CN109614873A - Train safety sign detection method, server and storage medium neural network based - Google Patents
Train safety sign detection method, server and storage medium neural network based Download PDFInfo
- Publication number
- CN109614873A CN109614873A CN201811358358.6A CN201811358358A CN109614873A CN 109614873 A CN109614873 A CN 109614873A CN 201811358358 A CN201811358358 A CN 201811358358A CN 109614873 A CN109614873 A CN 109614873A
- Authority
- CN
- China
- Prior art keywords
- train
- image
- safety sign
- neural network
- region
- 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
- G06V20/00—Scenes; Scene-specific elements
-
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Train Traffic Observation, Control, And Security (AREA)
- Image Analysis (AREA)
Abstract
The present embodiments relate to detection fields, disclose a kind of train safety sign detection method, server and storage medium neural network based.In the present invention, comprising the following steps: obtain the image in last vehicle of train region according to train image;The detection of train safety sign is carried out to the image in last vehicle of train region by neural network, and obtains testing result, in the case where determining testing result is train safety sign abnormal state, real-time alert.The image in last vehicle of train region is obtained by being analyzed and processed to train image, and the testing result to the train safety sign in last vehicle of train area image is obtained by neural network model, the erroneous judgement due to caused by being detected by the way of artificial observation is avoided, to improve the accuracy rate of train safety sign detection, staff's working efficiency and the investment for reducing station staff.
Description
Technical field
The present embodiments relate to detection field, in particular to a kind of train safety sign neural network based detection side
Method, server and storage medium.
Background technique
In railway transport field, the tail portion of train generally requires installation taillight or train tail apparatus, and by taillight or column tail
Device is as train safety sign.Train safety sign plays an important role in terms of train safety traffic, therefore station exists
When sending and receiving vehicle, generally requires staff and recognition detection is carried out to train safety sign, to guarantee the safety of vehicle driving.
At least there are the following problems in the prior art for inventor's discovery: in the prior art, station generally puts into a large amount of work
Make personnel, and identification inspection is carried out to safety sign by visually observing for staff, but continuous observation often makes
People becomes tired out, therefore is carrying out identifying the case where being easy to appear erroneous judgement when checking, to reduce the accuracy rate of recognition detection.
Summary of the invention
Embodiment of the present invention is designed to provide a kind of train safety sign detection method neural network based, clothes
Business device and storage medium, so that improving the accuracy rate of detection when detecting to train safety sign, staff works and imitates
Rate and the investment for reducing station staff.
In order to solve the above technical problems, embodiments of the present invention provide a kind of train safety post neural network based
Will detection method, comprising the following steps: the image in last vehicle of train region is obtained according to train image;To the figure in last vehicle of train region
As carrying out the detection of train safety sign, and obtain testing result, wherein testing result includes that train safety sign state is normal
With train safety sign abnormal state;In the case where determining testing result is train safety sign abnormal state, issue in real time
Warning message.
Embodiments of the present invention additionally provide a kind of server, comprising: at least one graphics processor;And with extremely
The central processing unit of few graphics processor communication connection;And the memory being connect at least one processor communication;With
And the communication component with the communication connection of at least one graphics processor, communication component send and receive under the control of a processor
Data;Wherein, memory is stored with the instruction that can be executed by least one graphics processor, and instruction is held by least one processor
Row, so that at least one processor is able to carry out train safety sign detection method neural network based as above.
Embodiments of the present invention additionally provide a kind of computer readable storage medium, are stored with computer program, special
Sign is, train safety sign detection method neural network based as above is realized when computer program is executed by processor.
Embodiment of the present invention obtains train tail in terms of existing technologies, by being analyzed and processed to train image
The image in portion region, and the detection knot to the train safety sign in last vehicle of train area image is obtained by neural network model
Fruit avoids the erroneous judgement due to caused by being detected by the way of artificial observation, to improve the inspection of train safety sign
The accuracy rate and staff's working efficiency of survey, while by replacing traditional eye recognition to detect machine recognition detection, from
And reduce the investment of railroader.
In addition, before according to the image in train image acquisition last vehicle of train region, further includes: determination receives train rail
The acquisition signal that the sensor of upper setting is sent;According to the train image of acquisition signal acquisition photographic device shooting, wherein acquisition
Signal is for indicating coverage of the train by photographic device.In the realization, in the coverage that train passes through photographic device
When, the sensor on train rail can send sampled signal to server in time, keep server right only for sampled signal institute
At the time of answering, the train image got is handled, to alleviate the processing pressure of server, improves the place of processor
Manage speed.
