NL2029643A

NL2029643A - Ai eye protection desk lamp with sitting posture correction reminder and implementation method thereof

Info

Publication number: NL2029643A
Application number: NL2029643A
Authority: NL
Inventors: Huang Quan; Wang Chenchen
Original assignee: Hangzhou Huixinda Tech Co Ltd
Priority date: 2020-11-20
Filing date: 2021-11-04
Publication date: 2022-05-23
Also published as: CN112483924A; NL2029643B1; CN112483924B

Abstract

The present invention provides an AI eye protection desk lamp with sitting posture correction reminder and an implementation method thereof. The method includes: A, light source adjustment, detecting the brightness of ambient light, and adjusting LED lighting intensity according to the brightness of the ambient light; and B, sitting posture correction, detecting the current sitting posture of a user, and sending a reminder to the user when it is determined that the sitting posture of the user is bad. Considering the sitting posture and light, the user is reminded in the case of bad sitting posture, and the LED lighting intensity is adjusted when the light is not suitable, which can better achieve the purpose of eye protection.

Description

AT EYE PROTECTION DESK LAMP WITH SITTING POSTURE CORRECTION REMINDER AND IMPLEMENTATION METHOD THEREOF

TECHNICAL FIELD The present invention relates to the technical field of eye protection desk lamps, and particularly to an AI eye protection desk lamp with sitting posture correction reminder and an imple- mentation method thereof.

BACKGROUND ART The importance of vision is obvious to anyone, but it is of- ten overlooked. The year-on-year increase in the incidence of myo- pia in students is a worldwide issue. According to statistics, the total number of myopia in China exceeds 300 million, which is 1/3 of the total number of world myopia. The number of myopia in ado- lescents ranks first in the world. In China, the myopia rate of students is increasing year by year, almost at a rate of 10% per year. In many cities, the myopia rates in the first, second and third years of junior high schools are respectively about 50%, 60%, and 70%. The myopia rate of can- didates for College Entrance Examination is more than 80% in many places, and even more than 90% in some areas.

However, as society attaches importance to education, current school-age children are facing tremendous academic pressure. School-age children usually need to spend a lot of time on study- ing. If school-age children do not develop the habit of maintain- ing a correct sitting posture during the long learning process, it is easy to cause different types of health problems such as spine curvature and myopia in school-age children, which seriously af- fects the normal study and life of school-age children. All eye protection devices on the market only concern the impact of light on the eyes, and do not effectively detect and correct the sitting posture of children. Hence, there is an urgent need in the prior art for a method and apparatus capable of detecting and correcting the sitting posture of children, especially an integrated eye pro- tection lamp device capable of detecting and correcting the sit- ting posture of children.

In addition, the eye protection desk lamp in the prior art is usually composed of a touch module, an LED lighting module 7, an MCU control module, an ultrasonic ranging module and a voice broadcast module.

The “touch module” inputs control information to the “MCU control module” by means of touch keys, and the “MCU con- trol module” controls the on-off and brightness adjustment of the “LED lighting module” according to pre-edited code logics; the “ultrasonic ranging module” detects the distance between a user and an ultrasonic probe by means of the ultrasonic probe, and the “MCU control module” receives the distance detection data from the “ultrasonic ranging” module, compares the data with a threshold pre-programmed to approach the distance, and triggers an approach alarm, to indirectly achieve a goal of sitting posture detection; and the “voice broadcast module” is used for voice broadcast of the sitting posture reminder.

Since the light brightness adjustment in the prior art adopts manual touch control, the brightness of LED light cannot be auto- matically adjusted for eye protection according to the brightness of ambient light.

In addition, the sitting posture detection uses ultrasonic ranging technology, but this detection method is too rough, can only monitor the single dimension of the frontal dis- tance between the body and the desk lamp, and cannot effectively identify bad postures of the upper body of the human body (e.d. head-down, head tilt, body skew, etc.), so misjudgment easily oc- curs.

In order to solve the problem that the above solutions cannot effectively identify the bad postures of the upper body, people have improved the solutions.

