CN105574471B - The recognition methods of the method for uploading, user behavior of user behavior data and device - Google Patents
The recognition methods of the method for uploading, user behavior of user behavior data and device Download PDFInfo
- Publication number
- CN105574471B CN105574471B CN201410532912.3A CN201410532912A CN105574471B CN 105574471 B CN105574471 B CN 105574471B CN 201410532912 A CN201410532912 A CN 201410532912A CN 105574471 B CN105574471 B CN 105574471B
- Authority
- CN
- China
- Prior art keywords
- user
- behavior
- type
- dynamic frequency
- vector
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of method for uploading of user behavior data, the recognition methods of user behavior and devices, include: the acceleration information that acquisition terminal obtains user in the period of scheduled duration, the behavior type of the user in the period is determined and recorded according to the acceleration information；According to predetermined period, count the frequency of occurrences of the user's various actions type recorded in the predetermined period, and dynamic frequency vector is generated, each component of the dynamic frequency vector is the various actions type after sorting according to predetermined order, and the value of each component is the frequency of occurrences of corresponding various actions type；The dynamic frequency vector is uploaded to platform.The requirement of transmittability present invention reduces the requirement of the processing capacity to hardware and to acquisition terminal is finally realized and is identified on the basis of not increasing cost to the behavior of user.
Description
Technical field
The present invention relates to data service technical fields, method for uploading, user more particularly, to a kind of user behavior data
The recognition methods of behavior and device.
Background technique
Detailed User Activity identification can be formed activity log, be recorded the main work of user with comprehensive monitoring User Activity
It is dynamic, and according to the activity analysis of user as a result, carrying out exercise guidance and suggestion to user, can improve comprehensively and use on this basis
The physical condition at family.
Usually identify that user behavior, existing Activity recognition algorithm are to be based on adding by acceleration initial data in the prior art
Speed initial data carries out the identification of feature extraction and classifying device.It directlys adopt these algorithms and realizes strategy there are two types of behavioural analyses:
(1) enhance terminal transmission ability, initial data or feature are uploaded into back-end platform and calculated；(2) enhance terminal computing capability,
Terminal realizes Activity recognition；And in practical application, both strategies are all difficult to realize.The volume of transmitted data that wherein (1) needs is big,
It will increase system energy consumption and hardware cost；(2) stronger hardware processing capability is needed, will increase hardware cost.It is both uncomfortable
Close low cost, the low-power consumption hardware platform that current wearable device is widely used.
Summary of the invention
The object of the present invention is to provide a kind of method for uploading of user behavior data, the recognition methods of user behavior and dresses
It sets, overcomes the enhancing terminal transmission ability in behavioural analysis, it is this that initial data or feature are uploaded to back-end platform calculating
Defect existing for strategy, while enhancing terminal computing capability is overcome, it realizes in terminal and is lacked existing for this strategy of Activity recognition
It falls into.
To achieve the goals above, the present invention provides a kind of method for uploading of user behavior data, comprising: acquisition terminal
The acceleration information for obtaining user in the period of scheduled duration, being determined and recorded according to the acceleration information should in the period
The behavior type of user；According to predetermined period, the appearance frequency of the user's various actions type recorded in the predetermined period is counted
Rate, and dynamic frequency vector is generated, each component of the dynamic frequency vector is the various actions type after sorting according to predetermined order, and
The value of each component is the frequency of occurrences of corresponding various actions type；The dynamic frequency vector is uploaded to platform.
Optionally, the acceleration information includes 3-axis acceleration numerical value, described determining according to the acceleration information and remember
The behavior type for recording the user period Nei include: calculate the 3-axis acceleration numerical value square and value；Described in determination
The affiliated default value section with value, and by the corresponding behavior type in the default value section, as the use in the period
The behavior type at family, wherein different behavior types corresponds to different default value sections.
Optionally, the range of the scheduled duration is 1~60S.
According to another aspect of the invention, a kind of recognition methods of user behavior is provided, comprising: receive acquisition terminal
The dynamic frequency vector of periodicity sending, each component of the dynamic frequency vector are the various actions kinds after sorting according to predetermined order
Class, the value of each component are the frequency of occurrences of the corresponding various actions type in predetermined period, and the behavior type is according to pre-
In the timing long period determined by the acceleration information of user；The behavior of the user is identified based on the dynamic frequency vector.
