CN109920208A - Tumble prediction technique, device, electronic equipment and system - Google Patents
Tumble prediction technique, device, electronic equipment and system Download PDFInfo
- Publication number
- CN109920208A CN109920208A CN201910100602.7A CN201910100602A CN109920208A CN 109920208 A CN109920208 A CN 109920208A CN 201910100602 A CN201910100602 A CN 201910100602A CN 109920208 A CN109920208 A CN 109920208A
- Authority
- CN
- China
- Prior art keywords
- tumble
- target image
- framework information
- information feature
- guardianship
- 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
Landscapes
- Image Analysis (AREA)
Abstract
This application discloses a kind of tumble prediction technique, device, electronic equipment and systems, this method comprises: obtaining the target image of people under the guardianship；Extract the framework information feature in the target image；The framework information feature is predicted, judges whether there will be tumble event；If it is determined that there will be tumble event, then tumble early warning is exported.This method can obtain the image of people under the guardianship in real time, and feature extraction and prediction are carried out using target image of the deep learning to acquisition, the target image can not be obtained by depth camera, common camera is only needed to obtain image, in this way, largely can be with save the cost, in addition, the application can monitor whether people under the guardianship has tumble event in real time, effectively avoid children under guardianship's Fall injuries.
Description
Technical field
This application involves field of computer technology, more particularly, to a kind of tumble prediction technique, device, electronic equipment
And system.
Background technique
As China's Aging Problem is increasingly severe, endowment problem has become the focal issue of social concerns.Due to
Children can not accompany always parent at one's side, when Falls Among Old People, if cannot find in time, it will cause it is very serious after
Fruit.Therefore, being detected in time to the tumble behavior of old man and calling rescue becomes problem in the urgent need to address, i.e., for old age
The intelligent nursing demand of people is increasing.
Apply for content
In view of this, present applicant proposes a kind of tumble prediction technique, device, electronic equipment and system, it is above-mentioned to improve
Defect.
In a first aspect, the embodiment of the present application provides a kind of tumble prediction technique, this method comprises: obtaining people under the guardianship
Target image；Extract the framework information feature in the target image；Judgement, which is, to be predicted to the framework information feature
It is no to have tumble event；If it is determined that there will be tumble event, then tumble early warning is exported.
Second aspect, the embodiment of the present application provide a kind of tumble prediction meanss, which is applied to server or net
It closes, described device includes: to obtain module, characteristic extracting module, tumble prediction module and alarm module.Module is obtained for obtaining
The target image of people under the guardianship.Characteristic extracting module is used to extract the framework information feature in the target image.It falls pre-
Module is surveyed for predicting the framework information feature, judges whether there will be tumble event.Alarm module is used for
If it is determined that there will be tumble event, then tumble early warning is exported.
The third aspect, the embodiment of the present application provide a kind of tumble forecasting system, the system include image collecting device and
Server or gateway.Image collecting device is used to obtain the target image of people under the guardianship.Server or gateway are for mentioning
The framework information feature in the target image is taken, the framework information feature is predicted, judges whether there will be tumble
Event occurs, if it is decided that will have tumble event, then export tumble early warning.
Fourth aspect, the embodiment of the invention provides a kind of electronic equipment, which includes: one or more processing
Device；Memory, for storing one or more programs；One or more application program；It is wherein one or more of to apply journey
Sequence is stored in the memory and is configured as being executed by one or more of processors, one or more of programs
The method that the tumble prediction of the application any embodiment offer is provided.
5th aspect, the embodiment of the present application provide a kind of computer system, are stored in computer readable storage medium
Program code, said program code can be called the method for executing the tumble prediction that the application any embodiment provides by processor.
Compared with the existing technology, the embodiment of the present application proposes a kind of tumble prediction technique, device, electronic equipment and is
System, obtains the target image of people under the guardianship；Extract the framework information feature in the target image；It is special to the framework information
Sign is predicted, judges whether will there is tumble event；If it is determined that there will be tumble event, then it is pre- to export tumble
It is alert.The application judges to be by carrying out framework information feature extraction and the prediction of framework information feature to the target image got
It is no to have tumble event, tumble early warning can be exported immediately if predicting and there will be the generation of tumble event, so, it is possible reality
When monitor people under the guardianship, can effectively mitigate tumble give people under the guardianship's bring injury.
To can be more clearly understood the above objects, features, and advantages of the application, preferred embodiment is cited below particularly, and match
Appended attached drawing is closed, is described below in detail.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for this field
For technical staff, without creative efforts, it is also possible to obtain other drawings based on these drawings.
Fig. 1 shows a kind of structural schematic diagram of network system of the application proposition；
Fig. 2 shows a kind of tumble prediction technique flow charts that the application one embodiment proposes；
Fig. 3 shows human skeleton information schematic diagram in a kind of tumble prediction technique of the application one embodiment proposition；
Human body two states schematic diagram in a kind of tumble prediction technique that Fig. 4 the application one embodiment proposes；
Fig. 5 shows a kind of tumble prediction technique flow chart of another embodiment of the application proposition；
Fig. 6 shows a kind of tumble prediction technique flow chart of another embodiment of the application proposition；
Fig. 7 shows the tumble behavior prediction structure in a kind of tumble prediction technique that the another embodiment of the application proposes
Figure；
Fig. 8 shows the pass of two time series datas in a kind of tumble prediction technique that the another embodiment of the application proposes
It is 1 schematic diagram；
Fig. 9 shows the pass of two time series datas in a kind of tumble prediction technique that the another embodiment of the application proposes
It is 2 schematic diagram；
The tumble behavior prediction process that Figure 10 shows a kind of tumble prediction technique that the another embodiment of the application proposes is shown
It is intended to；
Figure 11 shows a kind of structural block diagram of tumble prediction meanss of the embodiment of the present application proposition；
Figure 12 shows a kind of structural block diagram of tumble forecasting system of the embodiment of the present application proposition；
Figure 13 shows a kind of detailed block diagram of tumble forecasting system of the embodiment of the present application proposition；
Figure 14 show the embodiment of the present application proposition for executing according to the tumble prediction technique of the embodiment of the present application
The structural block diagram of electronic equipment；
Figure 15 show the embodiment of the present application proposition for save or carry realize falling according to the embodiment of the present application
The storage unit of the program code of prediction technique.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description.Based on the embodiment in the application, those of ordinary skill in the art institute without creative efforts
The every other embodiment obtained, shall fall in the protection scope of this application.
It should also be noted that similar label or letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.
For the ease of application scheme is described in detail, will be first introduced below in conjunction with system of the attached drawing to the application.
The following embodiments of the application are can be applied to unless otherwise instructed in system 10 as shown in Figure 1, which can
To include terminal device 11, server/cloud server 12, gateway 13, electronic equipment 14 and router 15.Wherein, gateway 13
It can may be ZIGBEE gateway for gateway camera, when gateway 13 is gateway camera, be mainly used for target figure
Server/cloud server 12 is uploaded to by router 15 as being acquired, and by the target image of acquisition；When gateway 13 is
When ZIGBEE gateway, it is mainly used for receiving the target image that electronic equipment 14 is sent, and received target image is passed through road
Server/cloud server 12 is uploaded to by device 15.
