Display screen brightness automatic adjusting method, mobile phone and storage medium

Abstract

The invention relates to the technical field of mobile phone display brightness control, in particular to an automatic display screen brightness adjusting method, a mobile phone and a storage medium, and the method comprises the steps: recording adjustment data, carrying out the statistics of an adjustment frequency, transmitting the adjustment data to a server, comparing the adjustment frequency with a preset threshold value, and if the adjustment frequency exceeds the preset threshold value, stopping the adjustment. If the adjustment frequency exceeds the preset threshold value, obtaining the simplified model from the server, if the adjustment frequency does not exceed the preset threshold value, obtaining the attention score variation from the server, optimizing the local simplified model according to the attention score variation, and finally adjusting the brightness of the display screen according to the local simplified model. According to the invention, the server operates the preset neural network model to realize the individuation of the user, and the simplified model is used for locally and automatically adjusting the brightness of the mobile phone of the user, and is optimized according to the change condition of the attention score of the preset neural network model. The problem that in the prior art, a mobile phone display screen brightness adjusting model cannot give consideration to light weight and individuation is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile phone display brightness control, and particularly to a method for automatically adjusting the brightness of a display screen, a mobile phone, and a storage medium. Background Art

[0002] The function of adjusting the brightness of a mobile phone is of crucial significance for improving the user experience. In a changing usage environment, precise brightness control can effectively reduce visual fatigue and protect the user's eye health. Although a simple linear model has a fast response ability and can quickly adjust the screen brightness according to the change of ambient light, its accuracy is insufficient and it cannot adapt to different usage habits of different users.

[0003] Therefore, the brightness adjustment model based on artificial intelligence has gradually become a mainstream research direction at present. In contrast, the complex artificial intelligence-based model performs excellently in terms of accuracy, can better fit the usage habits of different users, and has good personalization. However, due to its complex calculation process, it will cause great pressure on the operating performance of the mobile phone, and may lead to problems such as overheating and faster power consumption of the mobile phone, thus restricting its wide promotion in practical applications to a certain extent.

[0004] Therefore, people need a method for adjusting the brightness of a mobile phone display screen that takes into account both lightweight and personalization. Summary of the Invention

[0005] Therefore, the present invention provides a method for automatically adjusting the brightness of a display screen, a mobile phone, and a storage medium, so as to solve the problem that the existing mobile phone display screen brightness adjustment model cannot take into account both lightweight and personalization.

[0006] The present invention provides a method for automatically adjusting the brightness of a display screen, including:

[0007] When the user actively adjusts the brightness of the display screen, record the adjustment data and count the adjustment frequency. The adjustment data includes ambient data and brightness data;

[0008] Send the adjustment data to a server, and the server is used to train a preset neural network model according to the adjustment data. The preset neural network model is used to calculate the attention score of each ambient data according to the adjustment data and calculate the brightness data according to the attention score;

[0009] Compare the adjustment frequency with a preset threshold;

[0010] If the adjustment frequency exceeds the preset threshold, obtain a simplified model from the server. The simplified model is used to calculate the brightness data according to the adjustment data, and the simplified model is obtained by fitting based on the calculation result of the preset neural network model;

[0011] If the adjustment frequency does not exceed the preset threshold, obtain the change amount of the attention score from the server, and optimize the local simplified model according to the change amount of the attention score; wherein, the change amount of the attention score is the change amount of the attention score during the backpropagation process of the preset neural network model;

[0012] Adjust the display screen brightness according to the local simplified model.

[0013] The present invention also provides a preferred solution: optimizing the local simplified model according to the change amount of the attention score includes:

[0014] Preliminarily adjust the local simplified model according to the change amount of the attention score to obtain a transition model;

[0015] Obtain the adjustment data corresponding to the change amount of the attention score as the target adjustment data, and the target adjustment data includes target environment data and target brightness data;

[0016] Input the target adjustment data into the transition model to obtain an output result;

[0017] Adjust the transition model according to the difference between the output result and the target brightness data to obtain an optimized simplified model.

[0018] The present invention also provides a preferred solution: the simplified model includes multiple coefficients, and each coefficient corresponds to environment data; preliminarily adjusting the local simplified model according to the change amount of the attention score to obtain a transition model includes:

[0019] Adjust the value of the coefficient corresponding to each environment data according to the change amount of the attention score corresponding to each environment data to obtain a transition model.

