CN108686363A - The method and apparatus of action evaluation - Google Patents
The method and apparatus of action evaluation Download PDFInfo
- Publication number
- CN108686363A CN108686363A CN201710232620.1A CN201710232620A CN108686363A CN 108686363 A CN108686363 A CN 108686363A CN 201710232620 A CN201710232620 A CN 201710232620A CN 108686363 A CN108686363 A CN 108686363A
- Authority
- CN
- China
- Prior art keywords
- evaluation
- feature
- status switch
- record
- action
- 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.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B71/00—Games or sports accessories not covered in groups A63B1/00 - A63B69/00
- A63B71/06—Indicating or scoring devices for games or players, or for other sports activities
- A63B71/0619—Displays, user interfaces and indicating devices, specially adapted for sport equipment, e.g. display mounted on treadmills
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2220/00—Measuring of physical parameters relating to sporting activity
- A63B2220/80—Special sensors, transducers or devices therefor
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2220/00—Measuring of physical parameters relating to sporting activity
- A63B2220/80—Special sensors, transducers or devices therefor
- A63B2220/83—Special sensors, transducers or devices therefor characterised by the position of the sensor
- A63B2220/836—Sensors arranged on the body of the user
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Physical Education & Sports Medicine (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to action evaluation method and apparatus.This method includes:The extraction feature from the record acquired by sensor, record include multiple data points of an action;Determine multiple elemental motions that extracted feature includes;The evaluation status switch for being evaluated multiple elemental motions that the feature extracted includes is determined using pre-stored evaluation model;And it is compared to identified evaluation status switch and pre-stored standard evaluation status switch to obtain the evaluation of estimate to record.Accurately each action can be evaluated with device according to the method for the present invention, to judge that some action of user does whether specification, and then know how to improve.Further, it is also possible to solve the data different problems that sensor acquisition comes caused by the direction difference held bat.
Description
Technical field
The present invention relates to sensor gathered data process fields, relate more specifically to a kind of method and dress of action evaluation
It sets.
Background technology
With the development of sensor, more and more sensors are widely applied to every field.In recent years, wearable
Sensor is greatly paid close attention in terms of training.Higher sexual valence can be provided using wearable sensors synkinesia training
The solution of ratio.
In training, especially in asf motor crowd, it is desirable to the sports level for improving oneself, need please to train into
Row targetedly instructs.High expense allows many people not afford to do.More and more universal wearable sensor solves this
The problem of sample.Wearable sensors generally include:Accelerometer, gyroscope and magnetometer.Accelerometer is can to perceive arbitrary side
Upward acceleration, accelerometer is obtained in some axial stressing conditions by measurement assembly as a result, the form of expression is axial
Acceleration magnitude and direction (XYZ).Gyroscope is by measuring in three-dimensional system of coordinate between the vertical axis and equipment of gyrorotor
Angle, and calculate angular speed, by angle and angular speed come differentiate object three dimensions motion state.Magnetometer is also named
Earth magnetism, magnetic strength device can be used for testing magnetic field intensity and direction, and the orientation of positioning device, the principle of magnetometer is with compass principle
It is similar, current device and the angle on the four direction of all directions can be measured.
In practical applications, since application, error correction, error compensation need, the sensor is often used in combination,
The speciality for making full use of each sensor makes final operation result more acurrate, for example is come simultaneously using magnetometer and accelerometer
Bearing meter is calculated, the azimuth information needs calculated can just be obtained in combination with magnetic direction and direction motion conditions.
By such wearable device be mounted on sports equipment on, such as on racket, shoes it is first-class.It can easily calculate
Go out the number of player motion, such as now common pedometer, can be worn on hand, be through shoe lining or be placed in pocket.Meter step
Device collects the data of sensor transmission and can calculate the number of user's walking by certain algorithm.Mounted on racket
On sensor can also accomplish to calculate the number of user movement at present, for example the number swung the bat, equipment also can calculate
User does the number of some specific action, such as the number served a ball, the number killed etc..
Presently, there are problem be, how by the data of sensor come judge user some action do whether specification,
How to improve, this is the most important demand of training.Each action can accurately only be evaluated, could all or
The effect of part substitution coach.Therefore how to evaluate the quality of action is a problem for needing to solve.
In addition, when needing sensor being mounted on portable equipment, such as racket, tennis racket etc..By
The data different problems that sensor acquisition comes caused by the direction difference held bat can not solve at present.
Therefore, it is necessary to a kind of action evaluation method and apparatus that can be solved the above problems.
Invention content
The brief overview about the present invention is given below, in order to provide the basic reason about certain aspects of the invention
Solution.It should be appreciated that this general introduction is not the exhaustive general introduction about the present invention.It is not intended to determine the key of the present invention
Or pith, nor is it intended to limit the scope of the present invention.Its purpose only provides certain concepts in simplified form, with
This is as the preamble in greater detail discussed later.
A primary object of the present invention is, provides a kind of action evaluation method, including:It is acquired from by sensor
Record in extraction feature, record comprising one action multiple data points;Determine multiple bases that extracted feature includes
This action;It is determined using pre-stored evaluation model for commenting multiple elemental motions that the feature extracted includes
The evaluation status switch of valence;And identified evaluation status switch and pre-stored standard evaluation status switch are compared
Relatively obtain the evaluation of estimate to record.
