CN110991346A - Suspected drug addict identification method and device and storage medium - Google Patents

Suspected drug addict identification method and device and storage medium Download PDF

Info

Publication number
CN110991346A
CN110991346A CN201911228487.8A CN201911228487A CN110991346A CN 110991346 A CN110991346 A CN 110991346A CN 201911228487 A CN201911228487 A CN 201911228487A CN 110991346 A CN110991346 A CN 110991346A
Authority
CN
China
Prior art keywords
neural network
deep neural
drug
network model
training
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
Application number
CN201911228487.8A
Other languages
Chinese (zh)
Inventor
林淑强
赵秀明
周成祖
鄢小征
陈志飞
杜新胜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xiamen Meiya Pico Information Co Ltd
Original Assignee
Xiamen Meiya Pico Information Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xiamen Meiya Pico Information Co Ltd filed Critical Xiamen Meiya Pico Information Co Ltd
Priority to CN201911228487.8A priority Critical patent/CN110991346A/en
Publication of CN110991346A publication Critical patent/CN110991346A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Tourism & Hospitality (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)
  • Collating Specific Patterns (AREA)

Abstract

The invention provides a method, a device and a storage medium for identifying suspected drug addicts, wherein the method comprises the following steps: an acquisition step, namely acquiring a face image img to be processed; and a detection step, namely identifying the face image img to be processed by using a trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, determining that the person corresponding to the face image is a suspected virus addict. The invention identifies drug addicts through facial features. The invention designs a special deep neural network algorithm by utilizing a deep learning technology, trains the neural network algorithm through a large number of facial pictures of drug addicts and non-drug addicts, enables the algorithm to learn the difference of facial features of the drug addicts and the non-drug addicts, solidifies the weight of the neural network algorithm, enables the algorithm to pertinently extract the facial features of the people, can classify the facial features, and distinguishes whether one person is suspected to be drug addicts, and is convenient and quick to identify.