[0020] The present invention also provides a preferred solution: adjusting the transition model according to the difference between the output result and the target brightness data to obtain an optimized simplified model includes:

[0021] Calculate the difference between the output result and the target brightness data;

[0022] Calculate the partial derivative of each coefficient in the transition model according to the difference;

[0023] Adjust the value of each coefficient in the transition model according to the partial derivative of each coefficient to obtain an optimized simplified model.

[0024] The present invention also provides a preferred solution: the simplified model is;

[0025] B 1 =∑(wX+b);

[0026] Wherein, B 1It represents the output result of the simplified model, X represents an environmental data, w is the coefficient corresponding to this type of adjustment data, and b is the bias corresponding to this type of adjustment data;

[0027] The transition model is:

[0028] B 2 = ∑((w + f(s))X + b + g(s));

[0029] Among them, B 2 represents the output result of the transition model, s represents the change in the attention score corresponding to this type of adjustment data, and f() and g() are different preset linear functions;

[0030] The optimized simplified model is;

[0031]

[0032] Among them, B 3 represents the output result of the optimized simplified model, is the partial derivative symbol, and α is the preset adjustment rate.

[0033] The present invention also provides a preferred solution: When the user adjusts the brightness of the display screen, record the adjustment data and count the adjustment frequency, including:

[0034] When the user adjusts the brightness of the display screen, obtain the brightness adjustment record;

[0035] Set an analysis window with a dynamic length in the brightness adjustment record, count the proportion of the brightness adjustment records in the analysis window that are actively adjusted by the user, and use this proportion as the adjustment frequency.

[0036] The present invention also provides a preferred solution: It also includes:

[0037] When the brightness of the mobile phone display screen is adjusted, determine whether this adjustment is an active adjustment by the user or an automatic adjustment by the system;

[0038] If this adjustment is an active adjustment by the user, reduce the length of the analysis window;

[0039] If this adjustment is an automatic adjustment by the system, increase the length of the analysis window.

[0040] The present invention also provides a preferred solution: The environmental data includes at least one of environmental light intensity, time, temperature, mobile phone backlight intensity, current application software, screen pixel distribution, face recognition data, fingerprint recognition data, and voiceprint recognition data; The preset neural network model includes an input layer, a query vector analysis layer, a key vector analysis layer, a softmax layer, a value vector analysis layer, a linear layer, and an output layer. Among them, the input layer is used to input environmental data. The input ends of the query vector analysis layer, the key vector analysis layer, and the value vector analysis layer are all connected to the input layer, and are respectively used to calculate the query vector, key vector, and value vector of each environmental data. The output ends of the query vector analysis layer and the key vector analysis layer are both connected to the softmax layer. The softmax layer is used to calculate the attention score of one environmental data relative to other environmental data. The output end of the softmax layer and the output end of the value vector analysis layer are both connected to the linear layer. The linear layer is connected to the output layer, and the output layer is used to output brightness data.

[0041] The present invention also provides a mobile phone, including:

[0042] A data recording module, configured to record adjustment data and count the adjustment frequency when the user actively adjusts the display screen brightness. The adjustment data includes environmental data and brightness data;

[0043] A data uploading module, configured to send the adjustment data to a server. The server is used to train a preset neural network model according to the adjustment data. The preset neural network model is used to calculate the attention score of each environmental data according to the adjustment data, and calculate the brightness data according to the attention score;

[0044] A comparison and optimization module, configured to compare the adjustment frequency with a preset threshold;

[0045] If the adjustment frequency exceeds the preset threshold, obtain a simplified model from the server. The simplified model is used to calculate the brightness data according to the adjustment data, and the simplified model is obtained by fitting based on the calculation result of the preset neural network model;

[0046] If the adjustment frequency does not exceed the preset threshold, obtain the change amount of the attention score from the server, and optimize the local simplified model according to the change amount of the attention score; wherein, the change amount of the attention score is the change amount of the attention score during the backpropagation process of the preset neural network model;

[0047] A brightness adjustment module, configured to adjust the display screen brightness according to the local simplified model.

[0048] The present invention also provides a computer-readable storage medium, used to store computer-readable programs or instructions. When the programs or instructions are executed by a processor, the steps in the automatic display screen brightness adjustment method described in any one of the above can be implemented.