In addition, carrying out the detection of train safety sign to the image in last vehicle of train region by neural network model, and obtain
Testing result is obtained, is specifically included: feature extraction is carried out using image of the preset neural network model to last vehicle of train region;Root
It is detected according to the feature of extraction, obtains testing result;Wherein, testing result includes that train safety sign state is normal or train
Safety sign abnormal state.
In addition, also being wrapped before carrying out feature extraction using image of the preset neural network model to last vehicle of train region
It includes: default neural network model being trained using multiple sample images, obtain and preset neural network model known to parameter,
Wherein, known to the state of the train safety sign in each sample image.
In addition, obtaining the image in last vehicle of train region according to train image, specifically include: image increasing is carried out to train image
Strength reason obtains enhancing image;The image in last vehicle of train region is obtained according to enhancing image.In the realization because may be
Accessed train image under different illumination conditions, therefore the train image obtained can occur not because of the difference of illumination
Clear problem can be such that obtained train image is more clear after image enhancement processing, to further improve
The accuracy of detection.
In addition, obtaining the image in last vehicle of train region according to enhancing image, specifically include: by carrying out figure to enhancing image
As dividing processing, the image in last vehicle of train region is obtained, wherein last vehicle of train region is for installing train safety sign.The reality
In existing, by being split processing to the enhancing image got, the last vehicle of train region of installation train safety sign is obtained
Image, and only the image in the region being partitioned into is detected, to reduce the range of detection, improve the speed of detection
Degree.
In addition, determine testing result be train safety sign abnormal state in the case where, real-time alert it
Afterwards, further includes: warning message is defeated by the terminal device for having communication connection with server.
In addition, carry out the detection of train safety sign to the image in last vehicle of train region, and after obtaining testing result, also
It include: to receive the last vehicle of train area with mark information of user's input in the case where warning message and actual scene are not inconsistent
The image in domain, and neural network model is trained according to the image in the last vehicle of train region with mark information.
Detailed description of the invention
One or more embodiments are illustrated by the picture in corresponding attached drawing, these exemplary theorys
The bright restriction not constituted to embodiment, the element in attached drawing with same reference numbers label are expressed as similar element, remove
Non- to have special statement, composition does not limit the figure in attached drawing.
Fig. 1 is the flow chart of train safety sign detection method neural network based in the application first embodiment;
Fig. 2 is that the application scenarios of train safety sign detection method neural network based in the application first embodiment show
It is intended to;
Fig. 3 is the flow chart of train safety sign detection method neural network based in the application second embodiment;
Fig. 4 is the box signal of train safety sign detection device neural network based in the application 3rd embodiment
Figure;
Fig. 5 is the box signal of train safety sign detection device neural network based in the application fourth embodiment
Figure;
Fig. 6 is the structural schematic diagram of server in the 5th embodiment of the application.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention
Each embodiment be explained in detail.However, it will be understood by those skilled in the art that in each embodiment party of the present invention
In formula, in order to make the reader understand this application better, many technical details are proposed.But even if without these technical details
And various changes and modifications based on the following respective embodiments, the application technical solution claimed also may be implemented.
The first embodiment of the present invention is related to a kind of train safety sign detection methods neural network based, are applied to
Server.Detailed process is as shown in Figure 1, comprising the following steps:
Step 101, the image in last vehicle of train region is obtained according to train image.
Specifically, in present embodiment, the application scenarios schematic diagram of present embodiment is as shown in Figure 2.It enters the station in train
Photographic device is arranged in throat region in preceding pre-determined distance, and sensor is arranged at the guide rail of throat region, when train passes through
When sensor, sensor can get the acquisition signal of train process, and train is also the shooting model in photographic device at this time
In enclosing.The concrete mode that server obtains train image, which can be, obtains photographic device shooting by the interchanger in station
Train image can be transferred to company by the way of optical fiber transmission mode or wireless transmission by train image, specific transmission mode
Machine, and by handing over connector to be transferred to interchanger, the type of connector specifically can be RJ45, and server can be at this time
The train image in interchanger is obtained by wired or wireless transmission mode in station local area network, being connected with interchanger is end
End equipment and workbench, terminal device specifically can be staff's smartwatch entrained with oneself, and workbench specifically can be with
It is a computer, and the workbench in the present embodiment can be used as a kind of terminal device of specific type.
