CN110569732A - Safety belt detection method based on driver monitoring system and corresponding equipment - Google Patents
Safety belt detection method based on driver monitoring system and corresponding equipment Download PDFInfo
- Publication number
- CN110569732A CN110569732A CN201910734872.3A CN201910734872A CN110569732A CN 110569732 A CN110569732 A CN 110569732A CN 201910734872 A CN201910734872 A CN 201910734872A CN 110569732 A CN110569732 A CN 110569732A
- Authority
- CN
- China
- Prior art keywords
- seat belt
- driver
- image
- safety belt
- dms
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- 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
- G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
- G06V20/597—Recognising the driver's state or behaviour, e.g. attention or drowsiness
-
- 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
- G06V40/168—Feature extraction; Face representation
-
- 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
Abstract
the invention relates to a safety belt detection method based on a Driver Monitoring System (DMS), which comprises the following steps: acquiring a DMS registration image of a driver and carrying out sectional labeling on a safety belt worn by the driver in the DMS registration image; binding the identity information of the driver with the safety belt; acquiring a target DMS image and determining a pinch angle value of a first portion of a seat belt from a vertical direction based on a seat belt first partial image in the target DMS image; comparing the included angle value of the first part of the safety belt and the vertical direction with a preset included angle threshold value so as to judge whether the included angle value is smaller than the preset included angle threshold value; if so, determining that the driver does not wear the safety belt normally. Correspondingly, the invention also relates to safety belt detection equipment based on the DMS. According to the method and the equipment, the interference of external factors can be effectively reduced, and the real-time performance and the accuracy of safety belt detection are effectively improved.
Description
Technical Field
The invention relates to the field of intelligent transportation, in particular to the field of intelligent driving. In particular, the invention relates to a seat belt detection method based on a driver monitoring system and a corresponding device. It will be apparent to those skilled in the art that the present invention is applicable to other fields as well.
Background
With the development and popularization of Driver Monitoring systems (DMS for short), whether to wear a safety belt becomes an important index for Monitoring the driving behavior safety of drivers. If the safety belt is worn correctly, the casualty rate of a driver in a frontal crash accident can be reduced by 75 percent, and the casualty rate in a rollover accident can be reduced by 80 percent. Therefore, the safety belt is supervised and urged to be worn by the driver in the driving process, and the safety belt has important significance for guaranteeing the life safety of the driver.
At present, the mainstream safety belt detection method for drivers mainly relies on an electronic police system of a traffic gate, a high-definition camera is used for collecting images of the driving state of the vehicles passing through the traffic gate, and whether the drivers wear the safety belts or not is analyzed through an image processing technology. For example, chinese patent application publication No. CN106650567A provides a method of first acquiring an image of the driving condition of a vehicle on a road, then determining a seat belt detection area by image segmentation and image detection and extracting a straight line segment in the seat belt detection area, and finally determining whether a driver contains a seat belt based on the straight line segment. However, the method has low accuracy and poor real-time performance, is easily influenced by external factors and conditions such as external light, a shooting angle, a vehicle running speed and weather conditions, and cannot detect and monitor whether a driver wears a safety belt accurately and in high real-time performance.
Disclosure of Invention
In view of the above-mentioned drawbacks in the prior art, the present invention provides a method and a corresponding device for detecting a seat belt based on a Driver Monitoring System (DMS), which can solve the problem in the prior art that the accuracy and real-time of seat belt detection are easily affected by external factors and conditions.
According to a first aspect of the present invention, there is provided a DMS-based seat belt detection method, comprising the steps of: (1) acquiring a DMS registration image of a driver and carrying out sectional labeling on a safety belt worn by the driver in the DMS registration image; (2) binding the identity information of the driver with the safety belt; (3) acquiring a target DMS image and determining a included angle value of a first part of a safety belt from a vertical direction based on a first partial image of the safety belt in the target DMS image, wherein the first part is from the uppermost end of the safety belt to a boundary between the safety belt and a shoulder of a driver; (4) comparing the included angle value of the first part of the safety belt and the vertical direction with a preset included angle threshold value so as to judge whether the included angle value is smaller than the preset included angle threshold value; and (5) under the condition that the included angle value is smaller than the preset included angle threshold value, judging that the driver does not wear the safety belt normally.
According to a second aspect of the present invention, the present invention provides a DMS-based seat belt detection apparatus capable of implementing the above-described seat belt detection method.
Compared with the prior art, the safety belt detection method based on the DMS and the corresponding system or device have at least the following advantages:
(1) According to the method or the equipment, the safety belt wearing state of the driver is segmented and labeled based on the DMS, and the infrared image of the DMS equipment is adopted, so that the interference of external factors and conditions (such as light rays) on the safety belt detection is overcome, and the safety belt detection accuracy is improved.
(2) according to the method or the equipment, whether the driver wears the safety belt or not is detected step by step in an easy-to-go sequence, and the real-time performance and the accuracy of safety belt detection are improved.
(3) according to the method or the equipment, historical video data of the driver safety belt collected by the DMS are fully utilized, whether the driver wears the safety belt or not is detected in a time domain dimension, and the accuracy of safety belt detection is further improved.
Drawings
the present invention will now be described, by way of example only, and not by way of limitation, with reference to the accompanying drawings, in which:
Fig. 1 shows a flow chart of an exemplary DMS-based seat belt detection method according to the invention.
Figure 2 illustrates a flow chart of another DMS-based seat belt detection method according to the present invention.
Fig. 3 is a schematic diagram illustrating the segmentation of the seat belt worn by the driver in the DMS registration image.
fig. 4 shows a schematic block diagram of a DMS-based seat belt detection device according to the invention.
Fig. 5 illustrates a schematic block diagram of a computer apparatus according to the present invention.
Detailed Description
The principles and spirit of the present invention will be described with reference to a number of exemplary embodiments. It is to be understood that these embodiments are given only for the purpose of enabling those skilled in the art to better understand and to implement the present invention, and are not to be construed as limiting the scope of the present invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As will be apparent to one skilled in the art, embodiments of the present invention may be embodied as a system, apparatus, method or computer program product. Thus, the present invention can be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.