According to an aspect of the present invention, a kind of action evaluation device is provided, including:Feature extraction unit is configured as
The extraction feature from the record acquired by sensor, record include multiple data points of an action;Elemental motion determines single
Member is configured to determine that multiple elemental motions that extracted feature includes;Status switch determination unit is evaluated, is configured as
It is determined using pre-stored evaluation model and is commented for what is evaluated multiple elemental motions that the feature extracted includes
Valence status switch;And evaluation unit, it is configured as identified evaluation status switch and pre-stored standard evaluating shape
State sequence is compared to obtain the evaluation of estimate to record.
In addition, the embodiments of the present invention also provide the computer programs for realizing the above method.
In addition, the embodiments of the present invention also provide the computer program product of at least computer-readable medium form,
Upper record is useful for realizing the computer program code of the above method.
By the detailed description below in conjunction with attached drawing to highly preferred embodiment of the present invention, these and other of the invention is excellent
Point will be apparent from.
Description of the drawings
Below with reference to the accompanying drawings illustrate embodiments of the invention, the above of the present invention and its can be more readily understood that
Its objects, features and advantages.Component in attached drawing is intended merely to show the principle of the present invention.In the accompanying drawings, identical or similar
Technical characteristic or component will be indicated using same or similar reference numeral.
Fig. 1 shows the simplified flowchart of action evaluation method according to the present invention;
Fig. 2 shows the flow charts of the example process of action evaluation method 200 according to an embodiment of the invention;
Fig. 3 is a kind of flow chart for the example process for showing the step S202 in Fig. 2;
Fig. 4 shows the rotation to pending segment;
Fig. 5 shows that training obtains a kind of flow chart of example process 500 of evaluation model;
Fig. 6 is the block diagram for the exemplary configuration for showing action evaluation device 600 according to another embodiment of the invention;
Fig. 7 is a kind of block diagram for the exemplary configuration for showing the feature extraction unit 602 in Fig. 6;And
Fig. 8 is the example arrangement for showing to can be used for implementing the computing device of the action evaluation method and apparatus of the present invention
Figure.
Specific implementation mode
The exemplary embodiment of the present invention is described hereinafter in connection with attached drawing.For clarity and conciseness,
All features of actual implementation mode are not described in the description.It should be understood, however, that developing any this actual implementation
Much decisions specific to embodiment must be made during example, to realize the objectives of developer, for example, symbol
Restrictive condition those of related to system and business is closed, and these restrictive conditions may have with the difference of embodiment
Changed.In addition, it will also be appreciated that although development is likely to be extremely complex and time-consuming, to having benefited from the disclosure
For those skilled in the art of content, this development is only routine task.
Herein, it is also necessary to which explanation is a bit, in order to avoid having obscured the present invention because of unnecessary details, in the accompanying drawings
The device structure closely related with scheme according to the present invention and/or processing step are illustrate only, and is omitted and the present invention
The little other details of relationship.
The present invention proposes a kind of action evaluation method, can by calculate the relative angle of each data slot by its
It rotates to normal coordinate system and carrys out extraction feature, it is determining included on the basis of the feature extracted observed to move substantially
Make, calculates the optimal evaluation status switch evaluated elemental motion, and calculate and compile with known standard evaluation status switch
Distance is collected to evaluate action.
Action evaluation method and apparatus according to an embodiment of the invention are described in detail below in conjunction with the accompanying drawings.Retouching hereinafter
It states and carries out in the following order:
1. action evaluation method
2. action evaluation device
3. to implement the computing device of methods and apparatus of the present invention
[1. action evaluation Fang Fa ]
Fig. 1 shows the simplified flowchart of action evaluation method according to the present invention.From figure 1 it will be seen that according to this hair
Bright action evaluation method uses sensor gathered data first, then carries out feature extraction, then carries out the identification of elemental motion
And evaluation status switch matching, it can finally be scored action according to evaluation status switch.
Fig. 2 shows the flow charts of the example process of action evaluation method 200 according to an embodiment of the invention.
The process of action evaluation method 200 is illustrated below in conjunction with Fig. 2.
First, in step S202, the extraction feature from the record acquired by sensor a, record is moved comprising one
The multiple data points made.
By wearable sensors acquire record include some action (such as:Shoulder rotation, lift elbow, is swung arm chest expanding)
Several data points.Due to usually used multiple sensors (accelerometer, gyroscope, magnetometer), and each sensor is usual
There are multiple dimensions again.Data point each in this way would generally include multiple dimensions.Such as most common accelerometer, gyro
Three instrument, magnetometer sensors, there are three axis for each sensor, therefore are respectively provided with x, y and z three-dimensional for each sensor
Reading.
Typically, the completion each acted is required for the experience regular hour.In data point each in this way there are one meetings
Time stab.For from this angle, the data each acted can regard time series as.It can use so traditional
Feature Extraction of Time Series method, such as Fast Fourier Transform (FFT), wavelet transformation etc..
Fig. 3 is to show that one kind of the step S202 (that is, the extraction feature from the record acquired by sensor) in Fig. 2 is shown
The flow chart of example property process.