The human body posture is recognized by a neural network algorithm to determine whether the human body is in a bad posture.

Although this method can solve the above problems, it requires a large number of libraries for class train- ing and recognition, has the problems of complex algorithm models, large calculation amount and like, and needs to be supported by hardware platforms with computing resources, resulting in high power consumption and high costs of products.

SUMMARY In light of the above problems, the objective of the present invention is to provide an AI eye protection desk lamp with sit- ting posture correction reminder and an implementation method thereof.

In order to achieve the above objective, the present inven- tion adopts the following technical solution: An AI eye protection desk lamp with sitting posture correc- tion reminder and an implementation method thereof, including: A. light source adjustment, detecting the brightness of ambi- ent light, and adjusting LED lighting intensity according to the brightness of the ambient light; and B. sitting posture correction, detecting the current sitting posture of a user, and sending a reminder to the user when it is determined that the sitting posture of the user is bad.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, the method A specifically includes: Al. detecting the brightness of ambient light, and roughly adjusting the LED lighting intensity according to the brightness of the ambient light so that the brightness of the ambient light reaches a brightness set value; A2. detecting the reflective brightness of a user eye-sighted target, and finely adjusting the lighting intensity up and down to a reflective brightness set value according to the reflective brightness; and A3. after the fine adjustment is finished, continuing to mon- itor the reflective brightness of the eye-sighted target, and fur- ther finely adjusting the LED lighting intensity when the reflec- tive brightness exceeds an eye protection brightness range, until the reflective brightness reaches the reflective brightness set value.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, in the method B, it is determined that the sitting posture is bad when the user is in a head tilt state, a head-down state, or an arm leaning state.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, in the method B, whether the user is currently in a head tilt state, a head-down state, or an arm leaning state is detected by the fol- lowing way: Bl. capturing a sitting posture state of the user through a camera, and extracting a head contour and a shoulder contour; Bl. when the deviation of the angle between the head contour and the shoulder contour from 90 degrees exceeds an angle devia- tion value, determining that the user is in a head tilt state; B2. extracting eyes, comparing the relative positions of the eyes and the shoulder contour, and when the distance between the two is less than a distance threshold, determining that the user is in a head-down state; and B3. when there is adhesion between the head contour and the shoulder contour, determining that the user is in an arm leaning state.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, in the method Bl, key features of the user's face are located by a face recognition algorithm to extract the head contour; the user's head and shoulders are located by a shape recognition algorithm to extract the shoulder contour.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, hu- man body detection is further included: performing human body recognition on an image captured by the camera, determining that there is no one currently when any human body is not recognized, and turning off the LED lighting after a preset time of determining that there is no one currently.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, fa- tigue detection is further included: performing face detection and eye detection on the image cap- tured by the camera, performing fatigue detection on the user based on a fatigue detection algorithm, and sending a reminder when the degree of fatigue reaches a fatigue threshold.

A learning concentration detection is further included: performing face detection and eye detection on the image cap- 5 tured by the camera, performing concentration detection on the us- er based on a concentration detection algorithm, and sending a re- minder when the degree of concentration is lower than a concentra- tion threshold.

In the aforementioned implementation method of the AI eye protection desk lamp with sitting posture correction reminder, the detection results are sent to a parent terminal intermittently, and a prompt message is sent to the parent terminal when the num- ber of reminders reaches a reminder number threshold.

An AI eye protection desk lamp with sitting posture correc- tion reminder, including a desk lamp with an LED lighting module, wherein the desk lamp also has a control module, a camera module, a voice broadcast module and touch keys; the control module is connected to the LED lighting module, the camera module, the voice broadcast module and the touch keys; the control module includes an image recognition module and a voice recognition module, and the image recognition module includes an ambient light detection module and a sitting posture detection module.

In the aforementioned AI eye protection desk lamp with sit- ting posture correction reminder, the control module is further connected with a voice collection module and a storage module.

The sitting posture detection module includes a face recogni- tion algorithm module for extracting a head contour and user eyes, and a shape recognition algorithm module for extracting a shoulder contour.

