WO2021138964A1

WO2021138964A1 - Read/write distance identification method based on smart watch

Info

Publication number: WO2021138964A1
Application number: PCT/CN2020/075568
Authority: WO
Inventors: 鄢家厚
Original assignee: 鄢家厚
Priority date: 2020-01-10
Filing date: 2020-02-17
Publication date: 2021-07-15
Also published as: CN111240481A; CN111240481B

Abstract

A read/write distance identification method based on a smart watch. The method comprises: step 1: constructing a sitting posture identification model; step 2: according to an acceleration sensor and a gyroscope sensor in a smart watch, performing detection to obtain first target data, and determining, on the basis of the first target data, the current read/write state of a target user (05) wearing the smart watch; step 3: according to a determination result, controlling a distance sensor group (01, 02) in the smart watch to perform detection to obtain second target data, and by means of the second target data, determining the current read/write distance, corresponding to the current read/write state, of the target user (05); and step 4: furthermore, on the basis of the constructed sitting posture identification model, identifying the first target data and the second target data, and determining and recording the current sitting posture of the target user (05). First target data and second target data are acquired to determine the read/write distance of the target user (05), thereby facilitating the determination of the current sitting posture of the target user (05).

Description

Read-write distance recognition method based on smart watch

Technical field

The present invention relates to the technical field of mobile sensing, and in particular to a method for recognizing a reading and writing distance based on a smart watch.

Background technique

With the advent of the information explosion era, both students and adults need to do a lot of desk work and study every day. However, in the process of work and study, the wrong sitting posture can cause fatigue and reduce vision, especially for students. Although teachers generally correct the sitting posture of students, it is only limited to the situation of the teacher in the classroom, and, There are also situations where students subconsciously appear incorrect sitting postures, and students may not care about these things, but they seriously affect the correct sitting posture of students. Therefore, on the basis of determining the sitting posture of students, the reading and writing distance of students It is particularly important to carry out the test.

Summary of the invention

The present invention provides a read-write distance recognition method based on a smart watch, which is used to determine the user's read-write distance by acquiring first target data and second target data, thereby facilitating the determination of the user's current sitting posture.

The embodiment of the present invention provides a method for recognizing a read-write distance based on a smart watch, which includes:

Build a seated recognition model;

The first target data is detected according to the acceleration sensor and the gyroscope sensor in the smart watch, and based on the first target data, the current reading and writing status of the target user wearing the smart watch is determined;

According to the judgment result, control the distance sensing group in the smart watch to detect the second target data, and judge the current reading and writing distance of the target user corresponding to the current reading and writing state through the second target data;

At the same time, based on the constructed sitting posture recognition model, the first target data and the second target data are recognized, and the current sitting posture of the target user is determined and recorded.

In a possible implementation manner, the current sitting posture includes: a sitting type, a head-down reading and writing type, and a hand-supporting cheek writing type;

Wherein, the step of recognizing the first target data and the second target data based on the constructed sitting posture recognition model, and determining the current sitting posture of the target user includes:

Judging whether the target hand of the target user is placed horizontally on the target object based on the gyroscope sensor in the smart watch;

Judging the head distance between the smart watch and the target head of the target user based on the distance sensing group in the smart watch;

Based on the sitting posture recognition model, if it is determined that the target hand of the target user is placed horizontally on the target object, and the head distance is greater than a preset distance, then the current sitting posture of the target user is an upright sitting type;

If it is determined that the target hand of the target user is horizontally placed on the target object, and the head distance is not greater than a preset distance, then the current sitting posture of the target user is a head-down reading and writing type;

If it is determined that the target hand of the target user is not placed horizontally on the target object, and the first angle with the target object is within the preset angle range, then the current sitting posture of the target user is hand-supported cheek writing type .

In a possible implementation manner, the step of judging the current reading and writing distance of the target user corresponding to the current reading and writing state based on the judgment result and through the second target data includes:

When the judgment result is that the target user is currently in a reading and writing state, controlling the distance sensing group to emit laser light, and calculating the current reading and writing distance of the target user according to the laser emission-reflection time difference;

When the judgment result is that the target user is not currently in a read-write state, control the smart watch not to perform any operation;

Wherein, the second target data is the current reading and writing distance calculated from the laser emission-reflection time difference.