Wherein, before the image for obtaining last vehicle of train region according to train image, it is thus necessary to determine that receive train rail
The acquisition signal that the sensor of upper setting is sent, according to the train image of acquisition signal acquisition photographic device shooting, wherein acquisition
Signal is for indicating coverage of the train by photographic device.Therefore, after receiving acquisition signal, it is ensured that from taking the photograph
It is train image as captured by device, the image without being other scenes, therefore, the acquisition sent by obtaining sensor
Signal, it is possible to reduce server is to the image processing operations of other scenes, to reduce the processing pressure of server.
Specifically, in the present embodiment, since server may be to obtain photographic device shooting under different scenes
Train image, and can be influenced by illumination under different scenes in the clarity of different scenes Train image and not
Together, it is generally the case that train image acquired in daytime is often higher than train image clarity acquired in night.
Therefore, it in order to solve the problems, such as that the image due to caused by different illumination conditions is not known, needs to carry out image to train image
Enhancing processing, Retinex algorithm, histogram can be used when enhancing the train image of acquisition by obtaining enhancing image
Equalization algorithm or Wavelet transformation algorithm are handled, to improve the whole and local contrast of train image, make to enhance image
With good visual effect, that is, it is more clear.Certainly, present embodiment is only with Retinex algorithm, histogram equalization algorithm
Or the explanation carried out for Wavelet Transformation Algorithm, it is also the protection in the application for other algorithms that can be realized image enhancement
In range.
Wherein, after obtaining enhancing image, the image in last vehicle of train region, specific implementation side are obtained according to enhancing image
Formula can obtain the image in last vehicle of train region, wherein train tail using by carrying out image dividing processing to enhancing image
For installing train safety sign, the train safety sign in the embodiment of the present application includes Tail light for train or column tail dress in portion region
It sets, and Tail light for train or train tail apparatus are generally mounted in last vehicle of train region, so by being split to enhancing image
Processing comes out the last vehicle of train region disconnecting for installing Tail light for train or train tail apparatus, and in subsequent processing only to the region
Interior image carries out analysis detection and improves the processing speed of detection to reduce the range of detection.In present embodiment,
When carrying out image dividing processing to enhancing image, using K-Means algorithm, the partitioning algorithm based on threshold value or area can be based on
The partitioning algorithm in domain, certainly, present embodiment are illustrated with above-mentioned algorithm, can be realized image point for other
The algorithm cut is also within the scope of protection of this application.
Step 102, the detection of train safety sign is carried out to the image in last vehicle of train region by neural network model, and
Obtain testing result.
Specifically, in the application embodiment, train safety sign is being carried out according to the image in last vehicle of train region
Detection when, concrete mode, which can be, to be carried out feature to the image in last vehicle of train region using preset neural network model and mentions
It takes;It is detected according to the feature of extraction, obtains testing result;Wherein, testing result includes that train safety sign state is normal
Or train safety sign abnormal state.Also, the neural network model type in present embodiment includes convolutional neural networks mould
Type is all certainly in this Shen as long as can be realized the detection to train safety sign for other types of neural network model
In protection scope please.
It should be noted that carrying out feature extraction using image of the preset neural network model to last vehicle of train region
Before, further includes: default neural network model is trained using multiple sample images, obtains and presets nerve known to parameter
Network model, wherein known to the state of the train safety sign in each sample image.
For example, the image pattern in 1000 parts of last vehicle of train regions is inputted default neural network model, and every a column
The image of vehicle tail region is all marked with that train safety sign state is normal and the mark information of train safety sign abnormal state,
By great amount of samples image known to train safety sign state, it can obtain and preset neural network model known to parameter.Cause
This, by the A image in the last vehicle of train region of train safety sign Status unknown, by presetting neural network known to input parameter
Model, so that it may obtain the testing result of train safety sign corresponding to A image.Certainly, be only in present embodiment with
It is illustrated for the image pattern in 1000 parts of last vehicle of train regions, in practical applications, the quantity of sample is not limited only to
1000 parts, 1000 parts can also be much larger than, only multiple samples to be offered can be realized the training to default neural network model,
It is all within the scope of protection of this application.
