CN112036352B - Training method of fatigue detection model, and fatigue driving detection method and device - Google Patents
Training method of fatigue detection model, and fatigue driving detection method and device Download PDFInfo
- Publication number
- CN112036352B CN112036352B CN202010936728.0A CN202010936728A CN112036352B CN 112036352 B CN112036352 B CN 112036352B CN 202010936728 A CN202010936728 A CN 202010936728A CN 112036352 B CN112036352 B CN 112036352B
- Authority
- CN
- China
- Prior art keywords
- fatigue
- target
- rule
- training
- order
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
- G06V20/597—Recognising the driver's state or behaviour, e.g. attention or drowsiness
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- G08B21/06—Alarms for ensuring the safety of persons indicating a condition of sleep, e.g. anti-dozing alarms
-
- 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
- G08B21/182—Level alarms, e.g. alarms responsive to variables exceeding a threshold
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Emergency Management (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The application provides a training method of a fatigue detection model, a fatigue driving detection method and a fatigue driving detection device, wherein the training method comprises the following steps: constructing a training sample based on the historical order, wherein the training sample comprises the fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics; taking fatigue characteristics as input of a decision tree model, and taking fatigue rules as output of the decision tree model; training the decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a corresponding fatigue rule; and according to the fatigue rule corresponding to the fatigue detection model, establishing a mapping relation between the fatigue characteristics and the fatigue degree so as to determine the target fatigue degree corresponding to the target order through the mapping relation. According to the method and the device, the detection efficiency and the detection accuracy are improved, the requirement of real-time detection can be met, the sample does not need to be labeled with the fatigue degree, and the training efficiency of the model is improved.
Description
Technical Field
The application relates to the technical field of fatigue detection, in particular to a training method of a fatigue detection model, a fatigue driving detection method and a fatigue driving detection device.
Background
In the traffic field, such as the field of network reservation, fatigue driving is a main cause of serious traffic accidents, and therefore, it is necessary to perform fatigue driving detection on a driver to ensure the driving safety of the driver and passengers.
At present, the method for detecting fatigue driving of a driver comprises the following steps: and monitoring statistics such as continuous online time length and working time length of the driver, and judging whether the driver is tired or not based on the statistics. However, in the above manner, the monitored statistic data is single, so that the working strength of the driver cannot be accurately described, and the fatigue state of the driver cannot be accurately represented, thereby causing poor detection accuracy; in addition, the above method has low detection efficiency, and cannot meet the requirement of real-time detection.
Disclosure of Invention
In view of the above, an object of the present application is to provide a training method for a fatigue detection model, a fatigue driving detection method and a device, wherein the fatigue detection model is trained by constructing a training sample including fatigue characteristics and corresponding accident results, and the fatigue state of a driver is detected based on the fatigue detection model, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, by means of the training samples, the fatigue degree of the samples does not need to be marked, and the training efficiency of the model is improved.
In a first aspect, an embodiment of the present application provides a training method for a fatigue detection model, where the training method includes:
constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics;
taking the fatigue characteristics as the input of a decision tree model, and taking fatigue rules as the output of the decision tree model; training a decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model;
establishing a mapping relation between the fatigue characteristics and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
In one possible embodiment, the fatigue characteristics include at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
In one possible embodiment, the constructing training samples based on historical orders includes:
constructing an initial training sample based on a historical order corresponding to a service provider, wherein the initial training sample comprises an initial fatigue characteristic and an accident result whether a traffic accident occurs or not corresponding to the initial fatigue characteristic;
aiming at each node of the decision tree model, obtaining an initial training sample corresponding to the node; aiming at each initial fatigue characteristic in the initial training sample, taking the initial fatigue characteristic as the input of the node to obtain a decision result corresponding to the node;
and if the decision result does not meet the preset condition, deleting the initial fatigue characteristics from the initial training sample to obtain a training sample.
In a possible implementation manner, training a decision tree model according to an output result obtained by the decision tree model based on an input fatigue feature and an accident result corresponding to the fatigue feature to obtain a fatigue rule corresponding to the fatigue detection model includes:
and determining that the corresponding accuracy rate meets a fatigue rule of a second preset threshold value when the output result of the fatigue detection model obtained based on the input fatigue characteristics is matched with the corresponding accident of the fatigue characteristics based on the influence of each fatigue characteristic on the output result of the fatigue detection model.
In a possible implementation manner, the establishing a mapping relationship between the fatigue characteristics and the fatigue degree according to the fatigue rule corresponding to the fatigue detection model includes:
when the fatigue degree corresponding to the training sample is slight fatigue, the fatigue detection model obtains a single fatigue rule based on the fatigue characteristics in the training sample; wherein, the single fatigue rules corresponding to different fatigue characteristics are different;
when the fatigue degree corresponding to the training sample is severe fatigue, the fatigue detection model obtains a combined fatigue rule based on the fatigue characteristics in the training sample; wherein, the combination fatigue rules corresponding to different fatigue characteristics are different.
In one possible embodiment, determining the combined fatigue rule comprises:
determining the time weight value of each single fatigue rule under each training positive sample according to the fatigue feature hit information of each single fatigue rule in each training positive sample; wherein the training positive sample is a training sample of which the corresponding accident result is a traffic accident;
determining a comprehensive time weight value corresponding to the single fatigue rule according to the time weight values of the single fatigue rule under a plurality of training positive samples;
selecting a first candidate fatigue rule and a second candidate fatigue rule from the single fatigue rules according to the comprehensive time weight value corresponding to each single fatigue rule; the comprehensive time weight value corresponding to the first candidate fatigue rule is smaller than a first threshold value, and the comprehensive time weight value corresponding to the second candidate fatigue rule is larger than a second threshold value; wherein the second threshold is greater than or equal to the first threshold;
determining the first candidate fatigue rule and the second candidate fatigue rule as the combined fatigue rule.
In a second aspect, an embodiment of the present application further provides a fatigue driving detection method, where the method includes:
acquiring target fatigue characteristics in a target order corresponding to a service provider;
inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model;
determining a target fatigue degree corresponding to the target order according to the target rule and a mapping relation between the fatigue characteristics and the fatigue degree which is established in advance;
and processing the service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
In one possible embodiment, the target fatigue characteristics include at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
In a possible implementation manner, the determining, according to the target rule and a mapping relationship between fatigue features and fatigue degrees established in advance, a target fatigue degree corresponding to the target order includes:
when any single fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is non-fatigue;
when any single fatigue rule is hit in the target rule but the combined fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is light fatigue; wherein the combined fatigue rule comprises at least two individual fatigue rules;
and when the target rule hits any fatigue rule and hits a combined fatigue rule, determining that the fatigue degree corresponding to the target order is severe fatigue.
In a possible implementation manner, the processing a service provider corresponding to a target order based on a target fatigue degree corresponding to the target order includes:
if the target fatigue degree corresponding to the target order is slight fatigue, carrying out first reminding on a service provider corresponding to the target order through first voice broadcasting;
and if the target fatigue degree corresponding to the target order is severe fatigue, performing second reminding on a service provider corresponding to the target order through second voice broadcasting, and suspending the order distribution to the service provider.
In one possible embodiment, said suspending the allocation of the order to the service provider comprises:
and if the target order is in a driving state, suspending the distribution of the first type of order matched with the driving state to the service provider.
In one possible embodiment, the suspending allocation of the order to the service provider further comprises:
and if the target order is in a driving end state, suspending the distribution of the second type order matched with the driving end state to the service provider.