Optionally, it is described the behavior of the user is identified based on the dynamic frequency vector after, the method also includes: by this
The behavior of user is sent to the display terminal of the user.
Optionally, the behavior that the user is identified based on the dynamic frequency vector include: to the dynamic frequency vector with it is various
The feature vector of behavior type is compared, by the behavior type with the dynamic the most similar feature vector of frequency vector, as
Behavior type of the user in the predetermined period.
Optionally, described that the dynamic frequency vector is compared with the feature vector of various actions type, it will be moved with described
The behavior type of the most similar feature vector of frequency vector as the user the predetermined period behavior type, comprising: meter
Calculate the maximum comparability of dynamic the frequency vector and described eigenvector；By the behavior type of the corresponding feature vector of maximum comparability
As the user the predetermined period behavior type.
Optionally, it before being compared to the dynamic frequency vector with the feature vector of various actions type, further presses
According to default weight, each component of the dynamic frequency vector is weighted, wherein there is each component preset importance to refer to
Mark, the higher component of importance index weight with higher.
According to another aspect of the invention, a kind of acquisition terminal is provided, comprising: module is obtained, it is predetermined for obtaining
The acceleration information of user in the period of duration, the row of the user in the period is determined and recorded according to the acceleration information
For type；Processing module, for according to predetermined period, counting the appearance frequency of the user's various actions type recorded in the period
Rate, and dynamic frequency vector is generated, each component of the dynamic frequency vector is the various actions type after sorting according to predetermined order, and
The value of each component is the frequency of occurrences of corresponding various actions type；First transmission module, for uploading the dynamic frequency vector
To platform.
Optionally, the acquisition module includes the first computing unit and the first comparing unit, and first computing unit is used
In calculate the 3-axis acceleration numerical value square and value；First comparing unit is pre- belonging to described and value for determining
If numerical intervals, and by the corresponding behavior type in the default value section, as the behavior type of the user in the period,
Wherein, different behavior types corresponds to different default value sections.
Optionally, the scheduled duration in the period for obtaining module acquisition scheduled duration in the acceleration information of user
Range be 1~60S.
According to another aspect of the invention, a kind of identifying system of user behavior is provided, comprising: receiving module is used
In the dynamic frequency vector for receiving acquisition terminal periodicity sending, each component of the dynamic frequency vector is after sorting according to predetermined order
Various actions type, the value of each component is the frequency of occurrences of the corresponding various actions type in predetermined period, the behavior
Type is in the period according to scheduled duration determined by the acceleration information of user；Identification module, for based on described dynamic
Frequency vector identifies the behavior of the user.
Optionally, the identifying system further includes the second transmission module, and second transmission module is used for the user's
Behavior is sent to the display terminal of the user.
Optionally, the identification module include the second comparing unit, second comparing unit be used for the dynamic frequency to
Amount is compared with the feature vector of various actions type, by the behavior kind with the dynamic the most similar feature vector of frequency vector
Class, as the user the predetermined period behavior type.
Optionally, second comparing unit is used to carry out the feature vector of the dynamic frequency vector and various actions type
Compare, by the behavior type with the dynamic the most similar feature vector of frequency vector, as the user in the predetermined period
Behavior type includes: the maximum comparability for calculating dynamic the frequency vector and described eigenvector；By the corresponding spy of maximum comparability
Levy vector behavior type as the user the predetermined period behavior type.
Optionally, the identification module further includes the second computing unit, and second computing unit is used for, to described dynamic
Before frequency vector is compared with the feature vector of various actions type, further according to default weight, to the dynamic frequency vector
Each component be weighted, wherein each component has preset importance index, and the higher component of importance index has
Higher weight.
The beneficial effects of the present invention are: the present invention is by carrying out processing shape to collected acceleration information in acquisition terminal
At dynamic frequency vector, dynamic frequency vector is then uploaded to platform, platform identifies the behavior of the user based on dynamic frequency vector, and the present invention is logical
The combination for crossing acquisition terminal and platform avoids and carries out complicated operation in acquisition terminal, reduces the processing capacity to hardware
Requirement, to reduce hardware cost, in addition, reducing the biography to acquisition terminal by the way that dynamic frequency vector is uploaded to platform
The requirement of Movement Capabilities, to reduce the energy consumption of acquisition terminal, finally realize on the basis of not increasing cost, to
The behavior at family is identified.