Specifically, in the present embodiment, neural network model is mentioned in the image according to the last vehicle of train region of input
The feature of train safety sign is taken out, for example, being directed to train tail apparatus, the frame for obtaining the train tail apparatus in last vehicle of train region is special
Sign, detects the bounding box features of train tail apparatus, however, it is determined that there are the bounding box features of train tail apparatus, it is determined that train tail apparatus is not
Missing is fallen, therefore train tail apparatus state is normal, however, it is determined that there is no the bounding box features of train tail apparatus, it is determined that train tail apparatus lacks
It loses, therefore train tail apparatus abnormal state.And it is directed to Tail light for train, obtain the bounding box features of the Tail light for train in last vehicle of train region
And brightness, however, it is determined that there are the bounding box features of Tail light for train, and the brightness of Tail light for train is more than the default of setting
Threshold value, it is determined that Tail light for train is bright, therefore Tail light for train state is normal;If it is determined that there are the bounding box features of Tail light for train, and
The brightness of Tail light for train is less than the preset threshold of setting, it is determined that Tail light for train is not bright, therefore Tail light for train state is different
Often, and by the abnormality it is known as Tail light for train first kind abnormality;If it is determined that the bounding box features of Tail light for train are not present, then
It determines that Tail light for train lacks, therefore Tail light for train abnormal state, and the abnormality is known as Tail light for train the second class exception shape
State.
It should be noted that neural network model is not limited solely to detect bounding box features or brightness, this
In embodiment, the explanation carried out for being only frame and the brightness by the frame of train tail apparatus and Tail light for train,
Neural network model can also be using the other feature of train tail apparatus or Tail light for train to train safety sign in practical application
State is detected, for example, shape feature, color characteristic or the area features of train safety sign, this embodiment party can be used
It is only to be illustrated by taking bounding box features and brightness as an example in formula, no matter using any kind of feature, as long as can be realized
Detection to train safety sign is all within the scope of protection of this application.
Step 103, in the case where determining testing result is train safety sign abnormal state, alarm signal is issued in real time
Breath.
It specifically, in the present embodiment, can be straight in the case where determining that testing result is the normal situation of train safety sign
Connecing will test result and is saved, and in the case where determining testing result is train safety sign abnormal state, it will test
It as a result, can also real-time alert while preservation.
At one in the specific implementation, in the case where determining train tail apparatus exception, " train tail apparatus lacks meeting alert
It loses ";In the case where determining Tail light for train exception, different warning messages can be issued according to different testing results, for example,
In the case where determining Tail light for train exception, then the classification of Tail light for train abnormal state is determined first, however, it is determined that for Tail light for train the
A kind of abnormality, then can alert " Tail light for train does not light " in real time, however, it is determined that abnormal for the second class of Tail light for train
State, then can real-time alert " Tail light for train missing ".
Compared with prior art, the train safety sign detection method neural network based of the application embodiment is led to
The image for being analyzed and processed to train image and obtaining last vehicle of train region is crossed, and is obtained by neural network model to train tail
The testing result of train safety sign in portion's area image, is avoided and is made by being detected by the way of artificial observation
At erroneous judgement, to improve the accuracy rate of train safety sign detection and the working efficiency of staff, while by by machine
Device recognition detection replaces traditional eye recognition to detect, to reduce the investment of station staff.
Second embodiment of the present invention is related to a kind of train safety sign detection method neural network based.This implementation
Example is further improved on the basis of first embodiment, specific improvements are as follows: increase after obtaining testing result
In the case where determining testing result is train safety sign exception, real-time alert, and warning message is transmitted
To the step on terminal device.Process such as Fig. 3 of train safety sign detection method neural network based in the present embodiment
It is shown.Specifically, in the present embodiment, including step 201 is to step 204, wherein step 201 to step 203 and first real
The step 101 applied in mode is roughly the same to step 103, and details are not described herein again, difference is mainly introduced below, not in this reality
The technical detail of detailed description in mode is applied, reference can be made to the method for train safety sign provided by first embodiment, herein not
It repeats again.
After step 201 to step 203, step 204 is executed.
In step 204, warning message is defeated by the terminal device that there is communication connection with server.