[0049] The beneficial effects of adopting the above embodiments are as follows:

[0050] The present invention provides a method for automatically adjusting the brightness of a display screen, a mobile phone, and a storage medium. When a user actively adjusts the brightness of the display screen, the adjustment data is recorded and the adjustment frequency is counted. Then, the adjustment data is sent to a server. After that, the adjustment frequency is compared with a preset threshold. If the adjustment frequency exceeds the preset threshold, a simplified model is obtained from the server. If the adjustment frequency does not exceed the preset threshold, the change amount of the attention score is obtained from the server, and the local simplified model is optimized according to the change amount of the attention score. Finally, the brightness of the display screen is adjusted according to the local simplified model. By running a preset neural network model on the server, the preset neural network model is used to calculate the brightness and utilize the attention score to achieve user personalization. The simplified model is obtained by fitting based on the calculation results of the preset neural network model and is used to automatically adjust the brightness locally on the user's mobile phone, realizing lightweight, while ensuring the rapidity of the screen brightness adjustment response. Most importantly, the local simplified model in the present invention can be optimized according to the change of the attention score of the preset neural network model, so that when the mobile phone adjusts the screen in most cases, it can complete the automatic adjustment that conforms to the user's usage habits without waiting for the server to respond. At the same time, when the local linear model is inaccurate, it can also be refitted to obtain a more accurate simplified model, realizing the perfect combination of the accuracy, personalization, and lightweight of the mobile phone brightness adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a flowchart of the method for automatically adjusting the brightness of the display screen provided by the present invention;

[0052] Figure 2 is Figure 1 a partial specific step diagram of step S105 in

[0053] Figure 3 is a module schematic diagram of the mobile phone provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] Combined with Figure 1 shown, the present invention provides a method for automatically adjusting the brightness of a display screen, including:

[0056] S101. When the user actively adjusts the display screen brightness, record the adjustment data and count the adjustment frequency. The adjustment data includes environmental data and brightness data.

[0057] S102. Send the adjustment data to the server. The server is used to train a preset neural network model according to the adjustment data. The preset neural network model is used to calculate the attention score of each environmental data according to the adjustment data and calculate the brightness data according to the attention score.

[0058] S103. Compare the adjustment frequency with a preset threshold.

[0059] S104. If the adjustment frequency exceeds the preset threshold, obtain a simplified model from the server. The simplified model is used to calculate the brightness data according to the adjustment data. The simplified model is obtained by fitting based on the calculation results of the preset neural network model.

[0060] S105. If the adjustment frequency does not exceed the preset threshold, obtain the change amount of the attention score from the server and optimize the local simplified model according to the change amount of the attention score. Wherein, the change amount of the attention score is the change amount of the attention score during the backpropagation process of the preset neural network model.

[0061] S106. Adjust the display screen brightness according to the local simplified model.

[0062] In the above process, the environmental data refers to any data that can affect the user's adjustment of the mobile phone brightness, such as ambient light intensity, time, temperature, mobile phone backlight intensity, current application software, screen pixel distribution, face recognition data, fingerprint recognition data, and voiceprint recognition data, etc., which can be flexibly set according to the actual situation.

[0063] The preset neural network model is a pre-trained model stored in the server. It has an attention mechanism and can analyze the combined effects of various environmental factors to ensure the accuracy of brightness analysis. For example, a certain user is used to browsing video software at a low brightness after turning off the lights at night, and will not actively adjust the brightness when alone in a dark environment or when watching videos alone. Then the preset neural network model based on the attention mechanism can analyze the association between the two environmental factors of brightness and software and handle this situation. The attention mechanism is an existing technology and will not be elaborated in this embodiment.

[0064] It can be understood that training a preset neural network model requires a large amount of data, while the data of user brightness adjustment may be less and cannot support the training of the preset neural network model. Therefore, the data of all users can be collected to pre-train a general model, and then on the basis of this general model, a preset neural network model can be separately established for each user with this general model as the prototype to achieve personalized training, so as to ensure reliability and save the computing resources of the server at the same time.