In a third aspect, an embodiment of the present application further provides a training device for a fatigue detection model, where the training device includes:
the construction module is used for constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics;
the training module is used for taking the fatigue characteristics as the input of a decision tree model and taking fatigue rules as the output of the decision tree model; training a decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model;
the mapping establishing module is used for establishing the mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
In a fourth aspect, an embodiment of the present application further provides a fatigue driving detection apparatus, where the apparatus includes:
the acquisition module is used for acquiring target fatigue characteristics in a target order corresponding to a service provider;
the first processing module is used for inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model;
the determining module is used for determining the target fatigue degree corresponding to the target order according to the target rule and the mapping relation between the fatigue characteristics and the fatigue degree which is established in advance;
and the second processing module is used for processing the service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
In a fifth aspect, an embodiment of the present application further provides an electronic device, including: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating via the bus when the electronic device is running, the processor executing the machine-readable instructions to perform the steps of the training method of the fatigue detection model according to any one of the first aspect.
In a sixth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the training method for a fatigue detection model according to any one of the first aspect.
In a seventh aspect, an embodiment of the present application further provides an electronic device, including: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, when the electronic device runs, the processor and the storage medium communicate through the bus, and the processor executes the machine-readable instructions to execute the steps of the fatigue driving detection method according to any one of the second aspect.
In an eighth aspect, the present embodiments further provide a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the fatigue driving detection method according to any one of the second aspects.
The embodiment of the application provides a training method of a fatigue detection model, a fatigue driving detection method and a fatigue driving detection device, wherein a training sample comprising a fatigue characteristic and an accident result corresponding to the fatigue characteristic is constructed through a historical order corresponding to a service provider, the fatigue detection model is trained based on the training sample, and a fatigue rule corresponding to the fatigue detection model is obtained; and then, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model, and determining the target fatigue degree corresponding to the target order and processing a service provider corresponding to the target order according to the target fatigue characteristics corresponding to the target order in the application process of the fatigue detection model through the mapping relation. According to the method, the fatigue driving detection is carried out by training the fatigue detection model, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the fatigue detection model is trained through the training samples, so that the fatigue degree of the samples does not need to be labeled, and the training efficiency of the model is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
FIG. 1a is a flow chart illustrating a training method of a fatigue detection model according to an embodiment of the present disclosure;
FIG. 1b is a schematic diagram illustrating a continuous fatigue signature provided by an embodiment of the present application;
FIG. 2a is a flow chart of another training method for a fatigue detection model provided by an embodiment of the present application;
2 b-2 d are schematic diagrams illustrating a process of screening initial fatigue features based on a decision tree model according to an embodiment of the present application;
FIG. 3 is a flow chart of another training method for a fatigue detection model provided by an embodiment of the present application;
FIG. 4 is a flow chart illustrating a method for detecting fatigue driving according to an embodiment of the present disclosure;
FIG. 5 is a schematic structural diagram illustrating a training apparatus of a fatigue detection model according to an embodiment of the present application;
fig. 6 is a schematic structural diagram illustrating a fatigue driving detection apparatus according to an embodiment of the present application;
fig. 7 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;
fig. 8 shows a schematic structural diagram of another electronic device provided in an embodiment of the present application.
Detailed Description
In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not used to limit the scope of protection of the present application. Additionally, it should be understood that the schematic drawings are not necessarily drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.
In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
To enable those skilled in the art to utilize the present disclosure, the following embodiments are presented in conjunction with a specific application scenario, "network appointment area". It will be apparent to those skilled in the art that the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the application. Although the present application is described primarily in the context of fatigue driving, it should be understood that this is only one exemplary embodiment.
It should be noted that in the embodiments of the present application, the term "comprising" is used to indicate the presence of the features stated hereinafter, but does not exclude the addition of further features.
The terms "passenger," "requestor," "service requestor," and "customer" are used interchangeably in this application to refer to an individual, entity, or tool that can request or order a service. The terms "driver," "provider," "service provider," and "provider" are used interchangeably in this application to refer to an individual, entity, or tool that can provide a service.
In consideration of the problems that the existing method for detecting fatigue driving of a driver is poor in detection accuracy, low in detection efficiency and incapable of meeting the requirement of real-time detection, the embodiment of the application provides a training method and a training device of a fatigue detection model and a method and a device for detecting fatigue driving, the fatigue detection model is trained by constructing a training sample comprising fatigue characteristics and corresponding accident results, the fatigue state of the driver is detected based on the fatigue detection model, and a service provider corresponding to a target order is processed based on the detection result, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, by means of the training samples, the fatigue degree of the samples does not need to be marked, and the training efficiency of the model is improved.
The training method of the fatigue detection model and the fatigue driving detection method provided by the embodiment of the application can be applied to terminal equipment and can also be applied to a server. The following describes the training method of the fatigue detection model and the fatigue driving detection method in detail by taking an example in which the above-described methods are applied to a server.
Referring to fig. 1a, a schematic flowchart of a training method for a fatigue detection model according to a first embodiment of the present application is shown, where the method may be executed by a server, and the training method specifically includes:
s101, constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics.
S102, taking the fatigue characteristics as input of a decision tree model, and taking fatigue rules as output of the decision tree model; and training the decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model.
S103, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
The embodiment of the application provides a training method of a fatigue detection model, wherein a training sample comprising a fatigue characteristic and an accident result corresponding to the fatigue characteristic is constructed through a historical order corresponding to a service provider, the fatigue detection model is trained based on the training sample, and a fatigue rule corresponding to the fatigue detection model is obtained; and then, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model, and determining the target fatigue degree corresponding to the target order and processing a service provider corresponding to the target order according to the target fatigue characteristics corresponding to the target order in the application process of the fatigue detection model through the mapping relation. According to the method, the fatigue driving detection is carried out by training the fatigue detection model, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the fatigue detection model is trained through the training samples, so that the fatigue degree of the samples does not need to be labeled, and the training efficiency of the model is improved.
The above exemplary steps of the embodiments of the present application are described below:
s101, constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics.
In the embodiment of the application, the training samples comprise positive samples and negative samples; and constructing a positive sample based on the fatigue driving traffic accident order, wherein the positive sample correspondingly comprises the fatigue characteristics and the accident result of the traffic accident corresponding to the fatigue characteristics. Constructing a negative sample based on other orders except the accident order on the same day of the traffic accident event, wherein the negative sample correspondingly comprises the fatigue characteristics and the accident result which is corresponding to the fatigue characteristics and does not have the traffic accident. Here, the negative sample may also be constructed by using other historical orders except the accident order in other time periods, and the negative sample is constructed by using other orders except the accident order on the same day when the traffic accident event occurs, so as to ensure the similarity of other order characteristics except the fatigue characteristic, and thus, the correlation between the fatigue characteristic and the accident result (i.e., the fatigue result) can be improved, and the accuracy of the trained fatigue detection model is further improved; wherein, other order characteristics can be: time characteristics, weather characteristics (e.g., all rainy days), date characteristics (e.g., all monday, all weekends, all holidays), and the like.
In the embodiment of the present application, for a specific scenario of fatigue driving, the fatigue characteristics include: basic characterization features of the service provider (i.e. driver) (basic action features of the service provider), continuous fatigue features of the service provider, order features. The three specific fatigue characteristics are described below:
1. the basic action features of the service provider (i.e. the basic characterization features of the service provider) include at least: the number of times of opening and closing the eyes, the length of time of opening and closing the eyes, the number of times of opening and closing the mouth, and the length of time of opening and closing the mouth of the service provider in the latest time period. For example, the number of times the driver opens and closes his eyes in 10 minutes, the length of time each eye opens and closes, the number of times the driver opens and closes his mouth in 10 minutes, the length of time each mouth opens and closes, etc.