In a possible implementation manner, the process of judging the head distance between the smart watch and the target head of the target user based on the distance sensing group in the smart watch further includes acquiring the smart watch and the target head. The steps of the optimal distance of the target head of the target user include:

Determining the first distance obtained by the first sensor and the second distance obtained by the second sensor in the distance sensing group in the smart watch;

Determining the horizontal measurement angle between the smart watch and the target object calculated by the gyroscope sensor in the smart watch;

According to the measurement angle, an optimal distance is selected from the first distance and the second distance as the head distance between the smart watch and the target head.

In a possible implementation manner, when it is determined that the target user wearing the smart watch is in a reading and writing state, it is detected whether the current sitting posture of the target user is correct, and if so, the smart watch does not perform any operation;

Otherwise, the vibration sensor inside the smart watch generates a vibration to indicate the current sitting posture error, and at the same time the display on the smart watch displays a prompt display;

Moreover, the smart watch further detects the current sitting posture of the target user a preset number of times at a preset time interval. If the preset number of detection results are all the current sitting posture errors, the smart watch The alarm in the system emits the first alarm sound.

In a possible implementation manner, the smart watch is also used to record the detected first target data, second target data, and the current sitting posture of the target user, and transmit the recording results to the parent terminal for display ；

In addition, when the user of the parent terminal is viewing the displayed recording result in real time, the parent terminal also receives a correction instruction input by the user of the parent terminal.

In a possible implementation manner, when it is judged that the target user is in a reading and writing state, timing statistics are performed on the current sitting posture of the target user;

When the current sitting posture of the target user changes, re-calculate the timing statistics of the changed current sitting posture, and when the current sitting posture of the target user does not change, continue timing statistics on the current sitting posture;

Based on the preset time axis, the timing statistics results are output to the parent terminal for display.

In a possible implementation manner, after constructing the sitting posture recognition model, in order to improve the accuracy of the sitting posture recognition model, the method further includes: performing correction processing on the sitting posture recognition model, and performing correction processing on the posture after correction processing. To verify the recognition model, the steps include:

Based on the camera and CCD module set on the smart watch, acquire scene images of preset read-write scenes where a number of historical users are located, and transmit the scene images to the server, the scene images including: the historical users Image information of the sitting posture, object image information of the target object;

Based on the server, perform image algorithm detection on the scene image to obtain scene calibration points, and obtain calibration coordinates of the scene calibration points based on a standard three-dimensional space;

Based on the image feature database in the server, obtain the scene feature points of the scene image, determine whether the calibration coordinates are consistent with the corresponding scene feature points, and if so, retain the corresponding scene calibration points;

Otherwise, the offset distance between the calibration coordinates and the corresponding scene feature points is determined based on the offset database in the server, and the calibration coordinates and the corresponding scene feature points are determined based on the preset offset correction algorithm. Perform correction processing until the calibration coordinates after correction processing are consistent with the corresponding scene feature points;

When all the calibration coordinates are consistent with the corresponding scene feature points, predict the authenticity of each calibration coordinate, and determine whether the calibration coordinates need to be deleted according to the result of the authenticity prediction;

If yes, delete the calibration coordinates, otherwise, keep the calibration coordinates;

At the same time, reconstruct the calibration image according to all the retained calibration coordinates, and perform deep learning training on the sitting posture recognition model based on the calibration image;

Acquire several scene images of the preset reading and writing scenes where the target user is located, and perform recognition verification based on the sitting posture recognition model after the deep learning training, and determine the recognition accuracy of the sitting posture recognition model after the deep learning training according to the recognition and verification results ；

When the accuracy is higher than the preset value, it is determined that the verification is successful, and the successfully verified sitting posture recognition model is transmitted to the smart watch to realize the recognition of the current sitting posture.

In a possible implementation manner, after recognizing the first target data and the second target data, determining the current sitting posture of the target user, and recording it, the method further includes:

Time the recorded retention time of the current sitting position of the target user, and when the time is greater than a preset time, control the smart watch’s display screen to display a reminder of too long eyes, or control the smart watch’s alarm Issue the second alarm message.

