WO2022042526A1 - 触发电子设备预载功能的方法、电子装置及存储介质 - Google Patents

触发电子设备预载功能的方法、电子装置及存储介质 Download PDF

Info

Publication number
WO2022042526A1
WO2022042526A1 PCT/CN2021/114252 CN2021114252W WO2022042526A1 WO 2022042526 A1 WO2022042526 A1 WO 2022042526A1 CN 2021114252 W CN2021114252 W CN 2021114252W WO 2022042526 A1 WO2022042526 A1 WO 2022042526A1
Authority
WO
WIPO (PCT)
Prior art keywords
gesture data
electronic device
negative sample
model
triggering
Prior art date
Application number
PCT/CN2021/114252
Other languages
English (en)
French (fr)
Inventor
吴耿晖
Original Assignee
深圳市万普拉斯科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市万普拉斯科技有限公司 filed Critical 深圳市万普拉斯科技有限公司
Publication of WO2022042526A1 publication Critical patent/WO2022042526A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/017—Gesture based interaction, e.g. based on a set of recognized hand gestures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present application relates to the technical field of terminal equipment, and in particular, to a method, an electronic device, and a storage medium for triggering a preload function of an electronic device.
  • the design of current electronic devices is becoming more and more user-friendly, and some preload functions are set according to the user's actions. For example, most of the current electronic devices The screen is automatically turned on when the electronic device is picked up or other possible usage patterns occur, allowing users to light up the screen without pressing a button to view the contents of the phone or trigger subsequent functions such as unlocking.
  • the existing preload function is mainly realized by manual design rules, and the basis for judging that the screen needs to be turned on is: whether the current gesture data obtained by using the sensor matches the preset fixed gesture data, and if so, the screen is turned on. or other preloaded features.
  • the fixed gesture data preset by the above method is fixed, so the triggering conditions of the electronic device for the preload function are fixed, and the user must use a fixed gesture to start the preload function, and the non-preset gesture cannot trigger the preload function. load function, thereby degrading the user experience.
  • the present application provides a method for triggering the preloading function of an electronic device, so as to solve the problem that the triggering condition of the existing electronic device for the preloading function is fixed, and the user wants to use a non-preset gesture to trigger the preloading function. implementation, thereby reducing the problem of user experience.
  • a first aspect of the present application provides a method for controlling an electronic device to trigger a preload function, including: collecting current gesture data of a user operating an electronic device; inputting the current gesture data into a basic model trained in advance using a first negative sample, the first negative sample Defined as fixed gesture data that does not trigger the preload function; if the basic model is used to identify the current gesture data as a non-negative sample, the current gesture data is input into the inference model trained in advance using the second negative sample, and the second negative
  • the sample is defined as the historical gesture data that does not trigger the preload function; if the current gesture data is identified as a negative sample by the inference model, the preload function of the electronic device is not triggered, and the current gesture data is used as a new history
  • the gesture data trains the inference model; if the current gesture data is not a negative sample, the current gesture is recognized as a positive sample, and the preloading function of the electronic device is triggered.
  • the method further includes: if the current gesture data is identified as a non-negative sample by using the basic model, inputting the current gesture data into a cluster model trained by using positive samples in advance, the positive sample The sample is defined as the historical gesture data that triggers the preload function of the electronic device; if the non-negative sample is identified as a positive sample by the cluster model, and the current gesture data is identified as a positive sample by the inference model, the electronic device is triggered preload function.
  • the training method of the inference model includes: collecting historical gesture data of the user operating the electronic device within a predetermined time period; inputting the historical gesture data into the basic model; if using the basic model to identify If the historical gesture data is a non-negative sample, use a sliding window algorithm to process the historical gesture data of the non-negative sample to obtain a processing result; input the processing result into the basic model, if the basic model is used to identify If the processing result is a negative sample, the historical gesture data corresponding to the processing result is marked as a negative sample; if the basic model is used to identify the processing result as a non-negative sample, the historical gesture corresponding to the processing result is marked as a negative sample.
  • the data is marked as a positive sample; the historical gesture data marked as a negative sample and its corresponding negative sample label are stored in a database to train the inference model as sample data.
  • the method for training the cluster model includes: training the cluster model using the historical gesture data marked as positive samples as sample data.
  • the training method for the inference model further includes: collecting historical gesture data that fluctuates greatly because the user does not trigger the preload function of the electronic device as sample data for training the inference model.
  • the method for setting the sliding window of the sliding window algorithm includes: setting pre-sliding windows of different sizes, and using each pre-sliding window of different sizes to run the sliding window algorithm to obtain different pre-sliding windows Processing result; select the minimum pre-sliding window that does not affect the accuracy of the basic model in the processing result as the sliding window.
  • the method further includes: judging whether the user is in a sleep state according to the usage of the electronic device; acquiring the power and charging state of the electronic device; In the state, the inference model and the cluster model are trained.