As shown in figure 3, in step S2022, the data point that record includes is divided into multiple segments.
Specifically, each record is divided into the segment (alternatively referred to as window) of several equal lengths, for example, segment (window
Mouthful) size be N mean each segment include N number of data point.
In collected each dimension data, per directional correlation of the one-dimensional data all with sensor itself.Sensor itself
Direction difference causes collected data also to differ, and the feature extracted also differs.For example, similarly lift elbow action,
In the case where sensor orientation is opposite, collected data are then that direction is opposite.It, can quilt if without any processing
It is classified as different actions.
Therefore, in step S2024, following processing is executed for each segment of segmentation:
The direction of the data point of each sensor is calculated first.
In one example, using following equation 1-3, to calculate N*3, (segment includes N number of data point, a data
Point has 3 dimensions) direction of data point in matrix:
Direction=avgbegin-avgendFormula 3
Wherein, Di is the data on time point i.In formula 1, avgbeginBe in time series before 25% data
Mean value, in formula 2, avgendIt is that the mean values of rear 25% data in time series uses avg in formula 3beginSubtract avgend
To calculate the direction (direction) of the N*3 matrixes.
Direction calculated above is calculated in segment, alternatively it is also possible to calculate direction between segment.Therefore exist
Two class directions:The direction between direction and segment in segment.Direction in the computational methods and segment in the direction between segment
Computational methods it is similar, this will not be repeated here.
It is then determined being perpendicularly to the direction (direction) and by the normal vector of the data point Di in the segment
(normal)。
Then, all data points in the N*3 matrixes are rotated so that the direction and biography of direction vector and normal vector
The direction of the respective shaft of sensor is consistent, i.e. the direction of direction vector and normal vector is respectively parallel to (0,0,1) and (1,0,0).
With reference to Fig. 4, the curve in left side show pending segment and the segment direction vector (direction) and
Normal vector (normal), the right sides Fig. 4 show that carrying out rotation to direction vector sum normal vector makes itself and the respective shaft of sensor
Direction is consistent.
It is then possible to which the data of all rotations to be connected together as to the feature of the segment sequentially in time.
Finally, in step S2026, spy that the feature of all segments is connected together as extracting from the record
Sign.
By step S202, feature has been extracted from record.
Then, in step S204, multiple elemental motions that extracted feature includes are determined.
In the present invention, elemental motion is defined as minimum unit or the son action of action, by combining elemental motion
It may be constructed all compound actions.
Preferably, if all features extracted are polymerized to Ganlei, such as K classes by the method using cluster.Herein, it clusters
Purpose be just intended to preferably obtain elemental motion.
For example, slam-shot (smash) action can have several elemental motions, such as:Shoulder rotation, stretching hand with arm,
Lift elbow is swung arm.
Furthermore it is possible to which all records are divided into several action classifications, such as M by the action label according to each record
Class.The action label of record can be obtained when acquiring training data, for example, can acquire the different action of several groups (swing the bat,
Kill), acquisition each time is as a record.
After obtaining elemental motion set by the method for cluster, it can be calculated by calculating the method for mutual information every
Mutual information between a elemental motion and action classification, and using the mutual information as the weight of each elemental motion.Substantially dynamic
The weight of work indicates significance level of the elemental motion for current action classification.
In one example, the mutual information between elemental motion and action classification can be calculated by following formula.
MI (B, C)=log (P (B|C)/P(B))
Wherein, B indicates that elemental motion, C indicate that action classification, MI (B, C) indicate between elemental motion and action classification
Mutual information.
Next, in step S206, determined using pre-stored evaluation model for being wrapped in the feature to being extracted
The evaluation status switch that the multiple elemental motions included are evaluated.
In the present invention, evaluation status switch be defined as include the sequence of elemental motion or including elemental motion and its
The sequence of evaluation of estimate.
Evaluation status switch including elemental motion for example can be " swivel, is swung arm lift elbow ", this evaluation status switch
The performance level of elemental motion is not considered, and only considers whether the elemental motion occurred.Including elemental motion and its evaluation of estimate
Evaluation status switch for example can be " swivel |5, lift Zhou |3, Hui Bei |5".In the evaluation status switch including evaluation of estimate,
"|" after number " 5 ", " 3 " be evaluation of estimate, evaluation of estimate is to represent the fine or not degree of elemental motion completion.
Here pre-stored evaluation model can be obtained by training.
Fig. 5 shows that training obtains a kind of flow chart of example process 500 of evaluation model.
First, in step S502, training data is acquired by sensor to construct training set.
Then, in step S504, following processing is executed for each of training set record:
The extraction feature from each record, a record include multiple data points of an action;Determine extracted spy
Multiple elemental motions in sign;Evaluation status switch is constituted based on multiple elemental motions or multiple elemental motions are carried out respectively
It scores and the scoring based on multiple elemental motions and to each elemental motion constitutes evaluation status switch.
After being carried out above-mentioned processing to each record, in step S506, it can obtain and be extracted from training set
All features set F:F1, f2 ..., fi } and all set Y for evaluating status switch:{y1,y2,…,yj}.
During evaluation of training model, from extraction feature in each record and determine multiple in extracted feature
Processing identical with the processing described above with respect to step S202 and S204 may be used to carry out in the processing of elemental motion.