Although the present invention provides the method steps or apparatus structures described in the following examples or figures, more or fewer method steps or modular units may be included in the method or apparatus based on conventional or non-inventive efforts. In a method step or device configuration where no causal relationship is logically necessary, the order of execution of the method steps or the modular configuration of the device is not limited to the order of execution or the modular configuration described in the embodiments and shown in the figures.
fig. 1 shows a flow chart of an exemplary DMS-based seat belt detection method according to the invention. Specifically, as shown in fig. 1, the DMS-based seat belt detection method 100 according to the present invention may include the steps of: (1) acquiring a target DMS image, as shown in step 101; (2) identifying a first portion of a seat belt in the target DMS image, wherein the first portion is from an uppermost end of the seat belt to a shoulder interface of the seat belt with a driver, as shown in step 102; (3) extracting a first partial image of the seat belt corresponding to the first portion from the target DMS image and determining a value of an angle of the first portion of the seat belt with respect to a vertical direction based on the first partial image of the seat belt, as shown in step 103; (4) judging whether the included angle value is smaller than a preset included angle threshold value or not, as shown in step 104; and (5) under the condition that the included angle value is smaller than the preset included angle threshold value, judging that the driver does not wear the safety belt normally, as shown in step 105.
Specifically, one skilled in the art may acquire a target DMS image from the DMS device and then identify a first portion of the seatbelt in the target DMS image, where the first portion is from an uppermost end of the seatbelt to a boundary between the seatbelt and the driver's shoulder. In one embodiment according to the present invention, the target DMS image is an infrared image. Here, the infrared image that DMS equipment gathered can greatly reduced external light and noise to the interference that the safety belt detected, effectively improves the rate of accuracy that the safety belt detected. Meanwhile, the infrared image is single-channel, and preprocessing is not needed, so that the efficiency and the real-time performance of subsequent image processing are improved. After identifying the first portion, a first partial image of the seat belt corresponding to the first portion may be extracted from the target DMS image. For example, in one embodiment according to the present invention, the image of the seat belt from the uppermost end of the seat belt, which may be the connection point of the seat belt to the cabin backrest or seat, or the uppermost end of the seat belt that is visible in the target DMS image, to the boundary between the seat belt and the shoulder of the driver may be labeled as the first partial image of the seat belt. And determining an included angle value between the first part of the safety belt and the vertical direction based on the first part image of the safety belt, further judging whether the included angle value is smaller than a preset included angle threshold value, and judging that the safety belt is not normally worn by a driver under the condition that the included angle value is smaller than the preset included angle threshold value.
In the above-mentioned embodiment, identifying the first portion of the seat belt in the DMS image and determining whether the driver does not wear the seat belt normally according to whether the angle value of the first portion of the seat belt with the vertical direction is smaller than the preset angle threshold value can simply, quickly, conveniently and reliably determine whether the driver does not wear the seat belt normally. Because the DMS device is installed in the vehicle, the interference of factors such as external light, weather conditions and the like can be effectively reduced by detecting the safety belt based on the DMS image, and the accuracy of safety belt detection is effectively improved.
Further, in an embodiment according to the present invention, the determining in step (3) the value of the angle of the first part of the seat belt to the vertical based on the first partial image of the seat belt may comprise the sub-steps of: (3-1) extracting an edge of the first partial image of the seat belt using a Canny edge detection algorithm; (3-2) recognizing a straight line in the first partial image of the seat belt using a Hough transform, thereby obtaining a plurality of straight line segments in the first partial image of the seat belt; (3-3) determining a length value of each of the plurality of straight line segments; (3-4) selecting a plurality of target straight-line segments of which the length values are larger than a preset length threshold value from the plurality of straight-line segments, and determining the included angle value between each target straight-line segment of the plurality of target straight-line segments and the vertical direction, so as to obtain the included angle value of each target straight-line segment; and (3-5) determining the included angle value of the first part of the safety belt with the vertical direction according to the included angle value of each target straight line segment and the length value of each target straight line segment. Through the specific implementation mode, the included angle between the first part of the safety belt and the vertical direction can be determined more accurately.
The Canny edge detection algorithm is a standard algorithm for edge detection that is widely used by those skilled in the art and was proposed by John Canny in 1986. The edge of the image refers to a part of the image with a significant brightness change in a local area, and the gray profile of the local area can be generally regarded as a step, i.e. a sharp change from one gray value in a small buffer area to another gray value with a larger gray value difference. The edge part of the image concentrates most information of the image, the determination and extraction of the image edge are very important for the identification and understanding of the whole image scene, and are also important features depended on by image segmentation, and the edge detection is mainly the measurement, detection and positioning of the gray scale change of the image. The image processed by the Canny detection algorithm is usually a gray scale image, so if the camera acquires a color image, graying is performed first. Graying a color image, namely carrying out weighted average according to sampling values of all channels of the image. In the above specific embodiment according to the present invention, the line fitting is performed using the Hough transform. The Hough transform is a method of finding straight lines, circles, and other simple shapes in an image. After edge detection of an image by a person skilled in the art, a simple shape, such as a straight line segment, in the image can be identified using the Hough transform. The basic principle of Hough transform is to detect whether a given image has a curve of a given nature by transforming lines in image space into concentration points in parameter space using the duality of points and lines. In the above embodiment according to the present invention, before the Hough transform is performed, Canny is used to detect the edge, and then the Hough transform operation is performed on the edge image.
Further, in an embodiment according to the invention, the sub-step (3-5) of determining the value of the angle of the first portion of the safety belt to the vertical direction from the value of the included angle of the respective target straight-line segment and the value of the length of the respective target straight-line segment comprises: determining a ratio of a length value of each target straight-line segment to a total length value of the plurality of target straight-line segments as a weight coefficient of each target straight-line segment; and multiplying and summing the included angle value of each target straight line segment by a corresponding weight coefficient to obtain the included angle value of the first part of the safety belt and the vertical direction. Here, it is easily understood that the larger the length value of the line segment is, the greater the possibility of being recognized as a straight line at the time of Hough transformation is, and accordingly, the larger the weight coefficient it occupies, i.e., the higher the weight is.
