CN109543659A - Risk behavior monitoring and pre-alarming method and system suitable for old user - Google Patents
Risk behavior monitoring and pre-alarming method and system suitable for old user Download PDFInfo
- Publication number
- CN109543659A CN109543659A CN201811590111.7A CN201811590111A CN109543659A CN 109543659 A CN109543659 A CN 109543659A CN 201811590111 A CN201811590111 A CN 201811590111A CN 109543659 A CN109543659 A CN 109543659A
- Authority
- CN
- China
- Prior art keywords
- old user
- data
- classification
- monitoring
- risk
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/174—Facial expression recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
- G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Business, Economics & Management (AREA)
- General Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Alarm Systems (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention provides a kind of risk behavior monitoring and pre-alarming methods and system suitable for old user.A kind of risk behavior monitoring and pre-alarming method suitable for old user, comprising: obtain daily behavioral data, social interaction data and the facial exercises monitoring data of old user；The daily behavioral data and the social interaction data are input to classification of risks device progress classification of risks and trained mood classifier carries out mood classification in advance by facial exercises monitoring data input；According to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning.In the present embodiment by determine old user type of emotion and and risk classifications, early warning result can be made more accurate.Also, in the present embodiment the real-time of monitoring can may be implemented, monitoring efficiency can be improved dynamically with real-time monitoring individual mood.
Description
Technical field
The present invention relates to early warning technology field more particularly to a kind of risk behavior monitoring and pre-alarming methods suitable for old user
And system.
Background technique
Currently, may include following manner for special population early warning: for example manager by periodically timing to user into
Row behavior observation and individual interview, then experience judges in special population with the presence or absence of abnormal behavior person.For another example, manager
Special population can be passed through using examining letter validity and correcting questionnaire according to theories of psychology design compilation matter/quantification problem
The result to fill in questionnaires infers whether there is abnormal behavior.
However, existing method for early warning has the disadvantage in that
First, observation, the subjectivity of Interview Method are strong, inaccurately, it is difficult to unified measurement standard, therefore be difficult to and accurately sentence
Disconnected identification.
Second, questionnaire scale is filled in needs and is taken considerable time, causes assessment efficiency relatively low.
Third, the assessment period is long, can not accomplish real-time monitoring.
Summary of the invention
For the defects in the prior art, the present invention provides a kind of risk behavior monitoring and warning sides suitable for old user
Method and system, for solving technical problem present in the relevant technologies.
In a first aspect, the embodiment of the invention provides a kind of risk behavior monitoring and pre-alarming method suitable for old user, institute
The method of stating includes:
Obtain daily behavioral data, social interaction data and the facial exercises monitoring data of old user；
By the daily behavioral data and the social interaction data be input to classification of risks device carry out classification of risks and
By facial exercises monitoring data input, trained mood classifier carries out mood classification in advance；
According to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning.
Optionally, the daily behavioral data includes every per day feed number, daily mean motion number and puts down daily
Rest number；The social interaction data include talking with number with other people frequency of exposure, with other people；The facial exercises monitoring
Data include regional change, facial temperature change and the breathing and changes in heart rate of facial exercises unit.
Optionally, the trained mood classifier training mode in advance is as follows:
(1) convolutional neural networks are constructed, it is maximum to construct 3 convolution & for the grayscale image for being 32*32 for input size
The Softmax layer of 1 p=0.5 is connected after pond layer, 1 full linking layer, full linking layer；
(2) 9 different interest region ROI are arranged according to face face structure, actively guide neural network concern and expression
Relevant region；
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and extract certificate photo
900 neutral Emotional Pictures handle to obtain 40500 picture training datas as training data, and by ROI；Test data by
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on internet are constituted；
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained；
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), tests and obtain final feelings
Thread classifier.
Optionally, 3 convolutional layers are respectively as follows: CNN-64:[32, and 32,64,64]；CNN-96:[48,48,96,200]；
CNN-128:[64,64,128,300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows: ReLU (x)=max (0, x)；
And
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows:
[0.0001,0.001,0.001,0.01,0.1]。
Optionally, the daily behavioral data and the social interaction data are input to classification of risks device and carry out risk point
Class includes:
The classification of risks device calculates separately every per day feed number, daily mean motion number, per per day rest
Number, with other people frequency of exposure and and other people talk with the difference of the average value of number, standard deviation and average and standard deviation；
The classification of risks device judges per per day feed number, daily mean motion number, often per day rest number,
Talk in number with the presence or absence of the difference more than or less than its average and standard deviation extremely with other people frequency of exposure and with other people
One item missing；
If so, the classification of risks device determines that classification of risks is that there may be risks.