In a possible implementation manner, an infrared camera is provided on the smart watch, and the infrared camera is used to capture an infrared image of the target user;

The smart watch is provided with a processor for judging the current user based on the infrared image when it is determined based on the gyro sensor in the smart watch that the target hand of the target user is horizontally placed on the target object The facial swing feature of the user and the corresponding prompt is issued according to the facial swing feature; wherein the judging the current user's facial swing feature based on the infrared image and the current reading and writing distance includes:

Acquire a total of N groups of infrared images, where the nth group of infrared images includes a first infrared image and a second infrared image, and the shooting time interval between the first infrared image and the second infrared image is less than or equal to the preset time interval ; N = 1, 2, 3,,, N;

Calculate the face swing coefficient of the current user in the shooting time interval corresponding to the nth group of infrared images according to the following formula (1):

Wherein, w _n(12) is the face swing coefficient of the current user in the shooting time interval corresponding to the nth group of infrared images; and the X _n1 (i,j) is the nth group of infrared images. The pixel value of a pixel in the i-th row and j-th column of an infrared image; said X _n2 (i,j) is the pixel value of the pixel in the i-th row and j-th column of the second infrared image in the n-th group of infrared images The I is the total number of rows of pixels in the infrared image captured by the infrared camera; the J is the total number of columns of pixels in the infrared image captured by the infrared camera;

Calculate the face swing coefficient W of the current user in the N groups of infrared images according to the following formula (2):

Where δ _n is the _{weight coefficient corresponding to w n(12)} , and the value range is (0,1);

When the W is equal to or greater than the preset action coefficient threshold, it is determined that the current user's facial wobbling feature is frequent facial wobbling, and the processor controls the speaker on the smart watch to emit sound prompts.

Other features and advantages of the present invention will be described in the following description, and partly become obvious from the description, or understood by implementing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings.

The technical solutions of the present invention will be further described in detail below through the accompanying drawings and embodiments.

Description of the drawings

The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the attached picture:

FIG. 1 is a flowchart of a method for recognizing read-write distance based on a smart watch in an embodiment of the present invention;

Figure 2 is a schematic diagram of a distance sensing group in an embodiment of the present invention;

Figure 3 is a top view of a smart watch in an embodiment of the invention;

Fig. 4 is a schematic diagram of a reading and writing gesture in an embodiment of the present invention.

Detailed ways

The preferred embodiments of the present invention will be described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described here are only used to illustrate and explain the present invention, and are not used to limit the present invention.

The embodiment of the present invention provides a method for recognizing a read-write distance based on a smart watch, as shown in FIG. 1, including:

Step 1: Build a sitting position recognition model;

Step 2: According to the acceleration sensor and the gyroscope sensor in the smart watch, the first target data is detected, and based on the first target data, the current reading and writing status of the target user wearing the smart watch is determined;

Step 3: According to the judgment result, control the distance sensing group in the smart watch to detect the second target data, and judge the current reading and writing of the target user corresponding to the current reading and writing state through the second target data distance;

Step 4: At the same time, based on the constructed sitting posture recognition model, the first target data and the second target data are recognized, and the current sitting posture of the target user is determined and recorded.

In the specific implementation process, it should be noted that the smart watch in the embodiment of the present invention is an existing conventional smart watch, which has all the functions of the existing conventional smart watch, such as GPS positioning, one-key alarm, Make phone calls, heart rate and blood pressure statistics, step counting, etc.

The above-mentioned construction of the sitting posture recognition model is to facilitate the subsequent determination of the user’s current sitting posture;

The above-mentioned target users can be students, etc.;

The above-mentioned first target data is acceleration data and gyroscope data to determine the current reading and writing status of the target user wearing the smart watch. For example, it can be based on the gyroscope sensor in the smart watch to determine whether the target hand of the target user is placed horizontally on the target. Related data on the object, such as: horizontal placement, not horizontal placement, etc.; the corresponding judgment result is that when the smart watch is placed horizontally and there is only a slight displacement within a certain period of time, it is judged that the target user has entered the reading and writing state; otherwise, the target is judged The user has not entered the read-write state.