Detailed description of the invention
The step process that Fig. 1 shows acquisition terminals in the embodiment of the present invention in the method for uploading of user behavior data
Figure；
Fig. 2 indicates step flow chart of the platform in the recognition methods of user behavior in the embodiment of the present invention；
Fig. 3 indicates that platform in the embodiment of the present invention identifies the step flow chart of the behavior of the user based on dynamic frequency vector；
Fig. 4 indicates the acceleration of the user in the embodiment of the present invention in an analysis result shown in RAD database
Data；
Fig. 5 indicates the recognition result obtained according to the acceleration information in Fig. 4；
Fig. 6 indicates the structural block diagram of acquisition terminal in the embodiment of the present invention；And
Fig. 7 indicates the structural block diagram of the identifying system in the embodiment of the present invention.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
As shown in Figure 1, for step of the acquisition terminal in the method for uploading of user behavior data in the embodiment of the present invention
Flow chart includes the following steps:
Step S101 obtains the acceleration information of user in the period of scheduled duration, is determined simultaneously according to acceleration information
Record the behavior type of the user period Nei.
In the present embodiment, acquisition terminal obtains the acceleration information of user in the period of scheduled duration, it is preferred that
Acquisition terminal, to reduce EMS memory occupation, the scheduled duration may range from 1~60S.
Preferably, acceleration information can be the numerical value in three axial directions of 3-axis acceleration, and acquisition terminal can be by right
The numerical value of three axial directions of 3-axis acceleration is handled to obtain the behavior type of the user.Specific processing can be, according to three
A axial acceleration value carries out preliminary treatment, obtains a processing result；Then, it is determined that the corresponding behavior of the processing result
Type, and the behavior type as the user in the period.
For example, preliminary treatment can be, calculate the axial numerical value of 3-axis acceleration three square and value.Acquisition is eventually
End first calculate the axial numerical value of 3-axis acceleration three square and value, it is assumed that scheduled duration 2S, then each 2S when
Between calculate in section a 3-axis acceleration three axial numerical value square and value.Above-mentioned preliminary treatment may also is that calculating three
Then three square roots axially obtained are carried out read group total again, obtain one by the square root of axle acceleration three axial numerical value
A and value；Either calculate the axial numerical value of 3-axis acceleration three square and after value, then the variance yields for calculating and being worth；Or
Be calculate the axial numerical value of 3-axis acceleration three absolute value and value, etc..That is, the processing of the numerical value to 3-axis acceleration,
It can voluntarily be selected according to specific environment and statistics experience, it is not limited to above citing.
If the mode for carrying out preliminary treatment to the acceleration value of three axial directions is to calculate 3-axis acceleration three axial numbers
Value square and value can determine default value section belonging to described and value then after calculating described and value, and will described in
The corresponding behavior type in default value section, the behavior type as the user in the period, wherein preset different
Behavior type corresponds to different default value sections.For example, being previously provided with different behavior type institutes in the acquisition terminal
Corresponding numerical intervals, be calculated 3-axis acceleration numerical value square and value after, by this and be worth and different numerical value areas
Between compare, determine should and value which numerical intervals be located at, it is assumed that should and value be located at the first numerical intervals, then the first numerical value area
Between corresponding behavior type be the affiliated behavior type of this and value.
If the mode for carrying out preliminary treatment to the acceleration value of three axial directions is to calculate 3-axis acceleration three axial numbers
Value square and value, then to the variance yields that this and value progress variance are calculated, then can should in the embodiment of the present invention
Variance yields is compared from different numerical intervals, determines which numerical intervals the variance yields is located at, it is assumed that the variance yields is located at
Second value section, then behavior type corresponding to second value section is behavior type belonging to the variance yields.Variance yields
Calculation amount it is smaller, and be more able to reflect out the changing rule of user's acceleration, the embodiment of the present invention can be excellent in acquisition terminal side
Selection of land determines the behavior type of user using variance yields.
Step S102 counts the appearance of the user's various actions type recorded in the predetermined period according to predetermined period
Frequency, and generate dynamic frequency vector.