2. The continuous fatigue characteristics of the service provider include at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; and the time interval between every two basic actions which are continuously triggered in the period of time is smaller than a first preset threshold value. As shown in fig. 1b, the continuous fatigue characteristics are calculated by recording the number, interval, and duration of the eyes and mouth opening and closing of each driver occurring continuously in a short time interval (e.g., 5 minutes), for example, the continuous fatigue characteristics include: the driver opens and closes the eyes or opens and closes the mouth 4 times continuously in 20 minutes, the time interval of each open and close eyes or opens and closes the mouth and basic movements before or after the basic movement in 20 minutes is less than the first preset threshold (for example, 5 minutes); the above continuous fatigue characteristics further include: a specific time interval between each of the opened and closed eyes or the opened and closed mouth and the basic movement before or after the basic movement within the 20 minutes (for example, 3 minutes or 4 minutes); the above continuous fatigue characteristics further include: the duration of each basic action is 20 minutes, for example, the first time the eyes are opened and closed, the duration is 2 minutes, the corresponding duration ratio is 2/20, and the instant duration ratio is 10%.
3. The order characteristics at least comprise order time and order distance; here, the order time includes a nighttime order, a afternoon order, a morning order (morning peak order, general morning order), and the like, and the order distance refers to a travel distance in the order.
In the embodiment of the application, as described above, in addition to the improvement of the accuracy of the basic characterization and identification of the driver, the analysis of the data of the heavy fatigue traffic accidents shows that most drivers with fatigue accidents have continuous yawning and dozing eye-closing actions for many times before the accidents happen, so that the invention also innovatively designs and develops the continuous fatigue characteristics of the drivers, the characteristics can be used for not only the scenes of the fatigue driving traffic accidents related to the invention, but also the effective input characteristics of other scenes needing fatigue judgment, and the detection accuracy of the trained fatigue detection model can be improved through the characteristics.
S102, taking the fatigue characteristics as input of a decision tree model, and taking fatigue rules as output of the decision tree model; and training the decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model.
In the embodiment of the present application, dt (X) represents a fatigue rule, X represents a fatigue feature of any training sample, X represents a training sample set, xi ∈ X, i ═ 1, 2, …, n; wherein i represents any training sample in the training sample set, and n represents the number of training samples in the training sample set.
In the embodiment of the application, an interpretable decision tree model is used, a training sample set X is used, fatigue characteristics X of the training sample are used as input of the decision tree model, and a fatigue rule DT (X) is used as output of the decision tree model to train the decision tree model. The specific training process is as follows: taking fatigue characteristics x of a training sample as input of a decision tree model, determining loss of the decision tree model according to an output result DT (x) obtained by the decision tree model based on the input fatigue characteristics x and an accident result corresponding to the fatigue characteristics x, and adjusting model parameters of the decision tree model based on the loss until the output result DT (x) obtained by the decision tree model based on the input fatigue characteristics x is matched with the accident result corresponding to the fatigue characteristics x (namely the loss meets a preset condition, for example, the loss value is smaller than a third preset threshold), so as to obtain a trained fatigue detection model; wherein the fatigue detection model comprises trained model parameters.
In the process of training a decision tree model, information gain (information gain) is used as a division standard, a feature set of each sample data set is determined according to the sample data sets, a training target of the model selects a test condition for maximizing the information gain each time to divide nodes, and an optimization algorithm is usually calculated by a greedy algorithm (greedy algorithm).
Wherein, the model parameters obtained by the output training are used as the input of the on-line partial calculation: for the off-line part, calculating all model parameters and storing the model parameters in a parameter file; and for the online part, after receiving the fatigue characterization data of the driver, acquiring the characteristic data in the request, loading the model parameters, and calculating to obtain the risk degree score of the order.
After the trained fatigue detection model is obtained, in addition, a fatigue rule corresponding to the fatigue detection model can be obtained, and the fatigue rule is a rule representing fatigue characteristics and accident results. In a specific embodiment, based on the influence of each fatigue feature on the output result of the fatigue detection model, when the output result obtained by the fatigue detection model based on the input fatigue feature matches with the accident corresponding to the fatigue feature, the corresponding accuracy rate meets a fatigue rule of a second preset threshold.
The fatigue rule may be one or multiple, and in general, the fatigue rule is multiple, and the accuracy corresponding to each fatigue rule meets a preset threshold (that is, the accuracy corresponding to each fatigue rule is greater than a second preset threshold).
S103, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
In the embodiment of the application, after the fatigue rule corresponding to the fatigue detection model is obtained, the mapping relation between the fatigue characteristics and the fatigue degree is established; wherein, the fatigue degree comprises non-fatigue, light fatigue and severe fatigue. According to the fatigue rule corresponding to the fatigue detection model, establishing a mapping relation between the fatigue characteristics and the fatigue degree, wherein the mapping relation comprises the following steps:
when the fatigue degree corresponding to the training sample is slight fatigue, the fatigue detection model obtains a single fatigue rule based on the fatigue characteristics in the training sample; wherein, the single fatigue rules corresponding to different fatigue characteristics are different;
when the fatigue degree corresponding to the training sample is severe fatigue, the fatigue detection model obtains a combined fatigue rule based on the fatigue characteristics in the training sample; wherein, the combination fatigue rules corresponding to different fatigue characteristics are different.
For example, non-fatigue is indicated by 0, mild fatigue is indicated by 1, and severe fatigue is indicated by 2.
Establishing a mapping f, defining p ═ f (x), wherein the establishing method of the mapping f adopts a decision tree model DT (x), and dt isie.DT is a single rule of the decision tree model (i.e. the above single fatigue rule), which is a single sample, p is the severity of fatigue of sample x (where 0 is non-fatigue, 1 is mild fatigue, and 2 is severe fatigue), and the set U is (DT)1,DT2),DT1,The rules are upgraded for the combination (i.e., the combination fatigue rules described above).
In the application stage of the fatigue detection model, the fatigue detection model obtains an output result based on the target fatigue characteristics corresponding to the target order, and the target fatigue degree corresponding to the target order can be determined based on the output result corresponding to the target order through the mapping relation.
Further, as shown in fig. 2a, the training method for a fatigue detection model provided in the embodiment of the present application, where the training sample is constructed based on a historical order, includes:
s201, constructing an initial training sample based on a historical order corresponding to a service provider, wherein the initial training sample comprises an initial fatigue characteristic and an accident result whether a traffic accident happens or not corresponding to the initial fatigue characteristic.
S202, aiming at each node of the decision tree model, obtaining an initial training sample corresponding to the node; and aiming at each initial fatigue characteristic in the initial training sample, taking the initial fatigue characteristic as the input of the node to obtain a decision result corresponding to the node.
And S203, if the decision result does not meet the preset condition, deleting the initial fatigue characteristics from the initial training sample to obtain a training sample.
With reference to steps 201 to 203, the initial training sample constructed in the embodiment of the present application includes an initial fatigue characteristic and an accident result corresponding to the initial fatigue characteristic. In practice, not all the initial fatigue features can improve the detection accuracy of the fatigue detection model, and therefore, in the embodiment of the present application, effective fatigue features need to be screened out from the initial fatigue features, and the fatigue detection model needs to be trained based on the effective fatigue features.
The method comprises the steps that a decision tree model comprises a plurality of nodes, for each node of the decision tree model, an initial training sample (namely a data set) corresponding to the node is obtained firstly, then, for each initial fatigue feature in the initial training sample, the initial fatigue feature is used as the input of the node, a decision result corresponding to the node is obtained, the decision result is a fatigue rule, a server judges whether the decision result is matched with a distribution rule and/or a conventional cognitive rule of a positive sample or a negative sample or not (namely whether the fatigue rule is matched with the distribution rule and/or the conventional cognitive rule of the negative sample or not), and if the decision result is not matched with the distribution rule and/or the conventional cognitive rule of the negative sample, the initial fatigue feature is deleted.
