CN109350072A - A kind of cadence detection method based on artificial neural network - Google Patents
A kind of cadence detection method based on artificial neural network Download PDFInfo
- Publication number
- CN109350072A CN109350072A CN201811359980.9A CN201811359980A CN109350072A CN 109350072 A CN109350072 A CN 109350072A CN 201811359980 A CN201811359980 A CN 201811359980A CN 109350072 A CN109350072 A CN 109350072A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- window
- acceleration
- label
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 23
- 230000001133 acceleration Effects 0.000 claims abstract description 72
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims abstract description 4
- 238000005070 sampling Methods 0.000 claims abstract description 3
- 238000002474 experimental method Methods 0.000 claims description 10
- 238000000034 method Methods 0.000 claims description 10
- 210000002569 neuron Anatomy 0.000 claims description 9
- 230000005021 gait Effects 0.000 claims description 8
- 238000012952 Resampling Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 210000004205 output neuron Anatomy 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 2
- 210000002364 input neuron Anatomy 0.000 claims 1
- 238000005516 engineering process Methods 0.000 description 5
- 210000005036 nerve Anatomy 0.000 description 4
- 238000007689 inspection Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000153 supplemental effect Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1118—Determining activity level
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/112—Gait analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Physiology (AREA)
- Veterinary Medicine (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Dentistry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Traffic Control Systems (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention discloses a kind of cadence detection method based on artificial neural network: step 1, acquisition neural network model training data；Step 2, data prediction；Step 3 carries out wave crest detection to the data that step 2 obtains, and is data creating label, and the label of wave crest is 1, and the label of non-wave crest or spurious peaks is 0；Step 4 intercepts training data using sliding window；The wave character of training data in step 5, prominent window；Step 6 takes label of the corresponding label of intermediate data as window data in window；Positive and negative sample proportion in step 7, balance training data；Step 8, neural network model are built and training；When step 9, the detection of online cadence, the acceleration information received is sequentially stored into sliding window, and to the data in window by step 2 and 5 processing；Step 9 treated data are inputted neural network model by step 10；Step 11, sampling time threshold value reject extra beans-and bullets shooter.
Description
[technical field]
The invention belongs to navigator fix fields, and in particular to a kind of cadence detection method based on artificial neural network.
[background technique]
Demand with the fast development of society and science and technology, in people's daily life indoors to based on location-based service
Just becoming increasing, indoor positioning technologies have also obtained extensive research, wherein pedestrian's reckoning (Pedestrian
Dead Reckoning, PDR) technology obtains a large amount of research by its exclusive advantage, and cadence detection method is PDR technology
Vital a part.Popularizing with smart phone simultaneously, people are also more next to health and physical exercise problem
It more pays close attention to, the step-recording method based on smart phone is also more and more, and cadence detection is the basis of step-recording method.
Currently, cadence detection method is based primarily upon acceleration realization, common method has smooth region detection, zero crossings
Detection, peak detection and auto-correlation etc..But they have some defects.The method of most common of them is peak detection, peak value
Detection made a move by detection wave crest or trough to judge whether pedestrian goes, it usually needs a series of supplemental threshold is screened
Peak value is rejected some because of Wave crest and wave trough caused by the reasons such as body arm extremely shake.Therefore peak detection is needed by a large amount of
Data and experiment train these threshold values.The threshold value trained is often fixation, dynamic to handle with fixed threshold value
Acceleration information tends to missing inspection or more inspections.
[summary of the invention]
The object of the present invention is to provide a kind of cadence detection method based on artificial neural network, utilizes neural network algorithm
Powerful capability of fitting handles acceleration information waveform, realizes the purpose of automatic detection cadence.
A kind of cadence detection method based on artificial neural network of the present invention, its steps are as follows:
Step 1, acquisition neural network model training data, pedestrian hold mobile phone and walk at different rates with step-length respectively
Repeatedly, hardware device collected acceleration information when record saves walking.