Optionally, according to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning, comprising:
Record the affective style and risk classifications of each individual within a preset period of time in old user；
Default prediction policy is called, determines whether the affective style and risk classifications meet the default prediction policy,
It alarms if meeting.
Second aspect, the embodiment of the invention provides a kind of risk behavior monitoring and warning system suitable for old user, institutes
The system of stating includes:
Monitoring data obtain module, and daily behavioral data, social interaction data and the face for obtaining old user are living
Dynamic monitoring data；
Early warning determining module, for the daily behavioral data and the social interaction data to be input to classification of risks device
It carries out classification of risks and trained mood classifier carries out mood classification in advance by facial exercises monitoring data input；
Early warning determining module, for the type of emotion and risk classifications according to the old user, it is determined whether carry out pre-
It is alert.
Optionally, the daily behavioral data includes every per day feed number, daily mean motion number and puts down daily
Rest number；The social interaction data include talking with number with other people frequency of exposure, with other people；The facial exercises monitoring
Data include regional change, facial temperature change and the breathing and changes in heart rate of facial exercises unit.
Optionally, the trained mood classifier training mode in advance is as follows:
(1) convolutional neural networks are constructed, it is maximum to construct 3 convolution & for the grayscale image for being 32*32 for input size
The Softmax layer of 1 p=0.5 is connected after pond layer, 1 full linking layer, full linking layer；
(2) 9 different interest region ROI are arranged according to face face structure, actively guide neural network concern and expression
Relevant region；
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and extract certificate photo
900 neutral Emotional Pictures handle to obtain 40500 picture training datas as training data, and by ROI；Test data by
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on internet are constituted；
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained；
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), tests and obtain final feelings
Thread classifier；Wherein
3 convolutional layers are respectively as follows: CNN-64:[32, and 32,64,64]；CNN-96:[48,48,96,200]；CNN-128:
[64,64,128,300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows: ReLU (x)=max (0, x)；
And
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows:
[0.0001,0.001,0.001,0.01,0.1]。
Optionally, the early warning determining module includes:
Affective style recording unit, for recording the affective style and wind of each individual within a preset period of time in old user
Dangerous type；
Emotion prewarning unit determines whether the affective style and risk classifications meet for calling default prediction policy
The default prediction policy is alarmed if meeting.
As shown from the above technical solution, by obtaining old user's face movement monitoring data in the embodiment of the present invention；So
Afterwards, by the facial exercises feature input of the old user recorded, trained mood classifier carries out mood classification in advance；Most
Afterwards, according to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning.In this way, passing through in the present embodiment
Face-image discriminance analysis emotional state, so as to so that early warning result is more accurate.Also, in the present embodiment can dynamic and
Real-time monitoring individual mood may be implemented the real-time of monitoring, monitoring efficiency can be improved.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other attached drawings are obtained according to these figures.
Fig. 1 is a kind of process for risk behavior monitoring and pre-alarming method suitable for old user that one embodiment of the invention provides
Schematic diagram；
Fig. 2~Fig. 3 is the risk behavior monitoring and warning system that the another kind that one embodiment of the invention provides is suitable for old user
The block diagram of system.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Currently, may include following manner for special population early warning: for example manager by periodically timing to user into
Row behavior observation and individual interview, then experience judges in special population with the presence or absence of abnormal behavior person.For another example, manager
Special population can be passed through using examining letter validity and correcting questionnaire according to theories of psychology design compilation matter/quantification problem
The result to fill in questionnaires infers whether there is abnormal behavior.
However, existing method for early warning has the disadvantage in that
First, observation, the subjectivity of Interview Method are strong, inaccurately, it is difficult to unified measurement standard, therefore be difficult to and accurately sentence
Disconnected identification.
Second, questionnaire scale is filled in needs and is taken considerable time, causes assessment efficiency relatively low.
Third, the assessment period is long, can not accomplish real-time monitoring.