Referring to fig. 2 b-2 d, when the decision tree model selects an unexplainable initial fatigue feature for model determination (as shown in fig. 2 b), the initial fatigue feature is deleted from the data set corresponding to the current node, the training sample is reconstructed and the decision tree model is trained (as shown in fig. 2 c), and so on until the classification feature selected by the decision tree model can be explained (as shown in fig. 2 d). For example, taking the feature "average duration of characterization interval" as an example, if the result automatically output by a certain node in the decision tree model is "average duration of characterization interval is greater than a threshold", however, the conventional cognitive rule is: the shorter the interval time of the appearance of the continuous fatigue representation is, the more likely the continuous fatigue representation is in a fatigue state; therefore, here the "characterize interval mean time" feature needs to be deleted from the node in the decision tree model. It should be noted that: the decision tree model performs condition judgment on each node to divide the data set into two subsets according to conditions, and overfitting of the model can be caused when the judgment condition of a certain node does not accord with the distribution of the positive samples. Because the number of positive samples is extremely small, the model is difficult to select reasonable judgment conditions at each node, and in order to avoid the problem, the screening algorithm is designed to delete the initial fatigue characteristics which do not meet the conditions.
The process of screening for effective initial fatigue characteristics described above is as follows: first, the decision tree is automatically trained to obtain the result (as shown in FIG. 2 b), wherein the model is used for feature in FIG. 2b2<F2 is determined to be a positive sample and is in feature with the original positive sample2The feature is found in the distribution2<Since the f2 criterion does not satisfy the original distribution, the feature is deleted from the S2 data set2The features are trained to obtain results (as shown in FIG. 2 c), and features are foundmIf the characteristics still do not accord with the conventional cognitive rules, continuing to delete featuremTraining is carried out to finally obtain featurenFeatures conform to the distribution (as shown in FIG. 2 d), feature is selectednThe characteristics are the same for different nodes of other layers.
Further, as shown in fig. 3, in the training method of the fatigue detection model provided in the embodiment of the present application, determining the combined fatigue rule includes:
s301, determining a time weight value of each single fatigue rule under each training positive sample according to fatigue feature hit information of each single fatigue rule in each training positive sample; and the training positive sample is a training sample of which the corresponding accident result is a traffic accident.
In the embodiment of the application, after the fatigue detection models are trained, fatigue rules corresponding to the fatigue detection models can be obtained. For each fatigue rule, obtaining fatigue feature hit information of the fatigue rule in each training positive sample (namely a positive sample constructed by an accident order), and then determining a time weight value of the fatigue rule under the training positive sample according to the fatigue feature hit information of the fatigue rule in each training positive sample. The fatigue feature hit information is a position where a representation of the fatigue rule (i.e., a basic action of the service provider) is hit, and the number of the position representations (i.e., the basic actions of the service provider) is, for example, the first representation (i.e., the basic action of the service provider) corresponds to position 1; the second token (i.e., the basic action of the service provider) corresponds to position 2.
For example, obtaining fatigue rules corresponding to the multiple fatigue detection models as fatigue rule 1 to fatigue rule 5, and the training positive sample includes: training positive sample 1-training positive sample 5. For fatigue rule 1, hit in the 2 nd, 4 th, 5 th, 6 th, and 7 th characterizations of training positive sample 1, and accordingly, the time weight value of fatigue rule 1 in training positive sample 1 is: (1/2+1/4+1/5+1/6+1/7)/5 is 0.252.
S302, determining a comprehensive time weight value corresponding to the single fatigue rule according to the time weight values of the single fatigue rule under the training positive samples.
In the embodiment of the application, each fatigue rule corresponds to a plurality of training samples. And aiming at each fatigue rule, calculating a comprehensive time weight value corresponding to the fatigue rule according to the time weight values of the fatigue rule under the corresponding training samples.
For example, the fatigue rule 1 corresponds to the training positive sample 1 to the training positive sample 5, and the time weight value of the fatigue rule 1 in the training positive sample 1 is 0.252, the time weight value of the fatigue rule 1 in the training positive sample 1 is 0.101 … …, and the like by the calculation method in step 301. Accordingly, the comprehensive time weight value corresponding to the fatigue rule 1 is 0.252+0.101+ the time weight value corresponding to the training positive sample 3 + the time weight value corresponding to the training positive sample 4+ the time weight value corresponding to the training positive sample 5.
S303, selecting a first candidate fatigue rule and a second candidate fatigue rule from the single fatigue rules according to the comprehensive time weight value corresponding to each single fatigue rule; the comprehensive time weight value corresponding to the first candidate fatigue rule is smaller than a first threshold value, and the comprehensive time weight value corresponding to the second candidate fatigue rule is larger than a second threshold value; wherein the second threshold is greater than or equal to the first threshold.
In the embodiment of the present application, a larger integrated time weight value indicates that the hit time point (i.e., the hit position) of the corresponding fatigue rule is earlier. In this way, a first candidate fatigue rule before the hit time point (i.e. smaller than the first threshold) and a second candidate fatigue rule after the hit time point (i.e. larger than the second threshold) are selected by the comprehensive time weight value of each fatigue rule to form a combined fatigue rule.
Here, the second threshold may be the same as or larger than the first threshold.
S304, determining the first candidate fatigue rule and the second candidate fatigue rule as the combined fatigue rule.
In the embodiment of the present application, in the manner of step 303, each selected group of the first candidate fatigue rule and the second candidate fatigue rule is a rule pair, that is, a pair of combined fatigue rules. And detecting severe fatigue through the combined fatigue rule.
The embodiment of the application provides a training method of a fatigue detection model, wherein a training sample comprising a fatigue characteristic and an accident result corresponding to the fatigue characteristic is constructed through a historical order corresponding to a service provider, the fatigue detection model is trained based on the training sample, and a fatigue rule corresponding to the fatigue detection model is obtained; and then, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model, and determining the target fatigue degree corresponding to the target order and processing a service provider corresponding to the target order according to the target fatigue characteristics corresponding to the target order in the application process of the fatigue detection model through the mapping relation. According to the method, the fatigue driving detection is carried out by training the fatigue detection model, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the fatigue detection model is trained through the training samples, so that the fatigue degree of the samples does not need to be labeled, and the training efficiency of the model is improved.
Referring to fig. 4, a flowchart of a method for detecting fatigue driving according to a second embodiment of the present application is shown, where the method may be executed by a server, and the method specifically includes:
s401, acquiring target fatigue characteristics in a target order corresponding to a service provider.
S402, inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model.
And S403, determining the target fatigue degree corresponding to the target order according to the target rule and the mapping relation between the fatigue characteristics and the fatigue degree established in advance.
S404, processing a service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
The embodiment of the application provides a fatigue driving detection method, which can determine a target fatigue degree corresponding to a target order based on a target fatigue characteristic corresponding to the target order through a fatigue detection model and a preset mapping relation between a fatigue rule and the fatigue degree, and intervene a service provider corresponding to the target order when determining that a driver is in a fatigue state, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the safety of the service provider and/or the service requester can be ensured by intervening the service provider corresponding to the target order.
The above exemplary steps of the embodiments of the present application are described below:
s401, acquiring target fatigue characteristics in a target order corresponding to a service provider.
In an embodiment of the present application, the target fatigue characteristics include at least one of: basic action characteristics of the service provider (i.e. basic characterization characteristics of the service provider); continuous fatigue characteristics of the service provider; order characteristics;
1. wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; for example, the number of times the driver opens and closes his eyes in 10 minutes, the length of time each eye opens and closes, the number of times the driver opens and closes his mouth in 10 minutes, the length of time each mouth opens and closes, etc.