Electronic equipment 14 may include image collecting device and intelligentized Furniture equipment, and wherein image collecting device can be to take the photograph
As head, the target image for being mainly used for obtaining target image in real time, and will acquire is transmitted to gateway 13, and smart home device
It can be anti-fall device, when tumble event occurs, it mainly plays protective action.Electronic equipment 14 can pass through gateway 13 and road
The control instruction that server/cloud server 12 is sent is received by device 15, which can control the execution of electronic equipment 14
The scene of electronic equipment corresponding with control instruction and/or the movement of automation.
Gateway 13 and terminal device 11 can be connect with router 15, and be linked into Ethernet by router 15
In, router 15 can pass through the access servers/cloud server such as 2G/3G/4G/5G, WIFI 12.Optionally, terminal device
11 can establish network connection by 2G/3G/4G/5G, WIFI etc. and server/cloud server 12, so as to obtain clothes
The data that business device/cloud server 12 issues.Server/cloud server 12 passes through wired or wireless network and terminal device 11
It connects, wherein, terminal device 11 may include: mobile phone, PC (personal computer) computer, tablet computer, notebook electricity
Brain, mobile internet device (MID, mobile internet device) or other kinds of equipment and/or platform.
Server/cloud server 12 in the present embodiment may include Cloud Server, network access server, database
Multiple servers such as server, authentication server.
With the fast development of deep learning, the method for deep learning can be used at present to extract video/image
The framework information of middle human body, but the less of behavior prediction is carried out using deep learning in the prior art.
Therefore, in order to overcome drawbacks described above, such as Fig. 2, the embodiment of the present application provides a kind of tumble prediction technique, as one
Kind of embodiment, the executing subject of this method can be above-mentioned server or gateway, the method comprising the steps of S110 to step
S140。
Step S110: the target image of people under the guardianship is obtained.
The target image of people under the guardianship is obtained, the target image of the people under the guardianship can be imaged directly by gateway
Head captured in real-time obtains, and can directly be passed through router after the target image that gateway camera gets people under the guardianship
It is uploaded to server or gateway.The target image of people under the guardianship can also be obtained in real time by image collecting device,
Image collecting device can be infrared camera, colour imagery shot etc., and the concrete type of image collecting device is implemented in the application
It is not intended as limiting in example.Image collecting device can be installed on monitored personnel often movable place, example in the present embodiment
Such as, the bedroom/parlor, home for destitute of old man, hospital etc..In the present embodiment people under the guardianship can be old man, child, patient or
Blind person etc., specially which kind of crowd here without clearly limiting, for example, have in family old man and old man action not side
Just, if children want preferably protection old man, old man can be people under the guardianship.In addition, can be in target image
Multiple people under the guardianship are existed simultaneously, for example, target image may just include multiple people under the guardianship simultaneously in home for destitute, when out
When existing such case, a people under the guardianship includes a region, and rear extended meeting is respectively to the bone of the people under the guardianship in these regions
Frame information is analyzed and is predicted.The resource that image collecting device mainly obtains in the present embodiment is video resource, Image Acquisition
The video resource of acquisition is uploaded to server by device, and perhaps gateway therefore server or gateway will receive video resource decoding
Generate target image.
It should be noted that server or gateway be after getting the target image of people under the guardianship, it can be to described
Target image is pre-processed, for example, denoising, greyscale transformation, filtering etc., it can be by mesh by carrying out pretreatment operation to image
Interference information in logo image weeds out, so convenient for extraction framework information feature.
Step S120: the framework information feature in the target image is extracted.
After server or gateway get the target image of people under the guardianship, the target image is input to training
In good Feature Selection Model, framework information feature extraction is carried out.The Feature Selection Model is stored in server or gateway
On, it is to be obtained by mass data collection training, mass data collection is the image comprising human skeleton information in the present embodiment
It constitutes, these images can be obtained by image collecting device, also be can use web crawlers and obtained scale from internet
Skeleton picture different, position is different, light conditions are different and posture is different, and picture concerned is automatically converted into fixed ruler
Very little, fixed format picture, for example, picture concerned to be automatically converted into the jpg picture of 32*32.According to the requirement of deep learning,
Each class object of the embodiment of the present application can acquire 10,000 or more image datas to obtain preferable effect.It needs
It is bright, if the data set of building is insufficient, more data sets can be obtained by image procossing, for example, can be to
The processing such as the data set of building is filtered plus makes an uproar.After getting mass data collection, the mass data collection is input to building
Good feature extraction network, training get feature extraction network model.
Fig. 3 shows human skeleton hum pattern in a kind of tumble prediction technique of the application one embodiment proposition, from Fig. 3
It can be seen that human skeleton information characteristics generally refer to human skeleton key point.Human skeleton key point may include head, neck
Sub, right shoulder, right elbow, right wrist, left shoulder, left elbow, left wrist, right hip, right knee, right ankle, left hip, left knee, left ankle, chest, nose, a left side
Eye, right eye, left ear, auris dextra, middle hip circumference, left hallux toe, left little toe, left heel, right hallux toe, right little toe and right crus of diaphragm with etc..Pass through
It is above-mentioned it can be concluded that, human skeleton key point can have it is multiple, it is specifically how many here without clearly limiting.Pass through feature
The feature that extraction network extracts these key points can get the framework information feature of human body.
Step S130: the framework information feature is predicted, judges whether there will be tumble event.
Server or gateway get can use after framework information feature tumble prediction model come to framework characteristic into
Then row prediction carries out judging whether to have tumble event is similar with Feature Selection Model to fall in advance according to the prediction
Model is surveyed to be also required to obtain by mass data collection training.Common tumble prediction model has Recognition with Recurrent Neural Network model, karr
How graceful filter forecasting model or combination forecasting etc. specifically train acquisition here just without being described in detail.
Step S140: if it is determined that there will be tumble event, then tumble early warning is exported.
If it is determined that server or gateway there will be tumble event, then tumble is exported in advance to the terminal of care provider
Equipment, and or send control instruction to anti-fall device, for controlling, the anti-fall device starting is anti-to fall function to the control instruction
Energy.
Terminal device can be the mobile phone of people under the guardianship relatives in the present embodiment, for example, when people under the guardianship is old man
When, terminal device can be the mobile phone of children.In addition, terminal device is also possible to the hand of the staff of some nurse mechanisms
Machine, or the monitoring device of nurse mechanism configuration, such as computer, monitoring and plate.When care provider receives warning information
After can go in time the case where checking people under the guardianship, so as to avoid people under the guardianship fall for a long time without people find
The case where, this may be that people under the guardianship obtains medical treatment in time and strives for more times.