It should be noted that in the present embodiment, since server can have communication connection to close with multiple terminal devices
System, the terminal device in present embodiment can be smartwatch or mobile phone, and terminal device be station staff institute with
What body carried.After warning message is transferred on terminal device by server, staff can be timely according to warning message
The measure of being correspondingly improved is made, such as after the warning message for receiving " train tail apparatus missing ", it can be in time in last vehicle of train
New train tail apparatus is installed, after the warning message for receiving " Tail light for train does not light ", Tail light for train can be carried out in time
Maintenance, to improve the safety of train driving.Certainly, as shown in Fig. 2, the terminal device in present embodiment can also be work
Make platform, workbench can be a computer, by the way that warning message is transferred to workbench.Workbench is receiving warning message
Afterwards, warning message can be broadcasted in platform, thus avoid staff terminal device entrained with oneself occur therefore
In the case where barrier, the case where warning message can not be received generation.
It should be noted that being arranged to further increase the accuracy of detection in the image to last vehicle of train region
The detection of vehicle safety sign, and after obtaining testing result, if user is according to the warning message of acquisition to train safety sign
When carrying out site examining and repairing, discovery warning message and in the case that actual scene is not inconsistent, the train that user can will detect
The image of tail region is marked.In the case where warning message and actual scene are not inconsistent, server can receive user's input
The last vehicle of train region with mark information image, and according to mark information last vehicle of train region image to mind
It is trained through network model, thus optimization neural network model.
Compared with prior art, the train safety sign detection method neural network based that present embodiment provides is led to
The image for being analyzed and processed to train image and obtaining last vehicle of train region is crossed, and is obtained by neural network model to train tail
The testing result of train safety sign in portion's area image, is avoided and is made by being detected by the way of artificial observation
At erroneous judgement, to improve the accuracy rate i.e. working efficiency of staff of train safety sign detection, while by by machine
Device recognition detection replaces traditional eye recognition to detect, to reduce the investment of railroader.And by by alarm signal
Breath is transferred on terminal device in time, in order to which staff is according to warning message acquired on terminal device, is taken in time
Corresponding corrective measure, to improve the safety of train driving.
The step of various methods divide above, be intended merely to describe it is clear, when realization can be merged into a step or
Certain steps are split, multiple steps are decomposed into, as long as including identical logical relation, all in the protection scope of this patent
It is interior;To adding inessential modification in algorithm or in process or introducing inessential design, but its algorithm is not changed
Core design with process is all in the protection scope of the patent.
Third embodiment of the invention is related to a kind of detection device of train safety sign, and specific structure is for example as shown in Figure 4.
As shown in figure 3, the detection device of train safety sign includes: to obtain module 301, detection module 302 and alarm mould
Block 303.
Wherein, module 301 is obtained, for obtaining the image in last vehicle of train region according to train image.
Detection module 302, for carrying out train safety sign by image of the neural network model to last vehicle of train region
Detection, and obtain testing result.
Alarm module 303, for being issued in real time in the case where determining testing result is train safety sign abnormal state
Warning message.
It is not difficult to find that present embodiment is Installation practice corresponding with first embodiment, present embodiment can be with
First embodiment is worked in coordination implementation.The relevant technical details mentioned in first embodiment still have in the present embodiment
Effect, in order to reduce repetition, which is not described herein again.Correspondingly, the relevant technical details mentioned in present embodiment are also applicable in
In first embodiment.
Four embodiment of the invention is related to a kind of detection device of train safety sign.The embodiment and third are implemented
Mode is roughly the same, and specific structure is as shown in Figure 5.Wherein, it mainly thes improvement is that: increasing on the basis of third embodiment
Transmission module 304 is added.
Wherein, module 301 is obtained, for obtaining the image in last vehicle of train region according to train image.
Detection module 302, for carrying out train safety sign by image of the neural network model to last vehicle of train region
Detection, and obtain testing result
Alarm module 303, for being issued in real time in the case where determining testing result is train safety sign abnormal state
Warning message.
Transmission module 304, for warning message to be defeated by the terminal device for having communication connection with server.
It is not difficult to find that present embodiment is Installation practice corresponding with second embodiment, present embodiment can be with
Second embodiment is worked in coordination implementation.The relevant technical details mentioned in second embodiment still have in the present embodiment
Effect, in order to reduce repetition, which is not described herein again.Correspondingly, the relevant technical details mentioned in present embodiment are also applicable in
In second embodiment.