2. The continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; the time interval between every two continuous triggered basic actions in the period of time is smaller than a first preset threshold value;
in the embodiment of the present application, during the actual order process, there may be multiple basic motion characteristics (i.e., basic characteristic characteristics of the service provider) corresponding to the target order over time, for example, at 10 minutes of the target order, there may be one basic motion characteristic (e.g., mouth open and close) corresponding to the target order, at 11 minutes of the target order, there may be one basic motion characteristic (e.g., mouth open and close) corresponding to the target order, at 14 minutes of the target order, and the like. Wherein the continuous fatigue characteristics are calculated by recording the times, intervals, time lengths and the like of the eyes and the mouths of the drivers continuously opening and closing within a short time interval (for example, 5 minutes), for example, the continuous fatigue characteristics include: the driver opens and closes the eyes or opens and closes the mouth 4 times continuously in 20 minutes, the time interval of each open and close eyes or opens and closes the mouth and basic movements before or after the basic movement in 20 minutes is less than the first preset threshold (for example, 5 minutes); the above continuous fatigue characteristics further include: a specific time interval between each of the opened and closed eyes or the opened and closed mouth and the basic movement before or after the basic movement within the 20 minutes (for example, 3 minutes or 4 minutes); the above continuous fatigue characteristics further include: the duration of each basic action is 20 minutes, for example, the first time the eyes are opened and closed, the duration is 2 minutes, the corresponding duration ratio is 2/20, and the instant duration ratio is 10%.
3. The order characteristics at least comprise order time and order distance. Here, the order time includes a nighttime order, a afternoon order, a morning order (morning peak order, general morning order), and the like, and the order distance refers to a travel distance in the order.
S402, inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model.
In the embodiment of the application, the fatigue detection model calculates the target fatigue characteristics based on the pre-trained model parameters and outputs the target rule.
And S403, determining the target fatigue degree corresponding to the target order according to the target rule and the mapping relation between the fatigue characteristics and the fatigue degree established in advance.
The mapping relationship between the fatigue characteristics and the fatigue degree comprises: when the fatigue degree corresponding to the training sample is mild fatigue, the fatigue detection model obtains a single fatigue rule based on the fatigue characteristics in the training sample; wherein, the single fatigue rules corresponding to different fatigue characteristics are different; when the fatigue degree corresponding to the training sample is severe fatigue, the fatigue detection model obtains a combined fatigue rule based on the fatigue characteristics in the training sample; wherein, the combination fatigue rules corresponding to different fatigue characteristics are different.
In the embodiment of the present application, the fatigue degree is divided into three levels, including: non-fatigue, mild fatigue, severe fatigue.
Specifically, the method for determining the target fatigue degree corresponding to the target order includes:
when the target rule does not hit any one fatigue rule, determining that the fatigue degree corresponding to the target order is non-fatigue; for example, the target rule is to open and close eyes every 20s without opening and closing mouth; correspondingly, the target rule does not hit any single fatigue rule, and in this case, the fatigue degree corresponding to the target order is determined to be non-fatigue.
When any single fatigue rule is hit but the combined fatigue rule is not hit by the target rule, determining that the fatigue degree corresponding to the target order is light fatigue; wherein the combined fatigue rule comprises at least two individual fatigue rules; for example, the target rule is a case where there are a plurality of times of opening and closing eyes and opening and closing mouths within a certain 1 minute, and accordingly, the target rule hits a single fatigue rule but does not hit a combined fatigue rule, and in this case, the fatigue degree corresponding to the target order is determined to be light fatigue.
And when any fatigue rule is hit by the target rule and the combined fatigue rule is hit by the target rule, determining that the fatigue degree corresponding to the target order is severe fatigue. For example, the target rule is a condition that eyes are opened and closed and mouths are opened and closed frequently within a period of time (for example, within 10 minutes), and accordingly, the target rule hits a single fatigue rule and a combined fatigue rule, and in this case, the fatigue degree corresponding to the target order is determined to be severe fatigue.
S404, processing a service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
In the embodiment of the application, when the service provider corresponding to the target order is determined to be in the fatigue state based on the target fatigue degree corresponding to the target order, the fatigue driving behavior of the service provider is subjected to real-time intervention and control. Here, the intervention management and control means may be: voice broadcast, for example, by broadcasting from Text To Speech (TTS) to remind the driver to take a rest; the driver's dispatch may also be masked to force the driver to rest.
In the embodiment of the application, when the driver uploads the representation data to hit the strategy model in the journey, TTS voice broadcast is called to the driver, and according to the model, the degree of fatigue severity is judged to broadcast mild fatigue and severe fatigue broadcast documents to the driver respectively. As shown in fig. 5, in the fatigue driving detection method provided in the embodiment of the present application, the processing a service provider corresponding to a target order based on a target fatigue degree corresponding to the target order includes:
firstly, if the target fatigue degree corresponding to the target order is light fatigue, carrying out first reminding on a service provider corresponding to the target order through first voice broadcasting.
In the embodiment of this application, mild tired lead to the voice broadcast can, specific, the file set that matches mild tired first voice broadcast can include:
1. the dripping is safe to remind, the driver needs to be vigilant when driving the car to doze, and the window can be properly opened for ventilation;
2. the dropping safety is reminded, the user is tired of stopping and wiping the face, and moves forward again after waking;
3. drip safety reminding is carried out, driving fatigue is frequent, and deep breathing is tried;
4. dripping safety reminding, and giving attention to windowing for ventilation and adjusting sitting posture to avoid fatigue driving after long-time driving;
5. drip safety reminding, deep breathing or music listening is helpful for relieving fatigue;
6. the dripping is safe to remind, fatigue driving needs to be avoided, and tea drinking and refreshing are the simplest and most convenient.
And secondly, if the target fatigue degree corresponding to the target order is severe fatigue, performing second reminding on a service provider corresponding to the target order through second voice broadcast, and suspending the order distribution to the service provider.
In the embodiment of the application, besides voice broadcasting, severe fatigue can also suspend the allocation of orders to the service providers, so as to force the corresponding service providers to rest.
Specifically, the set of documents matching the heavily fatigued second voice broadcast may include:
1. the dripping is particularly safe to remind, people are easy to feel tired when driving for a long time, and people are required to stop properly to have a rest after finishing the driving order and do not drive tired; 2. the dripping is safe and specially reminded, fatigue driving is too dangerous, feeling fatigue is stopped once, and rest are carried out before;
3. the dripping safety is specially reminded, the hidden danger of fatigue driving is more, the people need to pay attention to windowing and ventilation to keep clear-headed, and the fatigue driving is avoided;
4. the dripping safety is specially reminded, the vehicle is easy to fatigue after driving for too long time, the people need to pay attention to properly adjust the sitting posture, window opening is carried out in time for ventilation, and fatigue driving is avoided;
5. the dripping safety is specially reminded, the potential safety hazard of fatigue driving is high, the window is noticed to ventilate, the clear-headed state is kept, and the fatigue driving is avoided.
Further, in this embodiment of the present application, the suspending the allocation of the order to the service provider includes:
and if the target order is in a driving state, suspending the distribution of the first type of order matched with the driving state to the service provider.
And if the target order is in a driving end state, suspending the distribution of the second type order matched with the driving end state to the service provider.
In the embodiment of the application, when a driver is seriously tired, intervention is carried out according to the current order state, when the order is in a journey, a 'shielding order dispatching' strategy is triggered, and a car sharing order and a serial order dispatching are not dispatched for the tired driver; and when the order state is finished, triggering a forced vehicle receiving strategy to force the driver to stop for rest.