In the present embodiment, each component of the dynamic frequency vector is the various actions kind after sorting according to predetermined order
Class, the value of each component are the frequencies of occurrences of corresponding behavior type.It is illustrated herein with specific embodiment, if various actions kind
Class is respectively static, rides, rides, 5 kinds of behavior types of activity and exercise, then each component of the dynamic frequency vector is according to predetermined
Sequence is respectively static after sorting, ride, ride, activity and exercise, and the value of each component of the dynamic frequency vector is respectively quiet
Only, ride, ride, activity and take exercise 5 kinds of behavior types the frequency of occurrences, it is assumed that it is static, ride, ride, activity and exercise 5 kinds
The frequency of occurrences of behavior type is respectively D1, D2, D3, D4, D5, i.e., dynamic frequency vector D=(D1, D2, D3, D4, D5).
Optionally, predetermined period can be set as unit of hour, can also be set as unit of day, i.e., in advance
If the period is not limited to the specific time and limits, it is assumed that predetermined period 5Min, then acquisition terminal counts record in the 5Min
User's various actions type the frequency of occurrences, finally by the frequency of occurrences of user's various actions type formed dynamic frequency to
Amount.
The dynamic frequency vector is uploaded to platform by step S103, it should be noted that acquisition terminal can be that can wear
Wear formula equipment, such as pedometer.
As shown in Fig. 2, for step flow chart of the platform in the recognition methods of user behavior in the embodiment of the present invention, packet
Include following steps:
Step S201 receives the dynamic frequency vector of acquisition terminal periodicity sending, it should be noted that received dynamic frequency vector
It is the same dynamic frequency vector with the dynamic frequency vector generated in step S102.
Step S202 identifies the behavior of the user based on the dynamic frequency vector, after the information for obtaining the dynamic frequency vector, leads to
It crosses and the Activity recognition result that processing finally obtains the user is carried out to the dynamic frequency vector.
Preferably, after the behavior of identification user, the behavior of the user can also be sent to the display terminal of the user,
Wherein, display terminal can be computer, mobile phone etc..
As shown in figure 3, including the following steps: for the flow chart of step S202 in Fig. 2 further refined
Step S301 is weighted each component of the dynamic frequency vector.
In the present embodiment, the weight of each component of dynamic frequency vector is preset in platform, by each of the dynamic frequency vector
A component is weighted according to the weight of each component respectively；Preferably, each component has preset importance index, important
Then weight with higher, the i.e. different corresponding weight of behavior type of user are different the higher component of property index.Herein
It is illustrated with specific embodiment, it is assumed that various actions type is respectively static, rides, rides, 5 kinds of behavior kinds of activity and exercise
Class, due to appearance that is static, riding, ride, can all have static behavior in 5 kinds of behavior types of activity and exercise, static row
Minimum is contributed for the identification to user behavior, then for the importance index of static behavior with regard to minimum, weight is then minimum；On the contrary, taking exercise
Behavior generally only occurs in exercise behavior type, static, ride, ride, seldom will appear exercise in the behaviors type such as activity
Behavior, therefore temper behavior and the identification contribution of user behavior is the largest, then the importance index of behavior is tempered with regard to highest, power
It is heavy then maximum；When it is static, ride, ride, activity and after temper the weight of 5 kinds of behavior types and determine, then according to the weight to dynamic
Each component of frequency vector is weighted respectively.
Step S302 is compared the dynamic frequency vector with the feature vector of various actions type.
In the present embodiment, the feature vector of various actions type is preset in platform.It preferably, include point in platform
Class device stores the feature vector of various actions type in classifier, if described eigenvector did not carry out weighting, to institute
Frequency vector is stated to be compared with feature vector；If described eigenvector is the feature vector after weighting, to dynamic after weighting
Frequency vector is compared with the feature vector after weighting.
Step S303 calculates the maximum comparability of dynamic the frequency vector and described eigenvector, and maximum comparability is corresponding
Feature vector behavior type as the user predetermined period behavior type.
In the present embodiment, it is preferred that employed in the similitude for calculating the dynamic frequency vector and described eigenvector
Calculation method is the method for COS distance measured similarity, and the COS distance is defined as
Wherein, X is the dynamic frequency vector, x1For the one-component in the dynamic frequency vector, x2For in the dynamic frequency vector
Second component, and so on, xnFor n-th of vector in the dynamic frequency vector, Y is the predetermined various actions
The feature vector of type, y1For the one-component of described eigenvector, y2For second component of described eigenvector, according to
It is secondary to analogize, ynFor n-th of component of described eigenvector, θ be dynamic frequency vector X and the feature vector Y of various actions type it
Between angle.