Further, in an embodiment according to the present invention, in the case that the included angle value is not less than the preset included angle threshold value, the seat belt detection method includes the following first-stage sub-steps: (1) recognizing a facial feature of the driver and acquiring preset identification information associated with the facial feature, wherein the preset identification information is arranged on a second part of the safety belt from a boundary of the safety belt and the shoulder of the driver to the lowest end of the safety belt; (2) identifying a second portion of the seat belt in the target DMS image; (3) extracting a second partial image of the seat belt corresponding to the second portion from the target DMS image, and detecting a maximum confidence level that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt; (4) judging whether the maximum confidence coefficient is larger than a preset confidence coefficient threshold value; and (5) determining that the driver normally wears a seat belt if the maximum confidence is greater than the preset confidence threshold.
In an embodiment of the present invention, the driver identity information may be bound or associated with the preset identification information, that is, the driver and the preset identification information have a corresponding relationship or an associated relationship. In the case where it is determined that the included angle value is not less than the preset included angle threshold, it may be necessary to perform a subsequent step to further determine whether the driver is wearing the seat belt normally. The driver identification information may be driver identification information, such as facial features of the driver, collected and stored when the DMS device is installed in the cockpit or the driver's identity is registered. The predetermined identification information is provided on the second portion of the seat belt and may be a pattern printed or embossed on the second portion of the seat belt. In one embodiment according to the present invention, the preset identification information may be a bar code or a two-dimensional code coated or printed with an infrared-sensitive paint or paint. After acquiring the preset identification information corresponding to or associated with the driver, a belt second partial image corresponding to the second portion of the belt is extracted from the target DMS image. The maximum confidence level that the preset identification information appears in the second partial image of the seat belt can be detected based on the second partial image of the seat belt, so that whether the maximum confidence level is larger than a preset confidence level threshold value or not can be further judged. In the case where it is determined that the maximum confidence is greater than the preset confidence threshold, that is, the second portion of the seatbelt is present in the second partial image, it can be determined that the driver is wearing a seatbelt normally. Through the specific implementation mode, whether the driver wears the safety belt or not is detected step by step in the sequence from easy to difficult, the real-time performance and the accuracy of the safety belt detection method or corresponding equipment are improved, and whether the driver normally wears the safety belt or not can be judged more accurately and quickly.
It is readily understood by those skilled in the art that confidence indicates how trustworthy an event is. When detecting or judging whether the preset identification information appears in the second partial image of the safety belt, the confidence level is required to be used for giving a detection result or the credibility of a judgment result. The confidence indicates the accuracy of the detection result or the judgment result. For example, a classifier may give a probability of predicting a class, and whether the result of the classifier itself is reliable may be judged based on a confidence level.
in a preferred embodiment according to the present invention, the preset confidence threshold may be set to 85%. It will be readily understood by those skilled in the art that the preset confidence threshold may be set according to actual needs, and may also be set to 80%, 90%, 95%, etc.
Further, in an embodiment according to the present invention, the detecting, in the first-stage sub-step (3), the maximum confidence level that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt may include: acquiring a plurality of confidence degrees of the preset identification information in the second partial image of the safety belt by using sliding window detection; determining the confidence level with the largest value among the confidence levels as the maximum confidence level. There are many specific ways to implement the sliding window detection, the simplest of which is template matching detection. The template matching detection is capable of finding an image similar to the template image in the target DMS image from the template image. Specifically, in a specific embodiment according to the present invention, a template image (the template image may be, for example, a seat belt second partial image cut from a registered image in which a driver normally wears a seat belt) may be retrieved from a feature image database, for example, by face recognition, and then the seat belt second partial image is traversed and detected in a sliding window manner using the template image in an area of the seat belt second partial image in the target DMS image, so as to obtain a plurality of matching confidences of the seat belt second partial image in which the preset identification information appears. Furthermore, the sliding window detection may also be implemented using an SVM classifier or the like. And outputting a plurality of confidences in the sliding window detection process, wherein the confidence with the maximum value is the maximum confidence.
further, in an embodiment according to the present invention, in case the maximum confidence is not greater than the preset confidence threshold, the seat belt detection method further comprises the following second-stage substeps: (1) acquiring a trained classifier; (2) inputting the second partial image of the safety belt into the trained classifier for recognition so as to output a classification result; (3) and judging whether the driver wears the safety belt normally or not according to the classification result. For example, in a preferred embodiment according to the present invention, when the classification result is 1, it may be determined that the driver normally wears a seat belt, and when the classification result is 0, it may be determined that the driver does not normally wear a seat belt. Through the specific implementation mode, whether the driver wears the safety belt or not is detected step by step in the sequence from easy to difficult, the real-time performance and the accuracy of the safety belt detection method or corresponding equipment are improved, and whether the driver normally wears the safety belt or not can be judged more accurately and quickly.
Further, in a preferred embodiment according to the present invention, the second stage substep (1) of obtaining a classifier that has been trained comprises: acquiring historical video data of a current driver; collecting a plurality of positive samples in which the driver normally wears the seat belt and a plurality of negative samples in which the driver does not normally wear the seat belt from the historical video data; training a deep Convolutional Neural Network (CNN) based classifier using the plurality of positive samples and the plurality of negative samples. Through the specific implementation mode, historical video data collected by the DMS and used by a driver to wear or not wear a safety belt is fully utilized to train the classifier, whether the driver wears the safety belt or not can be detected in a time domain dimension, and therefore the accuracy of a safety belt detection method or corresponding equipment is further improved.
for example, in particular, in one embodiment according to the present invention, historical video data of the current driver is collected using the DMS, which may contain DMS images from the steps described above (i.e., from step (1) to first-level sub-step (5)) that it has not been possible to determine whether the driver is wearing the seatbelt, and which contains a video of the second portion of the seatbelt. For the historical video data, 3 static images are sampled every 5 seconds to obtain 15 static images, then the static images are normalized into images with the size of 56 × 14 pixels, so that an image sequence consisting of 15 static images with the size of 56 × 14 pixels is obtained, and then the image sequence consisting of the 15 static images is horizontally spliced into images with the size of 56 (14 × 15) pixels according to the time sequence to serve as an image sample. In this way, a plurality of positive samples and a plurality of negative samples are collected from the historical video data, for example, each of the positive and negative samples is at least 5000, wherein the positive and negative samples mean: the driver normally wears the seat belt in the positive sample, and the driver does not normally wear the seat belt in the negative sample. Training a deep Convolutional Neural Network (CNN) based classifier using the plurality of positive samples and the plurality of negative samples, the classifier outputting only two types of results. After the training is successful, the deep CNN-based classifier can be used to detect whether the driver in the target DMS image wears the seatbelt, and when the classifier output result is 1 (i.e., output is 1), it is determined that the driver wears the seatbelt, and when the classifier output result is another number such as 0, it is determined that the driver does not wear the seatbelt.