The embodiment of the application provides a fatigue driving detection method, which can determine a target fatigue degree corresponding to a target order based on a target fatigue characteristic corresponding to the target order through a fatigue detection model and a preset mapping relation between a fatigue rule and the fatigue degree, and intervene a service provider corresponding to the target order when determining that a driver is in a fatigue state, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the safety of the service provider and/or the service requester can be ensured by intervening the service provider corresponding to the target order.
Based on the same inventive concept, a training device of a fatigue detection model corresponding to the training method of the fatigue detection model is also provided in the third embodiment of the present application, and as the principle of solving the problem of the device in the third embodiment of the present application is similar to the training method of the fatigue detection model provided in the first embodiment of the present application, the implementation of the device can refer to the implementation of the method, and repeated details are not repeated.
Referring to fig. 5, a training device for a fatigue detection model according to a third embodiment of the present application is provided, where the training device includes:
a constructing module 501, configured to construct a training sample based on a historical order, where the training sample includes a fatigue feature and an accident result whether a traffic accident occurs or not corresponding to the fatigue feature;
a training module 502, configured to use the fatigue characteristics as an input of a decision tree model, and use a fatigue rule as an output of the decision tree model; training a decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model;
a mapping establishing module 503, configured to establish a mapping relationship between the fatigue feature and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
In one possible embodiment, the fatigue characteristics include at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
In one possible implementation, the constructing module 501 constructs training samples based on historical orders, including:
constructing an initial training sample based on a historical order corresponding to a service provider, wherein the initial training sample comprises an initial fatigue characteristic and an accident result whether a traffic accident occurs or not corresponding to the initial fatigue characteristic;
aiming at each node of the decision tree model, obtaining an initial training sample corresponding to the node; aiming at each initial fatigue characteristic in the initial training sample, taking the initial fatigue characteristic as the input of the node to obtain a decision result corresponding to the node;
and if the decision result does not meet the preset condition, deleting the initial fatigue characteristics from the initial training sample to obtain a training sample.
In a possible implementation manner, the training module 502 trains the decision tree model according to an output result obtained by the decision tree model based on the input fatigue feature and an accident result corresponding to the fatigue feature, and obtains a fatigue rule corresponding to the fatigue detection model, including:
and determining that the corresponding accuracy rate meets a fatigue rule of a second preset threshold value when the output result of the fatigue detection model obtained based on the input fatigue characteristics is matched with the corresponding accident of the fatigue characteristics based on the influence of each fatigue characteristic on the output result of the fatigue detection model.
In a possible implementation manner, the mapping establishing module 503 establishes a mapping relationship between the fatigue feature and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model, including:
when the fatigue degree corresponding to the training sample is slight fatigue, the fatigue detection model obtains a single fatigue rule based on the fatigue characteristics in the training sample; wherein, the single fatigue rules corresponding to different fatigue characteristics are different;
when the fatigue degree corresponding to the training sample is severe fatigue, the fatigue detection model obtains a combined fatigue rule based on the fatigue characteristics in the training sample; wherein, the combination fatigue rules corresponding to different fatigue characteristics are different.
In a possible embodiment, the training device further comprises:
the first determining module is used for determining the time weight value of each single fatigue rule under each training positive sample according to the fatigue feature hit information of each single fatigue rule in each training positive sample; wherein the training positive sample is a training sample of which the corresponding accident result is a traffic accident;
the second determining module is used for determining a comprehensive time weight value corresponding to the single fatigue rule according to the time weight values of the single fatigue rule under the plurality of training positive samples;
the selection module is used for selecting a first candidate fatigue rule and a second candidate fatigue rule from the single fatigue rules according to the comprehensive time weight value corresponding to each single fatigue rule; the comprehensive time weight value corresponding to the first candidate fatigue rule is smaller than a first threshold value, and the comprehensive time weight value corresponding to the second candidate fatigue rule is larger than a second threshold value; wherein the second threshold is greater than or equal to the first threshold;
a third determining module for determining the first candidate fatigue rule and the second candidate fatigue rule as the combined fatigue rule.
The embodiment of the application provides a training device of a fatigue detection model, which constructs a training sample comprising fatigue characteristics and accident results corresponding to the fatigue characteristics through a historical order corresponding to a service provider, trains the fatigue detection model based on the training sample, and obtains a fatigue rule corresponding to the fatigue detection model; and then, establishing a mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model, and determining the target fatigue degree corresponding to the target order and processing a service provider corresponding to the target order according to the target fatigue characteristics corresponding to the target order in the application process of the fatigue detection model through the mapping relation. According to the method, the fatigue driving detection is carried out by training the fatigue detection model, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the fatigue detection model is trained through the training samples, so that the fatigue degree of the samples does not need to be labeled, and the training efficiency of the model is improved.
Based on the same inventive concept, a fatigue driving detection device corresponding to the fatigue driving detection method is also provided in the fourth embodiment of the present application, and as the principle of solving the problem of the device in the fourth embodiment of the present application is similar to the above-mentioned fatigue driving detection method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and repeated details are not repeated.
Referring to fig. 6, a fatigue driving detecting apparatus according to a fourth embodiment of the present application includes:
an obtaining module 601, configured to obtain a target fatigue characteristic in a target order corresponding to a service provider;
the first processing module 602 is configured to input the target fatigue characteristics into a fatigue detection model trained in advance, so as to obtain a target rule corresponding to the fatigue detection model;
a determining module 603, configured to determine a target fatigue degree corresponding to the target order according to the target rule and a mapping relationship between fatigue characteristics and fatigue degrees established in advance;
the second processing module 604 is configured to process, based on the target fatigue degree corresponding to the target order, a service provider corresponding to the target order.
In one possible embodiment, the target fatigue characteristics include at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
In a possible implementation manner, the determining module 603 determines, according to the target rule and a mapping relationship between fatigue features and fatigue degrees established in advance, a target fatigue degree corresponding to the target order, including:
when any single fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is non-fatigue;
when any single fatigue rule is hit in the target rule but the combined fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is light fatigue; wherein the combined fatigue rule comprises at least two individual fatigue rules;
and when the target rule hits any fatigue rule and hits a combined fatigue rule, determining that the fatigue degree corresponding to the target order is severe fatigue.
In a possible implementation manner, the second processing module 604 processes, based on the target fatigue degree corresponding to the target order, the service provider corresponding to the target order, including:
if the target fatigue degree corresponding to the target order is slight fatigue, carrying out first reminding on a service provider corresponding to the target order through first voice broadcasting;
and if the target fatigue degree corresponding to the target order is severe fatigue, performing second reminding on a service provider corresponding to the target order through second voice broadcasting, and suspending the order distribution to the service provider.
In one possible embodiment, the second processing module 604 suspends the allocation of orders to the service provider, including:
and if the target order is in a driving state, suspending the distribution of the first type of order matched with the driving state to the service provider.
In one possible implementation, the second processing module 604 suspends allocating orders to the service provider, further comprising:
and if the target order is in a driving end state, suspending the distribution of the second type order matched with the driving end state to the service provider.
The embodiment of the application provides a fatigue driving detection device, which can determine the target fatigue degree corresponding to a target order based on the target fatigue characteristics corresponding to the target order through a fatigue detection model and a preset mapping relation between a fatigue rule and the fatigue degree, and intervene a service provider corresponding to the target order when determining that a driver is in a fatigue state, so that the detection efficiency and the detection accuracy are improved, and the requirement of real-time detection can be met; moreover, the safety of the service provider and/or the service requester can be ensured by intervening the service provider corresponding to the target order.
