CN106934368A - A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye - Google Patents
A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye Download PDFInfo
- Publication number
- CN106934368A CN106934368A CN201710147497.3A CN201710147497A CN106934368A CN 106934368 A CN106934368 A CN 106934368A CN 201710147497 A CN201710147497 A CN 201710147497A CN 106934368 A CN106934368 A CN 106934368A
- Authority
- CN
- China
- Prior art keywords
- data
- eye
- model
- sample
- random forest
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
Abstract
The invention discloses a kind of driving fatigue detecting system based on the dynamic achievement data of eye and recognition methods, four kinds of eye movement datas of frequency collection of 200HZ are pressed by data acquisition module；Then Data Analysis Services module is pre-processed the incoming calculation procedure of eye movement data；Then the dynamic achievement data of eye of collection driver and each data correspondence driver fatigue degree set up Random Forest model as initial data, and classification judgement is carried out to the eye movement data of driver by each decision tree in Random Forest model, provide result；The classification results that last comprehensive each decision tree is given, are voted using Random Forest model, the comprehensive ballot probability final result for being this subseries high.Using the Random Forest model of machine learning, training speed is fast, and classification estimated performance is excellent, can quickly recognize driver it is current whether fatigue, and Random Forest model is as data volume is continuously increased and updates, and can continue to optimize lifting and differentiate performance.
Description
Technical field
The invention belongs to intelligent transportation field, it is related to machine learning field, is specially excavated using random forests algorithm and driven
Sail the dynamic behavioral data of eye of people and quickly recognize system and the recognition methods of driver's driving fatigue state.
Background technology
In recent years, the recoverable amount of motor vehicle increases year by year, and traffic safety problem is also increasingly serious.In driving procedure
In, when driver is engaged in driving-activity for a long time, easily into fatigue state, influences it to perceive and judge and driver behavior, extremely
In the case of even can trigger traffic accident.
Occur in 136386 motor vehicle accidents in China, 2014, and cause 42847 people death, 141718 people injured, its
Middle about 14.3% is triggered by the driving fatigue of driver.Although the consequence that driving fatigue is produced is extremely serious, correlation theory is ground
Study carefully also more, but lack a kind of simple to operation and accurate reliably equipment or method energy accurate measurements/detection in reality and drive
Sail the tired situation of people.
For at present, for driver, whether the detection method in driving fatigue state mainly has two categories below：
First, the state of vehicle, such as steering angle, lateral direction of car position, yaw-rate, car speed are detected.However, these are joined
Number may be varied from because of different vehicles, the experience of driver, the geometric properties of road from the different of state, frequently can lead to
False alarm rate higher.
2nd, the index of correlation of human body, such as blood pressure, heart rate, electrocardio, body behavior, physiological change are detected.But due to brain
Electricity, the collection of heart rate data have stronger invasive for driver, therefore are often difficult to be driven in actual driving environment
People's receiving is sailed, the difference caused by different vehicle-states is larger in addition.
The content of the invention
The technical problems to be solved by the invention are for above-mentioned deficiency of the prior art, there is provided a kind of dynamic based on eye
The driving fatigue rapid detection system of behavioral data and recognition methods.The dynamic behavioral indicator of eye is strong with degree of fatigue correlation, data
Obtain relatively easy, while it is to drive the change of eye that fatigue is most directly reacted.
The present invention uses following technical scheme：
A kind of driving fatigue recognition methods based on the dynamic achievement data of eye, comprises the following steps：
S1, data acquisition module press frequency collection closed-eye time, frequency of wink, wink time and the PD of 200Hz
Four kinds of eye movement datas；
S2, eye movement data described in Data Analysis Services module receiving step S1, and by the incoming calculating journey of the eye movement data
Sequence, program is pre-processed to data, is that empty data are thrown aside to default, data；
S3, the dynamic achievement data of eye of collection driver and each data correspondence driver fatigue degree are used as original number
According to setting up Random Forest model.The real-time eye movement data of driver for gathering is processed by Random Forest model, it is random gloomy
Each decision tree carries out classification judgement to the eye movement data of driver in woods model, provides result；
The classification results that each decision tree is given in S4, combining step S3, are voted using the Random Forest model,
A comprehensive ballot probability class high is the final result of this subseries.