The image recognition module further includes an eye-sighted target brightness detection module for detecting the reflective brightness of an eye-view target.

The image recognition module includes a fatigue detection al- gorithm module for detecting user fatigue and a concentration de- tection algorithm module for detecting user learning concentra- tion.

The control module is connected with a wireless module, and is connected to a parent terminal through the wireless module.

The present invention has the following advantages:

1. The LED lighting brightness is automatically adjusted ac- cording to the ambient brightness to meet the requirements of eye protection;

2. The LED lighting brightness is further adjusted according to the eye-sighted target, which is not restricted by user habits {near or far light source, etc.) or by the material of the eye- sighted target, and can accurately provide the user with light of appropriate brightness;

3. The combination of feature points and contour algorithms can judge and detect bad sitting postures, with light algorithms, small calculation amount, and terminal calculation, and the result output is correspondingly fast and timely;

4. Bad sitting postures are indirectly detected on the basis of a shoulder contour combined with face key points and a head contour, so the method is simple and effective; and

5. The in-depth integration of visual behavior detection and optical lighting equipment realizes practical functions such as automatic adjustment of eye protection lighting, sitting posture detection, fatigue detection, and learning state recording.

BRIEFT DESCRIPTION OF THE DRAWINGS FIG. 1 is a flowchart of an implementation method of an AI eye protection desk lamp with sitting posture correction reminder according to the present invention; and FIG. 2 is a block diagram of a circuit structure of the AI eye protection desk lamp with sitting posture correction reminder according to the present invention.

Reference numerals, control module 1; image recognition mod- ule 11; voice recognition module 12; camera module 2; voice broad- cast module 3; touch keys 4; voice collection module 5; storage module 6; LED lighting module 7; wireless module 8.

DETAILED DESCRIPTION OF THE EMBODIMENTS The present invention will be further described in detail be- low in conjunction with the accompanying drawings and specific em-

bodiments.

As shown in FIG. 1, this embodiment discloses an implementa- tion method of an AI eye protection desk lamp with sitting posture correction reminder, including: A. Light source adjustment, the brightness of ambient light is detected, and LED lighting intensity is adjusted according to the brightness of the ambient light; B. Sitting posture correction, the current sitting posture of a user is detected, and a reminder is sent to the user when it is determined that the sitting posture of the user is bad.

Human body detection is performed during lighting: Human body recognition is performed on an image captured by a camera, it is determined that there is no one currently when any human body is not recognized, and the LED lighting is turned off after a preset time of determining that there is no one currently.

The method A specifically includes: Al. The brightness of ambient light is detected, and the LED lighting intensity is roughly adjusted according to the brightness of the ambient light so that the brightness of the ambient light reaches a brightness set value. Specifically, the detected bright- ness of the ambient light is compared with the stored brightness set value, and when the brightness of the ambient light does not reach the brightness set value, the lighting intensity is in- creased until it reaches the brightness set value. The brightness set value here is preferably 240 cd/m‘. For example, when the user needs lighting, the desk lamp is turned on by voice or manually, and the desk lamp automatically adjusts the illumination according to the brightness of the ambient light to reach the brightness set value. Specifically, the brightness of the ambient light can be determined by detecting the reflective brightness of a reflective surface of a surrounding object such as a desktop or a wall.

A2. The reflective brightness of a user eye-sighted target is detected, and the lighting intensity is finely adjusted up and down to a reflective brightness set value according to the reflec- tive brightness. If eyes or any eye-sighted target is not detect- ed, the lighting intensity adjusted in Al is maintained; if an eye-sighted target is detected, the lighting intensity is finely adjusted to the reflective brightness according to the reflective brightness of the eye-sighted target, so that the lighting inten- sity of the desk lamp can preferentially protect the eyes of the writing user. Too high or too low reflective brightness will cause damage to the eyes, so the reflective brightness set value is set to 180 cd/m’ here.