For a record, Utilization assessment model would generally obtain a variety of possible evaluation status switch Y1,
Y2 ..., Ym, wherein Y1, Y2, Ym are the subset of Y respectively.
It needs in these possible evaluation status switches, finds optimal evaluation status switch, i.e., in the spy extracted
Evaluation status switch Y* with maximum probability in the case that sign is f.
In one example, optimal status switch Y* can be calculate by the following formula:
Y*=arg max p (Y|f).
The Y* obtained herein is used to carry out record final evaluation.
Finally, in step S208, identified evaluation status switch and pre-stored standard are evaluated into status switch
It is compared to obtain the evaluation of estimate to record.
It specifically, can be by calculating Y* and pre-stored best after obtaining optimal evaluation status switch Y*
Or most standard evaluation status switch (Yp) between editing distance determine the evaluation of estimate to the record.
In the present invention, the weight of editing distance can pre-define or be concentrated through training in training and obtain.
The present invention proposes a kind of action evaluation method, can by calculate the relative angle of each data slot by its
It rotates to normal coordinate system and carrys out extraction feature, it is determining included on the basis of the feature extracted observed to move substantially
Make, calculates the optimal evaluation status switch evaluated elemental motion, and calculate and compile with known standard evaluation status switch
Distance is collected to evaluate action.
Action evaluation method according to the present invention can accurately evaluate each action, to judge user certain
It is a act do whether specification, and then know how to improve.Further, it is also possible to caused by solving the direction difference due to holding bat
The data different problems that sensor acquisition comes.
2. action evaluation device
Fig. 6 is the block diagram for the exemplary configuration for showing action evaluation device 600 according to another embodiment of the invention.
As shown in fig. 6, action evaluation device 600 includes feature extraction unit 602, elemental motion determination unit 604, evaluation
Status switch determination unit 606 and evaluation unit 608.
Wherein, feature extraction unit 602 is configured as the extraction feature from the record acquired by sensor, and record includes
Multiple data points of one action.
Elemental motion determination unit 604 is configured to determine that multiple elemental motions that extracted feature includes.
Evaluation status switch determination unit 606 is configured as that pre-stored evaluation model is utilized to determine for being extracted
The evaluation status switch evaluated of feature multiple elemental motions for including.
Evaluation unit 608 is configured as identified evaluation status switch and pre-stored standard evaluating status switch
It is compared to obtain the evaluation of estimate to record.
Fig. 7 is a kind of block diagram for the exemplary configuration for showing the feature extraction unit 602 in Fig. 6.
As shown in fig. 7, feature extraction unit 602 includes fragment segmentation subelement 6022,6024 and of segment processing subelement
Feature extraction subelement 6026.
Wherein, fragment segmentation subelement 6022 is configured as the data point that record includes being divided into multiple segments.
Segment processing subelement 6024 is configured as executing following processing for each segment:Calculate the data in the segment
The direction of point;It is determined perpendicular to direction and by the method phase vector of a data point in the segment;By the institute in the segment
There is data point to be rotated so that direction and the direction of normal vector are consistent with the direction of the respective shaft of sensor;And will own
The data connection of rotation is together as the feature of the segment.
Feature extraction subelement 6026 is configured as being connected together as extracting from record by the feature of all segments
Feature.
Wherein, segment processing subelement 6024 is configured to:It is flat with the data point of preceding first predetermined percentage
Mean value subtracts the average value of the data point of rear second predetermined percentage to calculate direction.
Wherein, elemental motion determination unit 604 is configured to:It is determined by clustering method in extracted feature
Multiple elemental motions.
Wherein, elemental motion determination unit 604 is configured to:Determine the belonging action classification of record;And
The weight of each elemental motion is determined based on action classification.
Wherein, elemental motion determination unit 604 is configured to:By calculating between elemental motion and action classification
Mutual information determine the weight of elemental motion.
Wherein, the element for evaluating status switch includes the elemental motion in extracted feature.
Wherein, the element for evaluating status switch includes elemental motion in extracted feature and comments the elemental motion
Point.
Wherein, evaluation status switch determination unit 606 is configured to:Utilization assessment model is determined for recording institute
Possible evaluation status switch;Determine the evaluation status switch with maximum probability in the case of the feature extracted.
Wherein, pre-stored evaluation model trains to obtain by following procedure:
Training data is acquired by sensor to construct training set;
Following processing is executed for each of training set record:The extraction feature from record, record include an action
Multiple data points;Determine multiple elemental motions in extracted feature;Evaluation state sequence is constituted based on multiple elemental motions
Row carry out multiple elemental motions scoring and the scoring composition based on multiple elemental motions and to each elemental motion respectively
Evaluate status switch;And
Obtain the set of all features extracted from training set and the set of all evaluation status switches.
The operation of various pieces about action evaluation device 600 and the details of function are referred to that Fig. 1-5 is combined to describe
The present invention action evaluation method embodiment, be not detailed herein.
It should be noted that Fig. 6 and the structure of action evaluation device 600 shown in Fig. 7 and its component units are only
It is exemplary, those skilled in the art can as needed modify to Fig. 6 and structure diagram shown in Fig. 7.