Present inventor considers: when individual is difficult to adapt to environment, it will usually the feelings of sad, low type occurs
Not-ready status；When individual, which is thought, has threat in environment, it will usually the emotional state of angry type occur.Also, when individual continues
In certain emotional state, and when deviateing larger with the general normal emotional state of its own, especially in moods such as indignation, sadnesss
Under state, there is high possibility that it is excited to make and the risk behaviors such as attack other people, commit suiside.Human emotion's state can be to autonomic nerve
Activity has an impact, to generate a series of external physiologic performance, more apparent performance specifically includes that facial expression (face
Portion's activity unit) variation, facial temperature change, respiratory variations etc..
For this purpose, Fig. 1 is this embodiment of the invention provides a kind of risk behavior monitoring and pre-alarming method suitable for old user
The flow diagram for inventing the risk behavior monitoring and pre-alarming method for being suitable for old user of embodiment offer, can be applied to intelligence
The electronic equipments such as energy equipment, personal computer, server.Referring to Fig. 1, a kind of risk behavior monitoring and warning suitable for old user
Method includes:
101, obtain daily behavioral data, social interaction data and the facial exercises monitoring data of old user；
102, the daily behavioral data and the social interaction data are input to classification of risks device and carry out classification of risks
And trained mood classifier carries out mood classification in advance by facial exercises monitoring data input；
103, according to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning.
Each step of 1, Fig. 2 and embodiment to the risk behavior monitoring and pre-alarming method for being suitable for old user with reference to the accompanying drawing
It is described in detail.
Firstly, introducing 101, daily behavioral data, social interaction data and the facial exercises for obtaining old user monitor number
According to the step of.
In the present embodiment, the behavioral data of record includes: (1) per per day feed number: C (eat)；(2) per per day
Times of exercise: C (move)；(3) per per day rest number: C (rest).
The social interaction data of record include: (1) and other people frequency of exposure C (connect)；(2) talk with number C with other people
(speech)。
Facial exercises monitoring data include regional change, facial temperature change and breathing and the heart rate of facial exercises unit
Variation.Wherein:
(1) regional change of facial exercises unit.Facial exercises unit is labeled as 18-20 activity point
(Landmarks), each activity point is described by one group of coordinate value:
Dn (Xn, Yn)；
Wherein, D indicates a certain activity point, and n is serial number, and Xn is the abscissa value of n-th of activity point, n-th of Yn
The ordinate value of activity point.
(2) facial temperature change.The variation of facial temperature, performance on the video images, are presented as face-image colour-difference
It is different, facial temperature change is described by image enhancement technique are as follows:
Δ C=(C (n+1)-Cn)；
Wherein, C indicates that image color, C (n+1) indicate that the image color value of (n+1) second, Cn indicate n-th second image
Color-values.
(3) breathing, changes in heart rate.Breathing, changes in heart rate are embodied in the color change of facial specific region, breathing, the heart
Rate variation can be calculated using pre-set formula.
Then, 102 are introduced, the daily behavioral data and the social interaction data are input to classification of risks device and carried out
Classification of risks and the step that the preparatory trained mood classifier of facial exercises monitoring data input is carried out to mood classification
Suddenly.
In the present embodiment, classification of risks device can be preset, can be by (1) per per day feed number: C (eat)；
(2) daily mean motion number: C (move)；(3) per per day rest number: C (rest)；(4) with other people frequency of exposure C
(connect)；(5) talk with number C (speech) with other people and be input to classification of risks device.Classification of risks device can calculate (1)
Per per day feed number: C (eat)；(2) daily mean motion number: C (move)；(3) per per day rest number: C
(rest)；(4) with other people frequency of exposure C (connect)；(5) talk with average value U, the standard deviation of number C (speech) with other people
S, and calculate difference Δ=U-S of average value U and standard deviation S.
When the C value that old user has at least one in (1) of a certain day~(5) is more than or less than U+/- Δ,
Classification of risks device classification are as follows: " there may be risks ".
In the present embodiment, by the input of facial exercises feature, trained mood classifier carries out mood classification in advance.Wherein,
Preparatory trained mood classifier training mode is as follows:
(1) convolutional neural networks are constructed, it is maximum to construct 3 convolution & for the grayscale image for being 32*32 for input size
The Softmax layer of 1 p=0.5 is connected after pond layer, 1 full linking layer, full linking layer；
CNN-64:[32,32,64,64]；
CNN-96:[48,48,96,200]；
CNN-128:[64,64,128,300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows:
ReLU (x)=max (0, x)；
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows:
[0.0001,0.001,0.001,0.01,0.1]。
(2) 9 difference interest region ROI (Region of Interesting) are arranged according to face face structure, actively
Neural network is guided to pay close attention to region relevant to expression.