Preferably, it is as follows the step of set up the Random Forest model in step S3：
S31,80% is taken in initial data as original training data, lead to from training sample set S using by bootstrap
Cross to repeat with putting back to randomly to extract N number of sample and be combined into a new training sample set, it is sub to generate K to repeat this flow
Sample set, S1, S2……SK；
S32, for each described subsample collection in step S1, it is random from all of characteristic variable M to select m feature work
Be subcharacter vector set, i.e. one sub- set of eigenvectors M of each sample set correspondence1, M2……MK, wherein m<M；
S33, the subsample collection selected according to step S1 and its corresponding subcharacter vector, generate K decision tree
Tree1, Tree2……Treek；
S34, all of decision tree are grouped together into random forest, and its judgement to data is the throwing of all decision trees
Ticket result, final classification results are the class decision high of the ballot probability of integrated decision-making Tree Classifier；
S35, by the use of initial data remaining 20% as the test set of model, using the Random Forest model pair for establishing
This partial data is tested, and test result and legitimate reading are contrasted, and determines the classification performance of model；
S36, arameter optimization is carried out to Random Forest model described in step S5 according to evaluation result, improve accuracy rate, it is described
Parameter includes the depth capacity max_depth of tree, according to Attribute transposition node when, each is set and divides minimum sample number min_
The minimum sample number min_samples_leaf of samples_split, leaf node, the Maximum sample size max_ of leaf tree
The characteristic divided when leaf_nodes and selection most suitable attribute is no more than this value max_features.
Preferably, in step S33, the purity measurement of the decision tree uses Gini indexes, and Gini indexes are bigger to represent pure
Spend lower, set KiIncluding n class sample records, the probability of each class is p1, p2... ... pn, then Gini indexes be：
Preferably, in step S34, each classification tree is binary tree, and its generation follows top-down recurrence mitogen
Then, i.e., training set is divided successively since root node, in binary tree, root node includes whole training datas, according to
Node purity minimum principle, is split into left sibling and right node, respectively a subset comprising training data, according to same rule
Then node continues to divide, and stops growing until meeting branch's stopping rule.
Preferably, in step S35, the classification performance of the model is evaluated using confusion matrix with ROC curve, described
Confusion matrix is commented using recognition accuracy Accuracy, recall rate Recall and tri- indexs of identification accuracy Precision
The degree of accuracy of valency model.
Preferably, the Model Identification accuracy rate Accuracy is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP
It is to be predicted as positive negative sample by model errors；FN is to be predicted as negative positive sample by model errors.
Preferably, the recall rate Recall is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP
It is to be predicted as positive negative sample by model errors；FN is to be predicted as negative positive sample by model errors.
Preferably, the accuracy of identification Precision is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP
It is to be predicted as positive negative sample by model errors；FN is to be predicted as negative positive sample by model errors.
Preferably, in step S4, the ballot probability vp of waking state classification1For：
The ballot probability vp of fatigue state classification0For：
Wherein, v0Represent the votes of waking state classification, v1Represent the votes of fatigue state classification.
A kind of driving fatigue detecting system based on the dynamic achievement data of eye, including data acquisition module, data transmission module
With data analysis and processing module, wherein, the data acquisition module is by the data transmission module and the data analysis
Reason module connection, is provided with eye tracking device in the data acquisition module, the eye tracking device is arranged on driver
Oblique upper, for gathering the dynamic achievement data of eye, the data transmission module is described for the dynamic achievement data of the eye to be transferred to
Data Analysis Services module, the Data Analysis Services module is received the dynamic achievement data of the eye and is relatively driven by data
Sail fatigue detecting.
Compared with prior art, the present invention at least has the advantages that：
The easy-to-use height of recognition methods of the present invention, data acquisition is more prone to compared to other technologies scheme, using machine learning
Random Forest model, model training speed is fast, is not likely to produce over-fitting situation, and classification estimated performance is excellent, can quickly recognize and drive
Sail people it is current whether fatigue, and Random Forest model can continue to optimize lifting as data volume is continuously increased and updates
Differentiate performance.