When calculating the maximum comparability of the dynamic frequency vector and described eigenvector, by the dynamic frequency vector and various rows
It is compared for the feature vector of type, that is, calculates separately angle between the dynamic frequency vector and the feature vector of various actions type
Cosine value, cosine value is bigger, then illustrates that similarity is bigger, after obtaining maximum similarity, by the corresponding feature of maximum comparability
For the behavior type of vector as the user in the behavior type of predetermined period, i.e. behavior type is last recognition result.
Citing is illustrated herein, it is assumed that various actions type is respectively static, rides, rides, 5 kinds of rows of activity and exercise
For type, the feature vector of static behavior is Y1, the feature vector of behavior is Y by bus2, the feature vector of driving behavior is Y3, living
The feature vector of dynamic behavior is Y4, the feature vector for tempering behavior is Y5, move frequency vector be D, then calculate separately dynamic frequency vector D with
The feature vector Y of static behavior1, by bus behavior feature vector Y2, driving behavior feature vector Y3, crawler behavior feature
Vector Y4, take exercise behavior feature vector Y5Between similarity, that is, calculate separately the feature of dynamic frequency vector D and static behavior to
Measure Y1, by bus behavior feature vector Y2, driving behavior feature vector Y3, crawler behavior feature vector Y4, temper behavior
Feature vector Y5Between angle cosine value, cosine value is bigger, then illustrates that similarity is bigger, takes the corresponding spy of maximum similarity
The behavior type for levying vector is behavior type, i.e. behavior type of the user in predetermined period belonging to the dynamic frequency vector,
As last Activity recognition result.
It should be noted that above-mentioned recognition result may be comprising individual accidental mistakes, it can be according to the experience of life
Recognition result is post-processed.For example, being mingled with 5 minutes strenuous exercise in multiple daily routines, usually judge by accident, it can
With removal.5 minutes to happen suddenly in long-time quiescing process are likely to be erroneous judgement by bus, can be removed.
Illustrate the beneficial effect of the recognition methods below with reference to example.
As shown in figure 4, Fig. 4 is the acceleration information of the user in an analysis result shown in RAD database, the number
Data are labeled according to can be used as prompt auxiliary user.It is about 5 minutes 2 hours when by can be seen that data acquisition in Fig. 4.
Actual conditions are 10 minutes sort articles and walk to parking lot (daily routines), drive within 35 minutes, play basketball within 25 minutes, 5 points
Clock walking drives, walks in family from parking lot within 5 minutes for 15 minutes, then static always.
Fig. 5 indicates that the recognition result obtained according to the acceleration information in Fig. 4, Fig. 5 divide user's various actions type
For it is static, ride, ride, activity and temper 5 kinds of behavior types, by Fig. 5 it can be seen that the result remember with user in conjunction with and
The annotation results of step count information are almost consistent, and primary mistake occurs at only 1.5 hours, i.e. last time activity.This section
Activity includes parking, walking in 5 minutes, takes elevator and sit down at home, due to 5 points to walk in family from parking lot
Clock activity is complicated, finally is identified as riding by the synthesis of mistake.
This mistake will be understood by, and due to including that multiple state is converted in 5 minutes, the excessive point of state is (i.e. from one
A state is switched to the period of another state) it will cause acquisition terminal identification generation mistake, and then platform can also generate mistake.
Test result on RAD data set shows that recognition accuracy reaches 94.21%.
As shown in fig. 6, for the structural block diagram of acquisition terminal in the embodiment of the present invention, acquisition terminal 400 includes:
Obtain module 401, the acceleration information of user in the period for obtaining scheduled duration, according to the acceleration degree
According to the behavior type for determining and recording the user in the period；
Processing module 402, for according to predetermined period, counting going out for the user's various actions type recorded in the period
Existing frequency, and dynamic frequency vector is generated, each component of the dynamic frequency vector is the various actions kind after sorting according to predetermined order
Class, and the value of each component is the frequency of occurrences of corresponding various actions type；
First transmission module 403, for the dynamic frequency vector to be uploaded to platform.
