) | 2018-03-06 |
Family
ID=61272758
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711080164.XA Pending CN107766835A (en) | 2017-11-06 | 2017-11-06 | traffic safety detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107766835A (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764169A (en) * | 2018-05-31 | 2018-11-06 | 厦门大学 | A kind of driver's Emotion identification based on machine learning and display device and method |
| CN108931288A (en) * | 2018-06-04 | 2018-12-04 | 云南力帆骏马车辆有限公司 | vehicle load monitoring method and device |
| CN108960107A (en) * | 2018-06-25 | 2018-12-07 | 安徽百诚慧通科技有限公司 | A kind of overcrowding recognition methods of small mini van and device |
| CN109508632A (en) * | 2018-09-30 | 2019-03-22 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN109858470A (en) * | 2019-03-06 | 2019-06-07 | 百度在线网络技术（北京）有限公司 | Method and apparatus for output information |
| 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 |
| CN110304072A (en) * | 2018-03-27 | 2019-10-08 | 通用汽车环球科技运作有限责任公司 | The occupancy of shared autonomous vehicle as security feature detects |
| CN110493297A (en) * | 2018-05-15 | 2019-11-22 | 上海博泰悦臻网络技术服务有限公司 | Children based on assistant driver seat prohibit seat method and cloud server |
| CN110738079A (en) * | 2018-07-19 | 2020-01-31 | 杭州海康威视数字技术股份有限公司 | Method and device for detecting abnormal number of front row personnel of motor vehicle and computer equipment |
| CN110909641A (en) * | 2019-11-13 | 2020-03-24 | 北京文安智能技术股份有限公司 | Method, device and system for detecting overload of motorcycle |
| CN111104845A (en) * | 2018-10-25 | 2020-05-05 | 佳能株式会社 | Detection apparatus, control method, and computer-readable recording medium |
| CN111243236A (en) * | 2020-01-17 | 2020-06-05 | 南京邮电大学 | Fatigue driving early warning method and system based on deep learning |
| CN111275008A (en) * | 2020-02-24 | 2020-06-12 | 浙江大华技术股份有限公司 | Method and device for detecting abnormality of target vehicle, storage medium, and electronic device |
| CN111832378A (en) * | 2019-08-13 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for identifying vehicle overtaking |
| CN111895447A (en) * | 2020-06-07 | 2020-11-06 | 李迎春 | Ignition control system based on people number analysis |
| CN112149482A (en) * | 2019-06-28 | 2020-12-29 | 深圳市商汤科技有限公司 | Method, device and equipment for detecting on-duty state of driver and computer storage medium |
| CN112633694A (en) * | 2020-12-24 | 2021-04-09 | 北京翔东智能科技有限公司 | Real-time geographic communication method based on space-time big data |
| CN112947740A (en) * | 2019-11-22 | 2021-06-11 | 深圳市超捷通讯有限公司 | Human-computer interaction method based on motion analysis and vehicle-mounted device |
| CN113053127A (en) * | 2020-11-26 | 2021-06-29 | 泰州芯源半导体科技有限公司 | Intelligent real-time state detection system and method |
| CN113923408A (en) * | 2021-09-29 | 2022-01-11 | 岚图汽车科技有限公司 | Back row detection and interaction system and method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013001479A (en) * | 2011-06-14 | 2013-01-07 | Mitsubishi Electric Corp | Passenger conveyor |
| CN105608422A (en) * | 2015-12-16 | 2016-05-25 | 安徽创世科技有限公司 | Intelligent monitoring detection method for overloading of passenger car |
| CN106241584A (en) * | 2016-08-23 | 2016-12-21 | 西尼电梯(杭州)有限公司 | A kind of intelligent video monitoring system based on staircase safety and method |
| CN106453570A (en) * | 2016-10-19 | 2017-02-22 | 北京速能数码网络技术有限公司 | Remote monitoring method and system of driving state, vehicle-mounted device and server |
-
2017
- 2017-11-06 CN CN201711080164.XA patent/CN107766835A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013001479A (en) * | 2011-06-14 | 2013-01-07 | Mitsubishi Electric Corp | Passenger conveyor |
| CN105608422A (en) * | 2015-12-16 | 2016-05-25 | 安徽创世科技有限公司 | Intelligent monitoring detection method for overloading of passenger car |
| CN106241584A (en) * | 2016-08-23 | 2016-12-21 | 西尼电梯(杭州)有限公司 | A kind of intelligent video monitoring system based on staircase safety and method |
| CN106453570A (en) * | 2016-10-19 | 2017-02-22 | 北京速能数码网络技术有限公司 | Remote monitoring method and system of driving state, vehicle-mounted device and server |
Non-Patent Citations (2)
| Title |
|---|
| 周云鹏: "基于面部视觉多特征融合的驾驶员疲劳检测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
| 周云鹏: "基于面部视觉多特征融合的驾驶员疲劳检测方法研究", 《中国优秀硕士学位论文全文数据库工程科技II辑》 * |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110304072A (en) * | 2018-03-27 | 2019-10-08 | 通用汽车环球科技运作有限责任公司 | The occupancy of shared autonomous vehicle as security feature detects |
| CN110493297A (en) * | 2018-05-15 | 2019-11-22 | 上海博泰悦臻网络技术服务有限公司 | Children based on assistant driver seat prohibit seat method and cloud server |
| CN108764169A (en) * | 2018-05-31 | 2018-11-06 | 厦门大学 | A kind of driver's Emotion identification based on machine learning and display device and method |
| CN108931288A (en) * | 2018-06-04 | 2018-12-04 | 云南力帆骏马车辆有限公司 | vehicle load monitoring method and device |
| CN108960107A (en) * | 2018-06-25 | 2018-12-07 | 安徽百诚慧通科技有限公司 | A kind of overcrowding recognition methods of small mini van and device |