[0065] It can be conceived that although the preset neural network can achieve personalization, its running cost is too high. Running it on a mobile phone may affect the user experience, and it is also unrealistic for the mobile phone to request data from the server every time the brightness is adjusted. Therefore, based on the preset neural network model, the embodiment simplifies the model of the preset neural network model into a simple model (for example, by fitting the input and output data of the preset neural network model to obtain a simple linear model), and then deploys this simplified model to the local area of the mobile phone, so that the mobile phone can quickly adjust without causing too much computational pressure on the mobile phone. In practice, it can be seen that the scenarios in which most users use mobile phones are relatively single. For example, according to the Pareto principle, users use 20% of the software on the mobile phone 80% of the time. Therefore, on the premise of conforming to the user's usage habits, a simple linear model can also meet the daily usage needs.

[0066] The initial simplified model in the mobile phone can be obtained by fitting the above general model. Similarly, because the number of times users adjust the brightness of the mobile phone is small, and optimizing the preset neural network model also requires a certain amount of data. Therefore, before the preset neural network model completes personalized training, the simplified model can be simply and preliminarily optimized locally to improve the user experience. When it is determined that the simple model is not accurate enough, then request the server to refit a more accurate simplified model.

[0067] This embodiment uses the adjustment frequency as the standard for judging the accuracy of the simplified model. The present invention defaults that the mobile phone will automatically adjust the brightness according to the simplified model in daily life, and the adjustment frequency refers to the proportion of the number of times the user actively adjusts the brightness to the total number of times the mobile phone adjusts the brightness. That is, every time the user adjusts the brightness, it can be regarded as that the current mobile phone brightness is unreasonable and the simplified model calculation is incorrect. Compared with other criteria for judging accuracy, the adjustment frequency can not only represent the accuracy rate, but also implicitly reflect the quantity of data that the mobile phone has collected, that is, when the adjustment frequency must be obtained on the basis that the user has adjusted the brightness a certain number of times. At this time, it can be considered that the optimization of the preset neural network model is reliable.

[0068] The optimization and simplification of the model on the mobile phone can also be carried out in any way. For example, according to the data at the last adjustment, the weight of one kind of environmental data with the largest current change rate in the simplified model is adjusted. And this embodiment provides a preferred way, that is, to reuse the attention analysis ability of the preset neural network model, and use the attention scores calculated by the preset neural network model to preliminarily optimize the simplified model. The attention score represents the correlation between each kind of environmental data and other environmental data. When the preset neural network performs backpropagation training, the change amount of each attention score will be obtained, and this change amount represents the deviation degree of the model analyzing one kind of environmental data, which can be used to optimize the local simplified model. In this way, the powerful computing power of the server is utilized. Under the condition of relatively small computing cost on the mobile phone, not only the correlation between various environmental data is analyzed, but also the influence degree of various environmental data is obtained, greatly improving the accuracy of optimizing the local simplified model.

[0069] It can be understood that the above-mentioned attention score can refer to any data that can characterize attention information obtained during the calculation of the attention mechanism. For example, in a preferred embodiment, the preset neural network model includes an input layer, a query vector analysis layer, a key vector analysis layer, a softmax layer, a value vector analysis layer, a linear layer, and an output layer. Among them, the input layer is used to input environmental data. The input ends of the query vector analysis layer, the key vector analysis layer, and the value vector analysis layer are all connected to the input layer, and are respectively used to calculate the query vector, the key vector, and the value vector of each kind of environmental data. The output ends of the query vector analysis layer and the key vector analysis layer are both connected to the softmax layer, and the softmax layer is used to calculate the attention score of one kind of environmental data relative to other environmental data. The output end of the softmax layer and the output end of the value vector analysis layer are both connected to the linear layer, and the linear layer is connected to the output layer, and the output layer is used to output brightness data.

[0070] In the above process, the result output by the softmax layer is used as the attention score. This data has been normalized and is relatively intuitive. In practice, the product of the query vector and the key vector, or the product result of the output of the softmax layer and the value vector can also be used as the attention score.

[0071] Further, in a preferred embodiment, the above step S101, when the user actively adjusts the brightness of the display screen, records the adjustment data and counts the adjustment frequency, specifically includes:

[0072] When the user adjusts the brightness of the display screen, obtain the brightness adjustment record;

[0073] Set an analysis window with a dynamic length in the brightness adjustment record, count the proportion of the brightness adjustment records in the analysis window that are actively adjusted by the user, and use this proportion as the adjustment frequency.

[0074] Specifically, in a preferred embodiment, in the above steps, the analysis window refers to the data range on the time scale for statistically adjusting the frequency. In this embodiment, the analysis window is designed to be dynamic to improve practicability and avoid extreme cases such as not requesting a new fitted simplified model for a long time or frequently requesting a new fitted simplified model.