It is noted that each module involved in present embodiment is logic module, and in practical applications, one
A logic unit can be a physical unit, be also possible to a part of a physical unit, can also be with multiple physics lists
The combination of member is realized.In addition, in order to protrude innovative part of the invention, it will not be with solution institute of the present invention in present embodiment
The technical issues of proposition, the less close unit of relationship introduced, but this does not indicate that there is no other single in present embodiment
Member.
Fifth embodiment of the invention is related to a kind of server, as shown in figure 5, including at least one graphics processor
(Graphics Processing Unit, GPU) 501;And the center with the communication connection of at least one graphics processor 501
Processor (Central Processing Unit, CPU) 504;And communicated to connect at least one graphics processor 501
Memory 502;And the communication component 503 with the communication connection of at least one graphics processor 501, communication component 503 is in figure
Data are sended and received under the control of processor;Wherein, memory 502, which is stored with, to be held by least one graphics processor 501
Capable instruction, instruction is executed by least one graphics processor 501, so that at least one graphics processor 501 is able to carry out
State the train safety sign detection method neural network based in embodiment.
In the present embodiment, for memory 502 is with readable and writable memory (Random Access Memory, RAM).Figure
Processor 501, memory 502, communication component 503 and central processing unit 504 can be connected by bus or other modes, figure
In 6 for being connected by bus.Memory 502 is used as a kind of non-volatile computer readable storage medium storing program for executing, can be used for storing
Non-volatile software program, non-volatile computer executable program and module, as realized in the embodiment of the present application based on mind
The program of train safety sign detection method through network is stored in memory 502.Graphics processor 501 is deposited by operation
Store up non-volatile software program, instruction and module in memory 502, thereby executing equipment various function application and
Above-mentioned train safety sign detection method neural network based is realized in data processing.Central processing unit 504 by with figure
Processor 501 communicate to connect, to graphics processor 501 realize above-mentioned train safety sign detection method neural network based into
The effect of row real time monitoring and overall control.
Memory 502 may include storing program area and storage data area, wherein storing program area can store operation system
Application program required for system, at least one function;It storage data area can the Save option list etc..In addition, memory can wrap
High-speed random access memory is included, can also include nonvolatile memory, for example, at least disk memory, a flash memories
Part or other non-volatile solid state memory parts.In some embodiments, it includes relative to processor 501 that memory 502 is optional
Remotely located memory, these remote memories can pass through network connection to external equipment.The example of above-mentioned network includes
But be not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
One or more program module is stored in memory 502, is held when by one or more graphics processor 501
When row, the train safety sign detection method neural network based in above-mentioned any means embodiment is executed.
Method provided by the embodiment of the present application can be performed in the said goods, has the corresponding functional module of execution method and has
Beneficial effect, the not technical detail of detailed description in the present embodiment, reference can be made to method provided by the embodiment of the present application.
The sixth embodiment of the application is related to a kind of computer readable storage medium, in the computer readable storage medium
It is stored with computer program, which can be realized when being executed by processor involved in any means embodiment of the present invention
Train safety sign detection method neural network based.
It will be understood by those skilled in the art that implementing the method for the above embodiments is that can pass through
Program is completed to instruct relevant hardware, which is stored in a storage medium, including some instructions are used so that one
A equipment (can be single-chip microcontroller, chip etc.) or processor (processor) execute each embodiment the method for the application
All or part of the steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can store journey
The medium of sequence code.
It will be understood by those skilled in the art that the respective embodiments described above are to realize specific embodiments of the present invention,
And in practical applications, can to it, various changes can be made in the form and details, without departing from the spirit and scope of the present invention.
Claims (10)
1. a kind of train safety sign detection method neural network based, which is characterized in that be applied to server, comprising:
The image in last vehicle of train region is obtained according to train image;
The detection of train safety sign is carried out to the image in the last vehicle of train region by neural network model, and is detected
As a result, wherein the testing result include the train safety sign state normally and the train safety sign abnormal state;
In the case where determining the testing result is the train safety sign abnormal state, real-time alert.
2. train safety sign detection method neural network based according to claim 1, which is characterized in that described
Before the image for obtaining last vehicle of train region according to train image, further includes:
Determine the acquisition signal that the sensor for receiving and being arranged on train rail is sent;
According to the train image of the acquisition signal acquisition photographic device shooting, wherein the acquisition signal is for indicating
The coverage that the train passes through the photographic device.
