CN113807169A - Method and system for detecting and early warning children in copilot - Google Patents
Method and system for detecting and early warning children in copilot Download PDFInfo
- Publication number
- CN113807169A CN113807169A CN202110905764.5A CN202110905764A CN113807169A CN 113807169 A CN113807169 A CN 113807169A CN 202110905764 A CN202110905764 A CN 202110905764A CN 113807169 A CN113807169 A CN 113807169A
- Authority
- CN
- China
- Prior art keywords
- child
- copilot
- passenger
- image
- face
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/24—Reminder alarms, e.g. anti-loss alarms
Abstract
The invention discloses a detection and early warning method for children in a copilot, which comprises the following steps: acquiring sample data of a child face area image of a copilot in a vehicle in advance; drawing a prior frame of the child face area based on the sample data to obtain a position coordinate of the prior frame; training a constructed MTCNN face detection model by adopting the sample data; wherein, in actual detection: acquiring a face area actual measurement image of a copilot in a vehicle; inputting the real-measured image of the face area into a trained MTCNN face detection model, and if the MTCNN face detection model is output: if a face signal is detected, the next step is carried out; and comparing the position coordinates of the real-measured image of the face area with the position coordinates of the prior frame to obtain the estimated value of the sitting height of the passenger in the passenger seat, and outputting an alarm signal if the estimated value of the sitting height of the passenger in the passenger seat is lower than a set first threshold value. Correspondingly, the invention also discloses a detection and early warning system for the children at the copilot.
Description
Technical Field
The invention relates to an auxiliary detection early warning method and system for a vehicle, in particular to a detection early warning method and system for a passenger in a copilot of the vehicle.
Background
In recent years, with the rapid development of economy in China, automobiles gradually enter thousands of households as main transportation means for transportation, the vehicle holding capacity is higher and higher, and the automobiles become an indispensable member of many families.
During vehicle driving, children riding in a copilot present a great safety hazard, but many consumers often ignore this. According to the summary report of death necropsy on Chongqing highways from 1996 to 2005, the rank ordering of the death people in traffic accidents is: pedestrian, 29.7%; passenger in the passenger seat accounts for 28.8%; driver, accounting for 21.1%; rear passenger, 17.8%. It follows that the mortality rate in the co-driver position is high. In addition, a large number of experiments prove that when a 10-year-old boy sits in the auxiliary driving position and is about 140 cm in height, the head of the boy is just close to the height of the airbag when the boy sits in the vehicle and leans forwards, which means that the head of the boy is easily hit by the airbag in case of an accident.
When a child sits at the position of a vehicle copilot and ties up an adult safety belt, the child can obviously feel that the safety belt is difficult to control the body of the child, the child can slide out of the safety belt easily, and if the restraining force is insufficient, the neck of the child can be tied by the safety belt when the child brakes. For children under 12 years old, the musculoskeletal structure of the children is fragile, and the impact force generated when the safety air bag is opened is likely to cause injuries such as chest fracture, intracranial hemorrhage and the like of the children.
Therefore, in order to warn a driver and ensure that a passenger does not take a child in a passenger seat, the invention is expected to obtain a passenger seat child detection early warning method and a passenger seat child detection early warning system.
The operation of the passenger detection early warning method for the copilot is simple and convenient and easy to implement, when the passenger in the copilot is judged to be a child, the passenger can output an alarm signal to remind a driver, the passenger detection early warning method has very wide applicability, can be applied to all vehicles, and has good popularization prospect and application value.
Disclosure of Invention
One of the objectives of the present invention is to provide a method for detecting and warning children in a copilot, which can collect real images of a face area of the copilot in a vehicle, analyze and compare the real images based on the face area, and determine whether a passenger in the copilot is a child.
The operation of the passenger detection early warning method for the copilot is simple and convenient and easy to implement, when the passenger in the copilot is judged to be a child, the passenger can output an alarm signal to remind a driver, the passenger detection early warning method has very wide applicability, can be applied to all vehicles, and has good popularization prospect and application value.
