TWI806036B - Electronic device and method for monitoring posture - Google Patents

Abstract

An electronic device and a method for monitoring a sitting posture are provided. The method includes: pre-storing a plurality of interpupillary distances (IPDs) corresponding to a plurality of groups respectively; capturing an image of a user by a camera; determining that the user corresponds to a first group of the plurality of groups by a first machine learning model, wherein the first group corresponds to a first IPD of the plurality of IPDs; obtaining an IPD in image of the user according to the image; calculating a distance between the user and the camera according to the first IPD and the IPD in image; and outputting an warning message in response to the distance exceeding a default range.

Description

Electronic device and method for monitoring posture

The invention relates to an electronic device and method for monitoring sitting posture.

At present, there are many jobs that need to rely on personnel to stay in the office for a long time to operate computers. Maintaining an improper sitting posture for a long time may not only reduce work efficiency, but may also have a negative impact on the health of personnel. In order to enable people to maintain a good sitting posture, warning systems that require multiple sitting postures have appeared on the market. For example, a traditional sitting posture warning system can detect whether the user's sitting posture is correct by using a hardware device installed on a desk, a chair, or the user. However, the above-mentioned sitting posture warning system needs to rely on additional hardware devices. Therefore, the above-mentioned sitting posture warning system is not only inconvenient to use, but also needs to consume relatively high cost.

Another traditional sitting position warning system can detect the user's sitting position through more than two video cameras. However, the increase in video cameras represents an increase in the cost of the posture warning system. In addition, the increase of the video camera also reduces the layout space of the components of the sitting posture warning system.

The invention provides an electronic device and method for monitoring sitting posture, which can use a single camera to monitor whether the user's sitting posture is correct.

An electronic device for monitoring sitting posture of the present invention includes a processor, a storage medium, a transceiver and a camera. The camera captures images of the user. The storage medium stores multiple modules and multiple eye distances respectively corresponding to multiple groups. The processor is coupled to the storage medium, the transceiver and the camera, and accesses and executes a plurality of modules, wherein the plurality of modules include a first machine learning model and a computing module. The first machine learning model judges according to the image that the user corresponds to the first group among the plurality of groups, wherein the first group corresponds to the first eye distance among the plurality of eye distances. The calculation module obtains the user's eye distance in the image according to the image, calculates the distance between the user and the camera according to the first eye distance and the image center eye distance, and outputs a warning message through the transceiver in response to the distance exceeding a preset range.

In an embodiment of the present invention, the aforementioned modules further include a feature extraction module. The feature extraction module obtains multiple feature points of the user's face according to the image, and the calculation module calculates the eye distance in the image according to the multiple feature points.

In an embodiment of the present invention, the above-mentioned calculation module judges the user's head movement according to a plurality of feature points, corrects the first eye distance according to the head movement to generate the first corrected eye distance, and according to the first corrected eye distance distance and the eye distance in the image to calculate the distance.

In an embodiment of the present invention, the above-mentioned head movement includes at least one of up-down turning, left-right yaw, and left-right turning, wherein the calculation module reduces the first eye distance according to the head movement to generate the first corrected eye distance .

In an embodiment of the present invention, the above-mentioned first machine learning model judges the race and age of the user according to the image, and judges that the user corresponds to the first group according to the race and age.

In an embodiment of the present invention, the above-mentioned modules further include a second machine learning model. The second machine learning model judges the user's head position and shoulder position according to the image, wherein the calculation module judges whether to output the second warning message through the transceiver according to the head position and shoulder position.

In an embodiment of the present invention, the above-mentioned calculation module obtains the horizontal axis of the head according to the position of the head, obtains the shoulder line according to the position of the shoulder, and determines whether to output the second warning message according to the horizontal axis of the head and the shoulder line.

In an embodiment of the present invention, the above-mentioned computing module determines to output the second warning message in response to the fact that the horizontal axis of the head is not parallel to the shoulder line.

A method for monitoring sitting posture of the present invention includes: pre-storing a plurality of eye distances corresponding to a plurality of groups; capturing an image of a user through a camera; judging by a first machine learning model that the user corresponds to a plurality of groups according to the image The first group in the group, wherein the first group corresponds to the first eye distance among the plurality of eye distances; obtain the user's eye distance in the image according to the image; calculate the user according to the first eye distance and the eye distance in the image distance from the camera; and outputting a warning message in response to the distance exceeding a preset range.