Further, random forest (random forest) be it is a kind of data differentiate using multiple classification trees and
The method of classification, pace of learning is fast；Without the concern for the conllinear sex chromosome mosaicism of independent variable, very high-dimensional data can be processed, and
And it goes without doing feature selecting；To the adaptable of data set, discrete data can be processed, can also process continuous data,
Data set is without standardization.It can also provide the prominence score of each variable while classifying to data.
Invention additionally discloses a kind of driving fatigue detecting system based on the dynamic achievement data of eye, by data acquisition module, number
Constituted according to transport module and data analysis and processing module, system architecture is simple, eye state is realized by obtaining the dynamic index of eye
Round-the-clock monitor in real time, data acquisition simple, intuitive influences small, it is easy to received by numerous drivers, Ke Yiyou to driver
Effect reduces the generation of traffic accident, for the security of the lives and property of the people provides powerful guarantee.
Below by drawings and Examples, technical scheme is described in further detail.
Brief description of the drawings
Fig. 1 is function structure chart of the invention；
Fig. 2 sets up artwork for Random Forest model；
Fig. 3 is certain decision tree deterministic process example in the Random Forest model for creating；
Fig. 4 is model data Treatment Analysis flow chart；
Fig. 5 is the certain embodiments figure of driver discriminating data result of the random forest to gathering.
Specific embodiment
Refer to Fig. 1, a kind of driving fatigue detecting system based on the dynamic achievement data of eye, including：Data acquisition module, number
According to transport module and data analysis and processing module.Frequency collection closed-eye time, blink of the data acquisition module according to 200HZ
Frequency, wink time, four groups of dynamic achievement datas of eye of PD.
The data acquisition module be using X3-120 eye tracking device of dynamic tracing system of Tobii Pro, it is described
Eye tracking device is located at driver oblique upper, and the dynamic finger of above-mentioned eye is gathered by X3-120 dynamic tracing system of Tobii Pro
Mark,, by four camera recording-related informations being arranged on front windshield, sample frequency is 200HZ for it, the device with
USB interface form is connected with notebook computer, and the real time data for being gathered is with txt form real-time storages.
The data transmission module is used to connect the data acquisition module and data analysis and processing module, by the data
The dynamic behavioral data of the eye of the driver that acquisition module is collected is transferred to Data Analysis Services module.
The Data Analysis Services module, the program for writing random forests algorithm using PYTHON programming languages carries out data
Treatment.PYTHON writes algorithm routine and TXT texts is formatted into treatment, becomes suitable for the data shape of program treatment
Formula.
Due to certain data uncertainty once with fluctuate it is larger, so analysis when by 5 minutes in data average value
As the input value of model, carry out processing with this drawing last result, and stored.
It is pointed out that the system is before formally Data Analysis Services are carried out, can be by the driving reality up to 3-5 hours
Four groups of dynamic achievement datas of eye and each data correspondence driver fatigue degree of collection certain amount driver are tested as this
The initial data of invention.
A kind of driving fatigue recognition methods based on the dynamic achievement data of eye of the present invention, comprises the following steps：
S1, data acquisition module are by frequency collection closed-eye time once per minute, frequency of wink, wink time and pupil
Four kinds of eye movement datas of diameter；
S2, eye movement data described in Data Analysis Services module receiving step S1, and by the incoming calculating journey of the eye movement data
Sequence, program is pre-processed to data, is that empty data are thrown aside to default, data；
S3, the dynamic achievement data of eye of collection driver and each data correspondence driver fatigue degree are used as original number
According to Random Forest model is set up, classification is carried out to the eye movement data of driver by each decision tree in Random Forest model and is sentenced
It is disconnected, result is given, wherein, the judgement example of decision tree is as shown in Figure 3；
First carry out driving experiment acquisition initial data：
On one section of specific highway, the age is chosen for the human pilot of 25-45 is continuously driven, remember within every and a half hours
Record its corresponding index and inquire about its degree of fatigue, according to Stamford fatigue gauge table (SSS) as module.