The seat belt detection method and the corresponding device according to the present invention will be described in detail with reference to a specific embodiment, however, it should be noted that the specific embodiment is only for better describing the present invention and should not be construed as limiting the present invention.
Figure 2 illustrates a flow chart of another DMS-based seat belt detection method according to the present invention. In addition, fig. 3 exemplarily shows a schematic diagram of segment labeling of a seat belt worn by a driver in a DMS registration image.
As shown in fig. 2, a DMS-based seat belt detection method 200 according to the present invention may include the steps of:
(1) a DMS registration image of the driver is acquired and the seat belt worn by the driver in the DMS registration image is segmented as shown in step 201 in fig. 2.
In one embodiment according to the present invention, the DMS registration image may be an image of a driver wearing a seat belt correctly, which is acquired when a DMS device is installed in a cabin of a vehicle or when facial features of the driver are registered. Based on the DMS registration image, the seat belt worn by the driver can be segmented. In a specific embodiment according to the present invention, the uppermost end of the seat belt, the boundary between the seat belt and the shoulder, and the lowermost end of the seat belt, such as the latch, may be determined based on the actual position of the seat belt in the DMS registration image. In certain particular cases, if the lock is not in the DMS registration image, the visible lowermost end of the seat belt may be marked. In one embodiment according to the present invention, when the DMS device is installed in the cabin of the vehicle or when the facial features of the driver are registered, the installation angle of the DMS device may be adjusted and an image of the registered driver wearing the seat belt correctly, that is, the DMS registration image, may be simultaneously acquired. Based on the DMS registration image, a seat belt worn by the driver in the DMS registration image may be segmented. As shown in fig. 3, the uppermost end of the harness (e.g. the connection point of the harness to the cabin backrest or seat) to the interface of the harness to the shoulder of the driver is denoted as the first part of the harness, and the interface of the harness to the shoulder of the driver to the buckle of the harness or the visibly lowermost end of the harness is denoted as the second part of the harness. In a specific embodiment according to the invention, a second part of the safety belt (here the second part of the actual safety belt) is provided with predetermined identification information, which may be, for example, a bar code or other identification pattern visible in the infrared, which is applied or printed using an infrared-sensitive paint or lacquer. The identification pattern may also be a bar code or a two-dimensional code.
(2) The driver's identity information is bound to the seat belt, as shown in step 202 of fig. 2.
the purpose of step 202 is to establish a correspondence or association between the preset identification information on the seat belt and the facial features of the driver, and to store the correspondence or association at the same time. In one embodiment according to the present invention, an ir-sensitive barcode or other identifying pattern feature code or feature vector provided on the second portion of the seat belt may be bound to the currently registered facial feature information of the driver so that a correlation is established between the two.
(3) The pinch angle value of the first portion of the belt from the vertical is determined based on the belt first partial image in the target DMS image, as shown in step 203 in fig. 2.
specifically, in step 203, a target DMS image is acquired from the DMS device, and then a first partial image of the seat belt corresponding to the first portion of the seat belt is extracted from the target DMS image. In a specific embodiment according to the present invention, a Canny edge detection algorithm is used to extract an edge of the first partial image of the safety belt, then a direct Hough transform is used to identify straight lines in the first partial image of the safety belt, so as to obtain a plurality of straight line segments in the first partial image of the safety belt, next, a plurality of target straight line segments of the plurality of straight line segments are selected, the length of each target straight line segment being greater than a preset length threshold (line _ threshold), and an included angle value arctan (Δ x/Δ y) between each target straight line segment and a vertical direction is calculated, where Δ x is a length value of the corresponding target straight line segment in an x-axis direction, and Δ y is a length value of the corresponding target straight line segment in a y-axis direction (i.e., a vertical direction). In a specific embodiment, the preset length threshold is set to line _ threshold ═ max (10,0.3 × part1_ height), where the number 10 represents 10 pixels, the number 0.3 represents a multiplication coefficient, and part1_ height represents the length of the first portion of the seat belt in the first portion of the seat belt (e.g., the length of the oblique side of the first portion of the seat belt), in pixels. The included angle value of the first part of the safety belt in the vertical direction is a weighted average value of the included angle values of the plurality of target straight line segments in the vertical direction, wherein the weight coefficient of each target straight line segment is a ratio of the length value of each target straight line segment to the total length value (namely the length accumulation sum) of all the plurality of target straight line segments.
(4) and comparing the included angle value of the first part of the safety belt and the vertical direction with a preset included angle threshold value so as to judge whether the included angle value is smaller than the preset included angle threshold value.
specifically, in the particular embodiment shown in fig. 2, after determining the pinch angle value (belt angle) of the first portion of the belt from the vertical direction based on the belt first partial image in the target DMS image, as shown in step 204 in fig. 2, it is determined whether the pinch angle value satisfies a preset condition: and (3) belt _ angle > -belt _ angle _ reg 0.7, wherein belt _ angle is an angle value between the first part of the safety belt based on the target DMS image and the vertical direction, and belt _ angle _ reg 0.7 is the preset angle threshold, and wherein belt _ angle _ reg is an angle value between the first part of the safety belt and the vertical direction calculated by using the DMS registration image to replace the target DMS image in the previous step (3), that is, an angle value between the first part of the safety belt based on the registration image and the vertical direction. If the angle between the first portion of the seat belt based on the target DMS image and the vertical direction satisfies the preset condition belt _ angle > -belt _ angle _ reg 0.7, the process proceeds to step 205, and if not, the process ends, and a determination result that the driver does not wear the seat belt is output, as shown in step 210 in fig. 2.