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and extract certificate photo
900 neutral Emotional Pictures handle to obtain 40500 picture training datas as training data, and by ROI；Test data by
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on internet are constituted.
It will be appreciated that the quantity of training data or test data can be adjusted according to concrete scene, herein not
It limits.
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained.
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), tests and obtain final feelings
Thread classifier.
In the present embodiment, the old user's face feature of video camera shooting is inputted into mood classifier, according to above-mentioned steps
(1)-(4) are trained, test obtained mood classifier.By mood classifier calculated, the available same day, at that time individual
The type of emotion of X, such as: (number X, sad).
Finally, 103 are introduced, according to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning
Step.
In the present embodiment, each individual affective style and risk class within a preset period of time in special population can recorde
Type.Then, default prediction policy is called, determines whether affective style and risk classifications meet default prediction policy, if meeting
Alarm.
Wherein presetting prediction policy can preset.For example, certain is individual, in three times per day observation, continuously go out three times
When existing sadness, angry mood, and classification of risks is divided to be to determine early warning there may be when risk.
For another example, certain individual is compared with other individuals, and in three times per day observation, type of emotion continuously presents obvious three times
When difference, and classification of risks is divided to be there may be when risk, such as: one day, in entire special population, 85% individual is continuous
Happy mood three times is presented, but angry mood is continuously presented in individual X three times, determines early warning；Or individual Y continuously three times present in
Disposition thread determines and prepares early warning.
So far, in the present embodiment by determine old user type of emotion and and risk classifications, early warning result can be made
It is more accurate.Also, in the present embodiment can dynamic and real-time monitoring individual mood, the real-time of monitoring may be implemented, can be with
Improve monitoring efficiency.
Second aspect, the embodiment of the invention provides a kind of risk behavior monitoring and warning system suitable for old user, ginsengs
See Fig. 2, the system comprises:
Monitoring data obtain module 201, for obtaining daily behavioral data, social interaction data and the face of old user
Movement monitoring data；
Mood categorization module 202, for the daily behavioral data and the social interaction data to be input to risk point
Class device carries out classification of risks and trained mood classifier carries out mood in advance by facial exercises monitoring data input
Classification；
Early warning determining module 203, for the type of emotion and risk classifications according to the old user, it is determined whether carry out
Early warning.
In some embodiments, the facial exercises monitoring data include the regional change of facial exercises unit, face temperature
Degree variation and breathing and changes in heart rate.
In some embodiments, the trained mood classifier training mode in advance is as follows:
(1) convolutional neural networks are constructed, it is maximum to construct 3 convolution & for the grayscale image for being 32*32 for input size
The Softmax layer of 1 p=0.5 is connected after pond layer, 1 full linking layer, full linking layer；
(2) 9 different interest region ROI are arranged according to face face structure, actively guide neural network concern and expression
Relevant region；
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and extract certificate photo
900 neutral Emotional Pictures handle to obtain 40500 picture training datas as training data, and by ROI；Test data by
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on internet are constituted；
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained；
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), tests and obtain final feelings
Thread classifier.
In some embodiments, 3 convolutional layers are respectively as follows: CNN-64:[32, and 32,64,64]；CNN-96:[48,48,96,
200]；CNN-128:[64,64,128,300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows: ReLU (x)=max (0, x)；
And
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows:
[0.0001,0.001,0.001,0.01,0.1]。
In some embodiments, referring to Fig. 3, the early warning determining module 203 includes:
Affective style recording unit 301, for recording the affective style of each individual within a preset period of time in old user
And risk classifications；
Emotion prewarning unit 302 determines whether the affective style and risk classifications are full for calling default prediction policy
The foot default prediction policy, alarms if meeting.
It should be noted that the risk behavior monitoring and warning system provided in an embodiment of the present invention for being suitable for old user with it is upper
The method of stating is one-to-one relationship, and the implementation detail of the above method is equally applicable to above system, and the embodiment of the present invention is no longer
Above system is described in detail.
In specification of the invention, numerous specific details are set forth.It is to be appreciated, however, that the embodiment of the present invention can be with
It practices without these specific details.In some instances, well known method, structure and skill is not been shown in detail
Art, so as not to obscure the understanding of this specification.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme should all cover within the scope of the claims and the description of the invention.