Wherein driver fatigue horizontal recording uses active degree of fatigue, and Subjective fatigue standard is Stamford fatigue gauge table
(SSS) degree of fatigue, is divided into seven classes, 1 is shown in Table.
The Stamford of table 1 fatigue gauge table (SSS)
Wherein, the estimation of degree of fatigue is carried out for convenience, can be by the state of driver according to the implication that each grade is represented
It is divided into two classes, " 1-3 " is waking state, " 4-7 " is fatigue state.For convenience of data are modeled with analysis, digital " 0 " table is used
Show " clear-headed " state, with digital " 1 " expression " fatigue " state.
The initial data obtained using experiment sets up random model of standing abreast, and its principle is as shown in Figure 2.
The first step：80% is taken in initial data as original training data, using by bootstrap (Bootstrap) so
A kind of resampling technique, by putting back to repeat randomly to extract from training sample set S N number of sample be combined into one it is new
Training sample set, this flow is repeated to generate K sub- sample sets, S1, S2……SK。
Second step：For each subsample collection, m feature is selected as subcharacter at random from all of characteristic variable M
One sub- set of eigenvectors M of vector set, i.e. each sample set correspondence1, M2……MK, wherein m<M.
3rd step：The subsample collection selected according to the first step and its corresponding subcharacter vector, generate K decision tree
Tree1, Tree2……Treek。
The purity measurement of decision tree uses Gini indexes, and Gini indexes are bigger to represent that purity is lower.Set KiIncluding n classes
Sample record, the probability of each class is p1, p2... ... pn, then Gini indexes be：
Wherein, an example decision tree of the invention is as shown in Figure 3.
4th step：All of decision tree is grouped together into random forest, and its judgement to data is all decision trees
Voting results, final classification results determine for the ballot probability of an integrated decision-making Tree Classifier class high.
Each classification tree is binary tree in random forest, and its generation follows top-down recurrence division principle, i.e., from
Root node starts successively to divide training set；In binary tree, root node includes whole training datas, according to node purity
Minimum principle, is split into left sibling and right node, and they include a subset of training data respectively, according to same rule section
Point continues to divide, and stops growing until meeting branch's stopping rule.
5th step：By the use of initial data remaining 20% as the test set of model, using the random forest mould for establishing
Type is tested this partial data, and test result and legitimate reading are contrasted, and determines the classification performance of model.
Wherein, the classification performance of model is evaluated using confusion matrix with ROC curve.
Wherein, confusion matrix recognition accuracy Accuracy, recall rate Recall and identification accuracy Precision tri-
Individual index carrys out the degree of accuracy of straight angle model.
Model Identification accuracy rate Accuracy：
Recall rate Recall：
Accuracy of identification Precision：
Each parameter represents that implication is as follows：
TP be True Positive (real) by model it is correctly predicted be positive positive sample；
TN be True Negative (really bear) by model it is correctly predicted be negative negative sample；
FP is that False Positive (vacation is just) are predicted as positive negative sample by model errors；
FN is that False Negative (vacation is negative) are predicted as negative positive sample by model errors.
6th step：Arameter optimization is carried out according to evaluation result, allows the accuracy rate of model to attain a yet higher goal.
Wherein arameter optimization mainly debugs following parameter：
(1)max_depth：The depth capacity of tree；
(2)min_samples_split：During according to Attribute transposition node, each is set and divides minimum sample number；
(3)min_samples_leaf：The minimum sample number of leaf node；
(4)max_leaf_nodes：The Maximum sample size of leaf tree；
(5)max_features：The characteristic divided during selection most suitable attribute is no more than this value.
As shown in figure 4, model adjustment well just can process the data of Real-time Collection on motor vehicle using it.
The classification results that each decision tree is given in S4, combining step S3, are voted using the Random Forest model,
A comprehensive ballot probability class high is the final result of this subseries.
The classification results that comprehensive each decision tree is given, Random Forest model is voted, comprehensive ballot probability it is high one
Class is the final result of this subseries." clear-headed " state is represented with digital " 0 ", with digital " 1 " expression " fatigue " state, classification
" 0 " ballot probability vp0With the ballot probability vp of classification " 1 "1It is as follows：
Wherein v0、v1It is classification " 0 ", the votes of " 1 ".
