CN107862302A - A kind of human motion detecting system and method based on semi-supervised learning - Google Patents
A kind of human motion detecting system and method based on semi-supervised learning Download PDFInfo
- Publication number
- CN107862302A CN107862302A CN201711228115.6A CN201711228115A CN107862302A CN 107862302 A CN107862302 A CN 107862302A CN 201711228115 A CN201711228115 A CN 201711228115A CN 107862302 A CN107862302 A CN 107862302A
- Authority
- CN
- China
- Prior art keywords
- state
- human motion
- semi
- methods
- classification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The invention discloses a kind of human motion detecting system and method based on semi-supervised learning, it is related to data message categorizing system.In the present invention:In sensor monitoring part, human motion state is accordingly monitored by sensor;In state parameter part, include the actuating range value of corresponding state, including the bound of state range value is adjusted;In semi-supervised learning part, including the analysis matching operation of data message that Study strategies and methods and Study strategies and methods arrive to Sensor monitoring.The present invention is by establishing default human action parameter area value, utilize sensor monitoring and video surveillance mode, Correct Analysis is carried out to human motion state, judge state classification accuracy of the Study strategies and methods to data sample, correcting mode is adjusted by parameter area, accurately state parameter scope is established out, so as to efficiently accurately carry out human motion state classification to a large amount of follow-up unknown samples.
Description
Technical field
The present invention relates to data message categorizing system field, more particularly to a kind of human motion inspection based on semi-supervised learning
Examining system and method.
Background technology
Machine learning is a kind of intelligence science, and the field is mainly studied is calculated based on the computer of sample data and passing experience
Method, simulated using computer and realize human perception, study, the behavior such as differentiation.Machine learning is that artificial intelligence field is ground
The core studied carefully, by being analyzed empirical data and the optimization of learning strategy, with the mesh of being optimal organization knowledge structure
's.
Semi-supervised learning is a kind of important form of machine learning, and data message sample is divided by semi-supervised learning
Class.Traditional classification carries out classification learning and has only focused on the training that labeled data collection is used for model, but the acquisition of flag data
The problems such as big, human input is larger is spent with extremely difficult, time, while data labeling process is one needs have experience to know
The process that the people of knowledge plays an active part in;On the contrary, Unlabeled data is often easier to collect, but not too many method and strategy
The information provided using these Unlabeled datas is provided.Semi-supervised learning take into account using a large amount of Unlabeled datas, in combination with
Marked data be used for establish the more preferable grader of performance;Therefore semi-supervised learning needs to use less manpower just can obtain
Larger grader precision improves, and it has obtained extensive concern in theoretical research and industrial practice.Semi-supervised learning is not only
Cover semisupervised classification, while be also applicable in all shape changeables such as semi-supervised clustering, Semi-Supervised Regression.
In life, many people can carry out corresponding activity, in active procedure is carried out, be broadly divided into standing activity, OK
Activity and running activities are walked, these activities are taken exercise and carry out corresponding sensing analysis, turns into and effectively carries out function of human body monitoring
Important reference foundation, and the mode classification of semi-supervised learning is used, parameter feedback preferably can be carried out to human body movement data
Amendment, so as to complete the accurate matching to follow-up a large amount of unknown samples.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of human motion detecting system based on semi-supervised learning and side
Method, by establishing default human action parameter area value, using sensor monitoring and video surveillance mode, to human motion state
Correct Analysis is carried out, judges state classification accuracy of the Study strategies and methods to data sample, is adjusted and corrected by parameter area
Mode, accurately state parameter scope is established out, so as to efficiently accurately carry out human motion shape to a large amount of follow-up unknown samples
State is classified.
In order to solve the above technical problems, the present invention is achieved by the following technical solutions:
The present invention provides a kind of human motion detecting system based on semi-supervised learning:
Including sensor monitoring part, state parameter part and semi-supervised learning part and video surveillance analysis part.
In sensor monitoring part:Human motion state is accordingly monitored by sensor.
In state parameter part:Actuating range value including corresponding state, including the bound of state range value is adjusted.
In semi-supervised learning part:The data message arrived including Study strategies and methods and Study strategies and methods to Sensor monitoring
Analysis matching operation.
In video surveillance analysis part:Human motion is accordingly monitored including video frequency monitoring system.
Wherein, state parameter part includes some default value parts and the restriction corresponding with preset value adjustment value part.