In one embodiment, anti-fall device can be air bag, when server or gateway predict to have and fall
When the generation of falling event, control instruction can be sent to anti-fall device while sending warning information to terminal device, passed through
The control instruction starts anti-fall device, and it is dynamic can to make protection in time for anti-fall device after people under the guardianship falls down
Make, can so mitigate the damage of people under the guardianship's body to a certain extent.
In addition, can also install audible and visible alarm in anti-fall device, anti-fall device can not only be risen after tumble event occurs
Protective effect can also notify neighbouring care provider, such care provider that can rush in time by audible and visible alarm, because of prison
Shield personnel there may come a time when the information that can ignore terminal device, or without carried terminal equipment when cannot find tumble thing in time
Part.It should be noted that audible and visible alarm also may be mounted at the place that care provider or people under the guardianship often enter and leave, when it
It can be controlled by server or gateway when being mounted on the care provider place that perhaps people under the guardianship often enters and leaves
Where system, audible and visible alarm are specifically installed here without clearly limiting, and user can install according to demands of individuals.
In conclusion carry out framework information feature extraction and tumble event prediction when need in advance training characteristics mention
Modulus type and tumble prediction model, two models can store on Cloud Server or gateway in the present embodiment, and take in cloud
Relevant operation is carried out on business device or gateway.It should be noted that Feature Selection Model and tumble prediction model also can store
And operate on image collecting device, for example, camera shooting generator terminal, has higher want to the performance of video camera in this case
It asks.
The embodiment of the present application proposes a kind of tumble prediction technique, without somatosensory device, it is only necessary to have one section of video
As input, so that it may which the human body in predictive picture whether there is a possibility that falling.In the scene for needing old man to nurse, peace
A common camera is filled, the behavior of old man can be monitored by this programme, when predicting old man in the presence of the possibility fallen, i.e.,
Can start some includes but is not limited to the anti-equipment for falling air bag etc., mitigates to fall and injure to old man's bring, and carries out early warning letter
The linkage of breath exports.
In order to preferably extract the framework information feature in target image, another embodiment of the application proposes a kind of tumble
Prediction technique, referring to Fig. 5, the method comprising the steps of as can be seen from Figure 5 S210 to step S240, is detailed below:
Step S210: the target image of people under the guardianship is obtained.
Step S220: the framework information feature in the target image is extracted.
Wherein, step S220 may include step S221 to step S222.
Step S221: extracting the target critical point in the target image, and the target critical point is that human skeleton is crucial
Point.
After getting the target image of people under the guardianship, the target image is input to the feature extraction net built
Feature extraction is carried out in network in model, that feature extraction Web vector graphic is convolutional neural networks (Convolution in the present embodiment
Neural network, CNN), convolutional neural networks are a kind of artificial neural networks based on deep learning theory, and advantage exists
Reduce the parameters inflation problem in traditional neural network in sharing using weight.
In one embodiment, using in the target image of image acquisition device not only include people under the guardianship,
Other objects under scene where may include people under the guardianship, for example, flowers and plants, stool, sofa, animal or other household electrical appliances
Deng, these objects can all be regarded as to background in this implementation, regard human body image as prospect, it can be by being carried out to background
Modeling after building up model, matches each pixel in target image to extract prospect, no if matching is then background
It is then prospect.After obtaining foreground image, the image of generation can be two-value region, and foreground area pixel value is 255, background area
Pixel value is 0, can so extract the human body image in target image.
Convolutional neural networks can be a kind of feedforward neural network, its artificial neuron can respond a part covering model
Interior surrounding cells are enclosed, have outstanding performance for large-scale image procossing.In general, the basic structure of convolutional neural networks includes two
Layer, one are characterized extract layer, and the input of each neuron and the part of preceding layer receive domain connection, and extract the spy of the part
Sign, after the local feature is extracted, the positional relationship of it and other feature vector is also decided therewith；The second is feature
Mapping layer, Feature Mapping layer use activation primitive, so that Feature Mapping has shift invariant.Therefore, convolutional neural networks are normal
For making the initial model of zone location model and Feature Selection Model.
The extraction of human skeleton information realizes that the implementation method includes based on the method for convolutional neural networks in the present embodiment
But be not limited to Mask R-CNN, DensePose etc..By taking Mask R-CNN as an example, it is inherited in Faster R-CNN the present embodiment,
It is added to a Mask Prediction Branch (Mask predicted branches) on Faster R-CNN, and improves
ROI Pooling proposes ROI Align.Mask R-CNN can be used to do " target detection ", " object instance segmentation ", " mesh
Mark critical point detection " etc., the method that the present embodiment is used to obtain human skeleton information is namely based on the detection of target critical point and realizes
's.
The application can use the target critical point in convolutional neural networks acquisition human body image, and the target critical point refers to
Be human skeleton key point, human skeleton key point can be 15 in the present embodiment, and crucial dot sequency is: 1,2 necks
Son, 3 right shoulders, 4 right elbows, 5 right wrists, 6 left shoulders, 7 left elbows, 8 left wrists, 9 right hips, 10 right knees, 11 right ankles, 12 left hips, 13 left knees, 14 is left
Ankle, 15 chests, i.e. human body key point may include 15 key points.Human skeleton key point is also possible to 18, key point
Sequence is: 1 nose, 2 necks, 3 right shoulders, 4 right elbows, 5 right wrists, 6 left shoulders, 7 left elbows, 8 left wrists, 9 right hips, 10 right knees, 11 right ankles, 12
Left hip, 13 left knees, 14 left ankles, 15 left eyes, 16 right eyes, 17 left ears, 18 auris dextras.Similarly, human skeleton key point can be 25,
Its crucial dot sequency is: 1 nose, 2 necks, 3 right shoulders, 4 right elbows, 5 right wrists, 6 left shoulders, 7 left elbows, 8 left wrists, and hip circumference in 9,10 is right
Hip, 11 right knees, 12 right ankles, 13 left hips, 14 left knees, 15 left ankles, 16 left eyes, 17 right eyes, 18 left ears, 19 auris dextras, 20 left hallux toes, 21
Left little toe, 22 left heels, 23 right hallux toes, 24 right little toes, 25 rights crus of diaphragm with.It can be concluded that, human skeleton key point can by above-mentioned
It is multiple to have, it can it is 15,18 or 25, is also possible to others, and the sequence of these multiple key points is also
Different, human skeleton key point is specifically how many here without clearly limiting.
Step S222: framework information feature is determined based on acquired multiple target critical points.
After getting the skeleton key point of children under guardianship in target image, need to determine based on the multiple target critical point
Framework information feature, the framework information feature can be the information for the skeleton in skeleton image region to be described,
The including but not limited to relevant various fundamentals of skeleton, for example, skeleton movement, skeleton profile, frame position, skeleton texture
Deng.
Step S230: the framework information feature is predicted, judges whether there will be tumble event.
Step S240: if it is determined that there will be tumble event, then tumble early warning is exported.