As shown in Figure 5, this figure be in an example of the invention 56 eye movement datas of driver by of the invention
The voting results of Random Forest model classification.It is a probability results that model finally draws, probability state high is final defeated
Go out result.
Above content is only explanation technological thought of the invention, it is impossible to limit protection scope of the present invention with this, every to press
According to technological thought proposed by the present invention, any change done on the basis of technical scheme each falls within claims of the present invention
Protection domain within.
Claims (10)
1. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye, it is characterised in that comprise the following steps：
S1, data acquisition module press four kinds of frequency collection closed-eye time, frequency of wink, wink time and the PD of 200Hz
Eye movement data；
S2, eye movement data described in Data Analysis Services module receiving step S1, and by the incoming calculation procedure of the eye movement data, journey
Ordered pair data are pre-processed, and are that empty data are thrown aside to default, data；
S3, the dynamic achievement data of eye of collection driver and each data correspondence driver fatigue degree are built as initial data
Vertical Random Forest model, is processed the eye movement data of Real-time Collection, Random Forest model by Random Forest model
In each decision tree classification judgement is carried out to the eye movement data of driver, provide result；
The classification results that each decision tree is given in S4, combining step S3, are voted using the Random Forest model, comprehensive
A ballot probability class high is the final result of this subseries.
2. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 1, it is characterised in that step
It is as follows the step of set up the Random Forest model in rapid S3：
S31, take in initial data 80% as original training data, using by bootstrap from training sample set S by putting
Repeatedly randomly extract N number of sample with returning and be combined into a new training sample set, repeat this flow to generate K subsample
Collection, S1, S2……SK；
S32, for each described subsample collection in step S1, select m feature as son at random from all of characteristic variable M
One sub- set of eigenvectors M of set of eigenvectors, i.e. each sample set correspondence1, M2……MK, wherein m<M；
S33, the subsample collection selected according to step S1 and its corresponding subcharacter vector, generate K decision tree Tree1,
Tree2……Treek；
S34, all of decision tree are grouped together into random forest, and its judgement to data is the ballot knot of all decision trees
Really, final classification results are the class decision high of the ballot probability of integrated decision-making Tree Classifier；
S35, by the use of initial data remaining 20% as the test set of model, using the Random Forest model for establishing to this portion
Divided data is tested, and test result and legitimate reading are contrasted, and determines the classification performance of model；
S36, arameter optimization is carried out to Random Forest model described in step S5 according to evaluation result, improve accuracy rate, the parameter
Including set depth capacity max_depth, according to Attribute transposition node when, each is set and divides minimum sample number min_
The minimum sample number min_samples_leaf of samples_split, leaf node, the Maximum sample size max_ of leaf tree
The characteristic divided when leaf_nodes and selection most suitable attribute is no more than this value max_features.
3. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 2, it is characterised in that：Step
In rapid S33, the purity measurement of the decision tree uses Gini indexes, and Gini indexes are bigger to represent that purity is lower, set KiIncluding n
Class sample record, the probability of each class is p1, p2... ... pn, then Gini indexes be：
4. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 2, it is characterised in that step
In rapid S34, each classification tree is binary tree, and its generation follows top-down recurrence division principle, i.e., since root node
Training set is divided successively, in binary tree, root node includes whole training datas, according to node purity minimum principle,
Left sibling and right node are split into, respectively a subset comprising training data, continue to divide according to same regular node, directly
Stopped growing to branch's stopping rule is met.
5. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 2, it is characterised in that step
In rapid S35, the classification performance of the model is evaluated using confusion matrix with ROC curve, and the confusion matrix is using identification
Accuracy rate Accuracy, recall rate Recall and tri- indexs of identification accuracy Precision carry out the degree of accuracy of evaluation model.
6. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 5, it is characterised in that institute
Stating Model Identification accuracy rate Accuracy is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP be by
Model errors are predicted as positive negative sample；FN is to be predicted as negative positive sample by model errors.
7. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 5, it is characterised in that institute
Stating recall rate Recall is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP be by
Model errors are predicted as positive negative sample；FN is to be predicted as negative positive sample by model errors.
8. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 5, it is characterised in that institute
Stating accuracy of identification Precision is：
Wherein, TP be by model it is correctly predicted be positive positive sample；TN be by model it is correctly predicted be negative negative sample；FP be by
Model errors are predicted as positive negative sample；FN is to be predicted as negative positive sample by model errors.
9. a kind of driving fatigue recognition methods based on the dynamic achievement data of eye according to claim 1, it is characterised in that step
In rapid S4, the ballot probability vp of waking state classification1For：
The ballot probability vp of fatigue state classification0For：
Wherein, v0Represent the votes of waking state classification, v1Represent the votes of fatigue state classification.
10. a kind of detecting system of utilization claim 1 methods described, it is characterised in that：Passed including data acquisition module, data
Defeated module and data analysis and processing module, wherein, the data acquisition module is by the data transmission module and the data
Analysis and processing module is connected, and eye tracking device is provided with the data acquisition module, and the eye tracking device is arranged on
Driver oblique upper, for gathering the dynamic achievement data of eye, the data transmission module is used for the dynamic achievement data transmission of the eye
To the Data Analysis Services module, the Data Analysis Services module is received the dynamic achievement data of the eye and is compared by data
Carry out driving fatigue detection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710147497.3A CN106934368A (en) | 2017-03-13 | 2017-03-13 | A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710147497.3A CN106934368A (en) | 2017-03-13 | 2017-03-13 | A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106934368A true CN106934368A (en) | 2017-07-07 |
Family
ID=59432941
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710147497.3A Pending CN106934368A (en) | 2017-03-13 | 2017-03-13 | A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106934368A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107450524A (en) * | 2017-08-30 | 2017-12-08 | 深圳市智物联网络有限公司 | Predict the method, apparatus and computer-readable recording medium of industrial equipment failure |
| CN107468206A (en) * | 2017-08-01 | 2017-12-15 | 成都大学 | A kind of stereos copic viewing visual fatigue Forecasting Methodology based on pupil diameter |
| CN108464839A (en) * | 2018-03-22 | 2018-08-31 | 东华大学 | A kind of vehicle-mounted fatigue monitoring and early warning of driving system based on machine learning |
| CN108537289A (en) * | 2018-04-24 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | Training method, device and the storage medium of data identification model |
| CN109086964A (en) * | 2018-07-03 | 2018-12-25 | 南京邮电大学 | MR coverage rate influence factor determination method based on random forest |
| CN109377012A (en) * | 2018-09-26 | 2019-02-22 | 中国航天员科研训练中心 | A kind of dynamic man-machine function allocation system and unmanned plane |
| CN109409647A (en) * | 2018-09-10 | 2019-03-01 | 昆明理工大学 | A kind of analysis method of the salary level influence factor based on random forests algorithm |
| CN109460703A (en) * | 2018-09-14 | 2019-03-12 | 华南理工大学 | A kind of non-intrusion type fatigue driving recognition methods based on heart rate and facial characteristics |
| CN109512442A (en) * | 2018-12-21 | 2019-03-26 | 杭州电子科技大学 | A kind of EEG fatigue state classification method based on LightGBM |
| CN109616201A (en) * | 2018-11-09 | 2019-04-12 | 深圳职业技术学院 | A kind of giving fatigue pre-warning method based on small data dynamic prediction model |
| CN109841284A (en) * | 2017-11-24 | 2019-06-04 | 广东乐心医疗电子股份有限公司 | Sleep state discrimination model construction method and device and portable intelligent equipment |