Wherein, time synchronized module is set between Study strategies and methods and video frequency monitoring system.
Wherein, corresponding state classification is set in Study strategies and methods, and state classification includes the standing activity of human body, walking
State and state of running.
A kind of body movement detection method based on semi-supervised learning:
The first step, primary data sample is established, initial data is imported into Study strategies and methods, phase is carried out to primary data sample
The action decision answered, mark off the primary data of actuating range value;Second step, sensor is detected by Study strategies and methods
The data sample arrived carries out status information classification, and the sample data in the range of corresponding data is categorized into corresponding state classification
It is interior;3rd step, detection observation, the passage time method of synchronization, analytics are carried out to human motion state by video frequency monitoring system
Practise the state classification result of the data sample arrived in grader by Sensor monitoring;4th step, to the state of Study strategies and methods
Classification results are judged, respective stored is carried out to the status data sample of correct state classification;5th step, to making a fault point
The status data sample of class is analyzed, the human motion state arrived by video surveillance, by the status data sample of classification of slipping up
Originally it is divided into the range of corresponding correct state classification, and behaviour is adjusted correspondingly to the division data value of reset condition classification
Make.
Compared with prior art, the beneficial effects of the invention are as follows:
The present invention pre-sets out corresponding state classification, passes through biography by establishing default human action parameter area value
Sensor carries out sensor monitoring to human body sport parameter, and the state ownership of data sample is carried out by Study strategies and methods, then passes through
The video surveillance mode of time synchronized, Correct Analysis is carried out to human motion state, judges Study strategies and methods to data sample
State classification accuracy, correcting mode is adjusted by parameter area, accurately state parameter scope is established out, so as to high-efficiency precision
Accurate carries out human motion state classification to a large amount of follow-up unknown samples.
Brief description of the drawings
Fig. 1 is the human motion detecting system structural representation based on semi-supervised learning of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Specific embodiment one:
The present invention is a kind of human motion detecting system based on semi-supervised learning, including sensor monitoring part, state ginseng
Number part and semi-supervised learning part and video surveillance analysis part.
In sensor monitoring part:Human motion state is accordingly monitored by sensor.
In state parameter part:Actuating range value including corresponding state, including the bound of state range value is adjusted.
In semi-supervised learning part:The data message arrived including Study strategies and methods and Study strategies and methods to Sensor monitoring
Analysis matching operation.
In video surveillance analysis part:Human motion is accordingly monitored including video frequency monitoring system.
Further, state parameter part includes some default value parts and the restriction adjusted value portion corresponding with preset value
Point.
Further, time synchronized module is set between Study strategies and methods and video frequency monitoring system.
Further, corresponding state classification is set in Study strategies and methods, state classification include human body standing activity,
Walking states and state of running.
A kind of body movement detection method based on semi-supervised learning:
The first step, primary data sample is established, initial data is imported into Study strategies and methods, phase is carried out to primary data sample
The action decision answered, mark off the primary data of actuating range value;Second step, sensor is detected by Study strategies and methods
The data sample arrived carries out status information classification, and the sample data in the range of corresponding data is categorized into corresponding state classification
It is interior;3rd step, detection observation, the passage time method of synchronization, analytics are carried out to human motion state by video frequency monitoring system
Practise the state classification result of the data sample arrived in grader by Sensor monitoring;4th step, to the state of Study strategies and methods
Classification results are judged, respective stored is carried out to the status data sample of correct state classification;5th step, to making a fault point
The status data sample of class is analyzed, the human motion state arrived by video surveillance, by the status data sample of classification of slipping up
Originally it is divided into the range of corresponding correct state classification, and behaviour is adjusted correspondingly to the division data value of reset condition classification
Make.
Specific embodiment two:
As shown in figure 1, establishing parameter values A, B, C, D, E, and parameter values are transferred to study point
Class device, the Study strategies and methods value range initial to this five carry out state classification.Wherein, A parameter areas correspond to standing activity shape
State, B and C parameter areas correspond to walking states, D and E parameter areas are correspondingly run state.
When human body is moved, the sensor being installed on human body carries out sensor monitoring to physical activity;At the same time,
Video monitoring device synchronizes recording to the state of human body now;Meanwhile Study strategies and methods get what sensor transmissions were come
Data acquisition time and data sample are synchronized storage, Study strategies and methods by Sensor monitoring data sample, Study strategies and methods
Data sample is subjected to state classification.