Based on the above, the electronic device of the present invention can judge whether the user's sitting posture is correct based on information such as the user's eye distance, head position, and shoulder position, and can output a warning message to remind the user to adjust when the user's sitting posture is incorrect. sitting posture.

100: Electronic device

110: Processor

120: storage media

121:First machine learning model

122:Second machine learning model

123: Feature extraction module

124: Operation module

130: Transceiver

140: camera

20: first eye distance

30: Eye distance in the image

40: Distance

50: head position

51: Head horizontal axis

60: shoulder position

61: shoulder line

F: focal length

S601, S602, S603, S604, S605, S606: steps

FIG. 1 is a schematic diagram of an electronic device for monitoring sitting posture according to an embodiment of the present invention.

FIG. 2 is a schematic diagram illustrating a user's first eye distance according to an embodiment of the present invention.

FIG. 3 is a schematic diagram illustrating the distance between a user and a camera according to an embodiment of the present invention.

4A , 4B and 4C illustrate schematic diagrams of head movements according to an embodiment of the present invention.

FIG. 5 is a schematic diagram illustrating a user's head position and shoulder position according to an embodiment of the present invention.

FIG. 6 is a flow chart of a method for monitoring sitting posture according to an embodiment of the present invention.

In order to make the content of the present invention more comprehensible, the following specific embodiments are taken as examples in which the present invention can actually be implemented. In addition, wherever possible, elements/components/steps using the same reference numerals in the drawings and embodiments represent the same or similar parts.

Fig. 1 illustrates an electronic device for monitoring sitting posture according to an embodiment of the present invention Schematic diagram of setting 100. The electronic device 100 may include a processor 110 , a storage medium 120 , a transceiver 130 and a camera 140 . The electronic device 100 is, for example, a notebook computer, a desktop computer or a tablet computer, and the present invention is not limited thereto.

The processor 110 is, for example, a central processing unit (central processing unit, CPU), or other programmable general purpose or special purpose micro control unit (micro control unit, MCU), microprocessor (microprocessor), digital signal processing Digital Signal Processor (DSP), Programmable Controller, Application Specific Integrated Circuit (ASIC), Graphics Processing Unit (GPU), Image Signal Processor (ISP) ), image processing unit (image processing unit, IPU), arithmetic logic unit (arithmetic logic unit, ALU), complex programmable logic device (complex programmable logic device, CPLD), field programmable logic gate array (field programmable gate array , FPGA) or other similar components or a combination of the above components. The processor 110 can be coupled to the storage medium 120 , the transceiver 130 and the camera 140 , and access and execute multiple modules and various application programs stored in the storage medium 120 .

The storage medium 120 is, for example, any type of fixed or removable random access memory (random access memory, RAM), read-only memory (read-only memory, ROM), flash memory (flash memory) , hard disk drive (hard disk drive, HDD), solid state drive (solid state drive, SSD) or similar components or a combination of the above components, and are used to store multiple modules or various application programs that can be executed by the processor 110 . In this embodiment, the storage medium 120 can store It includes multiple modules such as the first machine learning model 121 , the second machine learning model 122 , the feature extraction module 123 and the computing module 124 , and their functions will be described later.

The transceiver 130 transmits and receives signals in a wireless or wired manner. The transceiver 130 may also perform operations such as low noise amplification, impedance matching, frequency mixing, up or down frequency conversion, filtering, amplification, and the like.

The camera 140 is used to capture images of users. For example, if the electronic device 100 is a notebook computer, the camera 140 may be a video lens (webcam) disposed around the display of the notebook computer. When the user operates the electronic device 100, the user will face the camera 140, and the camera 140 can capture the image of the user.

The interpupillary distance (IPD) is the distance between the pupil centers of the two eyes. According to statistics, people of different races or ages have different eye distances. For example, as we age, the distance between our eyes decreases. The storage medium 120 may pre-store multiple eye distances corresponding to multiple groups, wherein the multiple groups correspond to different races and/or ages.

The first machine learning model 121 can be used to determine which group among the multiple groups the user corresponds to according to the image of the user. For example, the first machine learning model 121 can determine the user's race and age according to the user's image, thereby judging that the user belongs to the first group corresponding to the race and age according to the race and age, The first group may correspond to the first eye distance among the plurality of eye distances stored in the storage medium 120 . Therefore, the first machine learning model 121 can determine that the eye distance of the user is the first eye distance, as shown in FIG. 2 . Figure 2 illustrates the use of Schematic diagram of the first eye distance 20 of the patient.