| CN110738079A (en) * | 2018-07-19 | 2020-01-31 | 杭州海康威视数字技术股份有限公司 | Method and device for detecting abnormal number of front row personnel of motor vehicle and computer equipment |
| CN109508632B (en) * | 2018-09-30 | 2023-04-07 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN109508632A (en) * | 2018-09-30 | 2019-03-22 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN111104845A (en) * | 2018-10-25 | 2020-05-05 | 佳能株式会社 | Detection apparatus, control method, and computer-readable recording medium |
| CN111104845B (en) * | 2018-10-25 | 2023-10-24 | 佳能株式会社 | Detection apparatus, control method, and computer-readable recording medium |
| CN109858470A (en) * | 2019-03-06 | 2019-06-07 | 百度在线网络技术（北京）有限公司 | Method and apparatus for output information |
| 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 |
| US11423676B2 (en) | 2019-06-28 | 2022-08-23 | Shenzhen Sensetime Technology Co., Ltd. | Method and apparatus for detecting on-duty state of driver, device, and computer storage medium |
| CN112149482A (en) * | 2019-06-28 | 2020-12-29 | 深圳市商汤科技有限公司 | Method, device and equipment for detecting on-duty state of driver and computer storage medium |
| WO2020258719A1 (en) * | 2019-06-28 | 2020-12-30 | 深圳市商汤科技有限公司 | Method, apparatus and device for detecting on-duty state of driver, and computer storage medium |
| CN111832378A (en) * | 2019-08-13 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for identifying vehicle overtaking |
| CN110909641A (en) * | 2019-11-13 | 2020-03-24 | 北京文安智能技术股份有限公司 | Method, device and system for detecting overload of motorcycle |
| CN112947740A (en) * | 2019-11-22 | 2021-06-11 | 深圳市超捷通讯有限公司 | Human-computer interaction method based on motion analysis and vehicle-mounted device |
| CN111243236A (en) * | 2020-01-17 | 2020-06-05 | 南京邮电大学 | Fatigue driving early warning method and system based on deep learning |
| CN111275008A (en) * | 2020-02-24 | 2020-06-12 | 浙江大华技术股份有限公司 | Method and device for detecting abnormality of target vehicle, storage medium, and electronic device |
| CN111275008B (en) * | 2020-02-24 | 2024-01-16 | 浙江大华技术股份有限公司 | Method and device for detecting abnormality of target vehicle, storage medium and electronic device |
| CN111895447A (en) * | 2020-06-07 | 2020-11-06 | 李迎春 | Ignition control system based on people number analysis |
| CN113053127A (en) * | 2020-11-26 | 2021-06-29 | 泰州芯源半导体科技有限公司 | Intelligent real-time state detection system and method |
| CN113053127B (en) * | 2020-11-26 | 2021-11-26 | 江苏奥都智能科技有限公司 | Intelligent real-time state detection system and method |
| CN112633694A (en) * | 2020-12-24 | 2021-04-09 | 北京翔东智能科技有限公司 | Real-time geographic communication method based on space-time big data |
| CN113923408A (en) * | 2021-09-29 | 2022-01-11 | 岚图汽车科技有限公司 | Back row detection and interaction system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107766835A (en) | traffic safety detection method and device | |
| CN110119676B (en) | Driver fatigue detection method based on neural network | |
| Rahman et al. | Real time drowsiness detection using eye blink monitoring | |
| CN104657752B (en) | A kind of seatbelt wearing recognition methods based on deep learning | |
| EP1732028B1 (en) | System and method for detecting an eye | |
| CN105354985B (en) | Fatigue driving monitoring apparatus and method | |
| CN111079474A (en) | Passenger state analysis method and device, vehicle, electronic device, and storage medium | |
| US7650034B2 (en) | Method of locating a human eye in a video image | |
| CN110766912A (en) | Driving early warning method, device and computer readable storage medium | |
| CN109543577A (en) | A kind of fatigue driving detection method for early warning based on facial expression feature | |
| CN106548132A (en) | The method for detecting fatigue driving of fusion eye state and heart rate detection | |
| Lashkov et al. | Driver dangerous state detection based on OpenCV & dlib libraries using mobile video processing | |
| CN103729646B (en) | Eye image validity detection method | |
| CN113838265A (en) | Fatigue driving early warning method and device and electronic equipment | |
| Rani et al. | Development of an Automated Tool for Driver Drowsiness Detection | |
| Mašanović et al. | Driver monitoring using the in-vehicle camera | |
| CN115937829A (en) | Method for detecting abnormal behaviors of operators in crane cab | |
| Joseph et al. | Real time drowsiness detection using Viola jones & KLT | |
| Koesdwiady et al. | Driver inattention detection system: A PSO-based multiview classification approach | |
| KR102338777B1 (en) | Method for monitering number of passengers in vehicle using camera | |
| Jain et al. | Driver drowsiness detection using DLIB | |
| Garg et al. | A Drowsy Driver Detection and security system | |
| Budiharto et al. | Design and Analysis of Fast Driver's Fatigue Estimation and Drowsiness Detection System using Android. | |
| Srivastava et al. | Driver drowsiness detection system with OpenCV & keras | |
| Radha et al. | Drowsiness Detection System using Visual Articulators |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180306 |
|
| RJ01 | Rejection of invention patent application after publication |