The current read-write distance determined above is determined after the user is determined to be in the read-write state, which can effectively save the power consumption of the smart watch;

The above-mentioned sitting posture recognition model based on the construction of the first target data and the second target data is recognized, the current sitting posture of the target user is determined, and recorded, in order to determine whether the corresponding sitting posture matches when the user is in the reading and writing state. Regulations, and continue to record them, in order to facilitate parents to supervise them and improve the enthusiasm of target users.

The above technical solution has the beneficial effect that it is used to determine the user's reading and writing distance by acquiring the first target data and the second target data, thereby facilitating the determination of the user's current sitting posture.

The embodiment of the present invention provides a reading and writing distance recognition method based on a smart watch, and the current sitting posture includes: a sitting type, a head-down reading and writing type, and a hand-supported cheek writing type;

The target object mentioned above can be a desk;

The above-mentioned preset distance is generally 30cm, and the preset distance can be effectively adjusted according to the age and height of the target user;

The aforementioned preset angle range is generally between [70°, 90°], and the aforementioned first angle is measured by a gyroscope sensor.

The beneficial effect of the above technical solution is that it can be effectively judged from the horizontal placement and the head distance to effectively determine whether it is a sitting type or a countersunk reading and writing type, and it can be effectively judged from both the horizontal placement and the first angle. It is a hand-supported cheek writing type, and its judgment method is simple, which improves the efficiency and efficiency of determining the current sitting posture.

The embodiment of the present invention provides a method for identifying a reading and writing distance based on a smart watch. The step of judging the current reading and writing distance of the target user corresponding to the current reading and writing state based on the judgment result and using the second target data includes:

The aforementioned distance sensor group may be a laser sensor or an infrared sensor, and the distance sensor group includes at least one sensor, generally two.

The beneficial effect of the above technical solution is that by controlling the distance sensing group to emit laser light, it provides convenience for obtaining the current reading and writing distance.

The embodiment of the present invention provides a method for recognizing a read-write distance based on a smart watch. The process of judging the head distance between the smart watch and the target head of the target user based on the distance sensing group in the smart watch further includes The steps of obtaining the optimal distance between the smart watch and the target head of the target user include:

As shown in Figure 2, the first sensor 01 and the second sensor 02 respectively include: a transmitting and receiving window, a receiving and sending window angle, and the receiving and sending window angle of the first sensor is 25°, and the receiving and sending window angle of the second sensor is 35°. °;

The above measurement angle is calculated based on the gyroscope sensor;

The above smart watch has two built-in distance sensors, and the two sensor angles are different, because each student's height is different, but there is a range of height. The smart watch will calculate the distance sensor according to the student's height segment when determining the position of the distance sensor installed on the watch. An appropriate angle and position ensure that students within this height range can accurately measure the distance.

The beneficial effect of the above technical solution is that by obtaining the optimal distance, it is convenient to effectively determine a more accurate distance sensor as an instrument for reading and writing distance measurement, and to ensure the accuracy of the reading and writing distance.

The embodiment of the present invention provides a method for recognizing the reading and writing distance based on a smart watch. When it is determined that the target user wearing the smart watch is in the reading and writing state, it is detected whether the current sitting posture of the target user is correct. If so, the smart watch does not execute anything. operating;

The above alarm can be implemented as a buzzer, horn, etc.;

As shown in Figure 3, it is a top view of the display screen and the distance sensing group;

Among them, 01 represents sensor A, which is the first sensor, 02 represents sensor B, which is the second sensor, and 10 represents the display screen;

As shown in Figure 4, a represents the correct reading and writing posture, and b represents the wrong reading and writing posture;

Among them, 03 represents the incident laser, 04 represents the reflected laser, and 05 represents the target user.

The above-mentioned preset times are generally 3 times, and the above-mentioned display is based on a display screen, for example, a text prompt;

The aforementioned preset time interval is, for example, 1 minute.

The above-mentioned current sitting posture error of the target user is generally caused by the target user being too close to the desk.