After Study strategies and methods carry out original state distribution to data sample, human body active state video is transferred, human body is lived
Dynamic state carries out auxiliary judgment, and the state classification to Study strategies and methods judges;If Study strategies and methods are to data sample
Classification is correct, then corresponding data is divided into values;If classification error of the Study strategies and methods to data sample, right
Data sample is divided into correct activity state classification, and the parameter area of corresponding state classification is adjusted, such as
The parameter area adjustment c carried out in Fig. 1 between B and D;After carrying out multiple sample training, Study strategies and methods can be accurately to rear
Continuous a large amount of Unlabeled datas carry out efficiently accurately state classification.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.
Claims (5)
- A kind of 1. human motion detecting system based on semi-supervised learning, it is characterised in that:Including sensor monitoring part, state parameter part and semi-supervised learning part and video surveillance analysis part;In sensor monitoring part:Human motion state is accordingly monitored by sensor;In state parameter part:Actuating range value including corresponding state, including the bound of state range value is adjusted;In semi-supervised learning part:Point of the data message arrived including Study strategies and methods and Study strategies and methods to Sensor monitoring Analyse matching operation;In video surveillance analysis part:Human motion is accordingly monitored including video frequency monitoring system.
- A kind of 2. human motion detecting system based on semi-supervised learning according to claim 1, it is characterised in that:It is described State parameter part includes some default value parts and the restriction corresponding with preset value adjustment value part.
- A kind of 3. human motion detecting system based on semi-supervised learning according to claim 1, it is characterised in that:It is described Time synchronized module is set between Study strategies and methods and video frequency monitoring system.
- A kind of 4. human motion detecting system based on semi-supervised learning according to claim 1, it is characterised in that:It is described Corresponding state classification is set in Study strategies and methods, and state classification includes the standing activity, walking states and shape of running of human body State.
- A kind of 5. body movement detection method based on semi-supervised learning as claimed in claim 1, it is characterised in that:The first step, primary data sample is established, initial data is imported into Study strategies and methods, primary data sample carried out corresponding Decision is acted, marks off the primary data of actuating range value;Second step, the data sample detected by Study strategies and methods to sensor carries out status information classification, by corresponding data In the range of sample data be categorized into corresponding in state classification;3rd step, detection observation, the passage time method of synchronization, analytics are carried out to human motion state by video frequency monitoring system Practise the state classification result of the data sample arrived in grader by Sensor monitoring;4th step, the state classification result to Study strategies and methods judge, the status data sample of correct state classification is entered Row respective stored;5th step, the status data sample for classification of making a fault is analyzed, the human motion state arrived by video surveillance, The status data sample for classification of slipping up is divided into the range of corresponding correct state classification, and to the division of reset condition classification Data value is adjusted correspondingly operation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711228115.6A CN107862302A (en) | 2017-11-29 | 2017-11-29 | A kind of human motion detecting system and method based on semi-supervised learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711228115.6A CN107862302A (en) | 2017-11-29 | 2017-11-29 | A kind of human motion detecting system and method based on semi-supervised learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107862302A true CN107862302A (en) | 2018-03-30 |
Family
ID=61704577
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711228115.6A Pending CN107862302A (en) | 2017-11-29 | 2017-11-29 | A kind of human motion detecting system and method based on semi-supervised learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107862302A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110378291A (en) * | 2019-07-22 | 2019-10-25 | 浙江大学 | A kind of characteristics of human body's parameter prediction method based on semi-supervised learning |
EP4048148A4 (en) * | 2019-10-25 | 2023-12-06 | Plethy, Inc. | Systems and methods for assessing gait, stability, and/or balance of a user |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103961109A (en) * | 2014-05-05 | 2014-08-06 | 北京航空航天大学 | Human body posture detection device based on acceleration signals and angular speed signals |