Description

Suspected drug addict identification method and device and storage medium
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a method and a device for identifying suspected drug addicts and a storage medium.
Background
People who take drugs have great harm, and people who take drugs for a long time have the appearance of red eyelid and nose, very small pupils, scattered and unfocused eyes, listlessness or other features which are difficult to observe. For a long time, the identification of drug addicts is mainly carried out by blood detection, urine detection, hair detection, urine detection and other methods.
The above-mentioned method for drug addict identification: blood test, urine test, hair test, urine examination, although the accuracy is high, have several inadequacies or shortcomings: 1. the person who needs to find the drug addict himself needs to consume a large amount of manpower and material resources in the process of finding the person and finding which persons need to find the drug addict himself; 2. the drug addicts need to cooperate, and some drug addicts have unstable emotion and low cooperation degree when addicts are addicted to drugs, which brings great inconvenience; 3. the detection process takes time and not all approaches can give results immediately. Due to the defects, the existing drug addict identification method cannot quickly identify all people, and many drug addicts cannot find the drug addicts in time even if the drug addicts take the drug for a long time, so that great harm is brought to the drug addicts and families of the drug addicts, and potential safety hazards are brought to social public safety. In recent years, the group for taking drugs is getting bigger and bigger, and the acceleration is fast, so that how to quickly and accurately discover drug addicts becomes a key problem.
Disclosure of Invention
The present invention provides the following technical solutions to overcome the above-mentioned drawbacks in the prior art.
A method of suspected drug addict identification, the method comprising:
an acquisition step, namely acquiring a face image img to be processed;
and a detection step, namely identifying the face image img to be processed by using a trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, determining that the person corresponding to the face image is a suspected virus addict.
Furthermore, after the person is determined to be a suspected drug addict, the contact information and/or the identity card number of the person are obtained through the network database and are sent to relevant departments for checking.
Further, the deep neural network model comprises an M convolutional layer, a pooling layer, a full link layer and a softmax layer which are connected in sequence.
Further, training the deep neural network model using a training sample set, comprising: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model.
Still further, the detecting step includes:
preprocessing the face image img to obtain an image face _ img with the size of N x NN*N；
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure BDA0002302877160000021
The vector
Figure BDA0002302877160000031
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension;
the feature vector is combined
Figure BDA0002302877160000032
Inputting a two-classifier composed of full connection layers for classification, and outputting a classification result:
Figure BDA0002302877160000033
wherein the full connection layer weight parameter is a K x 2 matrix
Figure BDA0002302877160000034
Offset parameter
Figure BDA0002302877160000035
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure BDA0002302877160000036
wherein the probability of drug class
Figure BDA0002302877160000037
Probability of non-drug-taking class
Figure BDA0002302877160000038
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1；
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict.
The invention also provides a suspected drug addict identification device, which comprises:
the acquisition unit is used for acquiring a face image img to be processed;
and the detection unit is used for identifying the face image img to be processed by using the trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, the person corresponding to the face image is a suspected virus addict.
Furthermore, after the person is determined to be a suspected drug addict, the contact information and/or the identity card number of the person are obtained through the network database and are sent to relevant departments for checking.
Further, the deep neural network model comprises an M convolutional layer, a pooling layer, a full link layer and a softmax layer which are connected in sequence.
Further, training the deep neural network model using a training sample set, comprising: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model.
Still further, the detecting unit performs operations including:
preprocessing the face image img to obtain an image face _ img with the size of N x NN*N；
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure BDA0002302877160000041
The vector
Figure BDA0002302877160000042
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension;
the feature vector is combined
Figure BDA0002302877160000043
Inputting a two-classifier composed of full connection layers for classification, and outputting a classification result:
Figure BDA0002302877160000044
wherein the full connection layer weight parameter is a K x 2 matrix
Figure BDA0002302877160000045
Offset parameter
Figure BDA0002302877160000046
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure BDA0002302877160000051
wherein the probability of drug class
Figure BDA0002302877160000052
Probability of non-drug-taking class
Figure BDA0002302877160000053
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1；
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict.
The present invention also proposes a computer-readable storage medium having stored thereon computer program code means for performing any of the above-mentioned means when said computer program code means is executed by a computer.
The invention has the technical effects that: the invention discloses a suspected drug addict identification method, which comprises the following steps: an acquisition step, namely acquiring a face image img to be processed; and a detection step, namely identifying the face image img to be processed by using a trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, determining that the person corresponding to the face image is a suspected virus addict. The invention identifies drug addicts through facial features. By utilizing a deep learning technology, a special deep neural network algorithm is designed, the neural network algorithm is trained through a large number of facial pictures of drug addicts and non-drug addicts, the algorithm can learn the difference of facial features of the drug addicts and the non-drug addicts, the weight of the neural network algorithm is solidified, the algorithm can extract the facial features of the person in a targeted manner, the facial features can be classified, and whether the person is suspected to be drug addicts or not is distinguished. The method has the following advantages: 1. the method is convenient and quick, and the drug addict does not need to be directly contacted; 2. meanwhile, a large number of face pictures of the security camera are utilized, so that all people can be identified and filtered, people with high possibility of identifying a drug absorption result are pushed to relevant departments, the relevant departments pay attention to offline investigation, and the drug absorbers are timely and fully identified through the existing methods of blood detection, urine detection, hair detection, urine detection and the like, so that the harm to individuals and families of the people is reduced, and the hidden danger of social public safety is reduced and eliminated.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings.
FIG. 1 is a flow chart of a method of suspected drug addict identification according to an embodiment of the invention.
FIG. 2 is a block diagram of a suspected drug addict identification device, according to an embodiment of the invention.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the related invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
FIG. 1 illustrates a suspected drug addict identification method of the present invention, comprising:
an acquisition step S101, acquiring a face image img to be processed; the face image img can be obtained in real time through the camera, and can also be processed through the shot picture, for example, the face image can be obtained from the video stream of the camera. For example, a large number of security cameras (installed in airports, stations, docks, etc.) can be used to obtain face images, and all people can be identified and filtered for suspected drug addicts.
And a detection step S102, identifying the face image img to be processed by using the trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, determining that the person corresponding to the face image is a suspected virus addict. The invention utilizes the deep learning technology, grasps the change of facial and expression characteristics brought by drug inhalation, the face of a person who inhales the drug for a long time can form certain common characteristics, designs a set of neural network algorithm, and after learning through facial characteristics of a large number of drug-inhaling persons and non-drug-inhaling persons, the algorithm can pertinently extract the characteristics of the face of the person and classify the extracted characteristics, thereby conveniently and quickly distinguishing whether the person inhales the drug or not through facial pictures.