| CN109858553A (en) * | 2019-01-31 | 2019-06-07 | 深圳市赛梅斯凯科技有限公司 | Monitoring model update method, updating device and the storage medium of driving condition |
| CN111081016A (en) * | 2019-12-18 | 2020-04-28 | 北京航空航天大学 | Urban traffic abnormity identification method based on complex network theory |
| CN111444657A (en) * | 2020-03-10 | 2020-07-24 | 五邑大学 | Method and device for constructing fatigue driving prediction model and storage medium |
| CN112699793A (en) * | 2020-12-29 | 2021-04-23 | 长安大学 | Fatigue driving detection optimization identification method based on random forest |
| CN113326733A (en) * | 2021-04-26 | 2021-08-31 | 吉林大学 | Eye movement point data classification model construction method and system |
| CN115204537A (en) * | 2022-09-17 | 2022-10-18 | 华北理工大学 | Student score prediction method based on Bagging |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102096810A (en) * | 2011-01-26 | 2011-06-15 | 北京中星微电子有限公司 | Method and device for detecting fatigue state of user before computer |
| CN104183091A (en) * | 2014-08-14 | 2014-12-03 | 苏州清研微视电子科技有限公司 | System for adjusting sensitivity of fatigue driving early warning system in self-adaptive mode |
| US20160082838A1 (en) * | 2014-09-21 | 2016-03-24 | Electro-Motive Diesel, Inc. | Operator fatigue monitoring system |
| CN106448265A (en) * | 2016-10-27 | 2017-02-22 | 广州微牌智能科技有限公司 | Collecting method and device of driver's driving behavior data |
-
2017
- 2017-03-13 CN CN201710147497.3A patent/CN106934368A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102096810A (en) * | 2011-01-26 | 2011-06-15 | 北京中星微电子有限公司 | Method and device for detecting fatigue state of user before computer |
| CN104183091A (en) * | 2014-08-14 | 2014-12-03 | 苏州清研微视电子科技有限公司 | System for adjusting sensitivity of fatigue driving early warning system in self-adaptive mode |
| US20160082838A1 (en) * | 2014-09-21 | 2016-03-24 | Electro-Motive Diesel, Inc. | Operator fatigue monitoring system |
| CN106448265A (en) * | 2016-10-27 | 2017-02-22 | 广州微牌智能科技有限公司 | Collecting method and device of driver's driving behavior data |
Non-Patent Citations (2)
| Title |
|---|
| 曹正凤: ""随机森林算法优化研究"", 《中国博士学位论文全文数据库（电子期刊）》 * |
| 牛清宁等: ""基于眼动特征的疲劳驾驶检测方法"", 《哈尔滨工程大学学报》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107468206A (en) * | 2017-08-01 | 2017-12-15 | 成都大学 | A kind of stereos copic viewing visual fatigue Forecasting Methodology based on pupil diameter |
| CN107450524A (en) * | 2017-08-30 | 2017-12-08 | 深圳市智物联网络有限公司 | Predict the method, apparatus and computer-readable recording medium of industrial equipment failure |
| CN109841284A (en) * | 2017-11-24 | 2019-06-04 | 广东乐心医疗电子股份有限公司 | Sleep state discrimination model construction method and device and portable intelligent equipment |
| CN108464839A (en) * | 2018-03-22 | 2018-08-31 | 东华大学 | A kind of vehicle-mounted fatigue monitoring and early warning of driving system based on machine learning |
| CN108537289A (en) * | 2018-04-24 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | Training method, device and the storage medium of data identification model |
| CN108537289B (en) * | 2018-04-24 | 2023-04-07 | 百度在线网络技术（北京）有限公司 | Training method and device of data recognition model and storage medium |
| CN109086964A (en) * | 2018-07-03 | 2018-12-25 | 南京邮电大学 | MR coverage rate influence factor determination method based on random forest |
| CN109409647A (en) * | 2018-09-10 | 2019-03-01 | 昆明理工大学 | A kind of analysis method of the salary level influence factor based on random forests algorithm |
| CN109460703A (en) * | 2018-09-14 | 2019-03-12 | 华南理工大学 | A kind of non-intrusion type fatigue driving recognition methods based on heart rate and facial characteristics |
| CN109377012A (en) * | 2018-09-26 | 2019-02-22 | 中国航天员科研训练中心 | A kind of dynamic man-machine function allocation system and unmanned plane |
| CN109616201A (en) * | 2018-11-09 | 2019-04-12 | 深圳职业技术学院 | A kind of giving fatigue pre-warning method based on small data dynamic prediction model |
| CN109616201B (en) * | 2018-11-09 | 2023-10-31 | 深圳职业技术大学 | Fatigue early warning method based on small data dynamic prediction model |