As shown in fig. 7, a fifth embodiment of the present application provides an electronic device 700, including: a processor 701, a memory 702 and a bus, wherein the memory 702 stores machine-readable instructions executable by the processor 701, when the electronic device runs, the processor 701 communicates with the memory 702 through the bus, and the processor 701 executes the machine-readable instructions to execute the steps of the training method of the fatigue detection model as in the first embodiment.
Specifically, the memory 702 and the processor 701 can be general-purpose memory and processor, which are not limited in particular, and when the processor 701 runs a computer program stored in the memory 702, the training method of the fatigue detection model in the first embodiment can be performed.
A sixth embodiment of the present application further provides a computer-readable storage medium, which stores thereon a computer program, which, when executed by a processor, performs the steps of the training method of the fatigue detection model in the first embodiment.
As shown in fig. 8, a seventh embodiment of the present application provides an electronic device 800, including: a processor 801, a memory 802 and a bus, wherein the memory 802 stores machine-readable instructions executable by the processor 801, when the electronic device is operated, the processor 801 communicates with the memory 802 through the bus, and the processor 801 executes the machine-readable instructions to execute the steps of the fatigue driving detection method as in the second embodiment.
Specifically, the memory 802 and the processor 801 can be general-purpose memories and processors, which are not specifically limited herein, and the fatigue driving detection method in the second embodiment described above can be executed when the processor 801 executes a computer program stored in the memory 802.
Corresponding to the fatigue driving detection method in the second embodiment, an eighth embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, where the computer program is executed by a processor to perform the steps of the fatigue driving detection method in the second embodiment.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to corresponding processes in the method embodiments, and are not described in detail in this application. In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and there may be other divisions in actual implementation, and for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or modules through some communication interfaces, and may be in an electrical, mechanical or other form.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (18)
1. A training method of a fatigue detection model, the training method comprising:
constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics;
taking the fatigue characteristics as the input of a decision tree model, and taking fatigue rules as the output of the decision tree model; training a decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model; in the training process of the decision tree model, an output result obtained by input fatigue characteristics and an accident result corresponding to the fatigue characteristics are used for determining the loss of the decision tree model;
establishing a mapping relation between the fatigue characteristics and the fatigue degree according to a fatigue rule corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
2. A method of training a fatigue detection model according to claim 1, wherein the fatigue features comprise at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
3. The training method of the fatigue detection model according to claim 2, wherein the constructing training samples based on historical orders comprises:
constructing an initial training sample based on a historical order corresponding to a service provider, wherein the initial training sample comprises an initial fatigue characteristic and an accident result whether a traffic accident occurs or not corresponding to the initial fatigue characteristic;
aiming at each node of the decision tree model, obtaining an initial training sample corresponding to the node; aiming at each initial fatigue characteristic in the initial training sample, taking the initial fatigue characteristic as the input of the node to obtain a decision result corresponding to the node;
and if the decision result does not meet the preset condition, deleting the initial fatigue characteristics from the initial training sample to obtain a training sample.
4. The training method of the fatigue detection model according to claim 1, wherein training the decision tree model according to the output result obtained by the decision tree model based on the input fatigue feature and the accident result corresponding to the fatigue feature to obtain the fatigue rule corresponding to the fatigue detection model comprises:
and determining that the corresponding accuracy rate meets a fatigue rule of a second preset threshold value when the output result of the fatigue detection model obtained based on the input fatigue characteristics is matched with the corresponding accident of the fatigue characteristics based on the influence of each fatigue characteristic on the output result of the fatigue detection model.
5. The training method of the fatigue detection model according to claim 1, wherein the establishing of the mapping relationship between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model comprises:
when the fatigue degree corresponding to the training sample is slight fatigue, the fatigue detection model obtains a single fatigue rule based on the fatigue characteristics in the training sample; wherein, the single fatigue rules corresponding to different fatigue characteristics are different;
when the fatigue degree corresponding to the training sample is severe fatigue, the fatigue detection model obtains a combined fatigue rule based on the fatigue characteristics in the training sample; wherein, the combination fatigue rules corresponding to different fatigue characteristics are different.
6. A training method for a fatigue detection model according to claim 5, wherein determining the combined fatigue rule comprises:
determining the time weight value of each single fatigue rule under each training positive sample according to the fatigue feature hit information of each single fatigue rule in each training positive sample; wherein the training positive sample is a training sample of which the corresponding accident result is a traffic accident;
determining a comprehensive time weight value corresponding to the single fatigue rule according to the time weight values of the single fatigue rule under a plurality of training positive samples;
selecting a first candidate fatigue rule and a second candidate fatigue rule from the single fatigue rules according to the comprehensive time weight value corresponding to each single fatigue rule; the comprehensive time weight value corresponding to the first candidate fatigue rule is smaller than a first threshold value, and the comprehensive time weight value corresponding to the second candidate fatigue rule is larger than a second threshold value; wherein the second threshold is greater than or equal to the first threshold;
determining the first candidate fatigue rule and the second candidate fatigue rule as the combined fatigue rule.
7. A method of detecting fatigue driving, the method comprising:
acquiring target fatigue characteristics in a target order corresponding to a service provider;
inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model; the fatigue detection model is obtained by training a decision tree model based on an output result obtained by inputting fatigue characteristics and an accident result corresponding to the fatigue characteristics; in the training process of the decision tree model, an output result obtained by the input fatigue characteristics and an accident result corresponding to the fatigue characteristics are used for determining the loss of the decision tree model;
determining a target fatigue degree corresponding to the target order according to the target rule and a mapping relation between the fatigue characteristics and the fatigue degree which is established in advance;
and processing the service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
8. The fatigue driving detection method according to claim 7, wherein the target fatigue characteristics include at least one of: basic action features of the service provider; continuous fatigue characteristics of the service provider; order characteristics;
wherein the basic action characteristics of the service provider at least comprise: the opening and closing times, the opening and closing eye duration, the opening and closing mouth times and the opening and closing mouth duration of the service provider in the recent time period; the continuous fatigue characteristics of the service provider comprise at least: the service provider continuously triggers the times of basic actions in a period of time, the time interval between every two basic actions and the duration of each basic action in the corresponding period of time; wherein, the time interval between every two basic actions which are continuously triggered in the period of time is less than a first preset threshold; the order characteristics at least comprise order time and order distance.
9. The method according to claim 7, wherein the determining a target fatigue degree corresponding to the target order according to the target rule and a mapping relationship between fatigue characteristics and fatigue degrees established in advance comprises:
when any single fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is non-fatigue;
when any single fatigue rule is hit in the target rule but the combined fatigue rule is not hit in the target rule, determining that the fatigue degree corresponding to the target order is light fatigue; wherein the combined fatigue rule comprises at least two individual fatigue rules;
and when the target rule hits any fatigue rule and hits a combined fatigue rule, determining that the fatigue degree corresponding to the target order is severe fatigue.
10. The fatigue driving detection method according to claim 7, wherein the processing a service provider corresponding to the target order based on the target fatigue degree corresponding to the target order comprises:
if the target fatigue degree corresponding to the target order is slight fatigue, carrying out first reminding on a service provider corresponding to the target order through first voice broadcasting;
and if the target fatigue degree corresponding to the target order is severe fatigue, performing second reminding on a service provider corresponding to the target order through second voice broadcasting, and suspending the order distribution to the service provider.
11. The fatigue driving detection method of claim 10, wherein said suspending the allocation of the order to the service provider comprises:
and if the target order is in a driving state, suspending the distribution of the first type of order matched with the driving state to the service provider.
12. The fatigue driving detection method of claim 10, wherein the suspending allocation of the order to the service provider further comprises:
and if the target order is in a driving end state, suspending the distribution of the second type order matched with the driving end state to the service provider.