(5) The maximum confidence level that the preset identification information is present in the second partial image of the seat belt is detected on the basis of the second partial image of the seat belt, as shown in step 205 in fig. 2. In step 205, identity information of the current driver is obtained by the face recognition module of the DMS device, for example, facial features of the current driver are recognized, and an infrared-sensitive barcode or other identifying pattern feature code or feature vector, which is arranged on the second part (part2) of the seat belt and is bound to the facial features of the current driver, is read according to the facial features of the current driver. In a specific embodiment according to the present invention, a plurality of confidence levels (confidence) of the occurrence of the preset identification information in the second partial image of the seat belt are obtained by using sliding window detection, and then the confidence level with the largest value among the plurality of confidence levels is determined as the maximum confidence level (max _ confidence). One implementation of the sliding window detection is template matching detection.
(6) comparing the maximum confidence level with a preset confidence level threshold to determine whether the maximum confidence level is greater than the preset confidence level threshold. Specifically, in the specific embodiment shown in fig. 2, as shown in step 206 in fig. 2, after determining the maximum confidence level, the maximum confidence level is compared with a preset confidence level threshold, and whether the maximum confidence level meets a preset condition is determined: max _ confidence < ═ confidence _ threshold, where max _ confidence is the maximum confidence and confidence _ threshold is a preset confidence threshold. If the maximum confidence level meets the preset condition, the process proceeds to step 207, and if not, the operation is ended, and a determination result that the driver normally wears the seat belt is output, as shown in step 209 in fig. 2. The preset confidence threshold may be set according to actual needs, for example, may be set to be between 80% and 95%, for example, may be set to be 80%, 85%, 90%, 95%.
(7) Stitching the plurality of frames of the belt second partial images to form a plurality of positive and negative samples, obtaining a trained classifier based on the plurality of positive and negative samples, and inputting the belt second partial image in the target DMS image into the trained classifier, as shown in step 207 in fig. 2.
Specifically, in the particular embodiment shown in fig. 2, as shown in step 207 in fig. 2, historical video data of the current driver is collected using the DMS, which may contain a DMS image from which it has not been determined whether the driver is wearing the seatbelt (i.e., from step (1) to step (6)) by the above-described steps, and which contains a video of the second portion of the seatbelt. For the historical video data, 3 static images are sampled every 5 seconds to obtain 15 static images, then the static images are normalized into images with the size of 56 × 14 pixels, so that an image sequence consisting of 15 static images with the size of 56 × 14 pixels is obtained, and then the image sequence consisting of the 15 static images is horizontally spliced into images with the size of 56 (14 × 15) pixels according to the time sequence to serve as an image sample. In this way, a plurality of positive samples and a plurality of negative samples are collected from the historical video data, for example, each of the positive and negative samples is at least 5000, wherein the positive and negative samples mean: the driver normally wears the seat belt in the positive sample, and the driver does not normally wear the seat belt in the negative sample. Training a deep Convolutional Neural Network (CNN) based classifier using the plurality of positive samples and the plurality of negative samples, the classifier outputting only two types of results. After successful training, the deep CNN based classifier can be used to detect whether the driver in the target DMS image is wearing the seat belt, and then input the seat belt second partial image in the target DMS image to be evaluated into the trained classifier, as shown in step 207 in fig. 2.
(8) the output of the trained classifier is compared to a preset number (e.g., 1) to determine whether the driver is wearing a seat belt normally, as shown in step 208 of fig. 2. For example, if the classifier output result is 1 (i.e., output is 1), it is determined that the driver wears a seat belt, as shown in step 209. If the classifier outputs a different number, such as 0, then it is determined that the driver is not wearing a seat belt, as shown in step 210.
after determining that the driver is wearing the seat belt in step 209 and determining that the driver is not wearing the seat belt in step 210, the method may return to step 203 to continue to acquire the target DMS image and perform a new round of seat belt detection.
Based on the same inventive concept, the invention also provides a safety belt detection device based on the DMS, and the details are shown in the following embodiments. Since the technical problems to be solved by the DMS-based seat belt detection apparatus according to the present invention, the corresponding technical solutions and technical principles are similar to those of the DMS-based seat belt detection method according to the present invention, the specific implementation and corresponding technical effects of the DMS-based seat belt detection apparatus may refer to the implementation and technical effects of the DMS-based seat belt detection method, and repeated details are not repeated herein. The term "unit" or "module" used hereinafter may be a combination of software and/or hardware capable of realizing a predetermined function.
Fig. 4 shows a schematic block diagram of a DMS-based seat belt detection device according to the invention. Specifically, as shown in fig. 4, the DMS-based seat belt detection apparatus 400 according to the present invention includes: an acquisition module 401, the acquisition module 401 configured to acquire a target DMS image; an identification module 402, the identification module 402 configured to identify a first portion of a seat belt in the target DMS image, wherein the first portion is an uppermost end of the seat belt to a point where the seat belt interfaces with a driver shoulder; a first determination module 403, the first determination module 403 being configured to extract a first partial image of the safety belt corresponding to the first portion from the target DMS image and determine a pinch angle value of the first portion of the safety belt from a vertical direction based on the first partial image of the safety belt; a second determining module 404, wherein the second determining module 404 is configured to determine whether the included angle value is smaller than a preset included angle threshold value; and a third determination module 405, where the third determination module 405 is configured to determine that the driver does not wear a seat belt normally if the included angle value is smaller than the preset included angle threshold value.