The present invention proposes a kind of action evaluation method and apparatus, can pass through the relative angle of each data slot of calculating
Degree rotates it to normal coordinate system and carrys out extraction feature, and included base is determined on the basis of the feature extracted observed
This action calculates the optimal evaluation status switch evaluated elemental motion, and evaluates status switch meter with known standard
Editing distance is calculated to evaluate action.
Action evaluation method and apparatus according to the present invention can accurately evaluate each action, to judge to use
Some action at family do whether specification, and then know how to improve.Further, it is also possible to solve since the direction held bat is different and
The data different problems that caused sensor acquisition comes.
[3. to implement the Ji Suanshebei ] of the present processes and device;
The basic principle that the present invention is described above in association with specific embodiment, however, it is desirable to, it is noted that this field
For those of ordinary skill, it is to be understood that the whole either any steps or component of methods and apparatus of the present invention, Ke Yi
Any computing device (including processor, storage medium etc.) either in the network of computing device with hardware, firmware, software or
Combination thereof is realized that this is that those of ordinary skill in the art use them in the case where having read the explanation of the present invention
Basic programming skill can be achieved with.
Therefore, the purpose of the present invention can also by run on any computing device a program or batch processing come
It realizes.The computing device can be well known fexible unit.Therefore, the purpose of the present invention can also include only by offer
The program product of the program code of the method or device is realized to realize.That is, such program product is also constituted
The present invention, and the storage medium for being stored with such program product also constitutes the present invention.Obviously, the storage medium can be
Any well known storage medium or any storage medium that developed in the future.
In the case where realizing the embodiment of the present invention by software and/or firmware, from storage medium or network to
The computer of specialized hardware structure, such as the installation of all-purpose computer shown in Fig. 8 800 constitute the program of the software, the computer
When being equipped with various programs, it is able to carry out various functions etc..
In fig. 8, central processing unit (CPU) 801 is according to the program stored in read-only memory (ROM) 802 or from depositing
The program that storage part 808 is loaded into random access memory (RAM) 803 executes various processing.In RAM 803, also according to need
Store the data required when CPU 801 executes various processing etc..CPU 801, ROM 802 and RAM 803 are via bus
804 links each other.Input/output interface 805 also link to bus 804.
Components described below link is to input/output interface 805:Importation 806 (including keyboard, mouse etc.), output section
Divide 807 (including display, such as cathode-ray tube (CRT), liquid crystal display (LCD) etc. and loud speakers etc.), storage section
808 (including hard disks etc.), communications portion 809 (including network interface card such as LAN card, modem etc.).Communications portion 809
Communication process is executed via network such as internet.As needed, driver 810 also can link to input/output interface 805.
Detachable media 811 such as disk, CD, magneto-optic disk, semiconductor memory etc. is installed in driver 810 as needed
On so that the computer program read out is mounted to as needed in storage section 808.
It is such as removable from network such as internet or storage medium in the case of series of processes above-mentioned by software realization
Unload the program that the installation of medium 811 constitutes software.
It will be understood by those of skill in the art that this storage medium be not limited to it is shown in Fig. 8 wherein have program stored therein,
Separately distribute with equipment to provide a user the detachable media 811 of program.The example of detachable media 811 includes disk
(including floppy disk (registered trademark)), CD (comprising compact disc read-only memory (CD-ROM) and digital versatile disc (DVD)), magneto-optic disk
(including mini-disk (MD) (registered trademark)) and semiconductor memory.Alternatively, storage medium can be ROM 802, storage section
Hard disk for including in 808 etc., wherein computer program stored, and user is distributed to together with the equipment comprising them.
The present invention also proposes a kind of program product for the instruction code being stored with machine-readable.Instruction code is read by machine
When taking and executing, can perform it is above-mentioned according to the method for the embodiment of the present invention.
Correspondingly, the storage medium of the program product for carrying the above-mentioned instruction code for being stored with machine-readable also wraps
It includes in disclosure of the invention.Storage medium includes but not limited to floppy disk, CD, magneto-optic disk, storage card, memory stick etc..
It should be appreciated by those skilled in the art that being exemplary what this was enumerated, the present invention is not limited thereto.
In the present specification, the statements such as " first ", " second " and " n-th " be in order to by described feature in word
On distinguish, the present invention is explicitly described.Therefore, it should not serve to that there is any limited meaning.
As an example, each step of the above method and all modules and/or unit of above equipment can
To be embodied as software, firmware, hardware or combinations thereof, and as the part in relevant device.Each composition mould in above-mentioned apparatus
Block, unit when being configured by way of software, firmware, hardware or combinations thereof workable specific means or mode be ability
Known to field technique personnel, details are not described herein.
As an example, by software or firmware realization, can from storage medium or network to
Computer (such as all-purpose computer 800 shown in Fig. 8) installation of specialized hardware structure constitutes the program of the software, the computer
When being equipped with various programs, it is able to carry out various functions etc..
In the feature above in the description of the specific embodiment of the invention, describing and/or showing for a kind of embodiment
It can be used in one or more other embodiments in a manner of same or similar, with the feature in other embodiment
It is combined, or substitute the feature in other embodiment.
It should be emphasized that term "comprises/comprising" refers to the presence of feature, element, step or component when being used herein, but simultaneously
It is not excluded for the presence or additional of other one or more features, element, step or component.