A3. After the fine adjustment is finished, the reflective brightness of the eye-sighted target continues to be monitored, and the LED lighting intensity is further finely adjusted when the reflective brightness exceeds an eye protection brightness range, until the reflective brightness reaches the reflective brightness set value. Too high or too low reflective brightness will cause damage to the eyes, so the eye protection brightness range is set to 120-240 cd/m?. The dimming of external light will change the reflective brightness. When the user moves the workbook or changes the material of the workbook, the reflective brightness also changes. The subsequent automatic adjustment is conducted beyond the range of 120-240 cd/m*, so visual fatigue caused by frequent adjustments will not occur. The LED lighting intensity can be ad- Jjusted immediately after the reflective brightness exceeds the eye protection brightness range, and can also be further adjusted af- ter a preset time of exceeding the eye protection brightness range. It can ensure that the user eye-sighted target is always within the eye protection brightness range to protect user eyes.

Further, when the eye-sighted target cannot be monitored for a period of time, the aforementioned steps A1-A3 can be repeated. That is, when the end of the user's work is detected, the light is automatically turned up to facilitate user activity in the room until the eye-sighted target is detected again.

In addition, the brightness of the ambient light can also be continuously monitored, and when the change in the brightness of the ambient light exceeds a brightness change threshold, steps Al- A3 are repeated. The above is the process of automatically adjust- ing the lighting intensity of the desk lamp, and the user can also manually control it through voice or touch keys during use.

Further, after collecting and studying the sitting postures of a large number of school-age children, it can be seen that most of the bad sitting postures are concentrated in 5 postures of “tilting the head, lowering the head, leaning on the arm, bending over, and arching the back”, and the “bending over and arching the back” are incorporated into the two postures of “lowering the head and leaning on the arm” by further training analysis on images, so finally the 5 main wrong sitting postures “tilting the head, low- ering the head, leaning on the arm, bending over, and arching the back” can be quickly and effectively determined by combined analy- sis on the contours of three parts “head, eyes and shoulders”.

Therefore, in the method B of this solution, it is determined that the sitting posture is bad when the user is in a head tilt state, a head-down state, or an arm leaning state.

Specifically, in the method B, whether the user is currently in a head tilt state, a head-down state, or an arm leaning state is detected by the following way: Bl. A sitting posture state of the user is captured through the camera, and a head contour and a shoulder contour are extract- ed; Bl. Head tilt detection, when the deviation of the angle be- tween the head contour and the shoulder contour from 90 degrees exceeds an angle deviation value (such as 15 degrees), it is de- termined that the user is in a head tilt state; B2. Head-down detection, eyes are extracted, the relative po- sitions of the eyes and the shoulder contour are compared, and when the distance between the two is less than a distance thresh- old, it is determined that the user is in a head-down state, the distance threshold being 1/2 or close to 1/2 of the user's entire head size; and B3. Arm leaning detection, when there is adhesion between the head contour and the shoulder contour, it is determined that the user is in an arm leaning state.

Specifically, in the method Bl, key features of the user's face, such as eyes, nasal tip, mouth corner points, eyebrows, and contour points of various parts of the face, are located by a face recognition algorithm to extract the head contour; and the user's head and shoulders are located by a shape recognition algorithm to extract the shoulder contour.

Further, this method includes fatigue detection: Face detection and eye detection are performed on the image captured by the camera, fatigue detection is performed on the user based on a fatigue detection algorithm, and a reminder is sent when the degree of fatigue reaches a fatigue threshold. The spe- cific method of fatigue detection can directly use the existing technology, which is not limited here.

Further, this method includes learning concentration detec- tion: Face detection and eye detection are performed on the image captured by the camera, concentration detection is performed on the user based on a concentration detection algorithm, and a re- minder is sent when the degree of concentration is lower than a concentration threshold. The user's concentration is detected by the concentration algorithm according to the face image, the key points of the face, and the coordinates of the key points of the face. The specific concentration detection method can directly use the existing technology, which is not limited here.