  • the preloading function is to light up the screen or preload a program, or start fingerprint unlocking or start face recognition unlocking.
  • the current gesture data and the historical gesture data are values obtained by using an accelerometer or a gyroscope of an electronic device.
  • a second aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, when the processor executes the computer program, the above-mentioned
  • the method for triggering a preload function of an electronic device according to any one of the above.
  • a third aspect of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements any one of the methods for triggering the preloading function of an electronic device described above.
  • the above-mentioned method, electronic device and storage medium for triggering the preload function of an electronic device in the present application use the fixed gesture data that does not trigger the preload function as the first negative sample instead of fixing the gesture data that can trigger the preload function. If the current gesture data does not conform to the first negative sample, it can be regarded as gesture data that can trigger the preload function. Therefore, the preload function can be triggered without fixing the gesture data that can trigger the preload function. And if the inference model is used to identify the current gesture data as a negative sample, the preload function of the electronic device will not be triggered, and the current gesture data is used as new historical gesture data to train the inference model, which can make the inference model memorize. More negative samples, even if the gesture data that has to trigger the preload function of the electronic device increases, can reduce the unpredictability caused by the increase of positive samples.
  • FIG. 1 is a schematic flowchart of a method for triggering a preload function of an electronic device according to an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a specific implementation of a method for triggering a preload function of an electronic device according to an embodiment of the present application
  • FIG. 3 is a schematic structural block diagram of an electronic device according to an embodiment of the present application.
  • an embodiment of the present application provides a method for triggering a preload function of an electronic device, including: S1. Collecting current gesture data of a user operating an electronic device; In the basic model, the first negative sample is defined as the fixed gesture data that does not trigger the preload function; S3.
  • the current gesture data is input into the inference trained in advance using the second negative sample model, the second negative sample is defined as the historical gesture data that does not trigger the preload function; S4, if the inference model is used to identify the current gesture data as a negative sample, the preload function of the electronic device is not triggered, and the current gesture data is used as a new The historical gesture data trains the inference model; S5, if the current gesture data is not a negative sample, identify the current gesture as a positive sample, and trigger the preloading function of the electronic device.
  • the fixed gesture data that does not trigger the preload function is used as the first negative sample instead of the gesture data that can trigger the preload function. Therefore, if the input current gesture data does not conform to the first negative sample, it can be seen It is gesture data that can trigger the preload function, so the preload function can be triggered without fixing the gesture data that can trigger the preload function. And if the inference model is used to identify the current gesture data as a negative sample, the preload function of the electronic device will not be triggered, and the current gesture data is used as new historical gesture data to train the inference model, which can make the inference model memorize. More negative samples, even if the gesture data that has to trigger the preload function of the electronic device increases, can reduce the unpredictability caused by the increase of positive samples.
  • steps S1, S2, S3, S4, and S5 can be used to complete the triggering of the preload function operation of the electronic device.
  • the current gesture data obtained by S1 can also be directly input into the inference model of S3. , which simplifies the process of triggering the preload function of the electronic device and increases the degree of freedom. Since the inference model can continuously learn, it can continuously update the trigger conditions of the electronic device for the preload function, enabling users to use non-preset gestures Trigger the preload function of the electronic device, thereby increasing the user experience.
  • the method for triggering the preloading function of the electronic device further includes: S6, if the current gesture data is identified as a non-negative sample by using the basic model, the current gesture data input adopts a cluster model trained in advance using positive samples, and the positive samples are defined as Trigger the historical gesture data of the preload function of the electronic device; S7. If the cluster model is used to identify the non-negative sample as a positive sample, and the inference model identifies the current gesture data as a positive sample, trigger the preload function of the electronic device.
  • the triplet loss function can also be used to let the data point in several ways, so that the inference model can output whether the current sensor data is a non-negative sample.
  • the training method of the inference model includes: collecting historical gesture data of the user operating the electronic device within a predetermined time period; inputting the historical gesture data into the basic model; if using the basic model to identify the historical gesture data as a non-negative sample, Then use the sliding window algorithm to process the historical gesture data of the non-negative samples to obtain the processing results; input the processing results into the basic model, if the basic model is used to identify the processing results as negative samples, then mark the historical gesture data corresponding to the processing results as negative.