In addition, the method for the present invention be not limited to specifications described in time sequencing execute, can also according to it
His time sequencing, concurrently or independently execute.Therefore, the execution sequence of method described in this specification is not to this hair
Bright technical scope is construed as limiting.
The present invention and its advantage it should be appreciated that in the essence without departing from the present invention being defined by the claims appended hereto
Various changes, replacement and transformation can be carried out in the case of god and range.Moreover, the scope of the present invention is not limited only to specification institute
The process of description, the specific embodiment of equipment, means, method and steps.One of ordinary skilled in the art is from the present invention's
Disclosure it will be readily understood that can be used according to the present invention execute the function essentially identical to corresponding embodiment in this or
Obtain the result essentially identical with it, existing and to be developed in the future process, equipment, means, method or step.Cause
This, the attached claims are intended to include such process, equipment, means, method or step in the range of them.
Based on above explanation, it is known that open at least to disclose following technical scheme:
Note 1, a kind of action evaluation method, including:
The extraction feature from the record acquired by sensor, the record include multiple data points of an action;
Determine multiple elemental motions that extracted feature includes;
Determine that multiple elemental motions for including to the feature extracted carry out using pre-stored evaluation model
The evaluation status switch of evaluation;And
Identified evaluation status switch and pre-stored standard evaluation status switch are compared to obtain to institute
State the evaluation of estimate of record.
Note 2, the method according to note 1, wherein extraction feature includes from the record acquired by sensor:
The data point for including that records is divided into multiple segments;
Following processing is executed for each segment:
Calculate the direction of the data point in the segment;
It is determined perpendicular to the direction and by the method phase vector of a data point in the segment;
All data points in the segment are rotated so that the direction and the direction of the normal vector and the biography
The direction of the respective shaft of sensor is consistent;And
By the data connection of all rotations together as the feature of the segment;And
The feature that the feature of all segments is connected together as extracting from the record.
Note 3, the method according to note 2, wherein the direction for calculating the data point in the segment includes:With preceding
The average value of the data point of one predetermined percentage subtracts the average value of the data point of rear second predetermined percentage to calculate the side
To.
Note 4, the method according to note 1, wherein determine that multiple elemental motions in extracted feature include:
Multiple elemental motions in extracted feature are determined by clustering method.
Note 5, method described in note 4, wherein determine that multiple elemental motions in extracted feature are further
Including:
Determine the action classification belonging to the record;And
The weight of each elemental motion is determined based on the action classification.
Note 6, the method according to note 5, wherein the weight of each elemental motion is determined based on the action classification
Including:
The weight of elemental motion is determined by calculating the mutual information between the elemental motion and the action classification.
Note 7, the method according to note 1, wherein the element of the evaluation status switch includes extracted feature
In elemental motion.
Note 8, the method according to note 1, wherein the element of the evaluation status switch includes extracted feature
In elemental motion and scoring to the elemental motion.
Note 9, the method according to note 1, wherein determined for being extracted using pre-stored evaluation model
The evaluation status switch evaluated of feature multiple elemental motions for including include:
Using evaluation model determination all possible evaluation status switch is recorded for described;
Determine the evaluation status switch with maximum probability in the case of the feature extracted.
Note 10, the method according to note 1, wherein the pre-stored evaluation model is instructed by following steps
It gets:
Training data is acquired by sensor to construct training set;
Following processing is executed for each of described training set record:
The extraction feature from record, the record include multiple data points of an action;
Determine multiple elemental motions in extracted feature;
Evaluation status switch is constituted based on the multiple elemental motion or the multiple elemental motion is commented respectively
Divide and the scoring based on multiple elemental motions and to each elemental motion constitutes evaluation status switch;And
Obtain the set of all features extracted from the training set and the set of all evaluation status switches.
Note 11, a kind of action evaluation device, including:
Feature extraction unit, is configured as the extraction feature from the record acquired by sensor, and the record includes one
Multiple data points of a action;
Elemental motion determination unit is configured to determine that multiple elemental motions that extracted feature includes;
Status switch determination unit is evaluated, is configured as utilize pre-stored evaluation model determination for being extracted
The evaluation status switch that multiple elemental motions that feature includes are evaluated;And
Evaluation unit, be configured as by identified evaluation status switch and pre-stored standard evaluate status switch into
Row relatively obtains the evaluation of estimate to the record.
Note 12, the device according to note 11, wherein the feature extraction unit includes:
Fragment segmentation subelement is configured as the data point for including that records being divided into multiple segments;
Segment handles subelement, is configured as executing following processing for each segment:
Calculate the direction of the data point in the segment;
It is determined perpendicular to the direction and by the method phase vector of a data point in the segment;
All data points in the segment are rotated so that the direction and the direction of the normal vector and the biography
The direction of the respective shaft of sensor is consistent;And
By the data connection of all rotations together as the feature of the segment;And
Feature extraction subelement is configured as being connected together as extracting from the record by the feature of all segments
Feature.
Note 13, the device according to note 12, wherein the segment processing subelement is configured to:With
The average value of the data point of preceding first predetermined percentage subtracts the average value of the data point of rear second predetermined percentage to calculate
State direction.
Note 14, the device according to note 11, wherein the elemental motion determination unit is configured to:
Multiple elemental motions in extracted feature are determined by clustering method.
Note 15, the device according to note 14, wherein the elemental motion determination unit is configured to:
Determine the action classification belonging to the record;And
The weight of each elemental motion is determined based on the action classification.
Note 16, the device according to note 15, wherein the elemental motion determination unit is configured to:
The weight of elemental motion is determined by calculating the mutual information between the elemental motion and the action classification.
Note 17, the device according to note 11, wherein the element of the evaluation status switch includes extracted spy
Elemental motion in sign.
Note 18, the device according to note 11, wherein the element of the evaluation status switch includes extracted spy
Elemental motion in sign and the scoring to the elemental motion.
Note 19, the device according to note 11, wherein the evaluation status switch determination unit is further configured
For:
Using evaluation model determination all possible evaluation status switch is recorded for described;
Determine the evaluation status switch with maximum probability in the case of the feature extracted.
Note 20, the device according to note 11, wherein the pre-stored evaluation model is instructed by following procedure
It gets:
Training data is acquired by sensor to construct training set;
Following processing is executed for each of described training set record:
The extraction feature from record, the record include multiple data points of an action;
Determine multiple elemental motions in extracted feature;
Evaluation status switch is constituted based on the multiple elemental motion or the multiple elemental motion is commented respectively
Divide and the scoring based on multiple elemental motions and to each elemental motion constitutes evaluation status switch;And
Obtain the set of all features extracted from the training set and the set of all evaluation status switches.
Claims (10)
1. a kind of action evaluation method, including:
The extraction feature from the record acquired by sensor, the record include multiple data points of an action;
Determine multiple elemental motions that extracted feature includes;
It is determined using pre-stored evaluation model for evaluating multiple elemental motions that the feature extracted includes
Evaluation status switch;And
Identified evaluation status switch and pre-stored standard evaluation status switch are compared to obtain to the note
The evaluation of estimate of record.
2. according to the method described in claim 1, wherein, extraction feature includes from the record acquired by sensor:
The data point for including that records is divided into multiple segments;
Following processing is executed for each segment:
Calculate the direction of the data point in the segment;
It is determined perpendicular to the direction and by the method phase vector of a data point in the segment;
All data points in the segment are rotated so that the direction and the direction of the normal vector and the sensor
Respective shaft direction it is consistent;And
By the data connection of all rotations together as the feature of the segment;And
The feature that the feature of all segments is connected together as extracting from the record.
3. according to the method described in claim 1, wherein it is determined that multiple elemental motions in the feature extracted include:
Multiple elemental motions in extracted feature are determined by clustering method.
4. according to the method described in claim 3, wherein it is determined that multiple elemental motions in the feature extracted are further wrapped
It includes:
Determine the action classification belonging to the record;And
The weight of each elemental motion is determined based on the action classification.
5. according to the method described in claim 4, wherein, the weight packet of each elemental motion is determined based on the action classification
It includes:
The weight of elemental motion is determined by calculating the mutual information between the elemental motion and the action classification.
6. according to the method described in claim 1, wherein, the element for evaluating status switch includes in extracted feature
Elemental motion.
7. according to the method described in claim 1, wherein, the element for evaluating status switch includes in extracted feature
Elemental motion and scoring to the elemental motion.
8. according to the method described in claim 1, wherein, being determined for the spy to being extracted using pre-stored evaluation model
The evaluation status switch that multiple elemental motions that sign includes are evaluated includes:
Using evaluation model determination all possible evaluation status switch is recorded for described;
Determine the evaluation status switch with maximum probability in the case of the feature extracted.
9. according to the method described in claim 1, wherein, the pre-stored evaluation model is trained by following steps
It arrives:
Training data is acquired by sensor to construct training set;
Following processing is executed for each of described training set record:
The extraction feature from record, the record include multiple data points of an action;
Determine multiple elemental motions in extracted feature;
Evaluation status switch is constituted based on the multiple elemental motion or is scored respectively simultaneously the multiple elemental motion
Scoring based on multiple elemental motions and to each elemental motion constitutes evaluation status switch;And
Obtain the set of all features extracted from the training set and the set of all evaluation status switches.