Claims (10)
1. a kind of risk behavior monitoring and pre-alarming method suitable for old user, which is characterized in that the described method includes:
Obtain daily behavioral data, social interaction data and the facial exercises monitoring data of old user；
The daily behavioral data and the social interaction data are input to classification of risks device and carry out classification of risks and by institute
Stating the input of facial exercises monitoring data, trained mood classifier carries out mood classification in advance；
According to the type of emotion and risk classifications of the old user, it is determined whether carry out early warning.
2. the risk behavior monitoring and pre-alarming method according to claim 1 for being suitable for old user, which is characterized in that described every
Day, behavioral data was including every per day feed number, daily mean motion number and per per day rest number；The society hands over
It include talking with number with other people frequency of exposure, with other people toward data；The facial exercises monitoring data include facial exercises unit
Regional change, facial temperature change and breathing and changes in heart rate.
3. the risk behavior monitoring and pre-alarming method according to claim 1 for being suitable for old user, which is characterized in that described pre-
First trained mood classifier training mode is as follows:
(1) convolutional neural networks are constructed, the grayscale image for being 32*32 for input size constructs 3 convolution & maximum ponds
The Softmax layer of 1 p=0.5 is connected after layer, 1 full linking layer, full linking layer；
(2) 9 different interest region ROI are arranged according to face face structure, actively guide neural network concern related to expression
Region；
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and are extracted certificate photo 900 and opened
Neutral Emotional Picture handles to obtain 40500 picture training datas as training data, and by ROI；Test data is by interconnecting
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on the net are constituted；
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained；
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), test and obtain final mood and divide
Class device.
4. the risk behavior monitoring and pre-alarming method according to claim 3 for being suitable for old user, which is characterized in that 3 volumes
Lamination is respectively as follows: CNN-64:[32, and 32,64,64]；CNN-96:[48,48,96,200]；CNN-128:[64,64,128,
300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows: ReLU (x)=max (0, x)；
And
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows: [0.0001,
0.001,0.001,0.01,0.1]。
5. the risk behavior monitoring and pre-alarming method according to claim 2 for being suitable for old user, which is characterized in that will be described
Daily behavioral data and the social interaction data are input to classification of risks device progress classification of risks
The classification of risks device calculate separately per per day feed number, daily mean motion number, often per day rest number,
With other people frequency of exposure and and other people talk with the difference of the average value of number, standard deviation and average and standard deviation；
The classification of risks device judges per per day feed number, daily mean motion number, often per day rest number and he
People's frequency of exposure and with other people talk in number with the presence or absence of more than or less than its average and standard deviation difference at least one
?；
If so, the classification of risks device determines that classification of risks is that there may be risks.
6. the risk behavior monitoring and pre-alarming method according to claim 1 for being suitable for old user, which is characterized in that according to institute
State the type of emotion and risk classifications of old user, it is determined whether carry out early warning, comprising:
Record the affective style and risk classifications of each individual within a preset period of time in old user；
Default prediction policy is called, determines whether the affective style and risk classifications meet the default prediction policy, if full
It is sufficient then alarm.
7. a kind of risk behavior monitoring and warning system suitable for old user, which is characterized in that the system comprises:
Monitoring data obtain module, for obtaining daily behavioral data, social interaction data and the facial exercises prison of old user
Measured data；
Early warning determining module is carried out for the daily behavioral data and the social interaction data to be input to classification of risks device
Classification of risks and trained mood classifier carries out mood classification in advance by facial exercises monitoring data input；
Early warning determining module, for the type of emotion and risk classifications according to the old user, it is determined whether carry out early warning.
8. the risk behavior monitoring and warning system according to claim 7 for being suitable for old user, which is characterized in that described every
Day, behavioral data was including every per day feed number, daily mean motion number and per per day rest number；The society hands over
It include talking with number with other people frequency of exposure, with other people toward data；The facial exercises monitoring data include facial exercises unit
Regional change, facial temperature change and breathing and changes in heart rate.