Optionally, obtaining module 401 includes the first computing unit and the first comparing unit, and first computing unit is used for
Calculate the 3-axis acceleration numerical value square and value；First comparing unit is default belonging to described and value for determining
Numerical intervals, and by the corresponding behavior type in the default value section, as the behavior type of the user in the period,
In, different behavior types corresponds to different default value sections.
Optionally, the scheduled duration in the period of the acquisition scheduled duration of module 401 in the acceleration information of user is obtained
Range be 1~60S.
As shown in fig. 7, for the structural block diagram of the identifying system of user behavior in the embodiment of the present invention, identifying system 500
Include:
Receiving module 501, for receiving the dynamic frequency vector of acquisition terminal periodicity sending, each point of the dynamic frequency vector
Amount is the various actions type after sorting according to predetermined order, and the value of each component is corresponding various actions type in predetermined period
The interior frequency of occurrences, the behavior type are in the period according to scheduled duration determined by the acceleration information of user；
Identification module 502, for identifying the behavior of the user based on the dynamic frequency vector.
Optionally, identifying system 500 further includes the second transmission module, and second transmission module is used for the row of the user
For the display terminal for being sent to the user.
Optionally, identification module 502 includes the second comparing unit, and second comparing unit is used for the dynamic frequency vector
It is compared with the feature vector of various actions type, by the behavior kind with the dynamic the most similar feature vector of frequency vector
Class, as the user the predetermined period behavior type.
Optionally, second comparing unit is used to carry out the feature vector of the dynamic frequency vector and various actions type
Compare, by the behavior type with the dynamic the most similar feature vector of frequency vector, as the user in the predetermined period
Behavior type includes: the maximum comparability for calculating dynamic the frequency vector and described eigenvector；By the corresponding spy of maximum comparability
Levy vector behavior type as the user the predetermined period behavior type.
Optionally, identification module 502 further includes the second computing unit, and second computing unit is used for, to described dynamic
Before frequency vector is compared with the feature vector of various actions type, further according to default weight, to the dynamic frequency vector
Each component be weighted, wherein each component has preset importance index, and the higher component of importance index has
Higher weight.
Above-described is the preferred embodiment of the present invention, it should be pointed out that the ordinary person of the art is come
It says, can also make several improvements and retouch under the premise of not departing from principle of the present invention, these improvements and modifications also exist
In protection scope of the present invention.
Claims (16)
1. a kind of method for uploading of user behavior data characterized by comprising
Acquisition terminal obtains the acceleration information of user in the period of scheduled duration, is determined and recorded according to the acceleration information
The behavior type of the user in the period；
According to predetermined period, the frequency of occurrences of the user's various actions type recorded in the predetermined period is counted, and is generated dynamic
Frequency vector, each component of the dynamic frequency vector are the various actions types after sorting according to predetermined order, and each component
Value is the frequency of occurrences of corresponding various actions type；
The dynamic frequency vector is uploaded to platform.
2. method for uploading as described in claim 1, which is characterized in that
The acceleration information includes 3-axis acceleration numerical value, described to be determined and recorded in the period according to the acceleration information
The behavior type of the user include: calculate the 3-axis acceleration numerical value square and value；It determines pre- belonging to described and value
If numerical intervals, and by the corresponding behavior type in the default value section, as the behavior type of the user in the period,
Wherein, different behavior types corresponds to different default value sections.
3. method for uploading as described in claim 1, which is characterized in that the range of the scheduled duration is 1~60S.
4. a kind of recognition methods of user behavior characterized by comprising
The dynamic frequency vector of acquisition terminal periodicity sending is received, each component of the dynamic frequency vector is sorted according to predetermined order
Various actions type afterwards, the value of each component are the frequency of occurrences of the corresponding various actions type in predetermined period, the row
It is in the period according to scheduled duration determined by the acceleration information of user for type；
The behavior of the user is identified based on the dynamic frequency vector.
5. recognition methods as claimed in claim 4, which is characterized in that identify the user's based on the dynamic frequency vector described
After behavior, the method also includes:
The behavior of the user is sent to the display terminal of the user.
6. recognition methods as claimed in claim 4, which is characterized in that the row for identifying the user based on the dynamic frequency vector
To include:
The dynamic frequency vector is compared with the feature vector of various actions type, it will be the most similar with the dynamic frequency vector
The behavior type of feature vector, as the user the predetermined period behavior type.