10. a kind of action evaluation device, including:
Feature extraction unit, is configured as the extraction feature from the record acquired by sensor, and the record is dynamic comprising one
The multiple data points made;
Elemental motion determination unit is configured to determine that multiple elemental motions that extracted feature includes;
Status switch determination unit is evaluated, is configured as that pre-stored evaluation model is utilized to determine for the feature to being extracted
The evaluation status switch that the multiple elemental motions for including are evaluated;And
Evaluation unit is configured as comparing identified evaluation status switch and pre-stored standard evaluation status switch
Relatively obtain the evaluation of estimate to the record.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710232620.1A CN108686363A (en) | 2017-04-11 | 2017-04-11 | The method and apparatus of action evaluation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710232620.1A CN108686363A (en) | 2017-04-11 | 2017-04-11 | The method and apparatus of action evaluation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108686363A true CN108686363A (en) | 2018-10-23 |
Family
ID=63843533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710232620.1A Pending CN108686363A (en) | 2017-04-11 | 2017-04-11 | The method and apparatus of action evaluation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108686363A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110471529A (en) * | 2019-08-07 | 2019-11-19 | 北京卡路里信息技术有限公司 | Act methods of marking and device |
| CN112597835A (en) * | 2020-12-11 | 2021-04-02 | 国汽(北京)智能网联汽车研究院有限公司 | Driving behavior evaluation method and device, electronic equipment and readable storage medium |
| CN113657173A (en) * | 2021-07-20 | 2021-11-16 | 北京搜狗科技发展有限公司 | Data processing method and device and data processing device |
| CN113657173B (en) * | 2021-07-20 | 2024-05-24 | 北京搜狗科技发展有限公司 | Data processing method and device for data processing |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102317978A (en) * | 2009-12-22 | 2012-01-11 | 松下电器产业株式会社 | Action analysis device and action analysis method |
| CN105749537A (en) * | 2016-02-02 | 2016-07-13 | 四川长虹电器股份有限公司 | Exercise auxiliary marking system |
| JP2016209431A (en) * | 2015-05-13 | 2016-12-15 | セイコーエプソン株式会社 | Swing analyzer, swing analysis method, swing analysis program and swing analysis system |
| US9526946B1 (en) * | 2008-08-29 | 2016-12-27 | Gary Zets | Enhanced system and method for vibrotactile guided therapy |
| CN106267734A (en) * | 2016-08-30 | 2017-01-04 | 石家庄铁路职业技术学院 | A kind of digitized body-building system |
-
2017
- 2017-04-11 CN CN201710232620.1A patent/CN108686363A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9526946B1 (en) * | 2008-08-29 | 2016-12-27 | Gary Zets | Enhanced system and method for vibrotactile guided therapy |
| CN102317978A (en) * | 2009-12-22 | 2012-01-11 | 松下电器产业株式会社 | Action analysis device and action analysis method |
| JP2016209431A (en) * | 2015-05-13 | 2016-12-15 | セイコーエプソン株式会社 | Swing analyzer, swing analysis method, swing analysis program and swing analysis system |
| CN105749537A (en) * | 2016-02-02 | 2016-07-13 | 四川长虹电器股份有限公司 | Exercise auxiliary marking system |
| CN106267734A (en) * | 2016-08-30 | 2017-01-04 | 石家庄铁路职业技术学院 | A kind of digitized body-building system |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110471529A (en) * | 2019-08-07 | 2019-11-19 | 北京卡路里信息技术有限公司 | Act methods of marking and device |
| CN112597835A (en) * | 2020-12-11 | 2021-04-02 | 国汽(北京)智能网联汽车研究院有限公司 | Driving behavior evaluation method and device, electronic equipment and readable storage medium |
| CN113657173A (en) * | 2021-07-20 | 2021-11-16 | 北京搜狗科技发展有限公司 | Data processing method and device and data processing device |
| CN113657173B (en) * | 2021-07-20 | 2024-05-24 | 北京搜狗科技发展有限公司 | Data processing method and device for data processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kurach et al. | A large-scale study on regularization and normalization in GANs | |
| US9224104B2 (en) | Generating data from imbalanced training data sets | |
| JP5346128B2 (en) | Object alignment in augmented reality | |
| US9836481B2 (en) | Image-based retrieval and searching | |
| US20130127860A1 (en) | Methods and Apparatus for Light Space Graphical Model in Shape from Shading | |
| CN108230353A (en) | Method for tracking target, system and electronic equipment | |
| US9833173B2 (en) | Matching system for correlating accelerometer data to known movements | |
| CN108280477A (en) | Method and apparatus for clustering image | |
| JP2013242761A (en) | Method, and controller and control program thereof, for updating policy parameters under markov decision process system environment | |
| CN108229532A (en) | Image-recognizing method, device and electronic equipment | |
| de Sousa Costa et al. | Classification of malignant and benign lung nodules using taxonomic diversity index and phylogenetic distance | |
| CN108686363A (en) | The method and apparatus of action evaluation | |
| CN109154938A (en) | Using discrete non-trace location data by the entity classification in digitized map | |
| CN103927530B (en) | The preparation method and application process, system of a kind of final classification device | |
| CN103927529B (en) | The preparation method and application process, system of a kind of final classification device | |
| CN113095416A (en) | Small sample SAR target classification method based on mixed loss and graph attention | |
| CN106096042A (en) | Data message sorting technique and system | |
| Kempa-Liehr et al. | Feature engineering workflow for activity recognition from synchronized inertial measurement units | |
| US20130121558A1 (en) | Point Selection in Bundle Adjustment | |
| Jiang et al. | Fast, accurate event classification on resource-lean embedded sensors | |
| JP2013195363A (en) | Movement state determination program and movement state determination apparatus | |
| Wang et al. | Support vector hazards machine: A counting process framework for learning risk scores for censored outcomes | |
| CN113283152B (en) | Adjustment method and driving method of driving electric signal, driving circuit and electronic equipment | |
| Hachaj et al. | Averaging three-dimensional time-varying sequences of rotations: application to preprocessing of motion capture data | |
| US11289202B2 (en) | Method and system to improve clinical workflow |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181023 |
|
| WD01 | Invention patent application deemed withdrawn after publication |