9. the risk behavior monitoring and warning system according to claim 7 for being suitable for old user, which is characterized in that described pre-
First trained mood classifier training mode is as follows:
(1) convolutional neural networks are constructed, the grayscale image for being 32*32 for input size constructs 3 convolution & maximum ponds
The Softmax layer of 1 p=0.5 is connected after layer, 1 full linking layer, full linking layer；
(2) 9 different interest region ROI are arranged according to face face structure, actively guide neural network concern related to expression
Region；
(3) glad, sad, indignation and surprised 4 class each 900 facial pictures are extracted from internet, and are extracted certificate photo 900 and opened
Neutral Emotional Picture handles to obtain 40500 picture training datas as training data, and by ROI；Test data is by interconnecting
Happiness, sadness, indignation, each 300 picture of surprised and neutral 5 class of mood downloaded on the net are constituted；
(4) by training, test, the classifier that accuracy rate is more than 98% is obtained；
(5) old user's face feature is inputted and is trained according to above-mentioned steps (1)-(4), test and obtain final mood and divide
Class device；Wherein
3 convolutional layers are respectively as follows: CNN-64:[32, and 32,64,64]；CNN-96:[48,48,96,200]；CNN-128:[64,
64,128,300]；
Other than Softmax layers, remaining each layer activation primitive is equal are as follows: ReLU (x)=max (0, x)；
And
Weight W initialization uses the zero-mean of Krizhevsky, constant standard deviation, each layer constant standard deviation are as follows: [0.0001,
0.001,0.001,0.01,0.1]。
10. the risk behavior monitoring and warning system according to claim 7 for being suitable for old user, which is characterized in that described
Early warning determining module includes:
Affective style recording unit, for recording each individual affective style and risk class within a preset period of time in old user
Type；
It is described to determine whether the affective style and risk classifications meet for calling default prediction policy for emotion prewarning unit
Default prediction policy, alarms if meeting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811590111.7A CN109543659B (en) | 2018-12-25 | 2018-12-25 | Risk behavior monitoring and early warning method and system suitable for old users |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811590111.7A CN109543659B (en) | 2018-12-25 | 2018-12-25 | Risk behavior monitoring and early warning method and system suitable for old users |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109543659A true CN109543659A (en) | 2019-03-29 |
| CN109543659B CN109543659B (en) | 2020-03-31 |
Family
ID=65858085
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811590111.7A Active CN109543659B (en) | 2018-12-25 | 2018-12-25 | Risk behavior monitoring and early warning method and system suitable for old users |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109543659B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110264668A (en) * | 2019-07-10 | 2019-09-20 | 四川长虹电器股份有限公司 | More tactful old men based on machine vision technique see maintaining method |
| CN115662631A (en) * | 2022-10-26 | 2023-01-31 | 上海柚纯数字科技有限公司 | AI intelligence discrimination-based nursing home management system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100325078A1 (en) * | 2009-06-22 | 2010-12-23 | Lee Ho-Sub | Device and method for recognizing emotion and intention of a user |
| CN102413871A (en) * | 2009-04-30 | 2012-04-11 | 麦德托尼克公司 | Patient state detection based on support vector machine based algorithm |
| CN104994335A (en) * | 2015-06-11 | 2015-10-21 | 广东欧珀移动通信有限公司 | Alarm method and terminal |
| CN106027978A (en) * | 2016-06-21 | 2016-10-12 | 南京工业大学 | Video monitoring abnormal behavior system and method for smart home old people care |
| CN106650621A (en) * | 2016-11-18 | 2017-05-10 | 广东技术师范学院 | Deep learning-based emotion recognition method and system |
| CN106777954A (en) * | 2016-12-09 | 2017-05-31 | 电子科技大学 | The intelligent guarding system and method for a kind of Empty nest elderly health |
| CN107320090A (en) * | 2017-06-28 | 2017-11-07 | 广东数相智能科技有限公司 | A kind of burst disease monitor system and method |
| CN107491726A (en) * | 2017-07-04 | 2017-12-19 | 重庆邮电大学 | A kind of real-time expression recognition method based on multi-channel parallel convolutional neural networks |
| CN208227195U (en) * | 2018-04-13 | 2018-12-11 | 西安科技大学 | A kind of old solitary people monitor device based on Expression Recognition |
-
2018