In one embodiment, after the person is determined to be a suspected drug addict, the contact way and/or the identity card number of the person are obtained through a network database and are sent to a relevant department for verification, the network database can be a database of a security department, a household registration department, a bank and the like, the contact way can be a micro signal, a mobile phone number, a qq number, and certainly can also be an address, a unit address and the like.
In one embodiment, the deep neural network model designed by the invention comprises an M-layer convolutional layer, a pooling layer, a full-link layer and a softmax layer which are connected in sequence. The deep neural network model needs to be trained before being used, and the deep neural network model adopts the following training mode: training a deep neural network model using a training sample set, comprising: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model. The invention divides the sample set into three parts, one part is trained, one part is verified, and the other part is tested, thereby selecting the optimal parameter of the deep neural network model, improving the training efficiency of the neural network, and being one of the important invention points of the invention.
In one embodiment, the detecting step S102 includes the following steps.
Preprocessing the face image img to obtain an image face _ img with the size of N x NN*N(ii) a Preprocessing includes face detection, alignment, resizing, and the like, which are basic techniques for image processing.
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure BDA0002302877160000081
The vector
Figure BDA0002302877160000082
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension; wherein L is more than or equal to 1 and less than or equal to M, and L, M are integers.
The feature vector is combined
Figure BDA0002302877160000083
Inputting a binary classifier composed of full connection layers for classification, and outputting a classification nodeAnd (4) fruit:
Figure BDA0002302877160000084
wherein the full connection layer weight parameter is a K x 2 matrix
Figure BDA0002302877160000085
Offset parameter
Figure BDA0002302877160000086
Wherein, WL+1、BL+1The data in (1) are parameters generated by training the neural network model.
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure BDA0002302877160000087
wherein the probability of drug class
Figure BDA0002302877160000088
Probability of non-drug-taking class
Figure BDA0002302877160000089
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1。
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict. Epsilon is generally set to a value larger than 0.5 and smaller than 1, and the higher epsilon is set, the higher the precision of judging whether to take a poison or not, preferably epsilon is set to 0.85.
The invention designs a special neural network model, trains corresponding parameters and provides a special identification method, which can identify and filter drug addicts of all people, pushes people with high probability of identifying drug addicts to relevant departments, allows the relevant departments to pay attention to offline investigation in a targeted manner, and timely and fully identifies the drug addicts through the existing methods of blood detection, urine detection, hair detection, urine detection and the like, thereby reducing the injury to the individuals and families of the people and eliminating the hidden danger of social public safety, which is another important invention point of the invention.
FIG. 2 illustrates a suspected person identification device of the present invention, comprising:
an acquisition unit 201 that acquires a face image img to be processed; the face image img can be obtained in real time through the camera, and can also be processed through the shot picture, for example, the face image can be obtained from the video stream of the camera. For example, a large number of security cameras (installed in airports, stations, docks, etc.) can be used to obtain face images, and all people can be identified and filtered for suspected drug addicts.
The detection unit 202 identifies the face image img to be processed by using the trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, the person corresponding to the face image is a suspected drug addict. The invention utilizes the deep learning technology, grasps the change of facial and expression characteristics brought by drug inhalation, the face of a person who inhales the drug for a long time can form certain common characteristics, designs a set of neural network algorithm, and after learning through facial characteristics of a large number of drug-inhaling persons and non-drug-inhaling persons, the algorithm can pertinently extract the characteristics of the face of the person and classify the extracted characteristics, thereby conveniently and quickly distinguishing whether the person inhales the drug or not through facial pictures.
In one embodiment, after the person is determined to be a suspected drug addict, the contact way and/or the identity card number of the person are obtained through a network database and are sent to a relevant department for verification, the network database can be a database of a security department, a household registration department, a bank and the like, the contact way can be a micro signal, a mobile phone number, a qq number, and certainly can also be an address, a unit address and the like.
In one embodiment, the deep neural network model designed by the invention comprises an M-layer convolutional layer, a pooling layer, a full-link layer and a softmax layer which are connected in sequence. The deep neural network model needs to be trained before being used, and the deep neural network model adopts the following training mode: training a deep neural network model using a training sample set, comprising: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model. The invention divides the sample set into three parts, one part is trained, one part is verified, and the other part is tested, thereby selecting the optimal parameter of the deep neural network model, improving the training efficiency of the neural network, and being one of the important invention points of the invention.
In one embodiment, the operations performed by the detection unit 202 include the following steps.
Preprocessing the face image img to obtain an image face _ img with the size of N x NN*N(ii) a Preprocessing includes face detection, alignment, resizing, and the like, which are basic techniques for image processing.
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure BDA0002302877160000111
The vector
Figure BDA0002302877160000112
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension; wherein L is more than or equal to 1 and less than or equal to M, and L, M are integers.
The feature vector is combined
Figure BDA0002302877160000113
Inputting a two-classifier composed of full connection layers for classification, and outputting a classification result:
Figure BDA0002302877160000114
wherein the full connection layer weight parameter is a K x 2 matrix
Figure BDA0002302877160000115
Offset parameter
Figure BDA0002302877160000116
Wherein, WL+1、BL+1The data in (1) are parameters generated by training the neural network model.
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure BDA0002302877160000117
wherein the probability of drug class
Figure BDA0002302877160000118
Probability of non-drug-taking class
Figure BDA0002302877160000119
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1。
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict. Epsilon is generally set to a value larger than 0.5 and smaller than 1, and the higher epsilon is set, the higher the precision of judging whether to take a poison or not, preferably epsilon is set to 0.85.
The invention designs a special neural network model, trains corresponding parameters and provides a special recognition device, which can recognize and filter drug addicts of all people, pushes people with high possibility of recognizing drug addicts to relevant departments, enables the relevant departments to pay attention to offline investigation in a targeted manner, and timely and fully recognizes the drug addicts through the existing devices such as blood detection, urine detection, hair detection, urine detection and the like, thereby reducing the injury to the individuals and families and eliminating the hidden danger of social public safety, which is another important invention point of the invention.
For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be essentially implemented or the portions that contribute to the prior art may be embodied in the form of a software product, which may be stored in a storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the apparatuses described in the embodiments or some portions of the embodiments of the present application.
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made thereto without departing from the spirit and scope of the invention and it is intended to cover in the claims the invention as defined in the appended claims.