The application can use convolutional neural networks and carry out feature extraction to the framework information in target image, in certain journey
The accuracy rate of prediction can be improved on degree, because convolutional neural networks can extract the feature for utmostly distinguishing things.Separately
Outside, convolutional neural networks can learn image in feature at all levels by the operation of convolution sum pondization automatically, this also complies with people
To the common sense of image understanding, people is Hierarchical abstraction when recognizing image, and what is first understood that is color and brightness, followed by side
The local details feature such as edge, angle point, straight line eventually forms whole followed by the information and structure of the complexity such as texture, geometry
The concept of a object, and convolutional neural networks are then appropriate has merged these features can so make it mention carrying out feature
It is more advantageous when taking, and the feature that it is extracted is also more accurate.It can be improved by the use to convolutional neural networks
To the accuracy rate of tumble event prediction, so as to carry out more accurately protection to people under the guardianship.
When predicting framework information, it will usually the phenomenon that judging by accident, for example, people under the guardianship sits down,
It may be mistaken for falling, in order to which the prediction to framework information is more accurate, the another embodiment of the application proposes one kind and falls
Prediction technique, referring to Fig. 6, the method comprising the steps of as seen in Figure 6 S310 to step S340, detailed following institute
Show:
Step S310: the target image of people under the guardianship is obtained.
Step S320: the framework information feature in the target image is extracted.
Wherein, step S320 may include step S321 to step S322.
Step S321: extracting the target critical point in the target image, and the target critical point is that human skeleton is crucial
Point.
Step S322: framework information feature is determined based on acquired multiple target critical points.
Step S321 to step S322 has been described in detail here with regard to no longer being repeated one by one because above-mentioned.
Step S330: the framework information feature is predicted, judges whether there will be tumble event.
Wherein, step S330 may include step S331 to step S333.
Step S331: obtaining the framework information feature, and the framework information feature includes the skeleton number of target time section
According to.
The skeleton data of the target time section is the coordinate position of human skeleton key point, can be known by above-mentioned introduction
Road human skeleton key point can be 15,18 or 25, be also possible to others, and the sequence of these multiple key points
And it is different, human skeleton key point is specifically how many here without clearly limiting.
The skeleton data of target time section is the coordinate position of human skeleton key point, the coordinate position in the present embodiment
Refer to that the location information of human body key point, format can be (xi,yi,scorei), wherein (xi,yi) it is human body skeletal joint
The coordinate position of point, i refer to the number of skeleton artis, the number of left shoulder are known that by human body key point introduction
It is 6, and the coordinate of 6 left shoulders can then be obtained by the detection of key point.For example, the coordinate of 6 left shoulders can be (261,294).
In addition, score refers to the state of human body key point, in the present embodiment there are three types of the states of human body key point, be respectively it is visible,
It is invisible and not in Tu Nei or unpredictable, i.e., as score=1 human body key point as it can be seen that human body is crucial when score=2
Point is invisible, and human body key point is not in Tu Nei or unpredictable when score=3.For example, in a figure left shoulder be blocked it is invisible
When, then score=2, therefore the coordinate position of 6 left shoulders is equal to (261,294,1) in this illustration, the coordinate of other key points
It is likewise, here with regard to no longer being repeated one by one.
Step S332: the skeleton data of the target time section is analyzed, and obtains the next of the target time section
Period corresponding skeleton data.
Tumble prediction model can be the realization of the method based on Recognition with Recurrent Neural Network, the implementation method packet in the present embodiment
It includes but is not limited to RNN (Recurrent Neural Network, recurrent neural network), LSTM (Long Short Term
Memory, shot and long term memory network) etc..By taking RNN as an example, it is mainly used for processing and forecasting sequence data.The so-called time
Sequence data refers to the data being collected in different time points, and this kind of data reflect a certain things, phenomenon etc. at any time
Variable condition or degree.RNN is the network comprising circulation, allows the persistence of information, its output depends on current input
And memory.
Fig. 7 gives tumble behavior prediction structure chart, and the x in Fig. 7 is input layer, and input layer comes from nerve net for receiving
Information outside network, the data being input in Recognition with Recurrent Neural Network in the present embodiment are images, and described image can be with X-Y scheme
As form is directly inputted in Recognition with Recurrent Neural Network.O is output layer, is mainly used for exporting the calculated result of network, output layer
It is the last layer of neural network, node number is identical as the classification number of bone, can be divided with softmax logistic regression
Class.H is intermediate hidden layer, is mainly used for solving the problems, such as the maximum of linearly inseparable, hidden layer and input layer and output layer
Difference is that he does not receive extraneous signal, also sends signal not directly to the external world.A and B is one section of continuous human skeleton number
According to being also human skeleton information time sequence data, by the analysis of A sections of human skeleton information time sequence datas, can obtain
To B sections of human skeleton information time sequence datas, i.e. what A was indicated is the human skeleton information time sequence data of input, and B table
What is shown is the human skeleton information time sequence data of output.
A sections of human skeleton information time sequence datas (hereinafter referred to as input time sequence A) and B sections of human bodies in the present embodiment
The relationship of skeleton data (hereinafter referred to as output time series B) has the following two kinds possible, and possible relationship 1 is as shown in figure 8, from Fig. 8
It can be seen that two time spans of input time sequence A and output time series B are identical all △ t, the knot of input time sequence A
The beam time is ta, the end time of output time series B is tb；△tpTo be arrived after the complete input prediction network of input time sequence A
Processing time needed for providing prediction result, the prediction network can be Recognition with Recurrent Neural Network；As seen from Figure 8, can
The time range of look-ahead tumble behavior is [△ tMin,△tMax], wherein
ΔtMin=tb-ta-ΔtpΔ t,
ΔtMax=tb-ta-Δtp,
By above-mentioned analysis it is recognised that if input time sequence A and output time series B meets above-mentioned possible relationship
1, then illustrate the function that Recognition with Recurrent Neural Network there can be prediction to fall.
In addition, input time sequence A and output time series B there is likely to be may relationship 2, in detail as shown in figure 9, and
Fig. 8 is similar, and two time spans of input time sequence A and output time series B are identical all △ t, the knot of input time sequence A
The beam time is ta, the end time of output time series B is tb；△tpTo be arrived after the complete input prediction network of input time sequence A
The processing time needed for providing prediction result, as seen from Figure 9, be capable of look-ahead tumble behavior time range be [0,
△tMax], wherein
ΔtMax=tb-ta-Δtp,
It in the case of above-mentioned this relationship, is overlapped because input time sequence A and output time series B exist, works as and fall
Backward is to occur in time point tcWhen before, this Recognition with Recurrent Neural Network is just without the forecast function of tumble behavior.Therefore, this reality
Example is applied in training prediction module, can guarantee to predict that network-Recognition with Recurrent Neural Network has forecast function first.
Step S333: judge whether there will be tumble event according to next period corresponding skeleton data.