| CN109512442A (en) * | 2018-12-21 | 2019-03-26 | 杭州电子科技大学 | A kind of EEG fatigue state classification method based on LightGBM |
| CN109858553A (en) * | 2019-01-31 | 2019-06-07 | 深圳市赛梅斯凯科技有限公司 | Monitoring model update method, updating device and the storage medium of driving condition |
| CN109858553B (en) * | 2019-01-31 | 2023-12-12 | 锦图计算技术(深圳)有限公司 | Method, device and storage medium for updating driving state monitoring model |
| CN111081016A (en) * | 2019-12-18 | 2020-04-28 | 北京航空航天大学 | Urban traffic abnormity identification method based on complex network theory |
| CN111081016B (en) * | 2019-12-18 | 2021-07-06 | 北京航空航天大学 | Urban traffic abnormity identification method based on complex network theory |
| CN111444657A (en) * | 2020-03-10 | 2020-07-24 | 五邑大学 | Method and device for constructing fatigue driving prediction model and storage medium |
| CN111444657B (en) * | 2020-03-10 | 2023-05-02 | 五邑大学 | Method and device for constructing fatigue driving prediction model and storage medium |
| CN112699793A (en) * | 2020-12-29 | 2021-04-23 | 长安大学 | Fatigue driving detection optimization identification method based on random forest |
| CN113326733A (en) * | 2021-04-26 | 2021-08-31 | 吉林大学 | Eye movement point data classification model construction method and system |
| CN115204537A (en) * | 2022-09-17 | 2022-10-18 | 华北理工大学 | Student score prediction method based on Bagging |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106934368A (en) | A kind of driving fatigue detecting system and recognition methods based on the dynamic achievement data of eye | |
| CN108248610B (en) | It is a kind of adaptively divert one's attention driving behavior monitoring and tampering devic | |
| Li et al. | Drunk driving detection based on classification of multivariate time series | |
| CN106956591A (en) | A kind of system for being used to judge that human pilot drives authority | |
| CN106740457A (en) | Vehicle lane-changing decision-making technique based on BP neural network model | |
| CN113743471B (en) | Driving evaluation method and system | |
| KR102184109B1 (en) | The system and the method for recognizing driver's condition of multimodal learning | |
| CN115064246B (en) | Depression evaluation system and equipment based on multi-mode information fusion | |
| CN112699793A (en) | Fatigue driving detection optimization identification method based on random forest | |
| CN107095671A (en) | Driving fatigue recognition methods and system | |
| CN106408032A (en) | Fatigue driving detection method based on corner of steering wheel | |
| Ramirez et al. | Inferring drivers’ visual focus attention through head-mounted inertial sensors | |
| CN109664894A (en) | Fatigue driving safety pre-warning system based on multi-source heterogeneous data perception | |
| CN113753059B (en) | Method for predicting takeover capacity of driver under automatic driving system | |
| CN114092922A (en) | Driver emotion recognition and behavior intervention method based on specificity | |
| El Masri et al. | Toward self-policing: Detecting drunk driving behaviors through sampling CAN bus data | |
| Chakraborty et al. | Automatic detection of driver's awareness with cognitive task from driving behavior | |
| CN108647657A (en) | A kind of high in the clouds instruction process evaluation method based on pluralistic behavior data | |
| CN109875584A (en) | The detection method and its warning system of physiological driver's fatigue | |
| Wong et al. | Head motion recognition using a smart helmet for motorcycle riders | |
| Biju et al. | Drowsy driver detection using two stage convolutional neural networks | |
| CN108537161A (en) | A kind of driving of view-based access control model characteristic is divert one's attention detection method | |
| CN113920491A (en) | Fatigue detection system, method, medium and detection device based on facial skeleton model | |
| Josephin et al. | A review on the measures and techniques adapted for the detection of driver drowsiness | |
| CN117272155A (en) | Intelligent watch-based driver road anger disease detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170707 |