7. recognition methods as claimed in claim 6, which is characterized in that
It is described that the dynamic frequency vector is compared with the feature vector of various actions type, it will be with dynamic frequency vector phase the most
As feature vector behavior type as the user the predetermined period behavior type, comprising:
Calculate the maximum comparability of dynamic the frequency vector and described eigenvector；
Using the behavior type of the corresponding feature vector of maximum comparability as the user the predetermined period behavior type.
8. recognition methods as claimed in claim 7, which is characterized in that
Before being compared to the dynamic frequency vector with the feature vector of various actions type, further according to default weight,
Each component of the dynamic frequency vector is weighted, wherein each component has preset importance index, importance index
Higher component weight with higher.
9. a kind of acquisition terminal characterized by comprising
Module is obtained, the acceleration information of user in the period for obtaining scheduled duration is determined according to the acceleration information
And record the behavior type of the user period Nei；
Processing module, for according to predetermined period, counting the frequency of occurrences of the user's various actions type recorded in the period,
And dynamic frequency vector is generated, each component of the dynamic frequency vector is the various actions type after sorting according to predetermined order, and each
The value of a component is the frequency of occurrences of corresponding various actions type；
First transmission module, for the dynamic frequency vector to be uploaded to platform.
10. acquisition terminal as claimed in claim 9, which is characterized in that the acceleration information includes 3-axis acceleration numerical value,
The acquisition module includes the first computing unit and the first comparing unit；
First computing unit be used for calculate the 3-axis acceleration numerical value square and value；
First comparing unit is for determining default value section belonging to described and value, and by the default value section pair
The behavior type answered, the behavior type as the user in the period, wherein different behavior types corresponds to different pre-
If numerical intervals.
11. acquisition terminal as claimed in claim 9, which is characterized in that the period for obtaining module and obtaining scheduled duration
The range of scheduled duration in the acceleration information of interior user is 1~60S.
12. a kind of identifying system of user behavior characterized by comprising
Receiving module, for receiving the dynamic frequency vector of acquisition terminal periodicity sending, each component of the dynamic frequency vector be by
According to the various actions type after predetermined order sequence, the value of each component is corresponding various actions type going out in predetermined period
Existing frequency, the behavior type are in the period according to scheduled duration determined by the acceleration information of user；
Identification module, for identifying the behavior of the user based on the dynamic frequency vector.
13. identifying system as claimed in claim 12, which is characterized in that it further include the second transmission module, second transmission
Module is used to for the behavior of the user being sent to the display terminal of the user.
14. identifying system as claimed in claim 12, which is characterized in that the identification module includes the second comparing unit, institute
The second comparing unit is stated to be used to be compared the dynamic frequency vector with the feature vector of various actions type, it will be with the dynamic frequency
The behavior type of the most similar feature vector of vector, as the user the predetermined period behavior type.
15. identifying system as claimed in claim 14, which is characterized in that second comparing unit be used for the dynamic frequency to
Amount is compared with the feature vector of various actions type, by the behavior kind with the dynamic the most similar feature vector of frequency vector
Class in the behavior type of the predetermined period includes: to calculate the dynamic frequency vector and described eigenvector most as the user
Big similitude；Using the behavior type of the corresponding feature vector of maximum comparability as the user the predetermined period behavior kind
Class.