In order to achieve the purpose, the invention provides a method for detecting and early warning a child in a copilot, which comprises the following steps:
acquiring sample data of a child face area image of a copilot in a vehicle in advance;
drawing a prior frame of the child face area based on the sample data to obtain a position coordinate of the prior frame; training a constructed MTCNN face detection model by adopting the sample data;
wherein, in actual detection:
acquiring a face area actual measurement image of a copilot in a vehicle;
inputting the real-measured image of the face area into a trained MTCNN face detection model, and if the MTCNN face detection model is output: if a face signal is detected, the next step is carried out;
and comparing the position coordinates of the real-measured image of the face area with the position coordinates of the prior frame to obtain the estimated sitting height value of the passenger in the passenger seat, and outputting an alarm signal if the estimated sitting height value of the passenger in the passenger seat is lower than a set first threshold value.
In the technical scheme of the invention, the real-measured image of the face area of the copilot in the vehicle can be acquired by adopting the method for detecting and early warning the children at the copilot, the position coordinate of the real-measured image of the face area is compared with the position coordinate of the prior frame for analysis and comparison, so as to obtain the estimated sitting height value of the copilot, whether the passenger at the copilot is a child or not can be judged by comparing the estimated sitting height value with the set threshold value, and when the passenger at the copilot is judged to be a child, an alarm signal can be output to remind the driver.
Further, in the method for detecting and warning children at the copilot seat, when the estimated sitting height value of the passenger at the copilot seat is not less than the set first threshold, the following steps are continued:
adopting an SSR-Net age estimation module to carry out age identification on the real-measured image of the face region;
and when the identified age is lower than a set second threshold value, outputting an alarm signal.
In the technical scheme of the invention, the method for detecting and early warning the children at the copilot position can also adopt an SSR-Net age estimation module to identify the age of the real-measured image of the face area and judge by using the identified age.
Further, in the method for detecting and warning children at the copilot seat, the invention further comprises the following steps: and when the identified age is lower than a set second threshold value, storing the child characteristic information of the face area actual measurement image corresponding to the identified age into a memory container.
Further, in the method for detecting and warning children at the copilot seat, the invention further comprises the following steps: when the estimated sitting height value of the passenger in the passenger seat is lower than a set first threshold value, storing the child characteristic information of the real-measured image of the face area corresponding to the estimated sitting height value into a memory container.
Further, in the method for detecting and warning children at the copilot seat, the invention further comprises the following steps: when the estimated sitting height value of the passenger at the passenger seat is larger than or equal to a set first threshold value, comparing the characteristic information of the actual image of the face area with the characteristic information of the children stored in the memory container to judge whether the actual image of the face area corresponds to the children, and if so, outputting an alarm signal.
Further, in the method for detecting and warning children at the copilot position, the euclidean distance between the feature information of the real image of the face region and the feature information of the children stored in the memory container is calculated to judge whether the real image of the face region corresponds to the children.
Accordingly, another object of the present invention is to provide a children detection and early warning system for a passenger seat, which can be used to implement the above-mentioned children detection and early warning method for a passenger seat of a vehicle, and can effectively detect and determine whether a passenger in the passenger seat of the vehicle is a child, and alarm when the passenger in the passenger seat is a child.
In order to achieve the above object, the present invention provides a detection and early warning system for children in a copilot, comprising:
the data acquisition unit is used for acquiring sample data of a child face area image of a copilot in the vehicle in advance and acquiring a face area actual measurement image of the copilot in the vehicle during actual detection;
a prior frame unit which draws a prior frame of the child face region based on the sample data and obtains a position coordinate of the prior frame;
the MTCNN face detection unit is trained by the sample data and performs face detection based on a real-time measured image of a face area;
the sitting height estimation unit compares the position coordinates of the real-measured image of the face area with the position coordinates of the prior frame to obtain a sitting height estimation value of the passenger at the copilot;
an alarm unit configured to: and if the estimated sitting height value of the passenger in the passenger seat is lower than the set first threshold value, outputting an alarm signal.
Further, in the system for detecting and warning a child at a copilot position, the system further comprises an SSR-Net age estimation unit, which identifies the age of the actual image of the face area, and when the identified age is lower than a second threshold, the alarm unit outputs an alarm signal.
Further, in the system for detecting and warning children at the copilot position, the invention further comprises a memory container for storing the characteristic information of the children, wherein when the age identified by the SSR-Net age estimation unit is lower than a second set threshold, the characteristic information of the children corresponding to the detected image of the face area is stored in the memory container; and/or when the estimated sitting height value of the passenger at the passenger seat estimated by the sitting height estimating unit is lower than a set first threshold value, storing the child characteristic information of the measured image of the face area corresponding to the estimated sitting height value into the memory container.
Further, in the system for detecting and warning children at the copilot seat of the present invention, the system further includes a feature information comparing unit for comparing the feature information of the actual image of the face area with the feature information of the children stored in the memory container to determine whether the actual image of the face area corresponds to a child, and when the determination is "yes", the warning unit outputs a warning signal.
Compared with the prior art, the method and the system for detecting and early warning the children at the copilot have the advantages and beneficial effects as follows:
the method for detecting and early warning the children at the copilot can acquire the real-measured image of the face area of the copilot in the vehicle, analyze and compare the real-measured image of the face area and judge whether the passenger at the copilot is the child.
The operation of the passenger detection early warning method for the copilot is simple and convenient and easy to implement, when the passenger in the copilot is judged to be a child, the passenger can output an alarm signal to remind a driver, the passenger detection early warning method has very wide applicability, can be applied to all vehicles, and has good popularization prospect and application value.
Correspondingly, the passenger seat child detection early warning system can be used for implementing the passenger seat child detection early warning method, can effectively detect and judge whether the passenger in the passenger seat of the vehicle is a child, and gives an alarm when the passenger in the passenger seat is a child, and has the advantages and beneficial effects.
Drawings
Fig. 1 is a general flowchart of a method for detecting and warning a child in a co-driver seat according to an embodiment of the present invention.
FIG. 2 is a flow chart schematically illustrating the preprocessing of sample data by the prior box unit according to an embodiment of the present invention.
FIG. 3 schematically illustrates a training process of an MTCNN face detection unit according to the present invention.
Fig. 4 schematically shows a flow chart of the usage of the sitting height estimation unit according to an embodiment of the invention.
FIG. 5 is a flow chart schematically illustrating the use of the memory container according to one embodiment of the present invention.
Fig. 6 schematically shows the training process of the SSR-Net age estimation unit according to the present invention.
Fig. 7 schematically shows a data conversion process of the face region real image input to the SSR-Net age estimation unit according to an embodiment of the present invention.
Fig. 8 schematically shows a flow chart of the usage of the feature information comparison unit according to an embodiment of the present invention.
Fig. 9 schematically shows a start-stop flow chart of the children detection early warning system in the copilot according to an embodiment of the invention.
Detailed Description
The method and system for detecting and warning a child in a co-driver seat according to the present invention will be further explained and illustrated with reference to the drawings and the specific embodiments of the present invention, which, however, should not be construed as unduly limiting the technical solution of the present invention.
In the invention, the invention designs a children detection early warning system at a copilot, and in the children detection early warning system at the copilot, the system can comprise: the system comprises a data acquisition unit, a priori frame unit, an MTCNN face detection unit, a sitting height estimation unit, an SSR-Net age estimation unit, a characteristic information comparison unit, an alarm unit and a memory container.
The various "units" in the above-described system may also be referred to as "modules," merely as differences in expression, for example: the "MTCNN face detection unit" may also be denoted as an "MTCNN face detection module".
The passenger-seat child detection early-warning system can implement the following process of the passenger-seat child detection early-warning method shown in fig. 1, and can effectively detect and judge whether the passenger in the passenger seat of the vehicle is a child and ensure that the passenger in the passenger seat gives an alarm when the passenger in the passenger seat is a child.
Fig. 1 is a general flowchart of a method for detecting and warning a child in a co-driver seat according to an embodiment of the present invention.
The MTCNN model shown in FIG. 1 is an MTCNN face detection unit in the early warning system of the invention; the height estimation module shown in fig. 1 is also the height estimation unit in the early warning system of the present invention; the "feature point analysis module" shown in fig. 1 is a "feature information comparison unit" in the early warning system of the present invention; the SSR-Net module shown in FIG. 1 is an SSR-Net age estimation unit in the early warning system of the invention.
As shown in fig. 1, in the present embodiment, in the method for detecting and warning children at a copilot according to the present invention, a data acquisition unit in the system may acquire sample data of a facial region image of a child at a copilot in a vehicle in advance; based on the sample data, the prior frame unit is further utilized to process the input sample data, so that the prior frame of the child face area can be drawn, and the position coordinate of the prior frame is obtained.
In addition, in the method for detecting and early warning the children at the copilot, the built MTCNN face detection unit in the system can be trained based on the sample data acquired in advance. The MTCNN face detection unit is built and trained, belongs to the known content in the prior art, and the specific training process of the MTCNN face detection unit can refer to the following figure 3, which is not described herein any more.
It should be noted that, during actual detection, the data acquisition unit in the system may acquire a real image of a face area of a passenger seat in a vehicle, and input the real image of the face area into the MTCNN face detection unit after training, and the MTCNN may detect the face area and feature points of the input. When the MTCNN face detection unit outputs: if the face signal is detected, the position coordinate of the real-measured image of the face area can be further compared with the position coordinate of the prior frame by using the sitting height estimation unit to obtain the sitting height estimation value of the passenger at the copilot. Whether the passenger in the passenger seat is a child or not can be judged by using the estimated sitting height value.
When the estimated sitting height value of the passenger at the passenger seat is lower than a set first threshold value, an alarm unit in the system can be controlled to output an alarm signal; when the estimated sitting height value of the passenger in the passenger seat is lower than the set first threshold, the child characteristic information of the real-measured image of the face area corresponding to the estimated sitting height value can be stored in a memory container of the system.
It should be noted that when the estimated sitting height of the passenger in the passenger seat is greater than or equal to the set first threshold, the SSR-Net age estimation unit may be further adopted to perform age identification on the real-measured image of the face area, and the identified age is used to determine whether the passenger in the passenger seat is a child. When the age identified by the SSR-Net age estimation unit is lower than a set second threshold (the age is less than 12 years), the alarm unit can be controlled to output an alarm signal, and the child characteristic information of the face area actual measurement image corresponding to the alarm signal is stored in a memory container; if the age identified by the SSR-Net age estimation unit is greater than or equal to the set second threshold (the age is less than 12 years), no additional operation is performed, and the system operates normally.
In the invention, a memory container arranged in the system can store the child characteristic information of the actual image of the face area, when the estimated sitting height value of the passenger at the passenger seat is larger than or equal to a set first threshold value, the characteristic information of the actual image of the face area can be compared with the child characteristic information stored in the memory container by using a characteristic information comparison unit in the system so as to judge whether the actual image of the face area corresponds to a child, if so, the historical child characteristic information is detected on the surface, and an alarm signal is output by using an alarm unit.
In conclusion, the passenger seat child detection early warning system can receive the image data in the vehicle in real time, cut the image data in the vehicle, and ensure that the cutting area is the passenger seat area by adjusting proper parameters; further, according to a face region actual measurement image of a copilot in the vehicle, deep learning reasoning is adopted, so that face detection, feature matching, sitting height estimation and age analysis can be carried out on passengers in the copilot region; and judging according to the inference result, and giving an alarm if the inference result is that the child is a child.
FIG. 2 is a flow chart schematically illustrating the preprocessing of sample data by the prior box unit according to an embodiment of the present invention.
As shown in fig. 2, in the present embodiment, the inside and outside parameters may be calibrated by using the in-vehicle camera, and then children of different ages may take different copilot positions, and the data acquisition unit may acquire related data, and adjust different front and rear seat data to acquire as much data as possible. Therefore, sample data of the child face area image of the copilot in the vehicle can be acquired in advance.
Based on sample data acquired in advance, a priori frame of a child face area can be drawn correspondingly by using a priori frame unit, the coordinates of the center point of the priori frame { (x1, y1), (x2, y2), (x3, y3) (x4, y4) … … (xn, yn) } can be roughly measured, and the size of the center point face frame of the face frame { (w1, h1), (w2, h2), (w3, h3), (w4, h4) … … (wn, hn) } can be measured and calculated.
Thus, the average center point position coordinates (x, y) of the prior frame and the maximum size of the prior frame (w0 ═ Max { wi }, h0 ═ Max { hi }) can be further calculated.
To ensure as much full coverage as possible of all possible areas, w0 and h0 may be adaptively enlarged by a factor of 5, i.e., w 5 w0 and h 5 h 0; then, cutting the original pixels, wherein the coordinates of the average central point of the area where the prior frame is located after cutting are (x, y), and the width and the height of the area are respectively represented as w and h; finally, the cropped pixels may be scaled such that the scaled pixel matrix is (12,12, 3).
FIG. 3 schematically illustrates a training process of an MTCNN face detection unit according to the present invention.
As shown in fig. 3, the MTCNN face detection unit needs to be set up and trained, and the set-up and training belongs to the content known in the prior art, and a specific training process of the MTCNN face detection unit of the present invention may refer to fig. 3, and may use sample data acquired by the data acquisition unit as a training data set for training, which is not described herein again.
Fig. 4 schematically shows a flow chart of the usage of the sitting height estimation unit according to an embodiment of the invention.
As shown in fig. 4, with reference to fig. 1, in the present embodiment, a data acquisition unit may be used to acquire a real image of a human face region of a passenger seat in a vehicle during real detection. At this time, the real-measured image of the face region needs to be input into a trained MTCNN face detection unit, if the MTCNN face detection unit feeds back an unmanned signal, it indicates that no passenger is in the copilot, and the early warning system provided by the invention exits; if the MTCNN face detection unit outputs "face signal detected", the sitting height estimation unit in the system may be further started to perform the next operation.
Referring to fig. 4, it can be seen that the MTCNN face detection unit can deduce position coordinates X _ p, Y _ p of a center point of a real-measured image of a face region and a 10-dimensional vector of feature points. By adopting the sitting height estimation unit, whether the passenger is a child or not can be inferred according to the position information of the actual measurement image of the face area.
In the sit-up height estimation unit, the average center point position coordinate Y of the prior frame needs to be compared with Y _ p. Note that due to the imaging principle, the y-axis is downward in the pixel coordinate system. Thus, if Y _ p satisfies the following inequality: if Y _ p is more than or equal to a x Y (a takes a certain value in 1.3-1.8), judging that the sitting height estimated value is more than or equal to a first threshold value; otherwise, judging that the sitting height estimated value is lower than the first threshold value.
If the judgment result is that the sitting height estimated value is lower than the first threshold value, the system gives that children exist in a copilot area, and an alarm unit is used for outputting an early warning signal; and if the estimated sitting height value is larger than or equal to the first threshold value, sending the 10-dimensional vector of the characteristic point into a memory container of the system.
It should be noted that the 10-dimensional vector of the feature point of the real image of the face region is the feature information of the real image of the face region, which can be sent to a memory container of the system.
FIG. 5 is a flow chart schematically illustrating the use of the memory container according to one embodiment of the present invention.
As shown in fig. 5, in the present invention, the front passenger seat child detection and early warning system has a memory function, and can establish a memory container vector, and set the maximum capacity MaxNum of the memory container to 10, and the specific value of the memory container can be adjusted according to the requirement.
At the time of starting the memory container, the current time of the system may be recorded and set to T1; then judging whether the memory container is full, if the number of the members in the container reaches MaxNum, the members in the memory container reach the upper limit, and popping up the foremost member of the memory container; if the number of container members does not reach MaxNum, then no pop operation will be performed.
Accordingly, the SSR-Net-based age estimation unit may perform age recognition on the face region actual image, and determine whether the recognized age is lower than a set second threshold, which may be set to 12 years old, and when the recognized age is lower than the set second threshold, determine that the child is a child, store the 10-dimensional vector of the feature point of the face region actual image corresponding to the child in the memory container, and record the current time T2 of the system. And if the judgment result of the SSR-Net age estimation unit is that the child is a child, the memory container is not updated.
In this embodiment, it may be determined whether T2 and T1 satisfy T2-T1 > 1 year, if yes, the memory container is emptied and T2 is assigned to T1; otherwise, the memory container is not emptied. This is done by setting a time for emptying the container one year, considering that the driver's children may grow up.
Fig. 6 schematically shows the training process of the SSR-Net age estimation unit according to the present invention.
As shown in fig. 6, in the present invention, an SSR-Net age estimation unit is also required to be built and trained to identify the age of the real measurement image of the face area, and the age is used to determine whether the single-rider is a child.
The specific training process of the SSR-Net age estimation unit of the present invention can refer to fig. 6, and the construction and training of the SSR-Net age estimation unit belong to the known content in the prior art, and are not described herein again.
Fig. 7 schematically shows a data conversion process of the face region real image input to the SSR-Net age estimation unit according to an embodiment of the present invention.
As shown in fig. 7, in this embodiment, when the MTCNN face detection unit determines that a face signal is detected, a face frame region (x, y, w, h) is cut from a face region real measurement image to generate new face frame region frame image (img) data, the face frame region frame image is scaled so that the pixel matrix size is (36, 36, 3), and the data is input to the trained SSR-Net age estimation unit shown in fig. 6, so that age estimation can be performed.
Fig. 8 schematically shows a flow chart of the usage of the feature information comparison unit according to an embodiment of the present invention.
As shown in fig. 8, in combination with fig. 1, in the present invention, the characteristic information comparison unit can calculate the euclidean distance between the characteristic information of the actual image of the face region and the characteristic information of the child stored in the memory container, and perform comparison and judgment to determine whether the actual image of the face region corresponds to the child.
It should be noted that, in the present invention, the euclidean distance threshold between the feature information of the actual image of the face area and the child feature information stored in the memory container may be set as d _ limit, which may be set empirically or obtained by performing prior acquisition and calculation with an MTCNN face detection unit.
In the invention, the data of different slightly changed sitting postures of the same child in different copilot areas can be collected in advance, and the collected image data is input into the trained MTCNN face detection unit for calculation. At this time, for each graph i, a 10-dimensional feature vector V _ pi is output; the feature vector array of the actual measurement image can be normalized, the mean value V _ mean of the feature vector array is calculated, Euclidean distances between all vectors V _ pi and the mean value vector V _ mean are calculated one by one, and the minimum value d _ min of the distances is solved as min { | | V _ pi-V _ mean | }.
In this case, the euclidean distance threshold d _ limit may be set to d _ min, or may be obtained by adding the weight attenuation coefficient k to the actual effect and setting d _ limit to k × d _ min.
Accordingly, as shown in fig. 6, when the feature information comparison unit is used to determine whether the image actually measured in the face region corresponds to a child, it is first determined whether the storage container stores child feature information, and it is determined whether len (vector) is equal to 0, and if the storage container is empty, it indicates that len (vector) is equal to 0, and the feature information comparison unit is exited.
If the storage container is not empty, the feature point vector set { V _ i } in the storage container may be normalized, and the feature point vector V _ p of the face region actual measurement image generated by the MTCNN may be normalized. And comparing the Euclidean distances of V _ p and { V _ i } one by one, and obtaining the minimum Euclidean distance dp _ min.
By comparing the calculated minimum Euclidean distance dp _ min with the Euclidean distance threshold d _ limit, whether the actually-measured image of the face region corresponds to a child can be judged, and if dp _ min meets the following inequality: d _ limit is less than or equal to dp _ min, the feedback result is matching, the judgment is yes, the actual image of the face area is a child, and the alarm unit outputs an alarm signal; otherwise, the feedback result is not matched, and the judgment is no.
Fig. 9 schematically shows a start-stop flow chart of the children detection early warning system in the copilot according to an embodiment of the invention.
As shown in fig. 9, in the present embodiment, a time threshold may be preset, for example, the front passenger seat child detection and early warning system according to the present invention is started within 1 minute of starting the vehicle, so as to implement a front passenger seat child early warning analysis process. The passenger in the copilot area is subjected to face detection, feature matching, sitting height estimation and age analysis by adopting the copilot child detection early warning system, so that a conclusion is obtained. If the copilot has no children, the copilot children detection early warning system is quitted; if the children exist in the copilot, the early warning unit is started, and the system sends out an alarm to remind the driver.
It should be noted that, while sending out the alarm prompt, the time of the system at this moment is recorded as T3, and if there is manual intervention at any time, the whole early warning system is exited; if no manual intervention exists, the whole early warning system can be automatically quitted after 5 minutes, namely the whole early warning system is quitted when the current time T4 of the control system meets T4-T3 > 5 minutes.
In conclusion, the method for detecting and early warning the children at the copilot can acquire the real-measured image of the face area of the copilot in the vehicle, analyze and compare the real-measured image of the face area and judge whether the passenger at the copilot is the child.
The operation of the passenger detection early warning method for the copilot is simple and convenient and easy to implement, when the passenger in the copilot is judged to be a child, the passenger can output an alarm signal to remind a driver, the passenger detection early warning method has very wide applicability, can be applied to all vehicles, and has good popularization prospect and application value.
Correspondingly, the passenger seat child detection early warning system can be used for implementing the passenger seat child detection early warning method, can effectively detect and judge whether the passenger in the passenger seat of the vehicle is a child, and gives an alarm when the passenger in the passenger seat is a child, and has the advantages and beneficial effects.