Further, this embodiment preferably sends the detection re- sults to a parent terminal intermittently. The “intermittently” may be regular or irregular. The detection result is subject to the irregularity. Every time a bad sitting posture, a fatigue state or bad concentration is detected, the detection result is sent to the parent terminal. Preferably, a prompt message, such as message push or vibration prompt, is sent to the parent terminal when the number of reminders reaches a reminder number threshold. In addition, a student terminal may be reminded by a voice remind- er, for example, when the head-down state is determined, a remind- er “Hello, your head is too low, please raise your head to main- tain a good sitting posture” will be sent.

Preferably, during use, when it is detected that the degree of concentration reaches a second concentration threshold, if the bad sitting posture or fatigue of a student is detected, any re- minder is not given, but the detection result is saved and sent to the parent terminal. In this way, the degree of concentration can be used as a first priority, and no sound is produced to disturb the student when the degree of concentration of the student is relatively high.

In this method, a set eye vision program is started to delay on and off when no one is detected; sitting posture detection, fa- tigue detection, and learning concentration detection are per- formed when someone is detected; and when a bad sitting posture, fatigue or insufficient concentration is detected, a voice remind- er is sent out, wherein the insufficient concentration may not be reminded, but directly saved in a storage module, so that parents can read child's learning concentration wirelessly.

Embodiment 2 As shown in FIG. 2, this embodiment discloses an AI eye pro- tection desk lamp with sitting posture correction reminder based on the method of Embodiment 1. The desk lamp includes an LED lighting module 7. The desk lamp also has a control module 1, a camera module 2, a voice broadcast module 3, touch keys 4, a voice collection module 5 and a storage module 6. The control module 1 is connected to the LED lighting module 7, the camera module 2, the voice broadcast module 3, the voice collection module 5, the storage module 6 and the touch keys 4. The control module 1 in- cludes an image recognition module 11 and a voice recognition mod- ule 12. The voice recognition module 12 has two functions: voice input recognition and voice output broadcast. The voice recogni- tion is about what the user says, and is used to control the on and off of the lamp, the adjustment of the lighting brightness, etc. The voice broadcast is used to prompt the user, such as: the user says “turn on the lamp”, the desk lamp is controlled to be turned on by the voice recognition module 12, and a prompt “the lamp is on” is fed back after the lamp is turned on; in addition, when the sitting posture recognition discovers an incorrect sit- ting posture of the user, a prompt “head is too low, please main- tain a good sitting posture” may be given, etc.

The image recognition module 11 includes an ambient light de- tection module and a sitting posture detection module.

The sitting posture detection module includes a face recogni- tion algorithm module for extracting a head contour and user eyes, and a shape recognition algorithm module for extracting a shoulder contour; Preferably, the image recognition module 11 further includes an eye-sighted target brightness detection module for detecting the reflective brightness of an eye-view target, a fatigue detec- tion algorithm module for detecting user fatigue, and a concentra- tion detection algorithm module for detecting user learning con- centration; In addition, the control module 1 is connected with a wire- less module 8, and is connected to a parent terminal through the wireless module 8.

The user can control the on and off of the lamp and the brightness of the lamp through voice and the touch keys 4.

The specific embodiments described herein are merely examples to illustrate the spirit of the present invention. Those skilled in the art to which the present invention belongs can make various modifications or additions or similar alternatives to the specific embodiments described, without deviating from the spirit of the present invention or exceeding the scope of the appended claims.

Claims

CONCLUSIONS

1. Implementation method of an Al desk lamp for eye protection with reminder for sitting posture correction, characterized in that it comprises: A. adjusting the light source, detecting the brightness of the ambient light, and adjusting the intensity of the LED -lighting in accordance with the brightness of the ambient light; and B. sitting posture correction, detecting a user's current sitting posture, and sending a reminder to the user when it is determined that the user's sitting posture is bad.