Here, as will be readily understood by those skilled in the art, the above-described DMS-based seat belt detection apparatus according to the present invention includes not only the structural and functional features of the respective components of the seat belt detection apparatus, but also the image data transmission and processing features of the target DMS image transmitted and processed between the respective components, so that the electrical connection relationship between the respective components of the seat belt detection apparatus is implied in the description of the above-described seat belt detection apparatus. Meanwhile, from the description of the seat belt detection method corresponding to the seat belt detection device (for example, the description of fig. 1 and fig. 2), a person skilled in the art can easily know the process of transmitting and processing the image data processed by the seat belt detection device between the components and the generated technical effect, so that, in order to avoid redundancy and unnecessary difficulty in understanding the specific embodiment, the applicant will adopt a suitably simplified manner in the following description of the seat belt detection device.
In an embodiment of the present invention, the function performed by the first determining module 403 to determine the angle value of the first portion of the seat belt with respect to the vertical direction based on the first partial image of the seat belt includes: extracting an edge of the first partial image of the safety belt by using a Canny edge detection algorithm; identifying straight lines in the first partial image of the safety belt by using Hough transformation so as to obtain a plurality of straight line segments in the first partial image of the safety belt; determining a length value for each of the plurality of straight line segments; selecting a plurality of target straight-line segments of which the length values are larger than a preset length threshold value from the plurality of straight-line segments, and determining the included angle value of each target straight-line segment in the plurality of target straight-line segments and the vertical direction so as to obtain the included angle value of each target straight-line segment; and determining the included angle value of the first part of the safety belt in the vertical direction according to the included angle value of each target straight line segment and the length value of each target straight line segment.
Further, in an embodiment of the present invention, the determining the angle value of the first portion of the safety belt from the vertical direction according to the included angle value of each target straight line segment and the length value of each target straight line segment includes: determining a ratio of a length value of each target straight-line segment to a total length value of the plurality of target straight-line segments as a weight coefficient of each target straight-line segment; and multiplying and summing the included angle value of each target straight line segment by a corresponding weight coefficient to obtain the included angle value of the first part of the safety belt and the vertical direction.
Further, in an embodiment according to the present invention, in case that the included angle value is not less than the preset included angle threshold value, the seat belt detecting apparatus 400 is further configured to: (1) recognizing a facial feature of the driver and acquiring preset identification information associated with the facial feature, wherein the preset identification information is arranged on a second part of the safety belt from a boundary of the safety belt and the shoulder of the driver to the lowest end of the safety belt; (2) identifying a second portion of the seat belt in the target DMS image; (3) extracting a second partial image of the seat belt corresponding to the second portion from the target DMS image, and detecting a maximum confidence level that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt; (4) judging whether the maximum confidence coefficient is larger than a preset confidence coefficient threshold value; and (5) determining that the driver normally wears a seat belt if the maximum confidence is greater than the preset confidence threshold.
Further, in an embodiment according to the present invention, in case the maximum confidence is not greater than the preset confidence threshold, the seat belt detection device 400 is further configured for: (1) acquiring a trained classifier; (2) inputting the second partial image of the safety belt into the trained classifier for recognition so as to output a classification result; (3) and judging whether the driver wears the safety belt normally or not according to the classification result.
further, in an embodiment according to the present invention, the obtaining the trained classifier includes: acquiring historical video data of a current driver; collecting a plurality of positive samples in which the driver normally wears the seat belt and a plurality of negative samples in which the driver does not normally wear the seat belt from the historical video data; training a deep Convolutional Neural Network (CNN) based classifier using the plurality of positive samples and the plurality of negative samples.
in a preferred embodiment of the present invention, the detecting the maximum confidence that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt includes: acquiring a plurality of confidence degrees of the preset identification information in the second partial image of the safety belt by using sliding window detection; determining the confidence level with the largest value among the confidence levels as the maximum confidence level.
In another preferred embodiment of the present invention, the detecting the maximum confidence that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt includes: acquiring a plurality of confidence degrees of the preset identification information in the second partial image of the safety belt by utilizing template matching detection; determining the confidence level with the largest value among the confidence levels as the maximum confidence level.
Further, in a preferred embodiment according to the present invention, the preset confidence threshold is 80%, 85%, 90% or 95%.
further, in a preferred embodiment according to the present invention, the target DMS image is an infrared image, and the preset identification information is a bar code or a two-dimensional code coated or printed with an infrared-sensitive paint or paint.
From the above detailed description of the seat belt detecting apparatus 400, it can be seen that at least the following technical effects can be achieved according to the above-described embodiments of the present invention: the first part of the safety belt in the DMS image is identified, whether the driver wears the safety belt normally or not is determined according to whether the included angle value of the first part of the safety belt and the vertical direction is smaller than a preset included angle threshold value or not, and whether the driver wears the safety belt normally or not can be judged conveniently and reliably. Because the DMS device is installed in the vehicle, the interference of factors such as external light, weather conditions and the like can be effectively reduced by detecting the safety belt based on the DMS image, and the accuracy of safety belt detection is effectively improved.
Furthermore, the present invention also provides a computer device comprising a processor and a memory for storing executable instructions, which when executed by the processor, enable implementation of the DMS-based seat belt detection method according to the present invention. Fig. 5 illustrates a schematic block diagram of a computer apparatus according to the present invention. As shown in fig. 5, a computer device 500 according to the present invention is capable of implementing the DMS-based seat belt detection methods 100 and 200 according to the present invention. In particular, the computer device 500 may comprise an input device 501, a processor 502 and a memory 503, wherein the memory 503 is for storing processor executable instructions. The processor 502 when executing the instructions may implement the steps of the seat belt detection method described in any of the above embodiments.
In the embodiments of the present invention, the specific functions and technical effects implemented by the computer device can be understood by referring to other embodiments, which are not described herein.
Furthermore, the present invention also provides a computer readable storage medium having stored thereon computer instructions which, when executed, enable implementation of the DMS-based seat belt detection method according to the present invention.
In the embodiments of the present invention, specific functions and technical effects realized by the program instructions stored in the computer storage medium can be understood by referring to other embodiments, and are not described in detail herein.