3. train safety sign detection method neural network based according to claim 1, which is characterized in that described logical
It crosses neural network model and carries out the detection of train safety sign to the image in the last vehicle of train region, and obtain testing result,
It specifically includes:
Feature extraction is carried out using image of the preset neural network model to the last vehicle of train region;
It is detected according to the feature of extraction, obtains testing result;Wherein, the testing result includes the train safety
Flag states are normal or the train safety sign abnormal state.
4. train safety sign detection method neural network based according to claim 3, which is characterized in that described to adopt
Before carrying out feature extraction with image of the preset neural network model to the last vehicle of train region, further includes:
The default neural network model is trained using multiple sample images, obtains the default nerve known to parameter
Network model, wherein known to the state of the train safety sign in each described sample image.
5. train safety sign detection method neural network based according to claim 1, which is characterized in that described
The image that last vehicle of train region is obtained according to train image, specifically includes:
Image enhancement processing is carried out to the train image, obtains enhancing image;
The image in last vehicle of train region is obtained according to the enhancing image.
6. train safety sign detection method neural network based according to claim 5, which is characterized in that described
The image that last vehicle of train region is obtained according to the enhancing image, specifically includes:
By carrying out image dividing processing to the enhancing image, the image in the last vehicle of train region is obtained, wherein the column
Vehicle tail region is for installing the train safety sign.
7. train safety sign detection method neural network based according to any one of claims 1 to 6, feature exist
In described in the case where determining the testing result is the train safety sign abnormal state, real-time alert
Later, further includes:
The warning message is defeated by the terminal device for having communication connection with the server.
8. train safety sign detection method neural network based according to claim 6, which is characterized in that described right
The image in the last vehicle of train region carries out the detection of train safety sign, and after obtaining testing result, further includes:
In the case where the warning message and actual scene are not inconsistent, the train with mark information of user's input is received
The image of tail region, and according to the image in the last vehicle of train region with mark information to the neural network mould
Type is trained.
9. a kind of server characterized by comprising at least one graphics processor;And
With the central processing unit of at least one described graphics processor communication connection;And
With the memory of at least one described graphics processor communication connection;And
With the communication component of at least one described graphics processor communication connection, the communication component is in the graphics processor
Data are sended and received under control;
Wherein, the memory is stored with the instruction that can be executed by least one described graphics processor, and described instruction is described
At least one graphics processor executes, so that at least one described graphics processor is able to carry out such as any one of claim 1 to 8
The train safety sign detection method neural network based.
10. a kind of computer readable storage medium, is stored with computer program, which is characterized in that the computer program is located
It manages when device executes and realizes the described in any item train safety sign detection methods neural network based of claim 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811358358.6A CN109614873A (en) | 2018-11-15 | 2018-11-15 | Train safety sign detection method, server and storage medium neural network based |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811358358.6A CN109614873A (en) | 2018-11-15 | 2018-11-15 | Train safety sign detection method, server and storage medium neural network based |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109614873A true CN109614873A (en) | 2019-04-12 |
Family
ID=66003057
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811358358.6A Pending CN109614873A (en) | 2018-11-15 | 2018-11-15 | Train safety sign detection method, server and storage medium neural network based |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109614873A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110276742A (en) * | 2019-05-07 | 2019-09-24 | 平安科技(深圳)有限公司 | Tail light for train monitoring method, device, terminal and storage medium |
CN110287769A (en) * | 2019-05-07 | 2019-09-27 | 平安科技(深圳)有限公司 | Train train tail apparatus monitoring method, system, terminal and storage medium |