[0075] Specifically, in a preferred embodiment, the dynamic adjustment method for the length of the above analysis window is as follows:

[0076] When the brightness of the mobile phone display screen is adjusted, it is judged whether this adjustment is an active adjustment by the user or an automatic adjustment by the system;

[0077] If this adjustment is an active adjustment by the user, the length of the analysis window is reduced;

[0078] If this adjustment is an automatic adjustment by the system, the length of the analysis window is increased.

[0079] In this embodiment, an automatic adjustment by the system indicates that the calculation of the simplified model is accurate, while an active adjustment by the user indicates that the calculation of the simplified model is inaccurate. When the calculation is accurate, the length of the analysis window will increase. At this time, the field of view for statistically adjusting the frequency will become wider, and at the same time, the specific calculated value of the adjustment frequency will further decrease, making the algorithm more stable. That is, the mobile phone is more inclined to adjust the local simplified model, avoiding frequently requesting a new fitted simplified model, thus not destroying the original brightness adjustment preference of the mobile phone, ensuring the user experience, and at the same time reducing the server pressure.

[0080] When the calculation is inaccurate, the length of the analysis window will be shortened, and in the case of frequent adjustment by the user, the statistically adjusted frequency will rise steeply. In this way, it can be ensured that the mobile phone will promptly request a more accurate model, improving the response speed and ensuring the user experience.

[0081] Further, as shown in Figure 2 In a preferred embodiment, in the above step S105, optimizing the local simplified model according to the change amount of the attention score specifically includes:

[0082] S201. Initially adjust the local simplified model according to the change amount of the attention score to obtain a transition model;

[0083] S202. Obtain the adjustment data corresponding to the change amount of the attention score as the target adjustment data. The target adjustment data includes target environment data and target brightness data;

[0084] S203. Input the target adjustment data into the transition model to obtain an output result;

[0085] S204. Adjust the transition model according to the difference between the output result and the target brightness data to obtain an optimized simplified model.

[0086] In this embodiment, the simplified model is optimized twice. One optimization is based on the change in attention scores, and the other is based on the target adjustment data for further optimization. The analysis is carried out from two dimensions to improve the accuracy of optimization.

[0087] Specifically, in a preferred embodiment, the simplified model includes multiple coefficients, and each coefficient corresponds to environmental data. The above step S201, according to the change in attention scores, preliminarily adjusts the local simplified model to obtain a transitional model, which specifically includes:

[0088] According to the change in attention scores corresponding to each environmental data, adjust the value of the coefficient corresponding to this environmental data to obtain a transitional model.

[0089] Further, in a preferred embodiment, the above step S204, according to the difference between the output result and the target brightness data, adjusts the transitional model to obtain an optimized simplified model, which specifically includes:

[0090] Calculate the difference between the output result and the target brightness data;

[0091] Calculate the partial derivative of each coefficient in the transitional model according to the difference;

[0092] According to the partial derivative of each coefficient, adjust the value of each coefficient in the transitional model to obtain an optimized simplified model.

[0093] In the above process, the method of optimizing the transitional model refers to the gradient descent method in the neural network, which has a fast operation speed, high accuracy, and low consumption of mobile phone computing resources.

[0094] Specifically, in a preferred embodiment, the specific steps of the above steps S201 - S204 are:

[0095] The simplified model is:

[0096] B 1 = ∑(wX + b);

[0097] Among them, B 1 represents the output result of the simplified model, X represents a kind of environmental data, w is the coefficient corresponding to this kind of adjustment data, and b is the bias corresponding to this kind of adjustment data;

[0098] The transitional model is:

[0099] B 2 = ∑((w + f(s))X + b + g(s));

[0100] Among them, B 2represents the output result of the transition model, s represents the change in the attention score corresponding to this type of adjustment data, f() and g() are different preset linear functions, and both can be obtained according to experiments or experience;

[0101] The optimized simplified model is;

[0102]

[0103] where B 3 represents the output result of the optimized simplified model, is the partial derivative symbol, α is the preset adjustment rate, and the preset adjustment rate can also be obtained according to experiments or experience.