All the above description is only a preferred embodiment of the present invention and should not be taken as limiting the invention. It will be apparent to those skilled in the art that various modifications and variations can be made in the embodiments of the present invention. Any modification, equivalent replacement, or improvement made without departing from the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (10)
1. A Driver Monitoring System (DMS) -based seat belt detection method, comprising the steps of:
(1) Acquiring a DMS registration image of a driver and carrying out sectional labeling on a safety belt worn by the driver in the DMS registration image;
(2) Binding the identity information of the driver with the safety belt;
(3) Acquiring a target DMS image and determining a included angle value of a first part of a safety belt from a vertical direction based on a first partial image of the safety belt in the target DMS image, wherein the first part is from the uppermost end of the safety belt to a boundary between the safety belt and a shoulder of a driver;
(4) Comparing the included angle value of the first part of the safety belt and the vertical direction with a preset included angle threshold value so as to judge whether the included angle value is smaller than the preset included angle threshold value; and
(5) and under the condition that the included angle value is smaller than the preset included angle threshold value, judging that the driver does not wear the safety belt normally.
2. the seat belt detection method according to claim 1, wherein the determining in step (3) the value of the angle of the first portion of the seat belt from the vertical based on the first partial image of the seat belt in the target DMS image comprises the sub-steps of:
(3-1) extracting an edge of the first partial image of the seat belt using a Canny edge detection algorithm;
(3-2) recognizing a straight line in the first partial image of the seat belt using a Hough transform, thereby obtaining a plurality of straight line segments in the first partial image of the seat belt;
(3-3) determining a length value of each of the plurality of straight line segments;
(3-4) selecting a plurality of target straight-line segments of which the length values are larger than a preset length threshold value from the plurality of straight-line segments, and determining the included angle value between each target straight-line segment of the plurality of target straight-line segments and the vertical direction, so as to obtain the included angle value of each target straight-line segment; and
(3-5) determining the included angle value of the first part of the safety belt with the vertical direction according to the included angle value of each target straight line segment and the length value of each target straight line segment.
3. the seat belt detection method according to claim 2, wherein the sub-step (3-5) of determining the value of the angle of the first portion of the seat belt with respect to the vertical direction from the value of the included angle of each target straight line segment and the value of the length of each target straight line segment comprises:
Determining a ratio of a length value of each target straight-line segment to a total length value of the plurality of target straight-line segments as a weight coefficient of each target straight-line segment;
And multiplying and summing the included angle value of each target straight line segment by a corresponding weight coefficient to obtain the included angle value of the first part of the safety belt and the vertical direction.
4. the seat belt detection method according to one of claims 1 to 3, wherein in the case where the included angle value is not less than the preset included angle threshold value, the seat belt detection method further comprises the following first-stage sub-steps:
(1) Recognizing a facial feature of the driver and acquiring preset identification information associated with the facial feature, wherein the preset identification information is arranged on a second part of the safety belt from a boundary of the safety belt and the shoulder of the driver to the lowest end of the safety belt;
(2) Identifying a second portion of the seat belt in the target DMS image;
(3) Extracting a second partial image of the seat belt corresponding to the second portion from the target DMS image, and detecting a maximum confidence level that the preset identification information appears in the second partial image of the seat belt based on the second partial image of the seat belt;
(4) Comparing the maximum confidence level with a preset confidence level threshold value so as to judge whether the maximum confidence level is larger than the preset confidence level threshold value; and
(5) And under the condition that the maximum confidence degree is larger than the preset confidence degree threshold value, judging that the driver normally wears a safety belt.
5. the seat belt detection method according to claim 4, wherein in case the maximum confidence is not greater than the preset confidence threshold, the seat belt detection method further comprises the following second-stage substeps:
(1) Acquiring a trained classifier;
(2) Inputting the second partial image of the safety belt into the trained classifier for recognition so as to output a classification result;
(3) And judging whether the driver wears the safety belt normally or not according to the classification result.
6. The seat belt detection method according to claim 5, characterized in that the second-stage substep (1) of obtaining a classifier that has been trained comprises:
Acquiring historical video data of a current driver;
Collecting a plurality of positive samples and a plurality of negative samples from the historical video data, wherein the plurality of positive samples and the plurality of negative samples are formed by splicing a plurality of frames of second partial images of the seat belt, and wherein the seat belt is normally worn by the driver in the positive samples and the seat belt is not normally worn by the driver in the negative samples;
Training a deep Convolutional Neural Network (CNN) based classifier using the plurality of positive samples and the plurality of negative samples.
7. The seat belt detection method according to claim 4, characterized in that the detection of the maximum confidence level of the presence of the preset identification information in the seat belt second partial image based on the seat belt second partial image in the first-stage sub-step (3) comprises:
Acquiring a plurality of confidence degrees of the preset identification information in the second partial image of the safety belt by utilizing template matching detection;
determining the confidence level with the largest value among the confidence levels as the maximum confidence level.
8. The seat belt detection method according to claim 4, characterized in that the preset confidence threshold is between 80% and 95%.
9. the seat belt detection method according to one of claims 1 to 3, wherein the target DMS image is an infrared image, and the preset identification information is a bar code or a two-dimensional code coated or printed with an infrared-sensitive paint or paint.