The tumble behavior prediction process that Figure 10 shows a kind of tumble prediction technique that the another embodiment of the application proposes is shown
It is intended to, in conjunction with Fig. 8 it is recognised that arriving output time series B, the output time sequence by the way that input time sequence A is available
Arranging B is corresponding skeleton data of next period, if the time series B of output is the state fallen, judgement has tumble
Event occurs, otherwise not having tumble event then, continues to obtain next period human skeleton information time sequence
Data simultaneously make corresponding analysis.As shown in Figure 10, a certain frame image prediction to final tumble behavior, this implementation can be passed through
Example can be by the height of two hip central point of human body from the ground to determine whether will have after getting output time series B
Tumble event occurs, and can be judged by height threshold, i.e., when the height of two hip central point of human body from the ground is less than height
When spending threshold value, then judgement will have tumble event.It should be noted that may also can when people squats down or sits down
There is the case where height of two hip central points from the ground is less than height threshold, so also needing first to judge human body central point before this
Decrease speed whether be greater than critical speed threshold value, if the decrease speed of human body central point be greater than critical speed threshold value and people
The height of two hip central point of body from the ground is less than height threshold, then may determine that tumble event.
Step S340: if it is determined that there will be tumble event, then tumble early warning is exported.
The application can predict human body tumble behavior by convolutional neural networks and Recognition with Recurrent Neural Network, it can
The human skeleton information in image is extracted using convolutional neural networks, and continuous human skeleton information can be inputted
Into Recognition with Recurrent Neural Network, at the same can use the Recognition with Recurrent Neural Network to the behavior of people under the guardianship's next period into
Row prediction.The present embodiment Behavior-based control is prejudged, can active carry out Prevention of fall and tumble information judgement, fall in mitigation
The linkage output that tumble early warning can be carried out while injury is brought to old man.
Please refer to Figure 11, a kind of tumble prediction meanss 400 that the embodiment of the present application proposes, the executing subject of the present apparatus can be with
It is server or gateway, which includes obtaining module 410, and characteristic extracting module 420, tumble prediction module 430 sends mould
Block 440.
Module 410 is obtained, with the target image for obtaining people under the guardianship.
Characteristic extracting module 420, for extracting the framework information feature in the target image.
Extracting the framework information feature in the target image includes: the target critical point extracted in the target image,
The target critical point is human skeleton key point；Framework information feature is determined based on acquired multiple target critical points.
Tumble prediction module 430 judges whether will there is tumble event for predicting the framework information feature
Occur.
The framework information feature is predicted, it may include: to obtain institute that tumble event will be had by, which judging whether,
Framework information feature is stated, the framework information feature includes the skeleton data of target time section；Analyze the target time section
The skeleton data, and obtain the corresponding skeleton data of next period of the target time section；According to the next time
The corresponding skeleton data of section judges whether will there is tumble event.The skeleton data of the target time section is human skeleton
The coordinate position of key point.
Alarm module 440, if it is determined that then exporting tumble early warning for that will have tumble event.
If it is determined that there will be tumble event, then the terminal device of warning information to care provider is exported, and or
Control instruction is sent to anti-fall device, the anti-fall device starting is anti-to fall function to the control instruction for controlling.
Figure 12 is please referred to, a kind of tumble forecasting system 500 that the embodiment of the present application proposes, which includes image collector
510 are set, servers/gateways 520.
Image collecting device 510, with the target image for obtaining people under the guardianship.
Servers/gateways 520, it is special to the framework information for extracting the framework information feature in the target image
Sign is predicted, judges whether will there is tumble event, if it is decided that will have tumble event, then it is pre- to export tumble
Information is alert.
Please refer to Figure 13, a kind of tumble forecasting system 500 that the embodiment of the present application proposes can also include router 530,
Anti-fall device 540 and terminal device 550.
Router 530, the target image for obtaining described image acquisition device 510 are uploaded to server/net
Close 520.
Anti-fall device 540, the control instruction sent for receiving the servers/gateways 520, and start and prevent falling function.
Terminal device 550, the tumble warning information exported for receiving the servers/gateways 520.
Please refer to Figure 14, the structural block diagram of the electronic equipment for the tumble prediction technique that the embodiment of the present application proposes.The electronics
Equipment 600, which can be smart phone, tablet computer, e-book etc., can run the electronic equipment of application program.In the application
Electronic equipment 600 may include one or more such as lower component: processor 610, memory 620 and one or more application
Program, wherein one or more application programs can be stored in memory 620 and be configured as being handled by one or more
Device 610 executes, and the multiple programs of instruction of one or system are configured to carry out the method as described in preceding method embodiment.
Processor 610 may include one or more processing core.Processor 610 is whole using various interfaces and connection
Various pieces in a electronic equipment 600, by run or execute the instruction being stored in memory 620, program, code set or
Instruction set, and the data being stored in memory 620 are called, execute the various functions and processing data of electronic equipment 600.It can
Selection of land, processor 610 can use Digital Signal Processing (Digital Signal Processing, DSP), field-programmable
Gate array (Field-Programmable Gate Array, FPGA), programmable logic array (Programmable Logic
Array, PLA) at least one of example, in hardware realize.Processor 610 can integrating central processor (Central
Processing Unit, CPU), in image processor (Graphics Processing Unit, GPU) and modem etc.
One or more of combinations.Wherein, the main processing operation system of CPU, user interface and application program etc.；GPU is for being responsible for
Show the rendering and drafting of content；Modem is for handling wireless communication.It is understood that above-mentioned modem
It can not be integrated into processor 610, be realized separately through one piece of communication chip.
Memory 620 may include random access memory (Random Access Memory, RAM), also may include read-only
Memory (Read-Only Memory).Memory 620 can be used for store instruction, program, code, code set or instruction set.It deposits
Reservoir 620 may include storing program area and storage data area, wherein storing program area can store for realizing operation, for real
Show the instruction (such as touch function, sound-playing function, image player function etc.) of at least one function, for realizing following each
The instruction etc. of a embodiment of the method.Storage data area can also store the data that terminal electronic device 600 is created in use
(such as phone directory, audio, video data, chat record data) etc..
Figure 15 is please referred to, what the embodiment of the present application proposed realizes falling according to the embodiment of the present application for saving or carrying
The storage unit of the program code of prediction technique.Program code, the journey are stored in the computer readable storage medium 700
Sequence code can be called by processor and execute method described in above method embodiment.
Computer readable storage medium 700 can be such as flash memory, EEPROM (electrically erasable programmable read-only memory),
The electronic memory of EPROM, hard disk or ROM etc.Optionally, computer readable storage medium 700 includes non-transient meter
Calculation machine readable medium (non-transitory computer-readable storage medium).Computer-readable storage
Medium 700 has the memory space for the program code 710 for executing any method and step in the above method.These program codes can
With from reading or be written in one or more computer program product in this one or more computer program product.
Program code 710 can for example be compressed in a suitable form.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.Device class is implemented
For example, since it is basically similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part illustrates.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the device that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including institute
State in process, method, article or the device of element that there is also other identical elements.