Step 2, pretreatment, the collected acceleration information of step 1 is pre-processed in accordance with the following steps:
3-axis acceleration is synthesized three axis total accelerations by step 21, and calculation method is as follows:
Wherein aiIndicate the total acceleration data of ith sample point,Indicate ith sample point 3-axis acceleration
Count x, y, the acceleration value of z-axis.
Step 22 carries out smoothing denoising to by step 21 treated acceleration information.
Step 3 finds wave crest in the pretreated data of step 2, and tagged to the acceleration value of then wave crest 1,
It is not the acceleration value tagged 0 of wave crest.By acceleration information waveform visualization, according to known step number, artificial will not
It is that the label 1 of the wave crest of beans-and bullets shooter is changed to 0.In this way, each acceleration value has corresponded to a label.Wherein beans-and bullets shooter, which refers to, to represent
Pedestrian has walked acceleration information time point of a step, as beans-and bullets shooter at the time of generally correspondence using wave crest.
Step 4 intercepts training data using sliding window.Sliding window includes the continuous acceleration information of odd number.Number
It is as follows according to form:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j]
Wherein j indicates the number of midpoint the right and left acceleration information in window, and data amount check 2j+1 is no more than in window
Acceleration information number in one gait cycle is not less than acceleration information number in half of gait cycle, pedestrian's nature row
Gait cycle when walking is generally at 0.40~0.70 second.Therefore, the size of j can be by number in accelerometer sample frequency and window
It is determined according to number range.
The wave character of acceleration information in step 5, prominent window, defines sign function sign:
It enables
Wherein | x | indicate the absolute value of x.
Step 6 takes label of the label (label that step 3 obtains) of central acceleration in window as window data.Number
It is as follows according to form:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j,labeli]
Step 7 carries out resampling or the positive and negative sample weights of adjustment, positive negative sample ratio in balance training data to data set
Example.
Step 8, neural network model are built and training.Artificial neural network uses BP neural network, mode input nerve
First number is acceleration information number in window.The hidden layer number of plies and each hidden layer neuron number are obtained by many experiments
, experiment takes the different number of plies and neuron number every time, takes so that the highest hidden layer number of model prediction accuracy and phase
Neuron number is answered, experiment number is unlimited.Output neuron output valve characterizes the probability that the input waveform is beans-and bullets shooter.It will be through
Cross step 2,3,4,5,6,7 processing obtain training data input neural network model be trained, obtain trained nerve
Network model.
When step 9, the detection of online cadence, according to step 4, acceleration degree that the hardware acceptance carried when pedestrian is walked arrives
According to being sequentially stored into sliding window, and the data in window are successively handled according to step 2 and step 5.
Step 10, by step 9 treated neural network model that data input step 8 obtains.If neural network model
Prediction probability is less than or equal to 0.5, then is directly determined as non-beans-and bullets shooter, if prediction probability is greater than 0.5, according to step 11 place
Reason.
Step 11 rejects extra beans-and bullets shooter.One time threshold δ is setΔt, δΔtTake 2~3 adjacent acceleration sampling times
Interval.With the corresponding time T of intermediate data in windowtSubtract the corresponding time T of a beans-and bullets shootert-1Obtain time difference Δ
T, if Δ t > δΔt, then central acceleration data judging is beans-and bullets shooter in window, is otherwise determined as non-beans-and bullets shooter.
The present invention provides a kind of cadence detection method based on artificial neural network, according to steps 1 and 2,3,4,5,6,7,8
Obtain cadence detection neural network model.When online cadence detection, according to step 9,10,11, cadence detection is completed.
A kind of cadence detection method based on artificial neural network proposed by the present invention, compared with prior art, advantage
It is: 1, on the basis of remaining acceleration information order of magnitude, introduces left and right acceleration value and intermediate acceleration in window
Angle value subtract each other after symbol, strengthen wave character, more efficient can extract walk brief acceleration data of pedestrian and reflect
Gait feature；2, the powerful capability of fitting of artificial neural network is utilized in model, while being aided with time threshold δΔt, reached more
Add the ability for accurately identifying cadence.If not departing from the scope of the invention to various changes of the invention and deformation, still fall within
Within the scope of claim and equivalent technology of the invention.
[Detailed description of the invention]
Fig. 1 is a kind of cadence detection method flow diagram based on artificial neural network of the present invention.
[specific embodiment]
The invention proposes a kind of cadence detection method based on artificial neural network, flow chart is as shown in Fig. 1, it
Include following 11 steps:
Step 1, acquisition neural network model training data, pedestrian hold mobile phone and walk at different rates with step-length respectively
Repeatedly, hardware device collected acceleration information when record saves walking.
Step 2, pretreatment, the collected acceleration information of step 1 is handled in accordance with the following steps:
3-axis acceleration is synthesized three axis total accelerations by step 21, and calculation method is as follows:
Wherein aiIndicate the total acceleration data of ith sample point,Three axis of ith sample point is respectively indicated to add
Speedometer x, y, the acceleration value of z-axis.
Step 22 carries out smoothing denoising to by step 21 treated acceleration information.
Step 3 finds wave crest in the pretreated data of step 2, and tagged to the acceleration value of then wave crest 1,
It is not the acceleration value tagged 0 of wave crest.Data waveform is visualized, according to known step number, artificial will not be beans-and bullets shooter
The label 1 of wave crest be changed to 0.In this way, each acceleration value has corresponded to a label.Wherein beans-and bullets shooter, which refers to, can represent pedestrian's row
The acceleration information time point for having walked a step, as beans-and bullets shooter at the time of generally correspondence using wave crest.
Step 4 intercepts training data using sliding window.Sliding window includes the continuous acceleration information of odd number.Number
It is as follows according to form:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j]
Wherein j indicates the number of midpoint the right and left acceleration information in window, and data amount check 2j+1 is no more than in window
Acceleration information number in walking period is not less than acceleration information number in half of walking period, when pedestrian walks naturally
Gait cycle is generally at 0.40~0.70 second.Therefore, the size of j can be by data amount check in accelerometer sample frequency and window
Range determines.
The wave character of acceleration information in step 5, prominent window, defines sign function sign:
It enables
Wherein | x | indicate the absolute value of x.
Step 6 takes mark of the corresponding label of central acceleration in window (label that step 3 obtains) as window data
Label.Data mode is as follows:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j,labeli]
Step 7, to data set resampling or the positive and negative sample weights of adjustment, positive and negative sample proportion in balance training data.
Step 8, neural network model are built and training.Artificial neural network uses BP neural network, mode input nerve
First number is acceleration information number in window.The hidden layer number of plies and each hidden layer neuron number are obtained by many experiments
, experiment takes the different number of plies and neuron number every time, takes so that the highest hidden layer number of model prediction accuracy and phase
Neuron number is answered, experiment number is unlimited.Output neuron output valve characterizes the probability that the input waveform is beans-and bullets shooter.It will be through
Cross step 2,3,4,5,6,7 processing obtain training data input neural network model be trained, obtain trained nerve
Network model.
When step 9, the detection of online cadence, according to step 4, acceleration degree that the hardware acceptance carried when pedestrian is walked arrives
According to being sequentially stored into sliding window, and the data in window are successively handled according to step 2 and step 5.
Step 10, by step 9 treated neural network model that data input step 8 obtains.If neural network model
Prediction probability is less than or equal to 0.5, then is directly determined as non-beans-and bullets shooter, if prediction probability is greater than 0.5, according to step 11 place
Reason.
Step 11, with the corresponding time T of intermediate data in windowtSubtract the corresponding time T of a beans-and bullets shootert-1When obtaining
Between poor Δ t, if Δ t > δΔt, then central acceleration data judging is beans-and bullets shooter in window, and records Tt.Otherwise it is determined as non-beans-and bullets shooter.
Claims (5)
1. a kind of cadence detection method based on artificial neural network, it is characterised in that: the method steps are as follows:
Step 1, acquisition neural network model training data, it is more with step-length walking at different rates respectively that pedestrian holds mobile phone
It is secondary, hardware device collected acceleration information when record saves walking；
Step 2, pretreatment；
Step 3 finds wave crest in the pretreated data of step 2, and tagged to the acceleration value of then wave crest 1, be not
The acceleration value of wave crest tagged 0；By acceleration information waveform visualization, according to known step number, artificial will not be step
The label 1 of the wave crest of point is changed to 0；In this way, each acceleration value has corresponded to a label；
Step 4 intercepts training data using sliding window；
The wave character of acceleration information in step 5, prominent window, defines sign function sign:
It enables
Wherein | x | indicate the absolute value of x；
Step 6 takes the label of central acceleration in window as the label of window data；Data mode is as follows:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j,labeli]
Step 7 carries out resampling or the positive and negative sample weights of adjustment, positive and negative sample proportion in balance training data to data set；
Step 8, neural network model are built and training；
When step 9, the detection of online cadence, according to step 4, acceleration information that the hardware acceptance carried when pedestrian is walked arrives according to
In secondary deposit sliding window, and the data in window are successively handled according to step 2 and step 5；
Step 10, by step 9 treated neural network model that data input step 8 obtains；If Neural Network model predictive
Probability is less than or equal to 0.5, then is directly determined as non-beans-and bullets shooter, if prediction probability is greater than 0.5, handles according to step 11；
Step 11 rejects extra beans-and bullets shooter.
2. a kind of cadence detection method based on artificial neural network according to claim 1, it is characterised in that: the step
Rapid 2, it specifically includes and pre-processes the collected acceleration information of step 1 in accordance with the following steps:
3-axis acceleration is synthesized three axis total accelerations by step 21, and calculation method is as follows:
Wherein aiIndicate the total acceleration data of ith sample point,Indicate ith sample point three axis accelerometer x,
Y, the acceleration value of z-axis；
Step 22 carries out smoothing denoising to by step 21 treated acceleration information.
3. a kind of cadence detection method based on artificial neural network according to claim 1, it is characterised in that: the step
Rapid 4 detailed process is as follows: sliding window includes the continuous acceleration information of odd number, and data mode is as follows:
[ai-j,…,ai-2,ai-1,ai,ai+1,ai+2,…,ai+j]
Wherein j indicates the number of midpoint the right and left acceleration information in window, and data amount check 2j+1 is no more than one in window
Acceleration information number in gait cycle is not less than acceleration information number in half of gait cycle, when pedestrian walks naturally
Gait cycle generally at 0.40~0.70 second；Therefore, the size of j can be by data in accelerometer sample frequency and window
Number range determines.
4. a kind of cadence detection method based on artificial neural network according to claim 1, it is characterised in that: the step
Rapid 8 detailed process is as follows: artificial neural network uses BP neural network, and mode input neuron number is acceleration in window
Data amount check；The hidden layer number of plies and each hidden layer neuron number are obtained by many experiments, and experiment takes different layers every time
Several and neuron number takes so that the highest hidden layer number of model prediction accuracy and corresponding neuron number, experiment
Number is unlimited；Output neuron output valve characterizes the probability that the input waveform is beans-and bullets shooter；It will be by step 2,3,4,5,6,7 processing
Obtained training data input neural network model is trained, and obtains trained neural network model.
5. a kind of cadence detection method based on artificial neural network according to claim 1, it is characterised in that: the step
Rapid 11 detailed process is as follows: one time threshold δ of settingΔt, δΔt2~3 adjacent acceleration sampling time intervals are taken, window is used
The corresponding time T of interior intermediate datatSubtract the corresponding time T of a beans-and bullets shootert-1Time difference Δ t is obtained, if Δ t > δΔt, then window
Central acceleration data judging is beans-and bullets shooter in mouthful, is otherwise determined as non-beans-and bullets shooter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811359980.9A CN109350072B (en) | 2018-11-15 | 2018-11-15 | Step frequency detection method based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811359980.9A CN109350072B (en) | 2018-11-15 | 2018-11-15 | Step frequency detection method based on artificial neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109350072A true CN109350072A (en) | 2019-02-19 |
| CN109350072B CN109350072B (en) | 2020-08-04 |
Family
ID=65345227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811359980.9A Active CN109350072B (en) | 2018-11-15 | 2018-11-15 | Step frequency detection method based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109350072B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110553643A (en) * | 2019-09-17 | 2019-12-10 | 电子科技大学 | pedestrian self-adaptive zero-speed updating point selection method based on neural network |
| WO2021237659A1 (en) * | 2020-05-29 | 2021-12-02 | Beijing Didi Infinity Technology And Development Co., Ltd. | Indoor navigation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070068244A1 (en) * | 2003-10-17 | 2007-03-29 | M.B.T.L. Limited | Measuring forces in athletics |
| EP1691170B1 (en) * | 2005-02-11 | 2009-06-10 | Samsung Electronics Co., Ltd. | Stride-based route guiding apparatus and method |
| CN102646198A (en) * | 2012-02-21 | 2012-08-22 | 温州大学 | Mode recognition method of mixed linear SVM (support vector machine) classifier with hierarchical structure |
| CN103371814A (en) * | 2012-04-14 | 2013-10-30 | 兰州大学 | Remote wireless electrocardiograph monitoring system and feature extraction method on basis of intelligent diagnosis |
| CN104567912A (en) * | 2015-02-02 | 2015-04-29 | 河海大学 | Method for realizing pedometer on Android mobile phone |
| CN107091650A (en) * | 2017-04-27 | 2017-08-25 | 重庆邮电大学 | A kind of software step-recording method based on mobile phone acceleration and range sensor |
| CN107462258A (en) * | 2017-07-13 | 2017-12-12 | 河海大学 | A kind of step-recording method based on mobile phone 3-axis acceleration sensor |
-
2018
- 2018-11-15 CN CN201811359980.9A patent/CN109350072B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070068244A1 (en) * | 2003-10-17 | 2007-03-29 | M.B.T.L. Limited | Measuring forces in athletics |
| EP1691170B1 (en) * | 2005-02-11 | 2009-06-10 | Samsung Electronics Co., Ltd. | Stride-based route guiding apparatus and method |
| CN102646198A (en) * | 2012-02-21 | 2012-08-22 | 温州大学 | Mode recognition method of mixed linear SVM (support vector machine) classifier with hierarchical structure |
| CN103371814A (en) * | 2012-04-14 | 2013-10-30 | 兰州大学 | Remote wireless electrocardiograph monitoring system and feature extraction method on basis of intelligent diagnosis |
| CN104567912A (en) * | 2015-02-02 | 2015-04-29 | 河海大学 | Method for realizing pedometer on Android mobile phone |
| CN107091650A (en) * | 2017-04-27 | 2017-08-25 | 重庆邮电大学 | A kind of software step-recording method based on mobile phone acceleration and range sensor |
| CN107462258A (en) * | 2017-07-13 | 2017-12-12 | 河海大学 | A kind of step-recording method based on mobile phone 3-axis acceleration sensor |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110553643A (en) * | 2019-09-17 | 2019-12-10 | 电子科技大学 | pedestrian self-adaptive zero-speed updating point selection method based on neural network |
| CN110553643B (en) * | 2019-09-17 | 2021-12-21 | 电子科技大学 | Pedestrian self-adaptive zero-speed updating point selection method based on neural network |
| WO2021237659A1 (en) * | 2020-05-29 | 2021-12-02 | Beijing Didi Infinity Technology And Development Co., Ltd. | Indoor navigation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109350072B (en) | 2020-08-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106096662B (en) | Human motion state identification based on acceleration transducer | |
| CN107240122A (en) | Video target tracking method based on space and time continuous correlation filtering | |
| CN110505583A (en) | A kind of path matching algorithm based on bayonet data and signaling data | |
| CN106599922A (en) | Transfer learning method and transfer learning system for large-scale data calibration | |
| CN107358250A (en) | Body gait recognition methods and system based on the fusion of two waveband radar micro-doppler | |
| CN112799128B (en) | Method for seismic signal detection and seismic phase extraction | |
| CN106970379B (en) | Based on Taylor series expansion to the distance-measuring and positioning method of indoor objects | |
| CN108537101B (en) | Pedestrian positioning method based on state recognition | |
| CN106779086A (en) | A kind of integrated learning approach and device based on Active Learning and model beta pruning | |
| CN102198003A (en) | Limb movement detection and evaluation network system and method | |
| CN114152980B (en) | Method and device for rapidly and automatically producing seismic source mechanism solution | |
| CN112653991A (en) | WLAN indoor positioning method of TebNet neural network model based on deep learning | |
| Yang et al. | GPS and acceleration data in multimode trip data recognition based on wavelet transform modulus maximum algorithm | |
| CN110210550A (en) | Image fine granularity recognition methods based on integrated study strategy | |
| CN107895014A (en) | A kind of time series bridge monitoring data analysing method based on MapReduce frameworks | |
| CN105654516A (en) | Method for detecting small moving object on ground on basis of satellite image with target significance | |
| CN109350072A (en) | A kind of cadence detection method based on artificial neural network | |
| CN109637126A (en) | A kind of traffic object identifying system and its method based on V2X terminal | |
| CN108629295A (en) | Corner terrestrial reference identification model training method, the recognition methods of corner terrestrial reference and device | |
| CN110044375A (en) | A kind of novel step-recording method based on accelerometer | |
| CN110163264B (en) | Walking pattern recognition method based on machine learning | |
| CN103605960B (en) | A kind of method for identifying traffic status merged based on different focal video image | |
| CN108717548A (en) | A kind of increased Activity recognition model update method of facing sensing device dynamic and system | |
| CN108154199B (en) | High-precision rapid single-class target detection method based on deep learning | |
| CN116758479B (en) | Coding deep learning-based intelligent agent activity recognition method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210201 Address after: 315100 b-211-47, Kexin building, 655 bachelor Road, Yinzhou District, Ningbo City, Zhejiang Province Patentee after: NINGBO ZHIZHENG WEIYING INFORMATION TECHNOLOGY Co.,Ltd. Address before: 100191 No. 37, Haidian District, Beijing, Xueyuan Road Patentee before: BEIHANG University |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211124 Address after: 314500 01, No. 4, South Zaoqiang street, No. 1, Nanmen Gongnong Road, Chongfu Town, Tongxiang City, Jiaxing City, Zhejiang Province Patentee after: Jiaxing Qiyuan Network Information Technology Co.,Ltd. Address before: 315100 b-211-47, Kexin building, 655 bachelor Road, Yinzhou District, Ningbo City, Zhejiang Province Patentee before: NINGBO ZHIZHENG WEIYING INFORMATION TECHNOLOGY Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240516 Address after: 100010 Zuoanmen Neizuoanyiyuan 1-5-1301, Dongcheng District, Beijing Patentee after: Zhou Sihua Country or region after: China Address before: 314500 01, No. 4, South Zaoqiang street, No. 1, Nanmen Gongnong Road, Chongfu Town, Tongxiang City, Jiaxing City, Zhejiang Province Patentee before: Jiaxing Qiyuan Network Information Technology Co.,Ltd. Country or region before: China |
|
| TR01 | Transfer of patent right |