The beneficial effect of the above technical solution is: through the combination of multiple reminding means such as display screen reminder, vibration reminder, alarm reminder, etc., it is convenient for the target user to correct the current sitting posture in time.

The embodiment of the present invention provides a method for recognizing a read-write distance based on a smart watch. The smart watch is also used to record the detected first target data, second target data, and the current sitting posture of the target user, and record the result Transmit to the parent terminal for display;

The smart watch is actually the watch side of the mobile phone. When entering the smart watch, it will synchronize the data in the smart phone. These data include time, address book, weather, call history, text messages, etc. The synchronization of these data is through Bluetooth, wifi , 2G/3G/4G/5G and other communication modules for communication transmission, using a smart watch to make a call is also in the synchronization process, when the watch is connected to the mobile phone, the watch end is through Bluetooth or wifi or 2G/3G/4G/5G network Operate the mobile phone, and because the existing smart watch has a distance limitation, the communication connection with the parent terminal is generally realized through Bluetooth or wifi signal;

At the same time, when a SIM card is installed on the smart watch, parents can use the 2G/3G/4G/5G network to realize the smart phone to control the smart watch vibration, text message display, buzzer, call reminder, etc.;

In addition, when parents use the 4G/5G network to connect to the smart watch, the parents can use video signals to watch the child's reading and writing status live, and can issue instructions to correct them.

Connect the smart watch in the 2G/3G/4G/5G communication mode. The smart watch can effectively count the child's daily reading and writing time, and after the reading and writing scene of the day is over, based on the time axis, it will generate a log of the reading and writing time for each period to send To parents, parents can judge their children's reading and writing intensity or the intensity of reading and writing learning pressure. The advantage is that it is convenient for parents to use the data to reasonably arrange their children's stress relief, exercise and outdoor activities.

When there is no SIM card installed on the smart watch, Bluetooth and wifi communication modules can be used for communication.

The above correction instruction is, for example, the current sitting posture correction instruction, and the corresponding correction instruction result is: smart watch vibration, short message display, buzzer, call prompt, etc.

The above recorded result can be the user's reading and writing time and the sitting posture during the reading and writing process. The sitting posture includes, for example, the normal sitting posture in the first time period, the reading and writing distance is too close in the second time period, and the third time The side-slung sitting posture is used in the segment, which has the advantage of making it easier for parents to understand more clearly.

The beneficial effect of the above technical solution is that by transmitting the recorded data to the parent terminal for display, it is convenient for the parent to know the sitting posture of the target user in time.

The embodiment of the present invention provides a method for recognizing a reading and writing distance based on a smart watch. When it is judged that the target user is in a reading and writing state, timing statistics are performed on the current sitting posture of the target user;

The above-mentioned setting of the preset time axis is to clearly express the timing statistics, so that the parental user can understand it thoroughly.

The above-mentioned parent terminal may be a smart phone, a notebook, etc., and the user of the parent terminal can view the timing statistics results on the WeChat applet or related APP.

The beneficial effect of the above technical solution is that by counting the current sitting posture, it is convenient to record the sitting posture of the target user in different time periods in time, and it is convenient for parents to carry out effective supervision.

The embodiment of the present invention provides a read-write distance recognition method based on a smart watch. After the sitting posture recognition model is constructed, in order to improve the accuracy of the sitting posture recognition model, the method further includes: correcting the sitting posture recognition model and correcting The processed posture recognition model is verified, and the steps include:

When all the calibration coordinates mentioned above are consistent with the corresponding scene feature points, the authenticity prediction of each calibration coordinate is to determine the uniqueness of the calibration coordinates. For example, the position information corresponding to the coordinate c1 includes the coordinates of the coordinates c2 and c3. Position information, at this time, the coordinate c1 is uniquely irreplaceable, and keeping the uniquely irreplaceable calibration coordinates can reduce the workload of calculating each coordinate and improve efficiency.

The above-mentioned reconstruction of the calibration image based on all the retained calibration coordinates, and the in-depth learning training of the sitting posture recognition model based on the calibration image is to improve the recognition accuracy of the sitting posture recognition model;

When the accuracy is higher than the preset value, it is determined that the verification is successful, where the preset value is 90% or more.