Claims (11)

1. A method for identifying suspected drug addicts, the method comprising:
an acquisition step, namely acquiring a face image img to be processed;
and a detection step, namely identifying the face image img to be processed by using a trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, determining that the person corresponding to the face image is a suspected virus addict.
2. The method as claimed in claim 1, wherein after the person is determined to be suspected drug addict, the contact information and/or identification number of the person is obtained from the network database and sent to the relevant department for verification.
3. The method of claim 1 or 2, wherein the deep neural network model comprises M convolutional layers, pooling layers, fully-connected layers, and softmax layers connected in sequence.
4. The method of claim 3, wherein training the deep neural network model using the training sample set comprises: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model.
5. The method of claim 4, wherein the detecting step comprises:
preprocessing the face image img to obtain an image face _ img with the size of N x NN*N；
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure FDA0002302877150000021
The vector
Figure FDA0002302877150000022
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension;
the feature vector is combined
Figure FDA0002302877150000023
Inputting a two-classifier composed of full connection layers for classification, and outputting a classification result:
Figure FDA0002302877150000024
wherein the full connection layer weight parameter is a K x 2 matrix
Figure FDA0002302877150000025
Offset parameter
Figure FDA0002302877150000026
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure FDA0002302877150000027
wherein the probability of drug class
Figure FDA0002302877150000028
Probability of non-drug-taking class
Figure FDA0002302877150000029
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1；
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict.
6. A device for suspected person identification, the device comprising:
the acquisition unit is used for acquiring a face image img to be processed;
and the detection unit is used for identifying the face image img to be processed by using the trained deep neural network model, and if the probability value output by the neural network is greater than a threshold value, the person corresponding to the face image is a suspected virus addict.
7. The apparatus of claim 6, wherein after the person is determined to be suspected drug addict, the contact information and/or identification number of the person is obtained from the network database and sent to the relevant department for verification.
8. The apparatus of claim 6 or 7, wherein the deep neural network model comprises M convolutional layers, a pooling layer, a fully-connected layer and a softmax layer which are connected in sequence.
9. The apparatus of claim 8, wherein training the deep neural network model using a training sample set comprises: randomly dividing an obtained training sample set into a training set, a verification set and a test set, training a built deep neural network model by using the training set, adjusting weights among convolution layers of the deep neural network model, calculating a loss function value aiming at each group of weights, wherein the loss function value tends to be the minimum value, after training is finished, saving multiple groups of weight data of the deep neural network, selecting one group of weight data from the multiple groups of weight data of the deep neural network by using the verification set to enable the recognition precision of the deep neural network model to be the highest, then testing whether the deep neural network model meets the recognition precision by using the test set, and if so, taking the deep neural network model as the trained deep neural network model.
10. The apparatus of claim 9, wherein the detection unit performs operations comprising:
preprocessing the face image img to obtain an image face _ img with the size of N x NN*N；
Image face _ imgN*NInputting into M convolutional layers and pooling layers, and outputting feature vector
Figure FDA0002302877150000031
The vector
Figure FDA0002302877150000032
Is a matrix of 1 xK, L represents the L-th layer of the neural network, and K represents that the vector dimension is K dimension;
the feature vector is combined
Figure FDA0002302877150000033
Inputting a two-classifier composed of full connection layers for classification, and outputting a classification result:
Figure FDA0002302877150000041
wherein the full connection layer weight parameter is a K x 2 matrix
Figure FDA0002302877150000042
Offset parameter
Figure FDA0002302877150000043
Mapping the classification result output by the classifier composed of the full connection layer to a (0,1) probability interval through a softmax layer, wherein the softmax layer function is as follows:
Figure FDA0002302877150000044
wherein the probability of drug class
Figure FDA0002302877150000045
Probability of non-drug-taking class
Figure FDA0002302877150000046
0＜PNon-drug taking,PDrug taking＜1；PNon-drug taking+PDrug taking＝1；
Judgment of PDrug takingAnd if so, determining that the person corresponding to the face image is a suspected virus addict.
11. A computer-readable storage medium, characterized in that the storage medium has stored thereon computer program code which, when executed by a computer, performs the apparatus of any of claims 1-5.
CN201911228487.8A 2019-12-04 2019-12-04 Suspected drug addict identification method and device and storage medium Pending CN110991346A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911228487.8A CN110991346A (en) 2019-12-04 2019-12-04 Suspected drug addict identification method and device and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911228487.8A CN110991346A (en) 2019-12-04 2019-12-04 Suspected drug addict identification method and device and storage medium