It should be noted that the combination of the features in the present application is not limited to the combination described in the claims of the present application or the combination described in the embodiments, and all the features described in the present application may be freely combined or combined in any manner unless contradicted by each other.
It should also be noted that the above-mentioned embodiments are only specific embodiments of the present invention. It is apparent that the present invention is not limited to the above embodiments and similar changes or modifications can be easily made by those skilled in the art from the disclosure of the present invention and shall fall within the scope of the present invention.
Claims (10)
1. A method for detecting and early warning a child in a copilot is characterized by comprising the following steps:
acquiring sample data of a child face area image of a copilot in a vehicle in advance;
drawing a prior frame of the child face area based on the sample data to obtain a position coordinate of the prior frame; training a constructed MTCNN face detection model by adopting the sample data;
wherein, in actual detection:
acquiring a face area actual measurement image of a copilot in a vehicle;
inputting the real-measured image of the face area into a trained MTCNN face detection model, and if the MTCNN face detection model is output: if a face signal is detected, the next step is carried out;
and comparing the position coordinates of the real-measured image of the face area with the position coordinates of the prior frame to obtain the estimated sitting height value of the passenger in the passenger seat, and outputting an alarm signal if the estimated sitting height value of the passenger in the passenger seat is lower than a set first threshold value.
2. The method for detecting and warning children at the copilot seat as claimed in claim 1, wherein when the estimated sitting height of the passenger at the copilot seat is greater than or equal to the first threshold, the following steps are continued:
adopting an SSR-Net age estimation module to carry out age identification on the real-measured image of the face region;
and when the identified age is lower than a set second threshold value, outputting an alarm signal.
3. The copilot child detection and warning method of claim 2, further comprising the steps of: and when the identified age is lower than a set second threshold value, storing the child characteristic information of the face area actual measurement image corresponding to the identified age into a memory container.
4. The copilot child detection and early warning method of claim 1, further comprising the steps of: when the estimated sitting height value of the passenger in the passenger seat is lower than a set first threshold value, storing the child characteristic information of the real-measured image of the face area corresponding to the estimated sitting height value into a memory container.
5. The copilot child detection and warning method of claim 3 or 4, further comprising the steps of: when the estimated sitting height value of the passenger at the passenger seat is larger than or equal to a set first threshold value, comparing the characteristic information of the actual image of the face area with the characteristic information of the children stored in the memory container to judge whether the actual image of the face area corresponds to the children, and if so, outputting an alarm signal.
6. The method as claimed in claim 5, wherein the Euclidean distance between the feature information of the real image of the face region and the feature information of the child stored in the memory container is calculated to determine whether the real image of the face region corresponds to the child.
7. The utility model provides a copilot position children detect early warning system which characterized in that includes:
the data acquisition unit is used for acquiring sample data of a child face area image of a copilot in the vehicle in advance and acquiring a face area actual measurement image of the copilot in the vehicle during actual detection;
a prior frame unit which draws a prior frame of the child face region based on the sample data and obtains a position coordinate of the prior frame;
the MTCNN face detection unit is trained by the sample data and performs face detection based on a real-time measured image of a face area;
the sitting height estimation unit compares the position coordinates of the real-measured image of the face area with the position coordinates of the prior frame to obtain a sitting height estimation value of the passenger at the copilot;
an alarm unit configured to: and if the estimated sitting height value of the passenger in the passenger seat is lower than the set first threshold value, outputting an alarm signal.
8. The system of claim 7, further comprising an SSR-Net age estimation unit for performing age recognition on the real image of the face region, wherein the alarm unit outputs an alarm signal when the recognized age is lower than a second threshold.
9. The passenger seat child detection and early warning system according to claim 7, further comprising a memory container for storing the child feature information, wherein when the age identified by the SSR-Net age estimation unit is lower than a second threshold, the child feature information of the measured image of the face region corresponding to the SSR-Net age estimation unit is stored in the memory container; and/or when the estimated sitting height value of the passenger at the passenger seat estimated by the sitting height estimating unit is lower than a set first threshold value, storing the child characteristic information of the measured image of the face area corresponding to the estimated sitting height value into the memory container.