  • the historical gesture data corresponding to the processing result will be marked as a positive sample; the historical gesture data marked as a negative sample and its corresponding negative sample label will be stored in the database. Train an inference model as sample data.
  • the training method of the cluster model includes: using the historical gesture data marked as positive samples as sample data to train the cluster model.
  • Clustering models can be trained using conventional and its learning or autoencoders or variational autoencoders.
  • the historical gesture data corresponding to the user's historical operations within a predetermined period of time will be collected. It is hoped that the user can also respond quickly when the user has to use the mobile phone with other habits (such as injury), and it is necessary to obtain historical gesture data of different lengths from 2-15 days to train the inference model separately, and the selection requires that the correct rate of the validation set is high enough (such as , 99%) and select the parameters with the highest overall training accuracy on the premise that the prediction results of the historical gesture data of the last day are high enough.
  • This method enables the user to trigger the preload function of the electronic device even after the user changes his habits.
  • the growth of positive samples is highly unpredictable, so the performance of the inference model on negative samples needs to be high enough to limit the extent to which positive samples change their performance, thereby triggering the preload function of electronic devices more accurately.
  • the method for setting the sliding window of the sliding window algorithm includes: setting pre-sliding windows of different sizes, and using each pre-sliding window of different sizes to run the sliding window algorithm to obtain processing results of different pre-sliding windows; In the processing results, the minimum pre-sliding window that does not affect the accuracy of the basic model is used as the sliding window.
  • the sliding window of the sliding window algorithm can also use the attention mechanism or convolution of deep learning to directly obtain important information or high-level information, and use this information to form a sliding window, so that the inference model can be closer to the end-to-end. Learn.
  • the method for triggering the preload function of the electronic device further includes: judging according to the usage of the electronic device. Whether the user is in a sleep state; obtain the power and charging state of the electronic device; train the inference model and the cluster model when the user is in a sleep state and the electronic device has sufficient power and is in a charging state.
  • the time that the user is in the sleep state can be obtained, so that the inference model and the cluster model can be trained when the user is in the sleep state; Ensure that the power of electronic devices is always sufficient, and reduce the probability of insufficient power of electronic devices.
  • the preload function can be to light up the screen, so as to realize the function of automatically brightening the screen.
  • the preload function may be to start a preload program, or to start fingerprint unlocking, or to start face recognition unlocking.
  • the gesture for operating the electronic device includes, but is not limited to, the gesture of the user holding or wearing the electronic device for touch operation, the gesture or posture of the user holding or wearing the electronic device, the user's gesture or gesture in a non-contact manner Gesture control electronics.
  • the current gesture data and historical gesture data may be gesture data represented by values of an accelerometer or an acceleration sensor of an electronic device. In other embodiments, the current gesture data and historical gesture data may also be gestures represented by values of a gyroscope or an orientation sensor.
  • the data can also be gesture data represented by a value of a gravimeter or a gravity sensor, or gesture data represented by a value of an ultrasonic sensor or a depth sensor.
  • the user's operation can be judged by the difference of the gesture data, for example: the gravimeter or the gravity sensor judges the change of the pointing of the electronic device or the way it learns, such as setting the user's operation to collect the gesture data to apply the inference model Training.
  • the electronic device in this embodiment may be a mobile phone, a tablet computer, or a handheld electronic product such as a handheld game console.
  • an embodiment of the present application further provides an electronic device, the electronic device includes: a memory 601, a processor 602, and a computer program stored in the memory 601 and running on the processor 602, and the processor 602 executes the computer program
  • the method for triggering the preloading function of an electronic device described in the foregoing is implemented.
  • the electronic device further includes: at least one input device 603 and at least one output device 604 .
  • the above-mentioned memory 601 , processor 602 , input device 603 and output device 604 are connected through a bus 605 .
  • the input device 603 may specifically be a camera, a touch panel, a physical button, a mouse, or the like.
  • the output device 604 may specifically be a display screen.
  • the memory 601 may be a high-speed random access memory (RAM, Random Access Memory) memory, or may be a non-volatile memory (non-volatile memory), such as a disk memory.
  • RAM Random Access Memory
  • non-volatile memory such as a disk memory.
  • Memory 601 is used to store a set of executable program codes, and processor 602 is coupled to memory 601 .
  • an embodiment of the present application further provides a computer-readable storage medium, which may be provided in the electronic device in each of the foregoing embodiments, and the computer-readable storage medium may be the foregoing memory 601.
  • a computer program is stored on the computer-readable storage medium, and when the program is executed by the processor 602, the method for triggering the preloading function of the electronic device described in the foregoing embodiments is implemented.
  • the computer-readable storage medium may also be a U disk, a removable hard disk, a read-only memory 601 (ROM, Read-Only Memory), a RAM, a magnetic disk or an optical disk and other media that can store program codes.
  • ROM Read-Only Memory
  • RAM Random Access Memory