[0104] Combined with Figure 3 as shown, the present invention also provides a mobile phone, including:

[0105] A data recording module 310, configured to record adjustment data and count the adjustment frequency when the user actively adjusts the display screen brightness, where the adjustment data includes environmental data and brightness data;

[0106] A data uploading module 320, configured to send the adjustment data to a server, where the server is used to train a preset neural network model according to the adjustment data, and the preset neural network model is used to calculate the attention score of each environmental data according to the adjustment data and calculate the brightness data according to the attention score;

[0107] A comparison and optimization module 330, configured to compare the adjustment frequency with a preset threshold;

[0108] If the adjustment frequency exceeds the preset threshold, obtain a simplified model from the server, where the simplified model is used to calculate the brightness data according to the adjustment data, and the simplified model is obtained by fitting based on the calculation results of the preset neural network model;

[0109] If the adjustment frequency does not exceed the preset threshold, obtain the change in the attention score from the server, and optimize the local simplified model according to the change in the attention score; where the change in the attention score is the change in the attention score during the backpropagation process of the preset neural network model;

[0110] A brightness adjustment module 340, configured to adjust the display screen brightness according to the local simplified model.

[0111] It should be noted here that: the corresponding modules provided in the above embodiments can implement the technical solutions described in the above method embodiments, and the specific implementation principles of the above modules or units can refer to the corresponding content in the above method embodiments, which will not be elaborated here.

[0112] This embodiment also provides a computer-readable storage medium, on which an automatic display screen brightness adjustment program is stored. When the automatic display screen brightness adjustment program is executed by a processor, the steps in the above embodiment can be implemented.

[0113] The present invention provides an automatic display screen brightness adjustment method, a mobile phone, and a storage medium. When the user actively adjusts the display screen brightness, the adjustment data is recorded and the adjustment frequency is counted, and then the adjustment data is sent to the server. After that, the adjustment frequency is compared with a preset threshold. If the adjustment frequency exceeds the preset threshold, a simplified model is obtained from the server. If the adjustment frequency does not exceed the preset threshold, the change amount of the attention score is obtained from the server, and the local simplified model is optimized according to the change amount of the attention score. Finally, the display screen brightness is adjusted according to the local simplified model. The present invention runs a preset neural network model on the server. The preset neural network model is used to calculate the brightness and utilize the attention score to achieve user personalization. The simplified model is obtained by fitting based on the calculation result of the preset neural network model and is used to automatically adjust the brightness locally on the user's mobile phone to achieve lightweight, while ensuring the rapidity of the screen brightness adjustment response. Most importantly, the local simplified model in the present invention can be optimized according to the change of the attention score of the preset neural network model, so that when the mobile phone adjusts the screen in most cases, it can complete the automatic adjustment that conforms to the user's usage habits without waiting for the server to respond. At the same time, when the local linear model is inaccurate, it can also be refitted to obtain a more accurate simplified model, realizing the perfect combination of the accuracy, personalization, and lightweight of the mobile phone brightness adjustment.

[0114] It should be noted that the various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other.

[0115] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automatically adjusting the brightness of a display screen, characterized in that: include: When the user actively adjusts the brightness of the display screen, the adjustment data is recorded and the adjustment frequency is counted. The adjustment data includes environmental data and brightness data. The adjustment data is sent to a server, and the server is used to train a preset neural network model according to the adjustment data. The preset neural network model is used to calculate the attention score of each environmental data according to the adjustment data, and calculate the brightness data according to the attention score; comparing the adjustment frequency with a preset threshold; If the adjustment frequency exceeds a preset threshold, a simplified model is obtained from the server, the simplified model is used to calculate the brightness data according to the adjustment data, and the simplified model is fitted based on the calculation result of the preset neural network model; If the adjustment frequency does not exceed the preset threshold, the attention score change is obtained from the server, and the local simplified model is optimized according to the attention score change; wherein the attention score change is the change of the attention score of the preset neural network model during the back propagation process; Adjust the display brightness based on the local simplified model.

2. The method for automatically adjusting display screen brightness according to claim 1, characterized in that: Optimize the local simplified model according to the change in attention score, including: According to the change in attention score, the local simplified model is preliminarily adjusted to obtain the transition model; Acquire adjustment data corresponding to the change in the attention score as target adjustment data, where the target adjustment data includes target environment data and target brightness data; Input the target adjustment data into the transition model to obtain the output result; According to the difference between the output result and the target brightness data, the transition model is adjusted to obtain an optimized simplified model.