This embodiment is based on the scene image acquired by the smart watch, which is trained and recognized by the server, and is pre-loaded and transmitted to the smart watch, which can effectively reduce the transmission delay of data. At the same time, it is based on the trained sitting posture recognition model and the smart watch hardware itself. Recognizing the current posture can effectively reduce the occupation of multiple network resources and improve efficiency.

The above-mentioned image algorithm detection is performed on the scene image to obtain the scene calibration point, and based on the standard three-dimensional space, the calibration coordinates of the scene calibration point are obtained, in order to continue the splitting of the positioning point of the scene image, so as to facilitate the correction of each point Processing to improve its accuracy, where the scene image is, for example, an image including the target user and desk; the scene calibration point is to determine and label the positioning point of the scene image;

In the above, based on the offset database, the offset distance between the calibration coordinates and the corresponding scene feature points is determined, and the calibration coordinates and the corresponding scene feature points are corrected based on a preset offset correction algorithm, Until the calibration coordinates after the correction process are consistent with the corresponding scene feature points;

The above-mentioned correction processing of the calibration coordinates and the corresponding scene feature points is to ensure that the calibration coordinates are consistent with the corresponding scene feature points and ensure their unity.

The beneficial effect of the above technical solution is that by correcting the sitting posture recognition model and verifying the posture recognition model after the correction processing, in order to improve the accuracy of the sitting posture recognition model, the first target data and the The accuracy of the identification of the second target data.

The embodiment of the present invention provides a read-write distance recognition method based on a smart watch, which recognizes the first target data and the second target data, determines the current sitting posture of the target user, and records it, and further includes:

The above-mentioned second alarm information is different from the above-mentioned first alarm information. For example, the alarm durations of the two are different. The first alarm information has an alarm duration of 3S, and the second alarm information has a duration of 10S. The interval is longer, which facilitates effective distinction and correspondence. Alarm information.

The beneficial effect of the above technical solution is that it is convenient to give effective reminders to the target user, and avoids the use of the eye time process, which may cause the possibility of vision loss.

The embodiment of the present invention provides a method for recognizing a read-write distance based on a smart watch, where an infrared camera is provided on the smart watch, and the infrared camera is used to capture an infrared image of the target user;

The beneficial effect of the above technical solution is that according to the above algorithm, it can be judged whether the current user's face swings frequently. Frequent face swing refers to the high frequency of the current user's head swinging left and right. For example, the user's wrist wearing a smart watch does not swing frequently, and The current user’s face is frequently swinging to the left and right, and he does not read carefully; when it is determined that the current user’s face is swinging frequently, the smart watch can be used to give a reminder to remind the current user to read and write at ease.

Furthermore, for the above specific embodiments based on formulas (1) and (2), it is also proposed that an infrared camera is provided on the smart watch, and the infrared camera is used to capture an infrared image of the target user;

The smart watch is provided with a processor for judging the current face orientation of the current user according to the infrared image and the current reading and writing distance, and issuing corresponding prompts according to the current face orientation; wherein, The judging the current face orientation of the current user according to the infrared image and the current reading and writing distance includes:

Extracting the infrared region from the infrared image to obtain the infrared region;

Determine the current head infrared area corresponding to the current user's head in the infrared area according to the contour feature of the infrared area;

Determine the current face orientation corresponding to the current head infrared region according to the pre-stored various facial orientations corresponding to the current read-write distance and the infrared region feature maps corresponding to each facial orientation; The various face orientations include face forward, face deflection and face backward;

The sending out corresponding prompts according to the current face orientation includes:

When the current face orientation is the face deflection or the face is facing backward, the speaker on the smart watch is controlled to emit a sound prompt.

The beneficial effects of this embodiment are: according to the above method, it can be judged whether the current user has a face deflection or face is facing backward, and when the current user's face is deflected or face is facing backward, it means that the current user is not studying carefully; A reminder is issued through the smart watch to remind the current user to read and write at ease.

Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. In this way, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

The read-write distance recognition method based on smart watches is characterized in that it includes:

Build a seated recognition model;

The first target data is detected according to the acceleration sensor and the gyroscope sensor in the smart watch, and based on the first target data, the current reading and writing status of the target user wearing the smart watch is determined;

According to the judgment result, controlling the distance sensing group in the smart watch to detect the second target data, and judging the current reading and writing distance of the target user corresponding to the current reading and writing state through the second target data;

At the same time, based on the constructed sitting posture recognition model, the first target data and the second target data are recognized, and the current sitting posture of the target user is determined and recorded.
The method for recognizing the reading and writing distance of claim 1, wherein the current sitting posture includes: sitting type, head-down reading and writing type, and hand-supported cheek writing type;

Wherein, the step of recognizing the first target data and the second target data based on the constructed sitting posture recognition model, and determining the current sitting posture of the target user includes:

Judging whether the target hand of the target user is placed horizontally on the target object based on the gyroscope sensor in the smart watch;

Judging the head distance between the smart watch and the target head of the target user based on the distance sensing group in the smart watch;

Based on the sitting posture recognition model, if it is determined that the target hand of the target user is placed horizontally on the target object, and the head distance is greater than a preset distance, then the current sitting posture of the target user is an upright sitting type;

If it is determined that the target hand of the target user is horizontally placed on the target object, and the head distance is not greater than a preset distance, then the current sitting posture of the target user is a head-down reading and writing type;

If it is determined that the target hand of the target user is not placed horizontally on the target object, and the first angle with the target object is within the preset angle range, then the current sitting posture of the target user is hand-supported cheek writing type .
The method for identifying the reading and writing distance according to claim 1, wherein the step of judging the current reading and writing distance of the target user corresponding to the current reading and writing state through the second target data based on the judgment result include:

When the judgment result is that the target user is currently in a reading and writing state, controlling the distance sensing group to emit laser light, and calculating the current reading and writing distance of the target user according to the laser emission-reflection time difference;

When the judgment result is that the target user is not currently in a read-write state, control the smart watch not to perform any operation;

Wherein, the second target data is the current reading and writing distance calculated from the laser emission-reflection time difference.
3. The method for recognizing a reading and writing distance according to claim 2, wherein in the process of judging the head distance between the smart watch and the target head of the target user based on the distance sensing group in the smart watch, It also includes obtaining the optimal distance between the smart watch and the target head of the target user, and the steps include:

Determining the first distance obtained by the first sensor and the second distance obtained by the second sensor in the distance sensing group in the smart watch;

Determining the horizontal measurement angle between the smart watch and the target object calculated by the gyroscope sensor in the smart watch;

According to the measurement angle, an optimal distance is selected from the first distance and the second distance as the head distance between the smart watch and the target head.
The method for recognizing the reading and writing distance of claim 1, wherein when it is determined that the target user wearing the smart watch is in a reading and writing state, it is detected whether the current sitting posture of the target user is correct, and if so, the smart watch is not Perform any operation;

Otherwise, the vibration sensor inside the smart watch generates a vibration to indicate the current sitting posture error, and at the same time the display on the smart watch displays a prompt display;

Moreover, the smart watch further detects the current sitting posture of the target user a preset number of times at a preset time interval. If the preset number of detection results are all the current sitting posture errors, the smart watch The alarm in the system emits the first alarm sound.
The method for recognizing a read-write distance according to claim 1, wherein:

The smart watch is also used to record the detected first target data, second target data, and the current sitting posture of the target user, and transmit the recording result to the parent terminal for display;

In addition, when the user of the parent terminal is viewing the displayed recording result in real time, the parent terminal also receives a correction instruction input by the user of the parent terminal.
4. The method for recognizing a reading and writing distance according to claim 1, wherein when it is determined that the target user is in a reading and writing state, timing statistics are performed on the current sitting posture of the target user;

When the current sitting posture of the target user changes, re-calculate the timing statistics of the changed current sitting posture, and when the current sitting posture of the target user does not change, continue timing statistics on the current sitting posture;