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • User Interface Of Digital Computer (AREA)
  • Telephone Function (AREA)

Abstract

一种触发电子设备预载功能的方法，包括：将采集的当前手势数据输入预先使用第一负样本训练的基础模型，第一负样本为非触发预载功能的手势数据；若基础模型识别当前手势数据为非负样本，则将当前手势数据输入采用预先使用第二负样本训练的推理模型，第二负样本为非触发预载功能的历史手势数据；若推理模型识别当前手势数据为负样本，则不触发预载功能，且将当前手势数据作为新的历史手势数据训练推理模型；若当前手势数据不为负样本，触发预载功能；使用负样本作为新的历史手势数据去训练推理模型，使得推理模型记忆了更多负样本，增加了判断不触发电子设备的预载功能的判定条件，从而使得用户可以使用非预先设置的手势触发预载功能。

Description

触发电子设备预载功能的方法、电子装置及存储介质
相关申请的交叉引用
本申请基于申请号为202010866621.3、申请日为2020年08月25日的中国专利申请提出，并要求该中国专利申请的优先权，该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本申请涉及终端设备技术领域，具体涉及一种触发电子设备预载功能的方法、电子装置及存储介质。
背景技术
为了增加用户体验，现在的电子设备设计的越来越人性化，根据用户的动作设置了一些预载功能，例如，目前的电子设备，大多具有抬手亮屏或类似的功能，即在用户将电子设备拿起或其他可能的使用行为模式发生时自动点亮屏幕，让用户无需通过按键即可点亮屏幕来查看手机内容或进行触发后续如解锁等功能。
现有的预载功能，主要通过人工设计规则来实现，而判断需要亮屏的依据则为：使用传感器得到的当前手势数据是否与预先设置的固定手势数据相匹配，若匹配，则点亮屏幕或其他预载功能。
然而，上述方法预先设置的固定手势数据是固定不变的，因此电子设备对于预载功能的触发条件是固定的，用户须使用固定手势才能出发预载功能，使用非预先设置的手势无法触发预载功能，从而降低了用户体验。
发明内容
鉴于此，本申请提供一种触发电子设备预载功能的方法，以解决现有的电子设备对于预载功能的触发条件是固定的，用户想要使用非预先设置的手势触发预载功能，无法实现，从而降低了用户体验的问题。
本申请第一方面提供一种控制电子设备触发预载功能的方法，包括：采集用户操作电子设备的当前手势数据；将当前手势数据输入预先使用第一负样本训练的基础模型，第一负样本定义为固定的非触发预载功能的手势数据；若利用所述基础模型识别当前手势数据为非负样本，则将所述当前手势数据输入预先使用第二负样本训练的推理模型，第二负样本定义为非触发预载功能的历史手势数据；若利用所述推理模型识别所述当前手势数据为负样本，则不触发电子设备的预载功能，且将所述当前手势数据作为新的历史手势数据训练推理模型；若所述当前手势数据不为负样本，则识别所述当前手势为正样本，触发电子设备的预载功能。
在一些可选实施方式中，所述方法还包括：若利用所述基础模型识别当前手势数据为非负样本，则将所述当前手势数据输入采用预先使用正样本训练的群集模型，所述正样本定义为触发电子设备的预载功能的历史手势数据；若使用所述群集模型识别所述非负样本为正样本，且所述推理模型识别所述当前手势数据为正样本，则触发电子设备的预载功能。
在一些可选实施方式中，所述推理模型的训练方法包括：采集预定时间段内用户操作电子设备的历史手势数据；将所述历史手势数据输入所述基础模型；若使用所述基础模型识别所述历史手势数据为非负样本，则使用滑动窗口算法对非负样本的所述历史手势数据进行处理，得到处理结果；将所述处理结果输入所述基础模型，若使用所述基础模型识别所述处理结果为负样本，则将所述处理结果对应的历史手势数据标记为负样本；若使用所述基础模型识别所述处理结果为非负样本，则将所述处理结果对应的 历史手势数据标记为正样本；将标记为负样本的所述历史手势数据及其对应的负样本标签存入数据库中，以作为样本数据训练所述推理模型。
在一些可选实施方式中，所述群集模型的训练方法包括：使用标记为正样本的所述历史手势数据作为样本数据训练群集模型。
在一些可选实施方式中，所述推理模型的训练方法还包括：采集用户未触发电子设备的预载功能而发生波动较大的历史手势数据为训练所述推理模型的样本数据。
在一些可选实施方式中，所述滑动窗口算法的滑动窗口的设置方法包括：设置不同大小的预滑动窗口，并使用每个不同大小的预滑动窗口运行滑动窗口算法，得到不同预滑动窗口的处理结果；选取处理结果中不影响所述基础模型的正确率、且最小的预滑动窗口作为滑动窗口。
在一些可选实施方式中，所述方法还包括：根据电子设备的使用情况判断用户是否处于睡眠状态；获取电子设备的电量及充电状态；在用户处于睡眠状态、且电子设备电量充足、处于充电状态下，训练所述推理模型及所述群集模型。
在一些可选实施方式中，所述预载功能为点亮屏幕或预载程序或启动指纹解锁或启动人脸识别解锁。
在一些可选实施方式中，所述当前手势数据及所述历史手势数据为使用电子设备的加速度计或陀螺仪得到的数值。
本申请第二方面提供一种电子装置，包括：存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时，实现上述中的任意一项所述的触发电子设备预载功能的方法。