- 2018-12-25 CN CN201811590111.7A patent/CN109543659B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102413871A (en) * | 2009-04-30 | 2012-04-11 | 麦德托尼克公司 | Patient state detection based on support vector machine based algorithm |
| US20100325078A1 (en) * | 2009-06-22 | 2010-12-23 | Lee Ho-Sub | Device and method for recognizing emotion and intention of a user |
| CN104994335A (en) * | 2015-06-11 | 2015-10-21 | 广东欧珀移动通信有限公司 | Alarm method and terminal |
| CN106027978A (en) * | 2016-06-21 | 2016-10-12 | 南京工业大学 | Video monitoring abnormal behavior system and method for smart home old people care |
| CN106650621A (en) * | 2016-11-18 | 2017-05-10 | 广东技术师范学院 | Deep learning-based emotion recognition method and system |
| CN106777954A (en) * | 2016-12-09 | 2017-05-31 | 电子科技大学 | The intelligent guarding system and method for a kind of Empty nest elderly health |
| CN107320090A (en) * | 2017-06-28 | 2017-11-07 | 广东数相智能科技有限公司 | A kind of burst disease monitor system and method |
| CN107491726A (en) * | 2017-07-04 | 2017-12-19 | 重庆邮电大学 | A kind of real-time expression recognition method based on multi-channel parallel convolutional neural networks |
| CN208227195U (en) * | 2018-04-13 | 2018-12-11 | 西安科技大学 | A kind of old solitary people monitor device based on Expression Recognition |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110264668A (en) * | 2019-07-10 | 2019-09-20 | 四川长虹电器股份有限公司 | More tactful old men based on machine vision technique see maintaining method |
| CN115662631A (en) * | 2022-10-26 | 2023-01-31 | 上海柚纯数字科技有限公司 | AI intelligence discrimination-based nursing home management system |
| CN115662631B (en) * | 2022-10-26 | 2023-11-17 | 上海柚纯数字科技有限公司 | Nursing home management system based on AI intelligent discrimination |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109543659B (en) | 2020-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Oh et al. | Cybersickness and its severity arising from virtual reality content: A comprehensive study | |
| Krumhuber et al. | Human and machine validation of 14 databases of dynamic facial expressions | |
| Aigrain et al. | Multimodal stress detection from multiple assessments | |
| Yannakakis et al. | Towards affective camera control in games | |
| Fiorentini et al. | Is there a dynamic advantage for facial expressions? | |
| US20180330249A1 (en) | Method and apparatus for immediate prediction of performance of media content | |
| US20220067519A1 (en) | Neural network synthesis architecture using encoder-decoder models | |
| US20130102854A1 (en) | Mental state evaluation learning for advertising | |
| Fujimura et al. | Effects of dynamic information in recognising facial expressions on dimensional and categorical judgments | |
| Gavrilescu et al. | Predicting the Sixteen Personality Factors (16PF) of an individual by analyzing facial features | |
| CN109543659A (en) | Risk behavior monitoring and pre-alarming method and system suitable for old user | |
| CN109498039A (en) | Personality assessment's method and device | |
| JP2018206341A (en) | Event evaluation support system, event evaluation support device, and event evaluation support program | |
| EP4208079A1 (en) | Artificial intelligence-guided visual neuromodulation for therapeutic or performance-enhancing effects | |
| Rangarajan et al. | A descriptive phenomenological approach to perception of affective quality in design inspiration | |
| Islam | A deep learning based framework for detecting and reducing onset of cybersickness | |
| Deng et al. | Experimental study on panic during simulated fire evacuation using psycho-and physiological metrics | |
| Cattaneo et al. | Sex differences in affective facial reactions are present in childhood | |
| Roohi et al. | Neural network based facial expression analysis of gameevents: a cautionary tale | |
| Ortiz et al. | Design of an Immersive Virtual Reality Framework to Enhance the Sense of Agency Using Affective Computing Technologies | |
| CN109635778A (en) | Risk behavior monitoring and pre-alarming method and system suitable for special population | |
| EP2920671A1 (en) | Automated thumbnail selection for online video | |
| Campbell et al. | Saliency-based bayesian modeling of dynamic viewing of static scenes | |
| Song et al. | Deep learning-based assessment of facial periodic affect in work-like settings | |
| Kassam | Assessment of emotional experience through facial expression |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220310 Address after: 266000 room 504, floor 5, building a, Shinan Software Park, No. 288, Ningxia road, Shinan District, Qingdao, Shandong Province Patentee after: Shandong Xinfa Technology Co.,Ltd. Address before: 100094 No. 1008-1105, first floor, block a, building 1, yard 2, Yongcheng North Road, Haidian District, Beijing Patentee before: BEIJING XINFA TECHNOLOGY Co.,Ltd. |