CN110481276A (en) * | 2019-09-03 | 2019-11-22 | 隋晓明 | A kind of Vehicular intelligent pass window, ventilation, taillight monitor system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20100010735A (en) * | 2008-07-23 | 2010-02-02 | 한국철도기술연구원 | Monitoring system in railway station for determining dangerous situation using neural network and method thereof |
EP3205528A1 (en) * | 2016-02-09 | 2017-08-16 | Sncf Reseau | Method, device and system for detecting fault(s) in a pantograph of a moving vehicle on a railway |
CN107122747A (en) * | 2017-04-28 | 2017-09-01 | 北京理工大学 | A kind of railway carriage state non-contact detection device and method |
CN107273802A (en) * | 2017-05-16 | 2017-10-20 | 武汉华目信息技术有限责任公司 | A kind of detection method and device of railroad train brake shoe drill ring failure |
CN108320582A (en) * | 2018-03-30 | 2018-07-24 | 合肥城市泊车投资管理有限公司 | A kind of parking management system having remaining parking stall statistical function |
-
2018
- 2018-11-15 CN CN201811358358.6A patent/CN109614873A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20100010735A (en) * | 2008-07-23 | 2010-02-02 | 한국철도기술연구원 | Monitoring system in railway station for determining dangerous situation using neural network and method thereof |
EP3205528A1 (en) * | 2016-02-09 | 2017-08-16 | Sncf Reseau | Method, device and system for detecting fault(s) in a pantograph of a moving vehicle on a railway |
CN107122747A (en) * | 2017-04-28 | 2017-09-01 | 北京理工大学 | A kind of railway carriage state non-contact detection device and method |
CN107273802A (en) * | 2017-05-16 | 2017-10-20 | 武汉华目信息技术有限责任公司 | A kind of detection method and device of railroad train brake shoe drill ring failure |
CN108320582A (en) * | 2018-03-30 | 2018-07-24 | 合肥城市泊车投资管理有限公司 | A kind of parking management system having remaining parking stall statistical function |
Non-Patent Citations (1)
Title |
---|
谢利明: "基于隐式形状模型的机车走行部关键部件检测算法研究", 《中国博士学位论文全文数据库》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110276742A (en) * | 2019-05-07 | 2019-09-24 | 平安科技(深圳)有限公司 | Tail light for train monitoring method, device, terminal and storage medium |
CN110287769A (en) * | 2019-05-07 | 2019-09-27 | 平安科技(深圳)有限公司 | Train train tail apparatus monitoring method, system, terminal and storage medium |
CN110276742B (en) * | 2019-05-07 | 2023-10-10 | 平安科技(深圳)有限公司 | Train tail lamp monitoring method, device, terminal and storage medium |
CN110287769B (en) * | 2019-05-07 | 2023-12-26 | 平安科技(深圳)有限公司 | Train tail device monitoring method, system, terminal and storage medium |
CN110481276A (en) * | 2019-09-03 | 2019-11-22 | 隋晓明 | A kind of Vehicular intelligent pass window, ventilation, taillight monitor system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106373426B (en) | Parking stall based on computer vision and violation road occupation for parking monitoring method | |
CN111080598B (en) | Bolt and nut missing detection method for coupler yoke key safety crane | |
CN109255288A (en) | A kind of road surface breakage detection method, device and terminal device | |
CN109614873A (en) | Train safety sign detection method, server and storage medium neural network based | |
CN110111581A (en) | Target identification method, device, computer equipment and storage medium | |
CN110674790B (en) | Abnormal scene processing method and system in video monitoring | |
CN109325531A (en) | Car damage identification method, device, equipment and storage medium based on image | |
CN111753612B (en) | Method and device for detecting casting object and storage medium | |
CN109635823B (en) | Method and device for identifying winding disorder rope and engineering machinery | |
CN104463204B (en) | Destination number statistical method | |
CN109460787A (en) | IDS Framework method for building up, device and data processing equipment | |
CN112052815A (en) | Behavior detection method and device and electronic equipment | |
CN112100039B (en) | Equipment fault alarm method and system | |
CN109580653A (en) | A kind of rail fastener image-pickup method | |
CN112613453A (en) | Method and system for checking violation of regulations on construction site of electric power infrastructure | |
CN116863297A (en) | Monitoring method, device, system, equipment and medium based on electronic fence | |
CN109094574A (en) | A kind of unmanned vehicle driving condition Measurement and Control System based on rack | |
CN115995037A (en) | Signal lamp state detection method, device, equipment, medium and product | |
CN103600752A (en) | Automatic detection device for special gondola car hook mistake and detection method of special gondola car hook mistake | |
CN110263622A (en) | Train fire monitoring method, apparatus, terminal and storage medium | |
CN112967224A (en) | Electronic circuit board detection system, method and medium based on artificial intelligence | |
CN110533627B (en) | Method for detecting breaking of net rope | |
CN109063544A (en) | A kind of method, system, computer equipment and the storage medium of dump truck detection | |
CN112037182A (en) | Locomotive running gear fault detection method and device based on time sequence image and storage medium | |
CN111931693A (en) | Traffic sign recognition method, device, terminal and medium based on artificial intelligence |
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: 20190412 |