Those of ordinary skill in the art will appreciate that realizing that all or part of the steps of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can store in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The above is only the preferred embodiment of the application, not makes any form of restriction to the application, though
Right the application has been disclosed in a preferred embodiment above, however is not limited to the application, any technology people for being familiar with this profession
Member, is not departing within the scope of technical scheme, when the technology contents using the disclosure above are modified or are modified
For the equivalent embodiment of equivalent variations, but all technical spirits pair without departing from technical scheme content, according to the application
Any simple modification, equivalent change and modification made by above embodiments, in the range of still falling within technical scheme.
Claims (10)
1. a kind of tumble prediction technique characterized by comprising
Obtain the target image of people under the guardianship；
Extract the framework information feature in the target image；
The framework information feature is predicted, judges whether there will be tumble event；
If it is determined that there will be tumble event, then tumble early warning is exported.
2. being wrapped the method according to claim 1, wherein extracting the framework information feature in the target image
It includes:
The target critical point in the target image is extracted, the target critical point is human skeleton key point；
Framework information feature is determined based on acquired multiple target critical points.
3. judging whether the method according to claim 1, wherein predicting the framework information feature
There will be tumble event, comprising:
The framework information feature is obtained, the framework information feature includes the skeleton data of target time section；
The skeleton data of the target time section is analyzed, and obtains the corresponding bone of next period of the target time section
Rack data；
Judge whether there will be tumble event according to next period corresponding skeleton data.
4. according to the method described in claim 3, it is characterized in that, the skeleton data of the target time section is human skeleton pass
The coordinate position of key point.
5. if it is determined that the method according to claim 1, wherein will have tumble event, then output is fallen
Early warning, comprising:
If it is determined that there will be tumble event, then the terminal device of warning message to care provider is sent, and/or then send
Control instruction is to anti-fall device, and for controlling, the anti-fall device starting is anti-to fall function to the control instruction.
6. a kind of tumble prediction meanss characterized by comprising
Module is obtained, for obtaining the target image of people under the guardianship；
Characteristic extracting module, for extracting the framework information feature in the target image；
Tumble prediction module judges whether will there is tumble event for predicting the framework information feature；
Alarm module, if it is determined that then exporting tumble early warning for that will have tumble event.
7. a kind of tumble forecasting system, which is characterized in that including image collecting device, server or gateway:
Described image acquisition device, for obtaining the target image of people under the guardianship；
The server or gateway, it is special to the framework information for extracting the framework information feature in the target image
Sign is predicted, judges whether will there is tumble event, if it is decided that will have tumble event, then it is pre- to export tumble
It is alert.
8. system according to claim 7, which is characterized in that the system also includes router, anti-fall device and terminals
Equipment:
The router, the target image for obtaining described image acquisition device are uploaded to server or gateway；
The anti-fall device, the control instruction sent for receiving the server or gateway, and start and prevent falling function；
The terminal device, the tumble early warning sent for receiving the server or gateway.
9. a kind of electronic equipment characterized by comprising
One or more processors；
Memory, for storing one or more programs；
One or more application program, wherein one or more of application programs are stored in the memory and are configured
To be executed by one or more of processors, one or more of programs are configured to carry out as claim 1-5 is any
Method described in.
10. a kind of computer-readable storage medium, which is characterized in that be stored with journey in the computer-readable storage medium
Sequence code, said program code can be called by processor and execute the method according to claim 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910100602.7A CN109920208A (en) | 2019-01-31 | 2019-01-31 | Tumble prediction technique, device, electronic equipment and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910100602.7A CN109920208A (en) | 2019-01-31 | 2019-01-31 | Tumble prediction technique, device, electronic equipment and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109920208A true CN109920208A (en) | 2019-06-21 |
Family
ID=66961293
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910100602.7A Pending CN109920208A (en) | 2019-01-31 | 2019-01-31 | Tumble prediction technique, device, electronic equipment and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109920208A (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110287923A (en) * | 2019-06-29 | 2019-09-27 | 腾讯科技（深圳）有限公司 | Human body attitude acquisition methods, device, computer equipment and storage medium |
| CN110327053A (en) * | 2019-07-12 | 2019-10-15 | 广东工业大学 | A kind of human body behavior safety monitoring method, equipment and system based on lift space |
| CN110390303A (en) * | 2019-07-24 | 2019-10-29 | 深圳前海达闼云端智能科技有限公司 | Tumble alarm method, electronic device, and computer-readable storage medium |
| CN110477925A (en) * | 2019-08-23 | 2019-11-22 | 广东省智能制造研究所 | A kind of fall detection for home for the aged old man and method for early warning and system |
| CN110532850A (en) * | 2019-07-02 | 2019-12-03 | 杭州电子科技大学 | A kind of fall detection method based on video artis and hybrid classifer |
| CN110598633A (en) * | 2019-09-12 | 2019-12-20 | 杭州品茗安控信息技术股份有限公司 | Tumble behavior identification method, device and system |
| CN110686353A (en) * | 2019-09-26 | 2020-01-14 | 珠海格力电器股份有限公司 | Image control method and device, storage medium and air conditioner |
| CN111028912A (en) * | 2019-12-23 | 2020-04-17 | 北京深测科技有限公司 | Environment-assisted life monitoring method and system |