In one embodiment, the computing module 122 can train and generate the first machine learning model 121 according to a known machine learning algorithm, which is not limited in the present invention. For example, the computing module 122 can obtain multiple pieces of training data through the transceiver 130, wherein each piece of training data in the multiple pieces of training data can include the image of the person, the label of the race of the person, and the label of the age of the person. The calculation module 122 can generate the first machine learning model 121 based on a supervised learning algorithm according to multiple training data.

On the other hand, the computing module 124 can obtain the eye distance in the user's image according to the user's image (that is, the user's eye distance in the image). Specifically, the feature extraction module 123 can acquire a plurality of feature points of the user's face according to the user's image. The calculation module 124 can calculate the eye distance of the user's image according to the plurality of feature points.

After obtaining the eye distance of the user in the image, the computing module 124 can calculate the distance between the user and the camera 140 according to the first eye distance 20 and the eye distance in the image. FIG. 3 shows a schematic diagram of the distance 40 between the user and the camera 140 according to an embodiment of the present invention, wherein F is the focal length of the camera 140 , 30 is the middle eye distance of the user's image, and 20 is the first eye distance. When the focal length F, the eye distance 30 in the image, and the first eye distance 20 are known, the calculation module 124 can calculate the distance 40 between the user and the camera 140 according to the following formula (1).

Distance 40=(first eye distance 20/image distance 30)*focal length F...(1)

The measurement of the eye distance may be affected by the user's head movement. Therefore, in one embodiment, the computing module 122 can adjust the distance according to the user's head movement 40 is calculated. Specifically, the calculation module 122 can judge the user's head movement according to the multiple feature points of the user's face generated by the feature extraction module 123 . 4A , 4B and 4C illustrate schematic diagrams of head movement according to an embodiment of the present invention. The head movement may include pitch movement as shown in FIG. 4A , roll movement as shown in FIG. 4B , and yaw movement as shown in FIG. 4C .

The computing module 122 can correct the first interocular distance 20 according to the user's head movement to generate the first corrected interocular distance. Taking left-right turning as an example, if the calculation module 122 judges that the user's head is turned left-right according to multiple feature points of the user's face, the calculation module 122 can reduce the first eye distance 20 according to the angle of the left-right turning to generate The first corrected eye distance. After generating the first corrected eye distance, the calculation module 122 can calculate the distance 40 according to the first corrected eye distance, as shown in formula (2).

Distance 40=(first corrected eye distance/eye distance 30 in the image)*focal length F...(2)

After the distance 40 is calculated, the computing module 122 can output a warning message through the transceiver 130 in response to the distance 40 exceeding the preset range, so as to remind the user to adjust the distance 40 between the user and the camera 140 through the warning message. For example, the storage medium 120 may pre-store a preset lower bound and a preset upper bound of the distance 40 . If the distance 40 is less than the preset lower bound, it means that the user is too close to the camera 140 . Accordingly, the computing module 122 can output a warning message to remind the user to stay away from the camera 140 to improve the sitting posture. If the distance 40 is greater than the preset upper limit, it means that the user is too far away from the camera 140 . Accordingly, the computing module 122 can output a warning message to remind the user to approach the camera 140 to improve the sitting posture.

In addition to the distance between the user and the camera 140 , the electronic device 100 can also determine whether the user's sitting posture needs to be improved according to the relative relationship between the user's head and shoulders. Specifically, the second machine learning model 122 can determine the user's head position 50 and shoulder position 60 according to the user's image, as shown in FIG. 5 . FIG. 5 shows a schematic diagram of a user's head position 50 and shoulder positions 60 according to an embodiment of the present invention.

In one embodiment, the calculation module 122 can train and generate the second machine learning model 122 according to a known machine learning algorithm, which is not limited in the present invention. For example, the computing module 122 can obtain multiple pieces of training data through the transceiver 130 , wherein each piece of training data in the multiple pieces of training data can include the person's image, the person's head position label, and the person's shoulder position label. The calculation module 122 can generate the second machine learning model 122 based on a supervised learning algorithm according to a plurality of training data.

After obtaining the user's head position 50 and shoulder position 60, the computing module can judge whether to output a second warning message according to the head position 50 and shoulder position 60, wherein the second warning message is used to remind the user to adjust the head position 50 or shoulder position 60 for improved sitting posture. Specifically, the calculation module 122 can obtain the head horizontal axis 51 according to the head position 50 , and obtain the shoulder line 61 according to the shoulder position 60 . Next, the computing module 122 can determine whether the horizontal axis 51 of the head is parallel to the shoulder line 61 . If the head horizontal axis 51 is not parallel to the shoulder line 61, the computing module 122 can output a second warning message through the transceiver 130, so as to prompt the user to adjust the head position 50 or the shoulder position 60 so that the head The horizontal axis 51 is parallel to the shoulder line 61 .