13. Training device for a fatigue detection model, characterized in that the training device comprises:
the construction module is used for constructing a training sample based on a historical order, wherein the training sample comprises fatigue characteristics and an accident result whether a traffic accident happens or not corresponding to the fatigue characteristics;
the training module is used for taking the fatigue characteristics as the input of a decision tree model and taking fatigue rules as the output of the decision tree model; training a decision tree model according to an output result obtained by the decision tree model based on the input fatigue characteristics and an accident result corresponding to the fatigue characteristics to obtain a trained fatigue detection model and a fatigue rule corresponding to the fatigue detection model; in the training process of the decision tree model, an output result obtained by input fatigue characteristics and an accident result corresponding to the fatigue characteristics are used for determining the loss of the decision tree model;
the mapping establishing module is used for establishing the mapping relation between the fatigue characteristics and the fatigue degree according to the fatigue rules corresponding to the fatigue detection model; the mapping relation is used for determining the target fatigue degree corresponding to the target order based on the output result of the fatigue detection model based on the target fatigue characteristics corresponding to the target order.
14. A fatigue driving detecting apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring target fatigue characteristics in a target order corresponding to a service provider;
the first processing module is used for inputting the target fatigue characteristics into a fatigue detection model trained in advance to obtain a target rule corresponding to the fatigue detection model; the fatigue detection model is obtained by training a decision tree model based on an output result obtained by inputting fatigue characteristics and an accident result corresponding to the fatigue characteristics; in the training process of the decision tree model, an output result obtained by the input fatigue characteristics and an accident result corresponding to the fatigue characteristics are used for determining the loss of the decision tree model;
the determining module is used for determining the target fatigue degree corresponding to the target order according to the target rule and the mapping relation between the fatigue characteristics and the fatigue degree which is established in advance;
and the second processing module is used for processing the service provider corresponding to the target order based on the target fatigue degree corresponding to the target order.
15. An electronic device, comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating via the bus when the electronic device is running, the processor executing the machine-readable instructions to perform the steps of the training method of the fatigue detection model according to any one of claims 1 to 6.
16. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the training method of a fatigue detection model according to any one of claims 1 to 6.
17. An electronic device, comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating via the bus when the electronic device is running, the processor executing the machine-readable instructions to perform the steps of the fatigue driving detection method according to any one of claims 7 to 12.
18. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, which computer program, when being executed by a processor, performs the steps of the fatigue driving detection method according to any one of claims 7 to 12.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010936728.0A CN112036352B (en) | 2020-09-08 | 2020-09-08 | Training method of fatigue detection model, and fatigue driving detection method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010936728.0A CN112036352B (en) | 2020-09-08 | 2020-09-08 | Training method of fatigue detection model, and fatigue driving detection method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112036352A CN112036352A (en) | 2020-12-04 |
CN112036352B true CN112036352B (en) | 2021-09-14 |
Family
ID=73583889
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010936728.0A Active CN112036352B (en) | 2020-09-08 | 2020-09-08 | Training method of fatigue detection model, and fatigue driving detection method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112036352B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102406507A (en) * | 2010-09-20 | 2012-04-11 | 天津职业技术师范大学 | Method for monitoring fatigue degree of driver based on human body physiological signal |
CN104240446A (en) * | 2014-09-26 | 2014-12-24 | 长春工业大学 | Fatigue driving warning system on basis of human face recognition |
CN104720799A (en) * | 2015-04-03 | 2015-06-24 | 上海帝仪科技有限公司 | Fatigue detection method and system based on low-frequency electroencephalogram signals |
CN105769120A (en) * | 2016-01-27 | 2016-07-20 | 深圳地平线机器人科技有限公司 | Fatigue driving detection method and device |
US10740634B1 (en) * | 2019-05-31 | 2020-08-11 | International Business Machines Corporation | Detection of decline in concentration based on anomaly detection |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190077409A1 (en) * | 2017-07-31 | 2019-03-14 | Alcohol Countermeasure Systems (International) Inc. | Non-intrusive assessment of fatigue in drivers using eye tracking |
-
2020
- 2020-09-08 CN CN202010936728.0A patent/CN112036352B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102406507A (en) * | 2010-09-20 | 2012-04-11 | 天津职业技术师范大学 | Method for monitoring fatigue degree of driver based on human body physiological signal |
CN104240446A (en) * | 2014-09-26 | 2014-12-24 | 长春工业大学 | Fatigue driving warning system on basis of human face recognition |
CN104720799A (en) * | 2015-04-03 | 2015-06-24 | 上海帝仪科技有限公司 | Fatigue detection method and system based on low-frequency electroencephalogram signals |
CN105769120A (en) * | 2016-01-27 | 2016-07-20 | 深圳地平线机器人科技有限公司 | Fatigue driving detection method and device |
US10740634B1 (en) * | 2019-05-31 | 2020-08-11 | International Business Machines Corporation | Detection of decline in concentration based on anomaly detection |
Non-Patent Citations (2)
Title |
---|
Fatigue Driving Prediction on Commercial Dangerous Goods Truck Using Location Data: The Relationship between Fatigue Driving and Driving Environment;Shifeng Niu;《Journal of Advanced Transportation》;20200731;全文 * |
基于多尺度池化卷积神经网络的疲劳检测方法研究;顾王欢;《计算机应用研究》;20191130;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112036352A (en) | 2020-12-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109572550B (en) | Driving track prediction method, system, computer equipment and storage medium | |
US9676395B2 (en) | Incapacitated driving detection and prevention | |
Ratcliff et al. | Modeling simple driving tasks with a one-boundary diffusion model | |
US20200134729A1 (en) | Information processing device, information processing system, information processing method, and program | |
CN107992958A (en) | Population super-limit prewarning method based on ARMA | |
CN110781873A (en) | Driver fatigue grade identification method based on bimodal feature fusion | |
CN106775665B (en) | Emotional state change information obtaining method and device based on emotional indexes | |
CN110901385A (en) | Active speed limiting method based on fatigue state of driver | |
CN113867527A (en) | Vehicle window control method and device, electronic equipment and storage medium | |
CN113312958B (en) | Method and device for adjusting dispatch priority based on driver state | |
CN117610932B (en) | Public transport operation risk management and control system based on artificial intelligence | |
CN114601478A (en) | Method, device and equipment for improving alertness of driver and readable storage medium | |
CN112036352B (en) | Training method of fatigue detection model, and fatigue driving detection method and device | |
US20210323559A1 (en) | Data processing device, human-machine interface system including the device, vehicle including the system, method for evaluating user discomfort, and computer-readable medium for carrying out the method | |
CN113320537A (en) | Vehicle control method and system | |
JP2010033549A (en) | Information provision device, information provision method, program and information provision system | |
JP2010271794A (en) | Driving behavior guiding system | |
CN109285344B (en) | Identification method and intelligent decision-making system for key monitoring objects of high-risk traffic personnel | |
CN115946702A (en) | Vehicle-mounted driver health detection method and system | |
Liu et al. | Identification of driver distraction based on SHRP2 naturalistic driving study | |
CN113095584A (en) | Intersection safety level short-time prediction method, system, terminal and readable storage medium based on traffic flow characteristics | |
CN117922615B (en) | Method and device for reducing adverse reactions of passengers in automatic driving public transportation risk avoidance scene | |
CN115179960B (en) | Multi-source data acquisition human-vehicle state comprehensive monitoring system and method | |
CN116130086B (en) | Motion sickness prediction system | |
CN118035691B (en) | Quantitative evaluation method, equipment and storage medium for intelligent driving perception quality |
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 |