10. A safety belt detection device based on a Driver Monitoring System (DMS), characterized in that it is capable of implementing the method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910734872.3A CN110569732A (en) | 2019-08-09 | 2019-08-09 | Safety belt detection method based on driver monitoring system and corresponding equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910734872.3A CN110569732A (en) | 2019-08-09 | 2019-08-09 | Safety belt detection method based on driver monitoring system and corresponding equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110569732A true CN110569732A (en) | 2019-12-13 |
Family
ID=68774943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910734872.3A Pending CN110569732A (en) | 2019-08-09 | 2019-08-09 | Safety belt detection method based on driver monitoring system and corresponding equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110569732A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458093A (en) * | 2019-08-09 | 2019-11-15 | 径卫视觉科技（上海）有限公司 | A kind of Safe belt detection method and corresponding equipment based on driver's monitoring system |
| EP4358040A1 (en) | 2022-10-18 | 2024-04-24 | Smart Eye AB | Seat belt detection |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009165517A (en) * | 2008-01-11 | 2009-07-30 | Nakanishi Metal Works Co Ltd | System for monitoring and alerting use condition of safety belt |
| CN204845863U (en) * | 2015-07-09 | 2015-12-09 | 北京汽车股份有限公司 | Car with safety belt detecting system |
| CN106203499A (en) * | 2016-07-07 | 2016-12-07 | 青岛大学 | A kind of Safe belt detection method based on degree of depth CNN |
| CN106427881A (en) * | 2016-07-30 | 2017-02-22 | 桐城市祥瑞机动车辆检测有限公司 | System for detecting whether car safety belt is worn correctly and detection method thereof |
| CN106485224A (en) * | 2016-10-13 | 2017-03-08 | 北京智芯原动科技有限公司 | A kind of seatbelt wearing recognition methodss and device |
| CN107529659A (en) * | 2017-07-14 | 2018-01-02 | 深圳云天励飞技术有限公司 | Seatbelt wearing detection method, device and electronic equipment |
| CN108099839A (en) * | 2017-12-25 | 2018-06-01 | 芜湖炬胜机电设备厂 | A kind of safety belt detection device |
| CN109460699A (en) * | 2018-09-03 | 2019-03-12 | 厦门瑞为信息技术有限公司 | A kind of pilot harness's wearing recognition methods based on deep learning |
| CN109878449A (en) * | 2019-04-03 | 2019-06-14 | 陈轶芃 | A kind of seatbelt wearing recognition methods |
-
2019
- 2019-08-09 CN CN201910734872.3A patent/CN110569732A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009165517A (en) * | 2008-01-11 | 2009-07-30 | Nakanishi Metal Works Co Ltd | System for monitoring and alerting use condition of safety belt |
| CN204845863U (en) * | 2015-07-09 | 2015-12-09 | 北京汽车股份有限公司 | Car with safety belt detecting system |
| CN106203499A (en) * | 2016-07-07 | 2016-12-07 | 青岛大学 | A kind of Safe belt detection method based on degree of depth CNN |
| CN106427881A (en) * | 2016-07-30 | 2017-02-22 | 桐城市祥瑞机动车辆检测有限公司 | System for detecting whether car safety belt is worn correctly and detection method thereof |
| CN106485224A (en) * | 2016-10-13 | 2017-03-08 | 北京智芯原动科技有限公司 | A kind of seatbelt wearing recognition methodss and device |
| CN107529659A (en) * | 2017-07-14 | 2018-01-02 | 深圳云天励飞技术有限公司 | Seatbelt wearing detection method, device and electronic equipment |
| CN108099839A (en) * | 2017-12-25 | 2018-06-01 | 芜湖炬胜机电设备厂 | A kind of safety belt detection device |
| CN109460699A (en) * | 2018-09-03 | 2019-03-12 | 厦门瑞为信息技术有限公司 | A kind of pilot harness's wearing recognition methods based on deep learning |
| CN109878449A (en) * | 2019-04-03 | 2019-06-14 | 陈轶芃 | A kind of seatbelt wearing recognition methods |
Non-Patent Citations (3)
| Title |
|---|
| FU YUAN WANG等: "Development of Safety Detection Integrated Equipment for Car Seat", 《APPLIED MECHANICS AND MATERIALS》 * |
| XH QIN等: "Efficient seat belt detection in a vehicle surveillance application", 《2014 9TH IEEE CONFERENCE ON INDUSTRIAL ELECTRONICS AND APPLICATIONS》 * |
| 胡冻三等: "基于单片机的客车安全带智能检测系统", 《电子世界》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458093A (en) * | 2019-08-09 | 2019-11-15 | 径卫视觉科技（上海）有限公司 | A kind of Safe belt detection method and corresponding equipment based on driver's monitoring system |
| CN110458093B (en) * | 2019-08-09 | 2024-04-09 | 径卫视觉科技（上海）有限公司 | Safety belt detection method based on driver monitoring system and corresponding equipment |
| EP4358040A1 (en) | 2022-10-18 | 2024-04-24 | Smart Eye AB | Seat belt detection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110458093B (en) | Safety belt detection method based on driver monitoring system and corresponding equipment | |
| JP5127392B2 (en) | Classification boundary determination method and classification boundary determination apparatus | |
| CN108446630B (en) | Intelligent monitoring method for airport runway, application server and computer storage medium | |
| EP1782335B1 (en) | Method for traffic sign detection | |
| CN105488453B (en) | A kind of driver based on image procossing does not fasten the safety belt detection recognition method | |
| CN105631414B (en) | A kind of vehicle-mounted multi-obstacle avoidance sorter and method based on Bayes classifier | |
| US8824742B2 (en) | Occupancy detection for managed lane enforcement based on localization and classification of windshield images | |
| KR101653278B1 (en) | Face tracking system using colar-based face detection method | |
| KR101176552B1 (en) | Method and apparatus for recognizing speed limit signs and method for recognizing image | |
| US11334767B2 (en) | System and method of perception error evaluation and correction by solving optimization problems under the probabilistic signal temporal logic based constraints | |
| CN110569732A (en) | Safety belt detection method based on driver monitoring system and corresponding equipment | |
| CN111860610A (en) | License plate overexposure and license plate vehicle identification method and device and computer storage medium | |
| KR101084594B1 (en) | Real time image recognition system, and method thereof | |
| CN113408364B (en) | Temporary license plate recognition method, system, device and storage medium | |
| CN113807220A (en) | Traffic event detection method and device, electronic equipment and readable storage medium | |
| Yi et al. | Safety belt wearing detection algorithm based on human joint points | |
| Harish et al. | Identification of lane lines using advanced machine learning | |
| Desai et al. | Real-time implementation of Indian license plate recognition system | |
| Wang et al. | A two-layer night-time vehicle detector | |
| Kodwani et al. | Automatic license plate recognition in real time videos using visual surveillance techniques | |
| CN111046876B (en) | License plate character rapid recognition method and system based on texture detection technology | |
| CN111626334B (en) | Key control target selection method for vehicle-mounted advanced auxiliary driving system | |
| KR101936108B1 (en) | Method and apparatus for detecting traffic sign | |
| CN111523417A (en) | Truck overload identification method and device based on image matching | |
| US11854252B1 (en) | Automated probabilistic axiom generation and incremental updates |
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 |