Publications (1)

Publication Number Publication Date
CN110991346A true CN110991346A (en) 2020-04-10

Family

ID=70090043

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911228487.8A Pending CN110991346A (en) 2019-12-04 2019-12-04 Suspected drug addict identification method and device and storage medium

Country Status (1)

Country Link
CN (1) CN110991346A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112668800A (en) * 2021-01-04 2021-04-16 荣联科技集团股份有限公司 Information processing method, apparatus, medium, and device
CN113128387A (en) * 2021-04-14 2021-07-16 广州大学 Drug addict drug addiction attack identification method based on facial expression feature analysis
CN113827186A (en) * 2021-06-12 2021-12-24 华南师范大学 Drug craving degree measuring method based on computer vision
CN114868386A (en) * 2020-12-03 2022-08-05 Oppo广东移动通信有限公司 Encoding method, decoding method, encoder, decoder, and electronic device
WO2023123926A1 (en) * 2021-12-28 2023-07-06 苏州浪潮智能科技有限公司 Artificial intelligence task processing method and apparatus, electronic device, and readable storage medium

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101996314A (en) * 2009-08-26 2011-03-30 厦门市美亚柏科信息股份有限公司 Content-based human body upper part sensitive image identification method and device
AU2015203172A1 (en) * 2010-09-20 2015-07-02 Security First Corp. Systems and methods for secure data sharing
US20160307057A1 (en) * 2015-04-20 2016-10-20 3M Innovative Properties Company Fully Automatic Tattoo Image Processing And Retrieval
CN107320081A (en) * 2017-07-03 2017-11-07 北京大学 To drug addict relapse the method and system of monitoring
CN107679451A (en) * 2017-08-25 2018-02-09 百度在线网络技术（北京）有限公司 Establish the method, apparatus, equipment and computer-readable storage medium of human face recognition model
CN110046606A (en) * 2019-04-26 2019-07-23 上海云从企业发展有限公司 Method for early warning and device
CN110151184A (en) * 2019-04-19 2019-08-23 昆明依利科特科技有限公司 Contactless drug addict's rapid detection system and method
CN110263633A (en) * 2019-05-13 2019-09-20 广州烽火众智数字技术有限公司 The personnel that are involved in drug traffic based on space time correlation detect method for early warning, system and storage medium
CN110399821A (en) * 2019-07-17 2019-11-01 上海师范大学 Customer satisfaction acquisition methods based on facial expression recognition