本申请第三方面提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时，实现上述中的任意一项所述的触发 电子设备预载功能的方法。
本申请上述触发电子设备预载功能的方法、电子装置及存储介质，将固定的非触发预载功能的手势数据作为第一负样本而不是去固定能够触发预载功能的手势数据，因此输入的当前手势数据如不符合第一负样本就可视为可触发预载功能的手势数据，因此不需要固定可触发预载功能的手势数据就可以触发预载功能。且若利用所述推理模型识别所述当前手势数据为负样本，不会触发电子设备的预载功能，且将所述当前手势数据作为新的历史手势数据去训练推理模型，能够使得推理模型记忆更多的负样本，即使得不触发电子设备的预载功能的手势数据增加，可以降低正样本的增长带来的不可预期性。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例触发电子设备预载功能的方法的流程示意图；
图2是本申请实施例触发电子设备预载功能的方法的具体实施方式的流程示意图；
图3是本申请实施例电子设备的结构示意框图。
具体实施方式
下面结合附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本申请一部分实施例，而非全部实施例。基于本申请中的实施例，本领域技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。在不冲突的情况下，下述 各个实施例及其技术特征可以相互组合。
请参阅图1，本申请实施例提供一种触发电子设备预载功能的方法，包括：S1、采集用户操作电子设备的当前手势数据；S2、将当前手势数据输入预先使用第一负样本训练的基础模型，第一负样本定义为固定的非触发预载功能的手势数据；S3、若利用基础模型识别当前手势数据为非负样本，则将当前手势数据输入预先使用第二负样本训练的推理模型，第二负样本定义为非触发预载功能的历史手势数据；S4、若利用推理模型识别当前手势数据为负样本，则不触发电子设备的预载功能，且将当前手势数据作为新的历史手势数据训练推理模型；S5、若当前手势数据不为负样本，则识别当前手势为正样本，触发电子设备的预载功能。
在上述步骤中，将固定的非触发预载功能的手势数据作为第一负样本而不是去固定能够触发预载功能的手势数据，因此输入的当前手势数据如不符合第一负样本就可视为可触发预载功能的手势数据，因此不需要固定可触发预载功能的手势数据就可以触发预载功能。且若利用所述推理模型识别所述当前手势数据为负样本，不会触发电子设备的预载功能，且将所述当前手势数据作为新的历史手势数据去训练推理模型，能够使得推理模型记忆更多的负样本，即使得不触发电子设备的预载功能的手势数据增加，可以降低正样本的增长带来的不可预期性。
在本实施例中，可以使用步骤S1、S2、S3、S4及S5完成触发电子设备的预载功能操作，在其他实施例中，还可以将S1获取的当前手势数据直接输入S3的推理模型内，这样就简化了触发电子设备预载功能的流程，并且增加了自由度，由于推理模型能够不断的学习，能够不断更新电子设备对于预载功能的触发条件，能够使得用户使用非预先设置的手势触发电子设备的预载功能，从而增加了用户体验。
另外，为了加强触发电子设备的预载功能的准确性，还可加入群集模 型，在判断是否触发当前手势数据被推理模型判断为正样本的数据是否曾经出现在学习的样本数据中，因此，请参阅图2，触发电子设备预载功能的方法还包括：S6、若利用基础模型识别当前手势数据为非负样本，则将当前手势数据输入采用预先使用正样本训练的群集模型，正样本定义为触发电子设备的预载功能的历史手势数据；S7、若使用群集模型识别非负样本为正样本，且推理模型识别当前手势数据为正样本，则触发电子设备的预载功能。
在其他实施例中，还可以使用triplet loss函数让资料点几种的方式让推理模型对当前传感器数据可以输出是否为非负样本。
在一些可选实施方式中，推理模型的训练方法包括：采集预定时间段内用户操作电子设备的历史手势数据；将历史手势数据输入基础模型；若使用基础模型识别历史手势数据为非负样本，则使用滑动窗口算法对非负样本的历史手势数据进行处理，得到处理结果；将处理结果输入基础模型，若使用基础模型识别处理结果为负样本，则将处理结果对应的历史手势数据标记为负样本；若使用基础模型识别处理结果为非负样本，则将处理结果对应的历史手势数据标记为正样本；将标记为负样本的历史手势数据及其对应的负样本标签存入数据库中，以作为样本数据训练推理模型。
群集模型的训练方法包括：使用标记为正样本的历史手势数据作为样本数据训练群集模型。群集模型可用传统及其学习或者自动编码器或变分自动编码器来进行训练。
对于用于判断用户当前操作是否为触发电子设备预载功能的操作对应的样本数据，会采集预定时间段内用户的历史操作对应的历史手势数据，由于使用者的使用习惯通常会固定，但又希望在使用者不得不用其他习惯使用手机时(如受伤)也能够快速反应,需要获取不同长度从2-15天的历史手势数据分别训练推理模型，在选择上要求验证集正确率够高(如，99％)并 且对最近1天的历史手势数据预测结果够高的前提下去选择整体训练正确率最高的参数，此做法能够使得用户能够在用户改变习惯后也能够触发电子设备的预载功能，另外正样本的增长具有较高的不可预期性，因此需要推理模型在负样本的表现足够高来加以限制正样本改变其表现的程度，从而更加准确地触发电子设备的预载功能。