FIG. 6 shows a flowchart of a method for monitoring sitting posture according to an embodiment of the present invention, wherein the method can be implemented by the electronic device 100 shown in FIG. 1 . In step S601, a plurality of eye distances respectively corresponding to a plurality of groups are prestored. In step S602, an image of the user is captured. In step S603, the first machine learning model judges according to the image that the user corresponds to the first group among the plurality of groups, wherein the first group corresponds to the first eye distance among the plurality of eye distances. In step S604, the user's eye distance in the image is obtained according to the image. In step S605, the distance between the user and the camera is calculated according to the first eye distance and the eye distance in the image. In step S606, a warning message is output in response to the distance exceeding a preset range.

In summary, the electronic device of the present invention can determine the user's race or age according to the user's facial features. After confirming the user's race or age, the electronic device can judge the distance between the user and the camera according to the eye distance, thereby judging whether the user's sitting posture is correct according to the distance. On the other hand, the electronic device can also judge whether the user's sitting posture is correct according to the user's head position and shoulder position. When the user's sitting posture is incorrect, the electronic device can output a warning message to prompt the user to change the sitting posture. Therefore, the electronic device of the present invention can assist the user to maintain a good sitting posture.

S601, S602, S603, S604, S605, S606: steps

Claims (6)

An electronic device for monitoring gestures, comprising: a camera for capturing images of a user; a transceiver; a storage medium for storing a plurality of modules and a plurality of eye distances corresponding to a plurality of groups; and a processor coupled to all The storage medium, the transceiver, and the camera, and access and execute the multiple modules, wherein the multiple modules include: a first machine learning model, judging from the image that the user corresponds to In the first group of the plurality of groups, wherein the first group corresponds to the first eye distance of the plurality of eye distances; the computing module obtains the user's distance from the image according to the image an eye distance in an image, calculating the distance between the user and the camera according to the first eye distance and the eye distance in the image, and outputting it through the transceiver in response to the distance exceeding a preset range a warning message; and a feature extraction module, which obtains a plurality of feature points of the user's face according to the image, wherein the calculation module calculates the eye distance in the image according to the plurality of feature points, according to The plurality of feature points determine the user's head movement, correct the first eye distance according to the head movement to generate a first corrected eye distance, and based on the first corrected eye distance and the image The middle eye distance is used to calculate the distance; the second machine learning model judges the user's head according to the image. head position and shoulder position, wherein the computing module determines whether to output a second warning message through the transceiver according to the head position and the shoulder position. The electronic device according to claim 1, wherein the head motion includes at least one of up-down flip, left-right yaw, and left-right flip, wherein the computing module reduces the first-eye view according to the head motion distance to generate the first corrected eye distance. The electronic device according to claim 1, wherein the first machine learning model judges the race and age of the user according to the image, and judges that the user corresponds to the first group. The electronic device according to claim 1, wherein the calculation module obtains the head horizontal axis according to the head position, obtains the shoulder line according to the shoulder position, and obtains the shoulder line according to the head horizontal axis and the shoulder line It is judged whether to output the second warning message. The electronic device according to claim 4, wherein the computing module determines to output the second warning message in response to the fact that the horizontal axis of the head is not parallel to the shoulder line. A method for monitoring gestures, comprising: pre-storing a plurality of eye distances respectively corresponding to a plurality of groups; capturing an image of a user through a camera; judging by a first machine learning model according to the image that the user corresponds to the The first group in the plurality of groups, wherein the first group corresponds to the first eye distance in the plurality of eye distances; obtaining the eye distance in the image of the user according to the image includes : Acquiring multiple feature points of the user's face according to the image; judging the head movement of the user according to the multiple feature points; correcting the first eye distance according to the head movement to generating a first corrected eye distance; and calculating the distance according to the first corrected eye distance and the eye distance in the image; calculating the distance between the user and the image based on the first eye distance and the eye distance in the image the distance between the cameras; and outputting a warning message in response to the distance exceeding a preset range; judging the head position and the shoulder position of the user according to the image by the second machine learning model, wherein according to the head position and the shoulder position to determine whether to output a second warning message.

Images