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101996314A (en) * 2009-08-26 2011-03-30 厦门市美亚柏科信息股份有限公司 Content-based human body upper part sensitive image identification method and device
AU2015203172A1 (en) * 2010-09-20 2015-07-02 Security First Corp. Systems and methods for secure data sharing
US20160307057A1 (en) * 2015-04-20 2016-10-20 3M Innovative Properties Company Fully Automatic Tattoo Image Processing And Retrieval
CN107320081A (en) * 2017-07-03 2017-11-07 北京大学 To drug addict relapse the method and system of monitoring
CN107679451A (en) * 2017-08-25 2018-02-09 百度在线网络技术（北京）有限公司 Establish the method, apparatus, equipment and computer-readable storage medium of human face recognition model
CN110151184A (en) * 2019-04-19 2019-08-23 昆明依利科特科技有限公司 Contactless drug addict's rapid detection system and method
CN110046606A (en) * 2019-04-26 2019-07-23 上海云从企业发展有限公司 Method for early warning and device
CN110263633A (en) * 2019-05-13 2019-09-20 广州烽火众智数字技术有限公司 The personnel that are involved in drug traffic based on space time correlation detect method for early warning, system and storage medium
CN110399821A (en) * 2019-07-17 2019-11-01 上海师范大学 Customer satisfaction acquisition methods based on facial expression recognition

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114868386A (en) * 2020-12-03 2022-08-05 Oppo广东移动通信有限公司 Encoding method, decoding method, encoder, decoder, and electronic device
CN114868386B (en) * 2020-12-03 2024-05-28 Oppo广东移动通信有限公司 Encoding method, decoding method, encoder, decoder, and electronic device
CN112668800A (en) * 2021-01-04 2021-04-16 荣联科技集团股份有限公司 Information processing method, apparatus, medium, and device
CN113128387A (en) * 2021-04-14 2021-07-16 广州大学 Drug addict drug addiction attack identification method based on facial expression feature analysis
CN113128387B (en) * 2021-04-14 2023-10-13 广州大学 Drug addiction attack recognition method for drug addicts based on facial expression feature analysis
CN113827186A (en) * 2021-06-12 2021-12-24 华南师范大学 Drug craving degree measuring method based on computer vision
WO2023123926A1 (en) * 2021-12-28 2023-07-06 苏州浪潮智能科技有限公司 Artificial intelligence task processing method and apparatus, electronic device, and readable storage medium

Similar Documents

Publication Publication Date Title
CN110991346A (en) Suspected drug addict identification method and device and storage medium
JP7402239B2 (en) Face recognition method, neural network training method, face recognition device, electronic device and computer readable storage medium
CN109858375B (en) Living body face detection method, terminal and computer readable storage medium
CN108182409A (en) Biopsy method, device, equipment and storage medium
CN110503000B (en) Teaching head-up rate measuring method based on face recognition technology
CN104036254A (en) Face recognition method
CN111881726A (en) Living body detection method and device and storage medium
CN113221655B (en) Face spoofing detection method based on feature space constraint
CN112381987A (en) Intelligent entrance guard epidemic prevention system based on face recognition
CN103443772B (en) The method of the individual sex checking based on multi-modal data analysis
CN109145704A (en) A kind of human face portrait recognition methods based on face character
RU2316051C2 (en) Method and system for automatically checking presence of a living human face in biometric safety systems
Nahar et al. Twins and Similar Faces Recognition Using Geometric and Photometric Features with Transfer Learning
Kadhim et al. A multimodal biometric database and case study for face recognition based deep learning
Latha et al. A novel method for person authentication using retinal images
Bigun et al. Combining biometric evidence for person authentication
CN110738985A (en) Cross-modal biometric feature recognition method and system based on voice signals
CN108197593B (en) Multi-size facial expression recognition method and device based on three-point positioning method
Gupta et al. Proposed iris recognition algorithm through image acquisition technique
Booysens et al. Exploration of ear biometrics using EfficientNet
Guefrachi et al. Deep learning based DeepFake video detection
CN112949363A (en) Face living body identification method and device
CN106599764A (en) Lip-shape-characteristic-based living body determination method and equipment
CN109766814A (en) Face verification method and system
Islam et al. Sign language recognition for Bangla alphabets using deep learning methods

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: 20200410

RJ01 Rejection of invention patent application after publication