为了增加样本数据的来源，在训练推理模型前，采集用户未触发电子设备的预载功能而发生波动较大的历史手势数据为训练推理模型的样本数据，从而增加了训练推理模型时的样本数据。
在本实施例中，滑动窗口算法的滑动窗口的设置方法包括：设置不同大小的预滑动窗口，并使用每个不同大小的预滑动窗口运行滑动窗口算法，得到不同预滑动窗口的处理结果；选取处理结果中不影响基础模型的正确率、且最小的预滑动窗口作为滑动窗口。
在其他实施例中，滑动窗口算法的滑动窗口还可以使用深度学习的attention机制或convolution来直接获取重要的信息或高阶信息，使用这些信息构成滑动窗口，从而使得推理模型能够更接近端到端的学习。
由于推理模型及群集模型的训练需要占用处理器等资源，使得训练过程较为耗费电子设备的电量，因此为了兼顾使用者体验，触发电子设备预载功能的方法还包括：根据电子设备的使用情况判断用户是否处于睡眠状态；获取电子设备的电量及充电状态；在用户处于睡眠状态、且电子设备电量充足、处于充电状态下，训练推理模型及群集模型。
通过统计用户使用电子设备的习惯，能够获得用户处于睡眠状态的时间，从而在用户处于睡眠状态时训练推理模型及群集模型；在电子设备电量充足、处于充电状态下训练推理模型及群集模型，能够保证电子设备的电量始终充足，降低电子设备电量不足的几率。
不同的响应规则能够对非负样本集正样本的当前手势数据响应出不同 的预载功能，在本实施例中，预载功能可以为点亮屏幕，从而实现自动亮屏的功能，在其他实施例中，预载功能可以为启动预载程序，还可以为启动指纹解锁，还可以为启动人脸识别解锁。
在本实施例中，操作电子设备的手势包括但不限于用户握持或配戴电子设备触控操作的手势、用户握持或配戴电子装置的姿势或姿态、用户以非接触方式通过手势或姿势控制电子装置。当前手势数据及历史手势数据可以为电子设备的加速度计或加速度传感器的数值代表的手势数据，在其他实施例中，当前手势数据及历史手势数据还可以为陀螺仪或方位传感器的数值代表的手势数据，还可以为重力计或重力传感器的数值代表的手势数据，也可以为超声波传感器或深度传感器的数值代表的手势数据。
而用户的操作，可以由手势数据的差异做判断，例如：重力计或重力传感器判断电子设备指向的改变或是由及其学习的方式判断，如设定用户操作以收集手势数据对推理模型加以训练。
本实施例的电子设备，可以为手机，还可以为平板电脑，还可以为掌上游戏机等手持电子产品。
请参阅图3，本申请实施例还提供一种电子装置，该电子装置包括：存储器601、处理器602及存储在存储器601上并可在处理器602上运行的计算机程序，处理器602执行该计算机程序时，实现前述中描述的触发电子设备预载功能的方法。
进一步的，该电子装置还包括：至少一个输入设备603以及至少一个输出设备604。
上述存储器601、处理器602、输入设备603以及输出设备604，通过总线605连接。
其中，输入设备603具体可为摄像头、触控面板、物理按键或者鼠标等等。输出设备604具体可为显示屏。
存储器601可以是高速随机存取记忆体(RAM，Random Access Memory)存储器，也可为非不稳定的存储器(non-volatile memory)，例如磁盘存储器。存储器601用于存储一组可执行程序代码，处理器602与存储器601耦合。
进一步的，本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质可以是设置于上述各实施例中的电子装置中，该计算机可读存储介质可以是前述中的存储器601。该计算机可读存储介质上存储有计算机程序，该程序被处理器602执行时实现前述实施例中描述的触发电子设备预载功能的方法。
进一步的，该计算机可读存储介质还可以是U盘、移动硬盘、只读存储器601(ROM，Read-Only Memory)、RAM、磁碟或者光盘等各种可以存储程序代码的介质。
尽管已经相对于一个或多个实现方式示出并描述了本申请，但是本领域技术人员基于对本说明书和附图的阅读和理解将会想到等价变型和修改。本申请包括所有这样的修改和变型，并且仅由所附权利要求的范围限制。特别地关于由上述组件执行的各种功能，用于描述这样的组件的术语旨在对应于执行所述组件的指定功能(例如其在功能上是等价的)的任意组件(除非另外指示)，即使在结构上与执行本文所示的本说明书的示范性实现方式中的功能的公开结构不等同。
即，以上所述仅为本申请的实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，例如各实施例之间技术特征的相互结合，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。
另外，在本申请的描述中，需要理解的是，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示 的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者更多个特征。在本申请的描述中，“多个”的含义是两个或两个以上，除非另有明确具体的限定。
为了使本领域任何技术人员能够实现和使用本申请，本申请给出了以上描述。在以上描述中，为了解释的目的而列出了各个细节。应当明白的是，本领域普通技术人员可以认识到，在不使用这些特定细节的情况下也可以实现本申请。在其它实施例中，不会对公知的结构和过程进行详细阐述，以避免不必要的细节使本申请的描述变得晦涩。因此，本申请并非旨在限于所示的实施例，而是与符合本申请所公开的原理和特征的最广范围相一致。

Claims (11)

  1. 一种触发电子设备预载功能的方法，包括：
    采集用户操作电子设备的当前手势数据；
    将当前手势数据输入预先使用第一负样本训练的基础模型，第一负样本定义为固定的非触发预载功能的手势数据；
    若利用所述基础模型识别当前手势数据为非负样本，则将所述当前手势数据输入预先使用第二负样本训练的推理模型，第二负样本定义为非触发预载功能的历史手势数据；
    若利用所述推理模型识别所述当前手势数据为负样本，则不触发电子设备的预载功能，且将所述当前手势数据作为新的历史手势数据训练推理模型；
    若所述当前手势数据不为负样本，则识别所述当前手势为正样本，触发电子设备的预载功能。
  2. 根据权利要求1所述的触发电子设备预载功能的方法，其中，
    所述方法还包括：
    若利用所述基础模型识别当前手势数据为非负样本，则将所述当前手势数据输入采用预先使用正样本训练的群集模型，所述正样本定义为触发电子设备的预载功能的历史手势数据；
    若使用所述群集模型识别所述非负样本为正样本，且所述推理模型识别所述当前手势数据为正样本，则触发电子设备的预载功能。
  3. 根据权利要求2所述的触发电子设备预载功能的方法，其中，
    所述推理模型的训练方法包括：
    采集预定时间段内用户操作电子设备的历史手势数据；
    将所述历史手势数据输入所述基础模型；
    若使用所述基础模型识别所述历史手势数据为非负样本，则使用滑 动窗口算法对非负样本的所述历史手势数据进行处理，得到处理结果；
    将所述处理结果输入所述基础模型，若使用所述基础模型识别所述处理结果为负样本，则将所述处理结果对应的历史手势数据标记为负样本；
    若使用所述基础模型识别所述处理结果为非负样本，则将所述处理结果对应的历史手势数据标记为正样本；
    将标记为负样本的所述历史手势数据及其对应的负样本标签存入数据库中，以作为样本数据训练所述推理模型。
  4. 根据权利要求3所述的触发电子设备预载功能的方法，其中，
    所述群集模型的训练方法包括：
    使用标记为正样本的所述历史手势数据作为样本数据训练群集模型。
  5. 根据权利要求3所述的触发电子设备预载功能的方法，其中，
    所述推理模型的训练方法还包括：
    采集用户未触发电子设备的预载功能而发生波动较大的历史手势数据为训练所述推理模型的样本数据。
  6. 根据权利要求3所述的触发电子设备预载功能的方法，其中，
    所述滑动窗口算法的滑动窗口的设置方法包括：
    设置不同大小的预滑动窗口，并使用每个不同大小的预滑动窗口运行滑动窗口算法，得到不同预滑动窗口的处理结果；
    选取处理结果中不影响所述基础模型的正确率、且最小的预滑动窗口作为滑动窗口。
  7. 根据权利要求2所述的触发电子设备预载功能的方法，其中，
    所述方法还包括：
    根据电子设备的使用情况判断用户是否处于睡眠状态；
    获取电子设备的电量及充电状态；
    在用户处于睡眠状态、且电子设备电量充足、处于充电状态下，训练所述推理模型及所述群集模型。
  8. 根据权利要求1所述的触发电子设备预载功能的方法，其中，
    所述预载功能为点亮屏幕或预载程序或启动指纹解锁或启动人脸识别解锁。
  9. 根据权利要求1所述的触发电子设备预载功能的方法，其中，
    所述当前手势数据及所述历史手势数据为使用电子设备的加速度计或陀螺仪得到的数值。
  10. 一种电子装置，包括：存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时，实现权利要求1至9中的任意一项所述方法。
  11. 一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时，实现权利要求1至9中的任意一项所述方法。
PCT/CN2021/114252 2020-08-25 2021-08-24 触发电子设备预载功能的方法、电子装置及存储介质 WO2022042526A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010866621.3A CN114115511B (zh) 2020-08-25 2020-08-25 触发电子设备预载功能的方法、电子装置及存储介质
CN202010866621.3 2020-08-25

Publications (1)

Publication Number Publication Date
WO2022042526A1 true WO2022042526A1 (zh) 2022-03-03

Family

ID=80352696

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/114252 WO2022042526A1 (zh) 2020-08-25 2021-08-24 触发电子设备预载功能的方法、电子装置及存储介质

Country Status (2)