Based on the preset time axis, the timing statistics results are output to the parent terminal for display.
The method for reading and writing distance recognition according to claim 1, wherein after the construction of the sitting posture recognition model, in order to improve the accuracy of the sitting posture recognition model, the method further comprises: correcting the sitting posture recognition model, and The steps of verifying the posture recognition model after correction processing include:

Based on the camera and CCD module set on the smart watch, acquire scene images of preset read-write scenes where a number of historical users are located, and transmit the scene images to the server, the scene images including: the historical users Image information of the sitting posture, object image information of the target object;

Based on the server, perform image algorithm detection on the scene image to obtain scene calibration points, and obtain calibration coordinates of the scene calibration points based on a standard three-dimensional space;

Based on the image feature database in the server, obtain the scene feature points of the scene image, determine whether the calibration coordinates are consistent with the corresponding scene feature points, and if so, retain the corresponding scene calibration points;

Otherwise, the offset distance between the calibration coordinates and the corresponding scene feature points is determined based on the offset database in the server, and the calibration coordinates and the corresponding scene feature points are determined based on the preset offset correction algorithm. Perform correction processing until the calibration coordinates after correction processing are consistent with the corresponding scene feature points;

When all the calibration coordinates are consistent with the corresponding scene feature points, predict the authenticity of each calibration coordinate, and determine whether the calibration coordinates need to be deleted according to the result of the authenticity prediction;

If yes, delete the calibration coordinates, otherwise, keep the calibration coordinates;

At the same time, reconstruct the calibration image according to all the retained calibration coordinates, and perform deep learning training on the sitting posture recognition model based on the calibration image;

Acquire several scene images of the preset reading and writing scenes where the target user is located, and perform recognition verification based on the sitting posture recognition model after the deep learning training, and determine the recognition accuracy of the sitting posture recognition model after the deep learning training according to the recognition and verification results ；

When the accuracy is higher than the preset value, it is determined that the verification is successful, and the successfully verified sitting posture recognition model is transmitted to the smart watch to recognize the current sitting posture.
5. The method for recognizing a reading and writing distance according to claim 1, wherein after recognizing the first target data and the second target data, determining the current sitting posture of the target user, and recording it, the method further comprises:

Time the recorded retention time of the current sitting position of the target user, and when the time is greater than a preset time, control the smart watch’s display screen to display a reminder of too long eyes, or control the smart watch’s alarm Issue the second alarm message.
The method for recognizing a read-write distance according to claim 1, wherein:

An infrared camera is provided on the smart watch, and the infrared camera is used to capture an infrared image of the target user;

The smart watch is provided with a processor for judging the current user based on the infrared image when it is determined based on the gyro sensor in the smart watch that the target hand of the target user is horizontally placed on the target object The facial swing feature of the user and the corresponding prompt is issued according to the facial swing feature; wherein the judging the current user's facial swing feature based on the infrared image and the current reading and writing distance includes:

Acquire a total of N groups of infrared images, where the nth group of infrared images includes a first infrared image and a second infrared image, and the shooting time interval between the first infrared image and the second infrared image is less than or equal to the preset time interval ; N = 1, 2, 3,,, N;

Calculate the face swing coefficient of the current user in the shooting time interval corresponding to the nth group of infrared images according to the following formula (1):

Wherein, w n(12) is the face swing coefficient of the current user in the shooting time interval corresponding to the nth group of infrared images; and the X n1 (i,j) is the nth group of infrared images. The pixel value of a pixel in the i-th row and j-th column of an infrared image; said X n2 (i,j) is the pixel value of the pixel in the i-th row and j-th column of the second infrared image in the n-th group of infrared images The I is the total number of rows of pixels in the infrared image captured by the infrared camera; the J is the total number of columns of pixels in the infrared image captured by the infrared camera;

Calculate the face swing coefficient W of the current user in the N groups of infrared images according to the following formula (2):

Where δ n is the weight coefficient corresponding to w n(12) , and the value range is (0,1);

When the W is equal to or greater than the preset action coefficient threshold, it is determined that the current user's facial wobbling feature is frequent facial wobbling, and the processor controls the speaker on the smart watch to emit a sound prompt.