3. The method for automatically adjusting display screen brightness according to claim 2, characterized in that: The simplified model includes multiple coefficients, each of which corresponds to environmental data; According to the change in attention score, the local simplified model is initially adjusted to obtain a transition model, including: According to the change in attention score corresponding to each environmental data, the value of the coefficient corresponding to the environmental data is adjusted to obtain a transition model.

4. The method for automatically adjusting display screen brightness according to claim 3, characterized in that: According to the difference between the output result and the target brightness data, the transition model is adjusted to obtain an optimized simplified model, including: Calculate the difference between the output result and the target brightness data; Calculate the partial derivative of each coefficient in the transition model based on the difference; According to the partial derivative of each coefficient, the value of each coefficient in the transition model is adjusted to obtain an optimized simplified model.

5. The method for automatically adjusting display screen brightness according to claim 4, characterized in that: The simplified model is; B1=∑(wX+b); Among them, B1 represents the output result of the simplified model, X represents a kind of environmental data, w is the coefficient corresponding to the adjustment data, and b is the bias corresponding to the adjustment data; The transition model is: B2=∑((w+f(s))X+b+g(s)); Among them, B2 represents the output result of the transition model, s represents the change in attention score corresponding to the adjustment data, and f() and g() are different preset linear functions respectively; The simplified model after optimization is: Among them, B3 represents the output result of the optimized simplified model. is the sign of partial derivative, and α is the preset adjustment rate.

6. The method for automatically adjusting display screen brightness according to claim 1, characterized in that: When the user adjusts the brightness of the display, the adjustment data is recorded and the adjustment frequency is counted, including: When the user adjusts the brightness of the display screen, obtain the brightness adjustment record; An analysis window of dynamic length is set in the brightness adjustment record, and the brightness adjustment record in the statistical analysis window is the proportion of the user's active adjustment, and the proportion is used as the adjustment frequency.

7. The method for automatically adjusting display screen brightness according to claim 6, characterized in that: Also includes: When the brightness of the mobile phone display is adjusted, determine whether the adjustment is made actively by the user or automatically by the system; If this adjustment is made by the user actively, the length of the analysis window is reduced; If this adjustment is made automatically by the system, the length of the analysis window will be increased.

8. The method for automatically adjusting display screen brightness according to claim 1, characterized in that: The environmental data includes at least one of ambient light intensity, time, temperature, mobile phone backlight intensity, current application software, screen pixel distribution, face recognition data, fingerprint recognition data and voiceprint recognition data; the preset neural network model includes an input layer, a query vector analysis layer, a key vector analysis layer, a softmax layer, a value vector analysis layer, a linear layer and an output layer, wherein the input layer is used to input environmental data, the input ends of the query vector analysis layer, the key vector analysis layer and the value vector analysis layer are all connected to the input layer, which are respectively used to calculate the query vector, key vector and value vector of each environmental data, the output ends of the query vector analysis layer and the key vector analysis layer are both connected to the softmax layer, the softmax layer is used to calculate the attention score of one environmental data relative to other environmental data, the output end of the softmax layer and the output end of the value vector analysis layer are both connected to the linear layer, the linear layer is connected to the output layer, and the output layer is used to output brightness data.

9. A mobile phone, characterized in that: include: A data recording module is used to record adjustment data and count the adjustment frequency when the user actively adjusts the brightness of the display screen. The adjustment data includes environmental data and brightness data. A data upload module, used to send the adjustment data to the server, the server is used to train a preset neural network model according to the adjustment data, the preset neural network model is used to calculate the attention score of each environmental data according to the adjustment data, and calculate the brightness data according to the attention score; A comparison optimization module, used for comparing the adjustment frequency with a preset threshold; If the adjustment frequency exceeds a preset threshold, a simplified model is obtained from the server, the simplified model is used to calculate the brightness data according to the adjustment data, and the simplified model is fitted based on the calculation result of the preset neural network model; If the adjustment frequency does not exceed the preset threshold, the attention score change is obtained from the server, and the local simplified model is optimized according to the attention score change; wherein the attention score change is the change of the attention score of the preset neural network model during the back propagation process; The brightness adjustment module is used to adjust the brightness of the display screen according to a local simplified model.

10. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps of the method for automatically adjusting the brightness of a display screen as described in any one of claims 1 to 8.

Images