The implementation method of the AI desk lamp for eye protection with reminder for sitting posture correction according to claim 1, characterized in that the method A comprises in particular: A1. detecting the brightness of the ambient light and roughly adjusting the LED illumination intensity according to the brightness of the ambient light so that the brightness of the ambient light reaches a set brightness value; A2. detecting the reflective brightness of a target perceived with the eyesight of a user, and fine-tuning the illumination intensity up and down to a set value for reflective brightness in accordance with the reflective brightness; and A3. after the fine-tuning is completed, continuously tracking the reflective brightness of the target perceived by the user's eyesight, and further fine-tuning the intensity of the LED illumination intensity when the reflective brightness exceeds a brightness range for eye protection, until the reflective brightness reaches the set value for reflective brightness.

The implementation method of the AI desk lamp for eye protection with reminder for sitting posture correction according to claim 2, characterized in that in the method B, it is determined that the sitting posture is bad when the user is in a head tilted state, in a a downward position of the head, or in a leaning position on one arm.

The implementation method of the AI desk lamp for eye protection with seat correction reminder according to claim 3, characterized in that in the method B, whether the user is currently in a head tilted state, in a down position of the head, or in an arm leaning state, is detected in the following manner: Bl. recording a sitting position of the user by means of a camera, and extracting a head contour and a shoulder contour; p. when the deviation of the angle between the head contour and the shoulder contour of 90 degrees exceeds an angular deviation value, determining that the user is in a head tilted state; B2. extracting the eyes, comparing the relative positions of the eyes and the shoulder contour, and when the distance between the two is less than a distance threshold, determining that the user is in a down position from the head; and B3. if there is adhesion between the main contour and the shoulder contour, determining that the user is in an arm leaning state.

The implementation method of the AI desk lamp for eye protection with reminder for sitting posture correction according to claim 4, characterized in that in the method B1, the main features of the user's face are located by a face recognition algorithm to extract the head contour; and the user's head and shoulders are located by a shape recognition algorithm to extract the shoulder contour.

The implementation method of the AI desk lamp for eye protection with reminder for sitting posture correction according to claim 5, characterized in that it further comprises detecting the human body: performing recognition of the human body on an image captured by the camera, determining that no one is presently there when a human body is not recognized, and turning off the LED lighting after determining that no one is presently present for a preset time.

The implementation method of the AI desk lamp for eye protection with seat correction reminder according to claim 6, characterized in that it further comprises fatigue detection: performing face detection and eye detection on the image captured by the camera, performing user fatigue detection on based on a fatigue detection algorithm, and sending a reminder when the amount of fatigue reaches a fatigue threshold; further comprising learning concentration detection: performing face detection and eye detection on the image captured by the camera, performing concentration detection on the user based on a concentration detection algorithm, and sending a reminder when the degree of concentration is lower than a concentration threshold.

The implementation method of the AI desk lamp for eye protection with seat correction reminder according to claim 7, characterized in that the detection results are intermittently sent to a parent terminal and a prompt message is sent to the parent terminal when the number of reminders exceeds a threshold value for the number of memories reached.

An eye protection AI desk lamp with a seat correction reminder according to any one of claims 1 to 8, comprising a desk lamp with an LED lighting module (7), characterized in that the desk lamp also includes a control module (1), a camera module - le (2), has a speech broadcast module (3) and touch keys (4);

the control module (1) being connected to the LED lighting module (7), the camera module (2), the voice broadcasting module (3) and the touch keys (4); wherein the control module (1) comprises an image recognition module (11) and a speech recognition module (12), and the image recognition module (11) comprises an ambient light detection module and a sitting posture detection module.

The AI desk lamp for eye protection with seat correction reminder according to claim 9, characterized in that the control module (1) is also connected to a speech collection module (5) and a storage module (6); wherein the sitting posture detection module comprises a face recognition algorithm module for extracting a head contour and user eyes, and a shape recognition algorithm module for extracting a shoulder contour; the image recognition module (11) further comprising a visually-observed target detection module for detecting the reflective brightness of the visually-sighted target; wherein the image recognition module (11) comprises a fatigue detection algorithm module for detecting user fatigue and a concentration detection algorithm module for detecting user learning concentration; and wherein the control module (1) is connected to a wireless module (8), and is connected to a parent terminal via the wireless module (8).