10. The method as claimed in claim 9, further comprising a feature information comparing unit for comparing the feature information of the actual image of the face area with the feature information of the child stored in the memory container to determine whether the actual image of the face area corresponds to the child, wherein if yes, the alarm unit outputs an alarm signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110905764.5A CN113807169A (en) | 2021-08-06 | 2021-08-06 | Method and system for detecting and early warning children in copilot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110905764.5A CN113807169A (en) | 2021-08-06 | 2021-08-06 | Method and system for detecting and early warning children in copilot |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113807169A true CN113807169A (en) | 2021-12-17 |
Family
ID=78942825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110905764.5A Pending CN113807169A (en) | 2021-08-06 | 2021-08-06 | Method and system for detecting and early warning children in copilot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113807169A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550666A (en) * | 2016-01-22 | 2016-05-04 | 大连楼兰科技股份有限公司 | Alarming system and method for preventing juveniles from driving cars |
| EP3493541A2 (en) * | 2017-12-01 | 2019-06-05 | Koninklijke KPN N.V. | Selecting an omnidirectional image for display |
| 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 |
| CN110781799A (en) * | 2019-10-22 | 2020-02-11 | 上海商汤智能科技有限公司 | Method and device for processing images in vehicle cabin |
| CN111310650A (en) * | 2020-02-13 | 2020-06-19 | 上海眼控科技股份有限公司 | Vehicle riding object classification method and device, computer equipment and storage medium |
| CN112001348A (en) * | 2020-08-31 | 2020-11-27 | 上海商汤临港智能科技有限公司 | Method and device for detecting passenger in vehicle cabin, electronic device and storage medium |
-
2021
- 2021-08-06 CN CN202110905764.5A patent/CN113807169A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550666A (en) * | 2016-01-22 | 2016-05-04 | 大连楼兰科技股份有限公司 | Alarming system and method for preventing juveniles from driving cars |
| EP3493541A2 (en) * | 2017-12-01 | 2019-06-05 | Koninklijke KPN N.V. | Selecting an omnidirectional image for display |
| 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 |
| CN110781799A (en) * | 2019-10-22 | 2020-02-11 | 上海商汤智能科技有限公司 | Method and device for processing images in vehicle cabin |
| CN111310650A (en) * | 2020-02-13 | 2020-06-19 | 上海眼控科技股份有限公司 | Vehicle riding object classification method and device, computer equipment and storage medium |
| CN112001348A (en) * | 2020-08-31 | 2020-11-27 | 上海商汤临港智能科技有限公司 | Method and device for detecting passenger in vehicle cabin, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6493620B2 (en) | Motor vehicle occupant detection system employing ellipse shape models and bayesian classification | |
| EP3053809B1 (en) | Parking assistance system | |
| EP0885782B1 (en) | Apparatus for detecting the presence of an occupant in a motor vehicle | |
| CN107585096A (en) | Anti- forgetting system for prompting, method and vehicle | |
| CN110509881A (en) | The internal control adjusted for safety belt | |
| US20190375312A1 (en) | Method and system for controlling a state of an occupant protection feature for a vehicle | |
| CN113459982B (en) | System and method for occupant classification and adjusting airbag deployment based thereon | |
| CN1642782A (en) | Device for the protection of a vehicle occupant | |
| CN113665440A (en) | Self-adaptive riding space adjusting system and method | |
| US11529925B2 (en) | Vehicle side airbag | |
| CN108891349A (en) | A kind of parking brake locking reminding method and device | |
| CN111361469A (en) | Automatic seat adjusting device and adjusting method | |
| CN110370876A (en) | Suspension height control method, interactive device, storage medium, controller, vehicle | |
| CN113807169A (en) | Method and system for detecting and early warning children in copilot | |
| US11731580B2 (en) | Methods and apparatus for activating multiple external airbags | |
| CN114347999B (en) | Passenger type identification method, system and device based on multi-feature fusion | |
| CN113022489B (en) | Air bag bounce control method | |
| CN108460355A (en) | Driver status and behavioral value system and method | |
| CN114475511B (en) | Vision-based airbag activation | |
| EP4303821A1 (en) | Seated passenger height estimation | |
| CN115123141A (en) | Vision-based passenger safety belt reminding device and method | |
| CN116968765B (en) | Lane departure warning method and system with self-adaptive warning time interval | |
| CN114619993B (en) | Automobile control method based on face recognition, system, equipment and storage medium thereof | |
| CN114581893A (en) | Method for calculating passenger sitting height and classifying cabin members based on OMS camera in vehicle | |
| CN116923421A (en) | Occupant neck injury evaluation device and method, computer program product, and vehicle |
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 |