CN105095863A (en) * | 2015-07-14 | 2015-11-25 | 西安电子科技大学 | Similarity-weight-semi-supervised-dictionary-learning-based human behavior identification method |
CN105361880A (en) * | 2015-11-30 | 2016-03-02 | 上海乃欣电子科技有限公司 | Muscle movement event recognition system and method |
CN105771087A (en) * | 2016-04-06 | 2016-07-20 | 上海乃欣电子科技有限公司 | Rehabilitation training system based on music and myoelectricity feedback simulating |
CN106897738A (en) * | 2017-01-22 | 2017-06-27 | 华南理工大学 | A kind of pedestrian detection method based on semi-supervised learning |
CN107202594A (en) * | 2017-05-11 | 2017-09-26 | 深圳合优科技有限公司 | A kind of data processing method and device based on MTK platforms |
-
2017
- 2017-11-29 CN CN201711228115.6A patent/CN107862302A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103961109A (en) * | 2014-05-05 | 2014-08-06 | 北京航空航天大学 | Human body posture detection device based on acceleration signals and angular speed signals |
CN105095863A (en) * | 2015-07-14 | 2015-11-25 | 西安电子科技大学 | Similarity-weight-semi-supervised-dictionary-learning-based human behavior identification method |
CN105361880A (en) * | 2015-11-30 | 2016-03-02 | 上海乃欣电子科技有限公司 | Muscle movement event recognition system and method |
CN105771087A (en) * | 2016-04-06 | 2016-07-20 | 上海乃欣电子科技有限公司 | Rehabilitation training system based on music and myoelectricity feedback simulating |
CN106897738A (en) * | 2017-01-22 | 2017-06-27 | 华南理工大学 | A kind of pedestrian detection method based on semi-supervised learning |
CN107202594A (en) * | 2017-05-11 | 2017-09-26 | 深圳合优科技有限公司 | A kind of data processing method and device based on MTK platforms |
Non-Patent Citations (1)
Title |
---|
张宜浩 等: "《基于半监督学习的个性化推荐算法研究》", 31 May 2016 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110378291A (en) * | 2019-07-22 | 2019-10-25 | 浙江大学 | A kind of characteristics of human body's parameter prediction method based on semi-supervised learning |
CN110378291B (en) * | 2019-07-22 | 2021-06-25 | 浙江大学 | Human body characteristic parameter prediction method based on semi-supervised learning |
EP4048148A4 (en) * | 2019-10-25 | 2023-12-06 | Plethy, Inc. | Systems and methods for assessing gait, stability, and/or balance of a user |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103324937A (en) | Method and device for labeling targets | |
CN109815785A (en) | A kind of face Emotion identification method based on double-current convolutional neural networks | |
CN107942940A (en) | A kind of detection method and device of the feed shaft assembly failure of the numerically-controlled machine tool based on instruction domain analysis | |
CN109658387A (en) | The detection method of the pantograph carbon slide defect of power train | |
CN103630244A (en) | Equipment fault diagnosis method and system of electric power system | |
CN108761237A (en) | Unmanned plane electric inspection process image vital electrical component diagnoses automatically and labeling system | |
CN107862302A (en) | A kind of human motion detecting system and method based on semi-supervised learning | |
US10713965B2 (en) | Negative learning behavior alert system | |
CN109859554A (en) | Adaptive english vocabulary learning classification pushes away topic device and computer learning system | |
CN102496061A (en) | Neural network sample selection method and device based on active learning | |
CN111753877B (en) | Product quality detection method based on deep neural network migration learning | |
Atikuzzaman et al. | Human activity recognition system from different poses with cnn | |
Dong et al. | Lightweight and efficient neural network with SPSA attention for wheat ear detection | |
Bordoloi et al. | Policy interactions with research trajectories: The case of cyber-physical convergence in manufacturing and industrials | |
CN111027941A (en) | Teaching experiment platform based on STM32 singlechip | |
CN105117980B (en) | A kind of automatic evaluation method of grid equipment state | |
CN109584208A (en) | A kind of method of inspection for industrial structure defect intelligent recognition model | |
CN107146027A (en) | A kind of factory's intelligent early-warning system | |
CN112147221A (en) | Steel rail screw hole crack identification method and system based on ultrasonic flaw detector data | |
Hurd et al. | Quality assurance in additive manufacturing through mobile computing | |
CN106250924B (en) | A kind of newly-increased category detection method based on multi-instance learning | |
CN114154571B (en) | Intelligent auxiliary labeling method and system for image | |
CN115630299A (en) | Rotary machine fault diagnosis method and system based on joint domain adaptive network | |
CN105373780A (en) | Footprint recognition method in rat gait analysis, device and system | |
CN113361520B (en) | Transmission line equipment defect detection method based on sample offset network |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180330 |
|
RJ01 | Rejection of invention patent application after publication |