| CN111047827A (en) * | 2019-12-03 | 2020-04-21 | 北京深测科技有限公司 | Intelligent monitoring method and system for environment-assisted life |
| CN111191499A (en) * | 2019-11-11 | 2020-05-22 | 广东工业大学 | Fall detection method and device based on minimum center line |
| CN111460908A (en) * | 2020-03-05 | 2020-07-28 | 中国地质大学（武汉） | Human body tumbling identification method and system based on OpenPose |
| CN111767812A (en) * | 2020-06-18 | 2020-10-13 | 浙江大华技术股份有限公司 | Fall detection method, fall detection device and storage device |
| CN112120703A (en) * | 2019-06-25 | 2020-12-25 | 株式会社日立制作所 | Fall risk assessment method and device |
| CN112287760A (en) * | 2020-09-26 | 2021-01-29 | 浙江汉德瑞智能科技有限公司 | Behavior monitoring-based airport figure risk quantification method |
| CN112364695A (en) * | 2020-10-13 | 2021-02-12 | 杭州城市大数据运营有限公司 | Behavior prediction method and device, computer equipment and storage medium |
| CN112998697A (en) * | 2021-02-22 | 2021-06-22 | 电子科技大学 | Tumble injury degree prediction method and system based on skeleton data and terminal |
| CN113392751A (en) * | 2021-06-10 | 2021-09-14 | 北京华捷艾米科技有限公司 | Tumbling detection method based on human body skeleton nodes and related device |
| CN113936425A (en) * | 2021-09-26 | 2022-01-14 | 上海深豹智能科技有限公司 | Home-based old-age falling detection system and operation method |
| CN114127816A (en) * | 2019-07-18 | 2022-03-01 | 阿特拉斯5D公司 | System and method for on-floor detection without the need for wearable items |
| CN114201029A (en) * | 2020-09-02 | 2022-03-18 | 中国电信股份有限公司 | Action detection method, device and storage medium |
| CN114269243A (en) * | 2020-03-19 | 2022-04-01 | 株式会社日立制作所 | Fall risk evaluation system |
| CN114333238A (en) * | 2021-12-30 | 2022-04-12 | 广东省第二人民医院（广东省卫生应急医院） | Emergency rescue communication system based on 5G |
| CN115868966A (en) * | 2022-12-12 | 2023-03-31 | 北京顺源辰辰科技发展有限公司 | Intelligent action assisting system and intelligent action assisting method |
| CN115969356A (en) * | 2022-12-12 | 2023-04-18 | 北京顺源辰辰科技发展有限公司 | Multimode behavior monitoring method and device based on intelligent sliding rail |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104850846A (en) * | 2015-06-02 | 2015-08-19 | 深圳大学 | Human behavior recognition method and human behavior recognition system based on depth neural network |
| CN105796112A (en) * | 2016-05-11 | 2016-07-27 | 江苏财经职业技术学院 | Old people falling detection monitoring system |
| CN206100025U (en) * | 2016-09-19 | 2017-04-12 | 泰达国际心血管病医院 | Healthy data monitoring intelligent gateway of physiology |
| CN106874847A (en) * | 2017-01-03 | 2017-06-20 | 深圳先进技术研究院 | Tumble Forecasting Methodology and tumble airbag restraint device based on artificial neural network |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
| EP3289975A1 (en) * | 2016-09-04 | 2018-03-07 | Essence Smartcare Ltd | Detecting falls and near falls by ultrasound |
| CN108053469A (en) * | 2017-12-26 | 2018-05-18 | 清华大学 | Complicated dynamic scene human body three-dimensional method for reconstructing and device under various visual angles camera |
| CN108527370A (en) * | 2018-04-16 | 2018-09-14 | 北京卫星环境工程研究所 | The man-machine co-melting safety control system of view-based access control model |
| CN108549841A (en) * | 2018-03-21 | 2018-09-18 | 南京邮电大学 | A kind of recognition methods of the Falls Among Old People behavior based on deep learning |
| CN108657029A (en) * | 2018-05-17 | 2018-10-16 | 华南理工大学 | A kind of driver's seat seat intelligent regulating system and method based on limbs length prediction |
| CN108830150A (en) * | 2018-05-07 | 2018-11-16 | 山东师范大学 | One kind being based on 3 D human body Attitude estimation method and device |
| CN108986169A (en) * | 2018-07-06 | 2018-12-11 | 北京字节跳动网络技术有限公司 | Method and apparatus for handling image |
-
2019
- 2019-01-31 CN CN201910100602.7A patent/CN109920208A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104850846A (en) * | 2015-06-02 | 2015-08-19 | 深圳大学 | Human behavior recognition method and human behavior recognition system based on depth neural network |
| CN105796112A (en) * | 2016-05-11 | 2016-07-27 | 江苏财经职业技术学院 | Old people falling detection monitoring system |
| EP3289975A1 (en) * | 2016-09-04 | 2018-03-07 | Essence Smartcare Ltd | Detecting falls and near falls by ultrasound |
| CN206100025U (en) * | 2016-09-19 | 2017-04-12 | 泰达国际心血管病医院 | Healthy data monitoring intelligent gateway of physiology |
| CN106874847A (en) * | 2017-01-03 | 2017-06-20 | 深圳先进技术研究院 | Tumble Forecasting Methodology and tumble airbag restraint device based on artificial neural network |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
| CN108053469A (en) * | 2017-12-26 | 2018-05-18 | 清华大学 | Complicated dynamic scene human body three-dimensional method for reconstructing and device under various visual angles camera |
| CN108549841A (en) * | 2018-03-21 | 2018-09-18 | 南京邮电大学 | A kind of recognition methods of the Falls Among Old People behavior based on deep learning |
| CN108527370A (en) * | 2018-04-16 | 2018-09-14 | 北京卫星环境工程研究所 | The man-machine co-melting safety control system of view-based access control model |
| CN108830150A (en) * | 2018-05-07 | 2018-11-16 | 山东师范大学 | One kind being based on 3 D human body Attitude estimation method and device |
| CN108657029A (en) * | 2018-05-17 | 2018-10-16 | 华南理工大学 | A kind of driver's seat seat intelligent regulating system and method based on limbs length prediction |
| CN108986169A (en) * | 2018-07-06 | 2018-12-11 | 北京字节跳动网络技术有限公司 | Method and apparatus for handling image |
Non-Patent Citations (10)
| Title |
|---|
| 李旋旋等: "一种基于Kinect的人体摔倒检测方法", 《软件导刊》 * |
| 李旋旋等: "一种基于Kinect的人体摔倒检测方法", 《软件导刊》, vol. 16, no. 7, 31 July 2017 (2017-07-31) * |
| 王君泽等: "基于Kinect骨架追踪的人体摔倒自动检测", 《上海交通大学学报》, vol. 49, no. 9, 30 September 2015 (2015-09-30) * |
| 罗娟等: "基于Kinect平台的独居老人室内智能监控系统设计", 《电脑知识与技术》 * |
| 罗娟等: "基于Kinect平台的独居老人室内智能监控系统设计", 《电脑知识与技术》, vol. 13, no. 13, 31 May 2017 (2017-05-31) * |
| 袁德璋: "基于贝叶斯网络人体摔倒预测的研究", no. 3, pages 4 - 17 * |
| 辛阳等: "大数据技术原理与实践", 北京邮电大学出版社, pages: 139 * |
| 黄河: "基于Kinect的人体跌倒识别的研究与实现" * |
| 黄河: "基于Kinect的人体跌倒识别的研究与实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 01, 15 January 2019 (2019-01-15) * |
| 黄河: "基于Kinect的人体跌倒识别的研究与实现", no. 1, pages 4 - 6 * |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112120703A (en) * | 2019-06-25 | 2020-12-25 | 株式会社日立制作所 | Fall risk assessment method and device |
| CN110287923B (en) * | 2019-06-29 | 2023-09-15 | 腾讯科技（深圳）有限公司 | Human body posture acquisition method, device, computer equipment and storage medium |
| CN110287923A (en) * | 2019-06-29 | 2019-09-27 | 腾讯科技（深圳）有限公司 | Human body attitude acquisition methods, device, computer equipment and storage medium |
| CN110532850B (en) * | 2019-07-02 | 2021-11-02 | 杭州电子科技大学 | Fall detection method based on video joint points and hybrid classifier |
| CN110532850A (en) * | 2019-07-02 | 2019-12-03 | 杭州电子科技大学 | A kind of fall detection method based on video artis and hybrid classifer |
| CN110327053A (en) * | 2019-07-12 | 2019-10-15 | 广东工业大学 | A kind of human body behavior safety monitoring method, equipment and system based on lift space |
| US11948399B2 (en) | 2019-07-18 | 2024-04-02 | Atlas5D, Inc. | Systems and methods for on-the-floor detection without the need for wearables |
| CN114127816A (en) * | 2019-07-18 | 2022-03-01 | 阿特拉斯5D公司 | System and method for on-floor detection without the need for wearable items |
| CN110390303A (en) * | 2019-07-24 | 2019-10-29 | 深圳前海达闼云端智能科技有限公司 | Tumble alarm method, electronic device, and computer-readable storage medium |
| CN110390303B (en) * | 2019-07-24 | 2022-04-08 | 达闼机器人有限公司 | Tumble alarm method, electronic device, and computer-readable storage medium |
| CN110477925A (en) * | 2019-08-23 | 2019-11-22 | 广东省智能制造研究所 | A kind of fall detection for home for the aged old man and method for early warning and system |
| CN110598633B (en) * | 2019-09-12 | 2023-04-07 | 品茗科技股份有限公司 | Tumble behavior identification method, device and system |
| CN110598633A (en) * | 2019-09-12 | 2019-12-20 | 杭州品茗安控信息技术股份有限公司 | Tumble behavior identification method, device and system |
| CN110686353A (en) * | 2019-09-26 | 2020-01-14 | 珠海格力电器股份有限公司 | Image control method and device, storage medium and air conditioner |
| CN111191499A (en) * | 2019-11-11 | 2020-05-22 | 广东工业大学 | Fall detection method and device based on minimum center line |
| CN111191499B (en) * | 2019-11-11 | 2023-05-23 | 广东工业大学 | Fall detection method and device based on minimum center line |
| CN111047827A (en) * | 2019-12-03 | 2020-04-21 | 北京深测科技有限公司 | Intelligent monitoring method and system for environment-assisted life |
| CN111028912A (en) * | 2019-12-23 | 2020-04-17 | 北京深测科技有限公司 | Environment-assisted life monitoring method and system |
| CN111460908A (en) * | 2020-03-05 | 2020-07-28 | 中国地质大学（武汉） | Human body tumbling identification method and system based on OpenPose |
| CN111460908B (en) * | 2020-03-05 | 2023-09-01 | 中国地质大学（武汉） | Human body fall recognition method and system based on OpenPose |
| CN114269243A (en) * | 2020-03-19 | 2022-04-01 | 株式会社日立制作所 | Fall risk evaluation system |
| CN111767812A (en) * | 2020-06-18 | 2020-10-13 | 浙江大华技术股份有限公司 | Fall detection method, fall detection device and storage device |
| CN111767812B (en) * | 2020-06-18 | 2023-04-21 | 浙江大华技术股份有限公司 | Fall detection method, fall detection device and storage device |
| CN114201029A (en) * | 2020-09-02 | 2022-03-18 | 中国电信股份有限公司 | Action detection method, device and storage medium |
| CN112287760A (en) * | 2020-09-26 | 2021-01-29 | 浙江汉德瑞智能科技有限公司 | Behavior monitoring-based airport figure risk quantification method |
| CN112364695A (en) * | 2020-10-13 | 2021-02-12 | 杭州城市大数据运营有限公司 | Behavior prediction method and device, computer equipment and storage medium |
| CN112998697B (en) * | 2021-02-22 | 2022-06-14 | 电子科技大学 | Tumble injury degree prediction method and system based on skeleton data and terminal |
| CN112998697A (en) * | 2021-02-22 | 2021-06-22 | 电子科技大学 | Tumble injury degree prediction method and system based on skeleton data and terminal |
| CN113392751A (en) * | 2021-06-10 | 2021-09-14 | 北京华捷艾米科技有限公司 | Tumbling detection method based on human body skeleton nodes and related device |
| CN113936425A (en) * | 2021-09-26 | 2022-01-14 | 上海深豹智能科技有限公司 | Home-based old-age falling detection system and operation method |
| CN113936425B (en) * | 2021-09-26 | 2024-03-26 | 上海深豹智能科技有限公司 | Fall detection system for home-based aged people and operation method |
| CN114333238A (en) * | 2021-12-30 | 2022-04-12 | 广东省第二人民医院（广东省卫生应急医院） | Emergency rescue communication system based on 5G |
| CN114333238B (en) * | 2021-12-30 | 2024-06-25 | 广东省第二人民医院（广东省卫生应急医院） | Emergency rescue communication system based on 5G |
| CN115969356A (en) * | 2022-12-12 | 2023-04-18 | 北京顺源辰辰科技发展有限公司 | Multimode behavior monitoring method and device based on intelligent sliding rail |
| CN115868966A (en) * | 2022-12-12 | 2023-03-31 | 北京顺源辰辰科技发展有限公司 | Intelligent action assisting system and intelligent action assisting method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109920208A (en) | Tumble prediction technique, device, electronic equipment and system | |
| Kulikajevas et al. | Detection of sitting posture using hierarchical image composition and deep learning | |
| CN112784662A (en) | Video-based fall risk evaluation system | |
| CN109726672B (en) | Tumbling detection method based on human body skeleton sequence and convolutional neural network | |
| CN110472612B (en) | Human behavior recognition method and electronic equipment | |
| CN108875708A (en) | Behavior analysis method, device, equipment, system and storage medium based on video | |
| CN109558865A (en) | A kind of abnormal state detection method to the special caregiver of need based on human body key point | |
| Baldewijns et al. | Bridging the gap between real‐life data and simulated data by providing a highly realistic fall dataset for evaluating camera‐based fall detection algorithms | |
| CN110427834A (en) | A kind of Activity recognition system and method based on skeleton data | |
| CN112949417A (en) | Tumble behavior identification method, equipment and system | |
| CN107480612A (en) | Recognition methods, device and the terminal device of figure action | |
| CN113111865B (en) | Fall behavior detection method and system based on deep learning | |
| CN114511931A (en) | Action recognition method, device and equipment based on video image and storage medium | |
| CN110490055A (en) | A kind of Weakly supervised Activity recognition localization method and device recoded based on three | |
| CN112036267A (en) | Target detection method, device, equipment and computer readable storage medium | |
| Ma et al. | Human motion gesture recognition based on computer vision | |
| Chen et al. | Monitoring the behaviours of pet cat based on YOLO model and raspberry Pi | |
| CN117475258A (en) | Training method of virtual fitting model, virtual fitting method and electronic equipment | |
| CN114241270A (en) | Intelligent monitoring method, system and device for home care | |
| Wu et al. | Video-based fall detection using human pose and constrained generative adversarial network | |
| CN114373225A (en) | Behavior recognition method and system based on human skeleton | |
| JP6822326B2 (en) | Watching support system and its control method | |
| CN109961038A (en) | A kind of children's safety monitoring method and system based on computerized algorithm | |
| CN110348395B (en) | Skeleton behavior identification method based on space-time relationship | |
| Bragaglia et al. | A distributed system using MS kinect and event calculus for adaptive physiotherapist rehabilitation |
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 |