Country Link
CN (1) CN114115511B (zh)
WO (1) WO2022042526A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117423445A (zh) * 2023-12-11 2024-01-19 广东壹健康健康产业集团股份有限公司 基于用户集群感知的智能指环控制方法及装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180137219A1 (en) * 2016-11-14 2018-05-17 General Electric Company Feature selection and feature synthesis methods for predictive modeling in a twinned physical system
CN108595013A (zh) * 2018-05-15 2018-09-28 Oppo广东移动通信有限公司 握持识别方法、装置、存储介质及电子设备
CN110163236A (zh) * 2018-10-15 2019-08-23 腾讯科技（深圳）有限公司 模型的训练方法和装置、存储介质、电子装置
US20190294927A1 (en) * 2018-06-16 2019-09-26 Moshe Guttmann Selective update of inference models
CN110765967A (zh) * 2019-10-30 2020-02-07 腾讯科技（深圳）有限公司 一种基于人工智能的动作识别方法和相关装置
CN111027442A (zh) * 2019-12-03 2020-04-17 腾讯科技（深圳）有限公司 用于行人重识别的模型训练方法、识别方法、装置及介质
CN111191590A (zh) * 2019-12-30 2020-05-22 Oppo（重庆）智能科技有限公司 模型训练方法、装置、存储介质及电子设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180137219A1 (en) * 2016-11-14 2018-05-17 General Electric Company Feature selection and feature synthesis methods for predictive modeling in a twinned physical system
CN108595013A (zh) * 2018-05-15 2018-09-28 Oppo广东移动通信有限公司 握持识别方法、装置、存储介质及电子设备
US20190294927A1 (en) * 2018-06-16 2019-09-26 Moshe Guttmann Selective update of inference models
CN110163236A (zh) * 2018-10-15 2019-08-23 腾讯科技（深圳）有限公司 模型的训练方法和装置、存储介质、电子装置
CN110765967A (zh) * 2019-10-30 2020-02-07 腾讯科技（深圳）有限公司 一种基于人工智能的动作识别方法和相关装置
CN111027442A (zh) * 2019-12-03 2020-04-17 腾讯科技（深圳）有限公司 用于行人重识别的模型训练方法、识别方法、装置及介质
CN111191590A (zh) * 2019-12-30 2020-05-22 Oppo（重庆）智能科技有限公司 模型训练方法、装置、存储介质及电子设备

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117423445A (zh) * 2023-12-11 2024-01-19 广东壹健康健康产业集团股份有限公司 基于用户集群感知的智能指环控制方法及装置
CN117423445B (zh) * 2023-12-11 2024-03-15 广东壹健康健康产业集团股份有限公司 基于用户集群感知的智能指环控制方法及装置

Also Published As

Publication number Publication date
CN114115511B (zh) 2023-05-02
CN114115511A (zh) 2022-03-01

Similar Documents

Publication Publication Date Title
JP6761837B2 (ja) ユーザの心理状態判断装置及び方法
US10219222B2 (en) Adjusting mobile device state based on user intentions and/or identity
US8712931B1 (en) Adaptive input interface
CN107729092B (zh) 一种用于电子书的自动翻页方法和系统
EP3702953B1 (en) Electronic device for obfuscating and decoding data and method for controlling same
CN103745235A (zh) 人脸识别方法、装置及终端设备
WO2017005207A1 (zh) 一种输入方法、输入装置、服务器和输入系统
CN108121736A (zh) 一种主题词确定模型的建立方法、装置及电子设备
CN110163086B (zh) 基于神经网络的健身动作识别方法、装置、设备及介质
JP6197613B2 (ja) 生体情報抽出装置、生体情報抽出方法、および生体情報抽出プログラム
CN107850950A (zh) 基于时间的分词
WO2022042526A1 (zh) 触发电子设备预载功能的方法、电子装置及存储介质
CN113190646A (zh) 一种用户名样本的标注方法、装置、电子设备及存储介质
CN107132927A (zh) 输入字符的识别方法及装置和用于识别输入字符的装置
CN107135494A (zh) 垃圾短信识别方法及装置
US11468153B2 (en) Terminal device management method and terminal device
CN107479703A (zh) 自动语音报时方法、设备及计算机可读存储介质
CN108108356A (zh) 一种文字翻译方法、装置及设备
AU2015252057B2 (en) Adjusting mobile device state based on user intentions and/or identity
CN113222075B (zh) 一种评价方法、装置和电子设备
CN109388252A (zh) 一种输入方法及装置
LU101615B1 (en) Systems and methods for handwriting recognition
CN113467692B (zh) 信息拦截方法、装置、设备、介质及程序产品
US20230063055A1 (en) Method for fingerprint enrollment, electronic device and computer-readable storage medium
KR20230081584A (ko) 자동 및 멀티 모달리티 정보 입력을 활용하는 일상 추적 시스템을 위한 인공지능 기반의 상황 추론 장치 및 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21860361

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21860361

Country of ref document: EP

Kind code of ref document: A1