16. identifying system as claimed in claim 15, which is characterized in that the identification module further includes the second computing unit,
Second computing unit is used for, before being compared to the dynamic frequency vector with the feature vector of various actions type, into
One step is weighted each component of the dynamic frequency vector according to default weight, wherein each component has preset important
Property index, the higher component of importance index weight with higher.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410532912.3A CN105574471B (en) | 2014-10-10 | 2014-10-10 | The recognition methods of the method for uploading, user behavior of user behavior data and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410532912.3A CN105574471B (en) | 2014-10-10 | 2014-10-10 | The recognition methods of the method for uploading, user behavior of user behavior data and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105574471A CN105574471A (en) | 2016-05-11 |
| CN105574471B true CN105574471B (en) | 2019-02-05 |
Family
ID=55884581
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410532912.3A Active CN105574471B (en) | 2014-10-10 | 2014-10-10 | The recognition methods of the method for uploading, user behavior of user behavior data and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105574471B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102627452B1 (en) * | 2017-04-14 | 2024-01-18 | 매직 립, 인코포레이티드 | Multi-mode eye tracking |
| CN107463689A (en) * | 2017-08-10 | 2017-12-12 | 深圳市金立通信设备有限公司 | Generation method, moving state identification method and the terminal in motion characteristic data storehouse |
| CN109272432B (en) * | 2018-08-08 | 2020-11-13 | 广州杰赛科技股份有限公司 | User behavior monitoring method and system, computer device and computer storage medium |
| CN112115978A (en) * | 2020-08-24 | 2020-12-22 | 中国人民健康保险股份有限公司 | Motion recognition method and device and storage medium |
| CN114168948A (en) * | 2021-12-17 | 2022-03-11 | 北京华清信安科技有限公司 | Network security situation comprehensive analysis method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102087722A (en) * | 2009-12-04 | 2011-06-08 | 索尼公司 | Learning apparatus and method, prediction apparatus and method, and program |
| CA2758710A1 (en) * | 2011-11-07 | 2013-05-07 | Martin Bonac | Cadence detection |
| CN103886341A (en) * | 2014-03-19 | 2014-06-25 | 国家电网公司 | Gait behavior recognition method based on feature combination |
| CN103908259A (en) * | 2014-04-03 | 2014-07-09 | 深圳市德凯瑞科技有限公司 | Intelligent wearable device and human motion monitoring and recognition method |
-
2014
- 2014-10-10 CN CN201410532912.3A patent/CN105574471B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102087722A (en) * | 2009-12-04 | 2011-06-08 | 索尼公司 | Learning apparatus and method, prediction apparatus and method, and program |
| CA2758710A1 (en) * | 2011-11-07 | 2013-05-07 | Martin Bonac | Cadence detection |
| CN103886341A (en) * | 2014-03-19 | 2014-06-25 | 国家电网公司 | Gait behavior recognition method based on feature combination |
| CN103908259A (en) * | 2014-04-03 | 2014-07-09 | 深圳市德凯瑞科技有限公司 | Intelligent wearable device and human motion monitoring and recognition method |
Non-Patent Citations (2)
| Title |
|---|
| A Motion Recognition Method by Constancy-Decision;Kazuya Murao;《International Symposium on Wearable Computers (ISWC) 2010》;20101213;第1-4页 |
| 一种基于三维加速度传感器的人体行为识别方法;徐川龙 等;《计算机系统应用》;20130630;第22卷(第6期);第132-135页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105574471A (en) | 2016-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574471B (en) | The recognition methods of the method for uploading, user behavior of user behavior data and device | |
| EP3058443B1 (en) | Calculating pace and energy expenditure from athletic movement attributes | |
| KR101690649B1 (en) | Activity classification in a multi-axis activity monitor device | |
| US10098549B2 (en) | Local model for calorimetry | |
| EP3073402B1 (en) | System and method to recognize activities performed by an individual | |
| RU2658282C2 (en) | Extending gameplay with physical activity monitoring device | |
| CN206026334U (en) | Motion amount detection device and intelligent wearable equipment comprising same | |
| CN106709235A (en) | Exercise training data processing method and device | |
| CN105358046A (en) | Automatic exercise segmentation and recognition | |
| CN103654759A (en) | Exercise information detecting apparatus and exercise information detecting method | |
| WO2010083562A1 (en) | Activity detection | |
| Das et al. | Strength training: A fitness application for indoor based exercise recognition and comfort analysis | |
| CN116561589B (en) | Attendance training management method and system based on intelligent wearable equipment | |
| CN104586402A (en) | Feature extracting method for body activities | |
| Sen et al. | Did you take a break today? detecting playing foosball using your smartwatch | |
| CN111938580A (en) | Method and device for testing oxygen uptake in daily exercise of human body and computer equipment | |
| CN107995181A (en) | A kind of auth method based on gait, device, equipment and storage medium | |
| CN109692462A (en) | A kind of physical training validity assessment method and system | |
| CN204950959U (en) | Wearable human motion gesture trajectory record and test and appraisal device | |
| Song | User Activity Tracker using Android Sensor | |
| SEN et al. | Did you take a break today? Detecting playing foosball using your smartwatch.(2016) |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |