CN112163592B - Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof - Google Patents
Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof Download PDFInfo
- Publication number
- CN112163592B CN112163592B CN202010870729.XA CN202010870729A CN112163592B CN 112163592 B CN112163592 B CN 112163592B CN 202010870729 A CN202010870729 A CN 202010870729A CN 112163592 B CN112163592 B CN 112163592B
- Authority
- CN
- China
- Prior art keywords
- state
- mobile phone
- pedestrian
- acceleration
- current
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000001133 acceleration Effects 0.000 claims abstract description 49
- 238000003062 neural network model Methods 0.000 claims abstract description 15
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 6
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 241000283070 Equus zebra Species 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000007619 statistical method Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 7
- 238000012360 testing method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000007792 addition Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000036544 posture Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
- G06F17/142—Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/148—Wavelet transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V40/25—Recognition of walking or running movements, e.g. gait recognition
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/24—Reminder alarms, e.g. anti-loss alarms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Emergency Management (AREA)
- Discrete Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a method for identifying and early warning a pedestrian by using a mobile phone state and a smart phone thereof, and establishes a pedestrian motion state identification neural network model; detecting current acceleration through an acceleration sensor arranged in the mobile phone, and identifying the pedestrian motion state through a pedestrian motion state identification neural network model; acquiring current road surface information through a mobile phone, and identifying a current environment by using a convolutional neural network model; and early warning is carried out according to the current acceleration, the pedestrian motion state, the current environment and the current screen state. According to the invention, the analysis and the processing of the combined acceleration value are carried out by means of statistical analysis, time domain analysis and frequency domain analysis, so that the motion state identification is accurately and rapidly carried out.
Description
Technical Field
The invention relates to the field of mobile phone identification or pedestrian traffic safety intervention, in particular to a method for identifying and early warning the state of a mobile phone used by a pedestrian and a smart phone thereof.
Background
Daily and addictive use of smart phones has led to increasingly common use of mobile phones by pedestrians in walking, and road traffic accidents caused by distraction of the mobile phones continue to rise. In order to solve the traffic safety problem caused by using mobile phones by pedestrians, part of cities are provided with 'mobile phone user special sidewalks', or flash lamps and laser beams are erected at crossroads, so that accidents such as accidental collision, falling injury and the like caused by 'low head families' using mobile phones are avoided. However, these measures are limited by road traffic resources and road facilities, and their operability, intervention effect and wide popularization are limited. Therefore, there is a need to develop a method for inducing low-head families to pay attention to traffic safety and normalizing self-travel without the need for external road traffic setting.
Because the smart phone is internally provided with the microsensor, various data such as acceleration, brightness, GPS, inclination angle, rolling angle, magnetic field and the like can be detected, the computing capacity of the smart phone is stronger and stronger, and the recognition of the pedestrian activity state and the activity scene can be realized after pattern recognition. In the identification process of the pedestrian motion state, noise reduction and filtering treatment are required to be carried out on the collected dynamic data, and the traditional statistical analysis method can show the characteristics of different states to a certain extent, but has the advantages of unobvious parameter characteristics and poor identification robustness.
Disclosure of Invention
Aiming at the defects existing in the prior art, the invention provides a method for identifying and early warning the state of a mobile phone used by a pedestrian and a smart mobile phone thereof, which are used for analyzing and processing the combined acceleration value by means of statistical analysis, time domain analysis and frequency domain analysis, accurately and rapidly identifying the motion state, identifying the operation behavior of the mobile phone by means of a bright screen state and identifying the pedestrian environment by means of an image processing technology.
The present invention achieves the above technical object by the following means.
A method for identifying and pre-warning the state of a pedestrian by using a mobile phone comprises the following steps:
Establishing a pedestrian motion state identification neural network model;
detecting current acceleration through an acceleration sensor arranged in the mobile phone, and identifying the pedestrian motion state through a pedestrian motion state identification neural network model;
Acquiring current road surface information through a mobile phone, and identifying a current environment by using a convolutional neural network model;
and early warning is carried out according to the current acceleration, the pedestrian motion state, the current environment and the current screen state.
Further, a pedestrian motion state identification neural network model is established, comprising the following steps:
taking N acceleration values as an acceleration sequence x (N), wherein N is [1, N ];
Calculating the average acceleration of the acceleration sequence x (n) And variance σ 2;
performing discrete Fourier transform on the acceleration sequence X (n) to obtain X (n), and processing the X (n) by using fast Fourier transform, wherein the method specifically comprises the following steps:
Wherein/>
The acceleration sequence x (n) is decomposed into information of a low-frequency part and a high-frequency part by using 5 layers of multi Bei Xixiao wave transformation, the approximation coefficient is c, the detail coefficient is { d 1,d2,d3,d4,d5 }, and the standard deviation sigma c of the approximation coefficient and the standard deviation sigma d of the third layer detail coefficient d 3 are obtained, wherein:
Wherein: i=1, 2, …, N; /(I)
Wherein: i=1, 2, …, N; /(I)
Will average accelerationVariance sigma 2,/>The standard deviation sigma c of the approximation coefficient standard deviation sigma c and the standard deviation sigma d of the third layer detail coefficient d 3 are used as input layers of the BP neural network, the output layers of the BP neural network are a state Y 1 in which pedestrians keep stationary or have no obvious movement, a walking state Y 2 and a running state Y 3, an excitation function of the output layers adopts a Thah hyperbolic tangent function, an hidden layer of the BP neural network comprises p nodes, and the/>A is an adjustment constant, n is the number of nodes of an input layer, and m is the number of nodes of an output layer;
and training the BP neural network.
Further, the current road surface information is acquired through the mobile phone, and the current environment is identified by utilizing the convolutional neural network model, specifically:
and shooting a picture under the current gesture of the pedestrian by using a mobile phone rearview camera, and identifying the picture under the current gesture by using a lightweight depth separable convolution network MobileNets model to obtain that the current environment is a zebra crossing or a common pavement or stairs.
Further, early warning is carried out according to the current acceleration, the pedestrian motion state, the current environment and the current screen state, specifically:
When the pedestrian motion state is identified as a walking state Y 2 or a running state Y 3 by the pedestrian motion state identification neural network model, and the mobile phone screen is in a bright screen state, an alarm is sent out according to the identified current environment.
Further, the alarm is one or a combination of a text popup window or a vibration or voice prompt.
A smart phone for a pedestrian using a method for recognizing and early warning the state of the mobile phone comprises an acceleration sensor, a pedestrian motion state recognition module, an image recognition module and an alarm module;
the acceleration sensor is arranged in the intelligent mobile phone and is used for detecting acceleration signals;
the pedestrian motion state identification module identifies the pedestrian motion state according to the acceleration signal;
the image recognition module recognizes the current environment according to the picture shot by the rear camera of the smart phone;
and the alarm module sends alarm information according to the screen state, the pedestrian motion state and the current environment.
The invention has the beneficial effects that:
1. the method for identifying and early warning the pedestrian by using the mobile phone state realizes the identification of the pedestrian motion state, whether the mobile phone is used or not and the pedestrian motion environment, and finally realizes the accurate identification of the pedestrian by using the mobile phone state in the dangerous scene;
2. According to the method for identifying and early warning the state of the mobile phone, the mobile phone signals are used for early warning and reminding the low-head group of paying attention to traffic safety, users do not need to purchase other equipment to standardize self travel, and external road traffic facilities are not needed to intervene and induce, so that the feasibility and the operability are good.
Drawings
Fig. 1 is a flowchart of a method for recognizing and early warning pedestrians using a mobile phone state according to the present invention.
Fig. 2 is a third layer detail coefficient diagram after the acceleration wavelet transform according to the present invention.
Fig. 3 is an approximation coefficient diagram according to the present invention.
Fig. 4 is a pre-warning flow according to the present invention.
Detailed Description
The invention will be further described with reference to the drawings and the specific embodiments, but the scope of the invention is not limited thereto.
As shown in fig. 1, a flowchart of a method for identifying and pre-warning pedestrians by using a mobile phone state according to the present invention includes the following steps:
s01: establishing a pedestrian motion state identification neural network model;
Taking 200 continuous sampling data as an acceleration sequence x (N), namely taking N=200 acceleration values as the acceleration sequence x (N), wherein N is [1, N ];
Calculating the average acceleration of the acceleration sequence x (n) And variance σ 2;
performing discrete Fourier transform on the acceleration sequence X (n) to obtain X (n), and processing the X (n) by using fast Fourier transform, wherein the method specifically comprises the following steps:
Wherein/>
The discrete fourier transform requires about N 2 multiplications and N (N-1) additions to be calculated, which is computationally intensive. The invention adopts Fast Fourier Transform (FFT) to process discrete Fourier transform, and the calculated amount can be reduced to (N/2) log 2 N multiplications and Nlog 2 N additions.
As shown in fig. 2 and 3, the acceleration sequence x (n) is decomposed into information of a low frequency part and a high frequency part by using 5 layers of multi Bei Xixiao wave transformation, the approximation coefficient is c and the detail coefficient is { d 1,d2,d3,d4,d5 }, and a standard deviation σ d of an approximation coefficient standard deviation σ c and a third layer detail coefficient d 3 is obtained, wherein:
Wherein: i=1, 2, …, N; /(I)
Wherein: i=1, 2, …, N; /(I)
Average accelerationVariance sigma 2,/>The standard deviation sigma c of the approximate coefficient standard deviation sigma c and the standard deviation sigma d of the third layer detail coefficient d 3 are used as characteristic parameters of the pedestrian motion state, the characteristic parameters of 5 pedestrian motion states are used as input layers of a BP neural network, an output layer of the BP neural network is a state Y 1 in which pedestrians keep stationary or have no obvious movement, a walking state Y 2 and a running state Y 3, an excitation function of the output layer adopts a Thah double tangent function, an implicit layer of the BP neural network comprises p nodes, and the/>A is an adjustment constant, n is the number of nodes of an input layer, and m is the number of nodes of an output layer;
Input vector:
intermediate layer vector: b= { B 1,B2,…,Bp}T
Output vector: y= { Y 1,Y2,Y3}T;
After the neural network model is established, randomly dividing a training set and a testing set according to a proportion, setting super parameters such as learning times, learning rate, target error value and the like, training the neural network model, and generating a prediction model after the training is terminated to meet the requirement; and identifying and testing the test set, wherein the model precision can reach more than 96% through data verification.
S02: detecting current acceleration through an acceleration sensor arranged in the mobile phone, and identifying the pedestrian motion state through a pedestrian motion state identification neural network model;
S03: judging the using state of the mobile phone according to the bright screen sensor, and judging that the mobile phone is being operated when the bright screen sensor value is ON and the signal lasts for more than T seconds; otherwise, the mobile phone state is judged to be not used.
S04: acquiring current road surface information through a mobile phone, and identifying a current environment by using a convolutional neural network model;
(1) The method comprises the steps that under various walking postures of pedestrians, ordinary road surfaces, zebra crossings and stair treads of a handheld mobile phone in static and moving states are shot by adopting a mobile phone rearview camera, 1 ten thousand of the ordinary road surfaces, zebra crossings and stair treads are shot, and the total data amount is about 3 ten thousand; and adjusting the size of the picture, calibrating the picture label to establish a picture library. Thus, an image set Z= { Z 1,…,ze }, and the class label of the image is W= { W 1,…,we }, wherein e is the number of samples, and the class label is zebra stripes, common pavement and stairs.
(2) Construction of convolutional neural network model
Constructing a convolution neural network with an I-layer structure, wherein the convolution neural network comprises an input layer Z, a standard convolution layer C, a separable convolution layer D 1、D2、D3…DH-3, a pooling layer P, a full-connection layer F and a Softmax layer from top to bottom in sequence; the standard convolution layer takes as input the characteristic map F of D F×DF x R and produces as output the characteristic map G of D G×DG x S. Where D F is the width and height of an input profile, R is the number of input channels, D G is the width and height of an output profile, and S is the number of output channels. The convolution kernel K in the convolution layer contains D K×DK x R x S parameters, and the output feature map of the standard convolution is expressed as:
Gp,q,s＝∑i,j,rKi,j,r,s·Fp+i-1,q+j-1,r
The calculated amount is as follows: d K·DK·R·S·DF·DF;DK represents the side length of a square convolution kernel.
The invention adopts a lightweight depth separable convolution network MobileNets model, divides the generating and combining steps into two steps, and uses the depth convolution K1 and the point-by-point convolution K2 respectively, thereby reducing the calculated amount. The output feature map of the depth convolution K1 is expressed as:
Wherein, For the convolution kernel of D K×DK xr, the depth convolution K1 is used to filter the input channel, and the K2 point convolution linearly combines the output of the depth convolution by the convolution kernel of 1*1 and produces a feature map; the total calculated amount of the depth separable convolution is: d K·DK·R·DF·DF+R·S·DF·DF.
The excitation function adopts a leakage ReLU function, a full-connection layer is obtained through feature graph expansion after pooling of the layer, probability is distributed to different picture categories by adopting a softmax function, and an index of the maximum value of the probability is selected as a predicted picture type.
And (3) giving an initial value to the convolutional neural network, obtaining initial weights and thresholds of the separable convolutional layer and the standard convolutional layer, randomly selecting 80% of three types of pictures in the picture set for training, and updating the weights and the thresholds until the convolutional neural network model converges. The remaining 20% of the pictures in the picture set are used for testing, and the identification accuracy can reach over 96% through verification.
S04: and early warning is carried out according to the current acceleration, the pedestrian motion state, the current environment and the current screen state.
As shown in fig. 4, every 10 consecutive judgment values are used as the basis for finally judging the movement state of the user, and when a certain movement state is judged to be half the number of times, the state is considered to be in the state; that is, 10 motion states are recognized by the pedestrian motion state recognition neural network model in 10 consecutive cycles, and if the number of times of the walking state Y2 is determined to be half, the state is considered to be present, and the cycle is generally 0.1s. When the pedestrian motion state is identified as a walking state Y2 or a running state Y3 by the pedestrian motion state identification neural network model and the mobile phone screen is in a continuous bright screen state, different forms of alarm are sent out according to the identified current environment, and the alarm can be one or a combination of text popup window, vibration or voice reminding. And comprehensively obtaining whether the user uses the mobile phone when moving, passing through the zebra stripes and going up and down stairs according to the movement judging result and the road surface identifying result, and if so, sending out different warning information by the system.
A smart phone for a pedestrian using a method for recognizing and early warning the state of the mobile phone comprises an acceleration sensor, a pedestrian motion state recognition module, an image recognition module and an alarm module;
the acceleration sensor is arranged in the intelligent mobile phone and is used for detecting acceleration signals;
the pedestrian motion state identification module identifies the pedestrian motion state according to the acceleration signal;
the image recognition module recognizes the current environment according to the picture shot by the rear camera of the smart phone;
and the alarm module sends alarm information according to the screen state, the pedestrian motion state and the current environment.
The pedestrian motion state identification module, the image identification module and the alarm module can be installed on the smart phone in the form of software app or can be some kind of applet.
The examples are preferred embodiments of the present invention, but the present invention is not limited to the above-described embodiments, and any obvious modifications, substitutions or variations that can be made by one skilled in the art without departing from the spirit of the present invention are within the scope of the present invention.
Claims (5)
1. The method for recognizing and early warning the pedestrian by using the mobile phone state is characterized by comprising the following steps of:
the method for establishing the pedestrian movement state identification neural network model comprises the following steps of:
taking N acceleration values as an acceleration sequence x (N), wherein N is [1, N ];
Calculating the average acceleration of the acceleration sequence x (n) And variance σ 2;
performing discrete Fourier transform on the acceleration sequence X (n) to obtain X (n), and processing the X (n) by using fast Fourier transform, wherein the method specifically comprises the following steps:
Wherein/>
The acceleration sequence x (n) is decomposed into information of a low-frequency part and a high-frequency part by using 5 layers of multi Bei Xixiao wave transformation, the approximation coefficient is c, the detail coefficient is { d 1,d2,d3,d4,d5 }, and the standard deviation sigma c of the approximation coefficient and the standard deviation sigma d of the third layer detail coefficient d 3 are obtained, wherein:
Wherein: i=1, 2, …, N; /(I)
Wherein: i=1, 2, …, N; /(I)
Will average accelerationVariance sigma 2,/>The standard deviation sigma c of the approximation coefficient standard deviation sigma c and the standard deviation sigma d of the third layer detail coefficient d 3 are used as input layers of the BP neural network, the output layers of the BP neural network are a state Y 1 in which pedestrians keep stationary or have no obvious movement, a walking state Y 2 and a running state Y 3, an excitation function of the output layers adopts a Thah hyperbolic tangent function, an hidden layer of the BP neural network comprises p nodes, and the/>A is an adjustment constant, n is the number of nodes of an input layer, and m is the number of nodes of an output layer;
training a BP neural network;
detecting current acceleration through an acceleration sensor arranged in the mobile phone, and identifying the pedestrian motion state through a pedestrian motion state identification neural network model;
Acquiring current road surface information through a mobile phone, and identifying a current environment by using a convolutional neural network model;
and early warning is carried out according to the current acceleration, the pedestrian motion state, the current environment and the current screen state.
2. The method for recognizing and pre-warning the pedestrian by using the mobile phone state according to claim 1, wherein the current road surface information is obtained by the mobile phone, and the current environment is recognized by using a convolutional neural network model, specifically:
and shooting a picture under the current gesture of the pedestrian by using a mobile phone rearview camera, and identifying the picture under the current gesture by using a lightweight depth separable convolution network MobileNets model to obtain that the current environment is a zebra crossing or a common pavement or stairs.
3. The method for recognizing and pre-warning pedestrians by using the mobile phone state according to claim 1, wherein the pre-warning is performed according to the current acceleration, the motion state of the pedestrians, the current environment and the current screen state, specifically:
when the pedestrian motion state is identified as a walking state Y 2 or a running state Y 3 by the pedestrian motion state identification neural network model, and the mobile phone screen is in a continuous screen-lighting state, an alarm is sent out according to the identified current environment.
4. A method of pedestrian status recognition and pre-warning using a mobile phone according to claim 3, wherein the warning is one or a combination of a text pop-up or a vibration or voice alert.
5. A smart phone according to the method for recognizing and pre-warning pedestrian using the state of the smart phone of claim 1, comprising an acceleration sensor, a pedestrian motion state recognition module, an image recognition module and an alarm module;
the acceleration sensor is arranged in the intelligent mobile phone and is used for detecting acceleration signals;
the pedestrian motion state identification module identifies the pedestrian motion state according to the acceleration signal;
the image recognition module recognizes the current environment according to the picture shot by the rear camera of the smart phone;
and the alarm module sends alarm information according to the screen state, the pedestrian motion state and the current environment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010870729.XA CN112163592B (en) | 2020-08-26 | 2020-08-26 | Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010870729.XA CN112163592B (en) | 2020-08-26 | 2020-08-26 | Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112163592A CN112163592A (en) | 2021-01-01 |
| CN112163592B true CN112163592B (en) | 2024-05-14 |
Family
ID=73860442
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010870729.XA Active CN112163592B (en) | 2020-08-26 | 2020-08-26 | Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112163592B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345846A (en) * | 2018-01-29 | 2018-07-31 | 华东师范大学 | A kind of Human bodys' response method and identifying system based on convolutional neural networks |
| CN109495654A (en) * | 2018-12-29 | 2019-03-19 | 武汉大学 | One kind perceiving pedestrains safety method based on smart phone |
| CN110674875A (en) * | 2019-09-25 | 2020-01-10 | 电子科技大学 | Pedestrian motion mode identification method based on deep hybrid model |
-
2020
- 2020-08-26 CN CN202010870729.XA patent/CN112163592B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345846A (en) * | 2018-01-29 | 2018-07-31 | 华东师范大学 | A kind of Human bodys' response method and identifying system based on convolutional neural networks |
| CN109495654A (en) * | 2018-12-29 | 2019-03-19 | 武汉大学 | One kind perceiving pedestrains safety method based on smart phone |
| CN110674875A (en) * | 2019-09-25 | 2020-01-10 | 电子科技大学 | Pedestrian motion mode identification method based on deep hybrid model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112163592A (en) | 2021-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110992683B (en) | Dynamic image perception-based intersection blind area early warning method and system | |
| Choi et al. | Car crash detection using ensemble deep learning and multimodal data from dashboard cameras | |
| Zhang et al. | Prediction of pedestrian-vehicle conflicts at signalized intersections based on long short-term memory neural network | |
| CN100545867C (en) | Aerial shooting traffic video frequency vehicle rapid checking method | |
| CN108319909B (en) | Driving behavior analysis method and system | |
| CN113345237A (en) | Lane-changing identification and prediction method, system, equipment and storage medium for extracting vehicle track by using roadside laser radar data | |
| Chebrolu et al. | Deep learning based pedestrian detection at all light conditions | |
| KR101908481B1 (en) | Device and method for pedestraian detection | |
| CN114495006A (en) | Detection method and device for left-behind object and storage medium | |
| Xue et al. | A context-aware framework for risky driving behavior evaluation based on trajectory data | |
| CN114267082A (en) | Bridge side falling behavior identification method based on deep understanding | |
| KR101736466B1 (en) | Apparatus and Method for context recognition based on acoustic information | |
| CN113158835A (en) | Traffic accident intelligent detection method based on deep learning | |
| Gupta et al. | Real-time traffic control and monitoring | |
| CN117372969B (en) | Monitoring scene-oriented abnormal event detection method | |
| CN112163592B (en) | Method for recognizing and early warning pedestrian state by using mobile phone and smart phone thereof | |
| CN104200202B (en) | A kind of upper half of human body detection method based on cumulative perceptron | |
| CN112215188B (en) | Traffic police gesture recognition method, device, equipment and storage medium | |
| CN114492544B (en) | Model training method and device and traffic incident occurrence probability evaluation method and device | |
| Marciniuk et al. | Machine learning applied to acoustic-based road traffic monitoring | |
| Tok et al. | Vector classification of commercial vehicles using a high fidelity inductive loop detection system | |
| CN110556024B (en) | Anti-collision auxiliary driving method and system and computer readable storage medium | |
| Zhang | Prediction of Pedestrians' Red Light Violations Using Deep Learning | |
| CN111507380A (en) | Image classification method, system and device based on clustering and storage medium | |
| Saif et al. | Szr5: A modern technique to detect potholes using convolution neural 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |