WO2019183752A1 - 一种车前积雪与结冰的检测报警方法、存储介质和服务器 - Google Patents

一种车前积雪与结冰的检测报警方法、存储介质和服务器 Download PDF

Info

Publication number
WO2019183752A1
WO2019183752A1 PCT/CN2018/080477 CN2018080477W WO2019183752A1 WO 2019183752 A1 WO2019183752 A1 WO 2019183752A1 CN 2018080477 W CN2018080477 W CN 2018080477W WO 2019183752 A1 WO2019183752 A1 WO 2019183752A1
Authority
WO
WIPO (PCT)
Prior art keywords
snow
ice
vehicle
deep learning
road surface
Prior art date
Application number
PCT/CN2018/080477
Other languages
English (en)
French (fr)
Inventor
周清华
刘光军
陈炎平
黄雯
Original Assignee
深圳市锐明技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市锐明技术股份有限公司 filed Critical 深圳市锐明技术股份有限公司
Priority to CN201880000213.0A priority Critical patent/CN108701396B/zh
Priority to PCT/CN2018/080477 priority patent/WO2019183752A1/zh
Publication of WO2019183752A1 publication Critical patent/WO2019183752A1/zh

Links

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B19/00—Alarms responsive to two or more different undesired or abnormal conditions, e.g. burglary and fire, abnormal temperature and abnormal rate of flow
    • G08B19/02—Alarm responsive to formation or anticipated formation of ice
    • 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
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • 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
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle

Definitions

  • the present invention relates to the field of image processing technologies, and in particular, to a method for detecting and alarming snow and ice in front of a vehicle, a storage medium, and a server.
  • the embodiment of the invention provides a method for detecting and alarming snow and ice in front of the vehicle, a storage medium and a server.
  • a method for detecting and alarming snow and ice in front of a vehicle including:
  • it also includes:
  • the preset time or the preset distance sends an early warning message to the vehicle that is going to travel.
  • the method further includes:
  • the recognition result is that there is snow or ice on the road surface, the weather condition of the area to which the location corresponding to the positioning information belongs is obtained;
  • the unsupervised deep learning model is pre-trained by the following steps:
  • the model parameters of the unsupervised deep learning model are adjusted, and the unsupervised deep learning model after the model parameter adjustment is used as an initial unsupervised deep learning model, and the execution is performed by each Converting the sample picture into an input vector input to the initial unsupervised deep learning model and subsequent steps;
  • it also includes:
  • the generated alarm information is stored in the designated alarm list
  • the alarm information required for the request is queried from the alarm list, and then the alarm information required for the request is fed back to the requesting party.
  • a second aspect of the present invention provides a detection and alarm device for snow and ice in front of a vehicle, comprising:
  • a shooting video acquisition module configured to acquire a first video of a camera on a vehicle to capture a scene in front of the vehicle
  • a picture intercepting module configured to intercept, from a video frame of the first video, a first picture that represents a road surface condition in the front scene
  • An identification module configured to input the first picture to an unsupervised deep learning model completed by using unsupervised deep learning pre-training, to obtain a recognition result output by the unsupervised deep learning model, where the recognition result is that there is a product on the road surface Snow or ice, or no snow or ice on the road;
  • the alarm module is configured to issue an alarm message if the recognition result is that there is snow or ice on the road surface.
  • the detecting and alarming device for snow and ice in front of the vehicle further comprises:
  • a positioning information acquiring module configured to acquire positioning information of the vehicle when the camera captures a first video of a scene in front of the vehicle;
  • An association storage module configured to store the positioning information in association with the identification result
  • the early warning module is configured to: if the recognition result is that there is snow or ice on the road surface, and detecting that the vehicle is going to travel through the location corresponding to the positioning information, the preset time or the preset distance is to be traveled The vehicle issues an early warning message.
  • the detecting and alarming device for snow and ice in front of the vehicle further comprises:
  • a weather condition obtaining module configured to obtain weather conditions of a region corresponding to the location corresponding to the positioning information if the recognition result is that snow or ice is present on the road surface
  • An ablation duration determining module configured to determine a length of time required for snow or icing ablation in the region to be located according to the weather condition
  • the ablation time determining module is configured to determine an ablation time of snow or ice on the location corresponding to the positioning information according to the determined duration;
  • the effectiveness modification module is configured to modify the state of the recognition result stored in association with the positioning information to be invalid when the ablation time arrives.
  • a third aspect of the present invention provides a server including a memory, a processor, and a computer program stored in the memory and operable on the processor, the processor implementing the computer when the computer program is executed The steps of the detection method of the front snow and icing.
  • a computer readable storage medium storing a computer program, the computer program being executed by a processor to implement the above-described method for detecting and alarming snow and ice in front of a vehicle A step of.
  • the first video of the scene of the vehicle is captured by acquiring the camera, and the picture representing the road condition is intercepted from the image, and the picture is input to the unsupervised deep learning model completed by the unsupervised deep learning pre-training to obtain the recognition result. Therefore, it is possible to obtain whether there is snow or ice on the front road surface of the vehicle, and if there is, an alarm message is issued, which realizes detection of snow and icing conditions of the road area and timely sends out alarm information, thereby greatly reducing snow and knots. The adverse effects of ice on the road.
  • FIG. 1 is a flow chart of an embodiment of a method for detecting and alarming snow and ice in front of a vehicle according to an embodiment of the present invention
  • FIG. 2 is a schematic flow chart of a pre-training unsupervised deep learning model in an application scenario according to an embodiment of the invention for detecting and alarming snow and ice in front of the vehicle;
  • FIG. 3 is a schematic flow chart of providing early warning information for other vehicles in an application scenario according to an embodiment of the invention for detecting and alarming snow and ice in front of the vehicle;
  • FIG. 4 is a schematic flow chart of estimating ice and snow ablation time in an application scenario according to an embodiment of the invention for detecting and alarming snow and ice in front of the vehicle;
  • FIG. 5 is a structural diagram of an embodiment of a device for detecting and alarming snow and ice in front of a vehicle according to an embodiment of the present invention
  • FIG. 6 is a schematic diagram of a server according to an embodiment of the present invention.
  • the embodiment of the invention provides a method for detecting and alarming snow and icing in front of a vehicle, a storage medium and a server, which are used for solving the problem of detecting the snow and icing conditions of the road area and issuing corresponding alarms in time.
  • the invention adopts an image method based on unsupervised deep learning to detect the snow and icing conditions of the road area in real time and issue a real-time alarm.
  • all vehicles installing the method or system are connected to the platform to form a city network.
  • Information sharing can be achieved among all vehicles in each city. After a vehicle detects the presence of snow and ice in a certain area, it can remind the other vehicles passing through the area in advance for a certain period of time to maximize the dangerous situation. Early warning is provided to prevent unnecessary accidents.
  • an embodiment of a method for detecting and alarming snow and ice in front of a vehicle in an embodiment of the present invention includes:
  • Step S101 Acquire a first video of a camera on the vehicle to capture a scene in front of the vehicle;
  • the executor of the embodiment may be a terminal device or a server.
  • the executor of the embodiment is a server, such as a cloud server platform.
  • the vehicle can be equipped with an ADAS (Advanced Driver Assistance System) camera at a suitable position on the front window of the vehicle for capturing video and video of the scene in front of the vehicle, so that the server can obtain the first video by communicating with the vehicle's ADAS system.
  • ADAS Advanced Driver Assistance System
  • the camera can be specifically mounted on the vertical centerline of the front window of the vehicle to prevent the wiper from affecting or causing damage to its work.
  • the communication module communicating with the server on the vehicle can be specifically connected to the MDVR (Mobile). Digital Video Recorders), and can be applied to 2G/3G/4G/5G network bandwidth.
  • the communication module is generally realized by the corresponding antenna connection device, and the wired network communication can also be used in the indoor test environment.
  • Step S102 extracting, from a video frame of the first video, a first picture that represents a road surface condition in the front scene;
  • each video frame in the first video may be acquired, and a picture including the front road surface condition, that is, the first picture described above, is intercepted from the video frame.
  • the camera is usually fixedly mounted on a vehicle, and the shooting angle of the camera is unchanged after installation.
  • the camera can be installed on the vertical center line of the front window of the vehicle, and the camera is facing the vehicle.
  • the road surface condition image contained in the first video captured by the camera occupies a fixed area in the video frame. Therefore, when the first picture representing the road surface condition is captured in the front scene, the fixed area image in the video frame of the first video may be intercepted, and the cut fixed area image is the first picture.
  • the intercepting of the first picture may include the following steps:
  • step S102 the image in the intercepted area is extracted from the video frame of the first video to obtain the first picture.
  • Step S103 input the first picture to an unsupervised deep learning model completed by using unsupervised deep learning pre-training, and obtain a recognition result output by the unsupervised deep learning model, where the recognition result is that snow exists on the road surface or Icing, or there is no snow or ice on the road;
  • the first picture may be input to an unsupervised deep learning model completed by unsupervised deep learning pre-training, and the unsupervised deep learning model identifies and determines the first picture,
  • the result of the recognition is that there is snow or ice on the road surface, or there is no snow or ice on the road surface.
  • the unsupervised deep learning model is obtained by training a large number of training samples in advance, and can identify and judge the road surface condition in the first picture, thereby knowing whether there is a product in the road surface in the first picture. Snow or icing.
  • Step S104 If the recognition result is that there is snow or ice on the road surface, an alarm message is issued.
  • the alarm information may be sent by an LED or LCD prompter on the vehicle, for example, a voice prompt or a screen prompt may be sent to the driver of the vehicle, so that the driver can detect the danger in the first time and timely perform the slowdown processing. .
  • the generated alarm information is stored in the designated alarm list; when receiving the request for querying the alarm information, The alarm list is queried for the alarm information required by the request, and then the alarm information required for the request is fed back to the requesting party.
  • the alarm information can be transmitted to the server, and the server adds the alarm information to the specified alarm list, and can select to classify the alarm information in the alarm list according to time or according to the alarm type. Arrange for easy access.
  • the server may also store the first picture and/or the first video as needed, wherein the number of pictures or the length of the video may be set as needed to be used as evidence for facilitating subsequent call of these alarm pictures or alarms according to needs. video.
  • the unsupervised deep learning model can be pre-trained by the following steps:
  • Step S201 pre-collecting a plurality of sample videos captured from the scene before the vehicle;
  • Step S202 extracting a sample picture representing a road surface condition from each of the sample videos
  • Step S203 converting each of the sample pictures into an input vector and inputting into an initial unsupervised deep learning model
  • Step S204 Encoding and decoding the input vector by using the initial unsupervised deep learning model to obtain an output vector
  • Step S205 calculating an output error between the output vector and the input vector
  • Step S206 If the output error does not meet the preset condition, adjust the model parameter of the unsupervised deep learning model, and use the unsupervised deep learning model adjusted by the model parameter as an initial unsupervised deep learning model, and return to execution. Converting each of the sample pictures into a step of inputting an input vector into an initial unsupervised deep learning model and subsequent steps;
  • Step S207 After the output error meets the preset condition, determining that the unsupervised deep learning model training is completed.
  • step S201 before training the unsupervised deep learning model, it is necessary to pre-collect a plurality of sample videos for training, which are taken from the front scene, including snow or ice on the road surface before the vehicle, and There is no snow or ice on the road front.
  • step S202 is similar to the content of the above step S102, and the principle is basically the same, and details are not described herein again.
  • the sample pictures are converted into a vector form and input into the unsupervised deep learning model, that is, the sample pictures are converted into input vectors and input to the Unsupervised deep learning model to facilitate subsequent calculations and processing.
  • the unsupervised deep learning model adopts an automatic coding mode, that is, after the initial unsupervised deep learning model acquires input vectors of the sample images in the above step S204, the input vectors are obtained by an automatic encoder. Converted into feature coding to complete the coding process; then the code is converted by the decoder into a data form that the unsupervised deep learning model can recognize, that is, the output vector is obtained by the decoding process of the decoder, so that the unsupervised deep learning model It is possible to learn the features to be extracted, that is, to learn a special expression of the sample picture.
  • an output error between the output vector and an input vector corresponding to the sample picture may be calculated, and it is determined whether the output error meets a preset condition.
  • step S206 if the output error does not meet the preset condition, the model parameters of the unsupervised deep learning model are adjusted, and the unsupervised deep learning model after the model parameter adjustment is used as the initial unsupervised deep learning.
  • the model returns to performing the steps of converting each sample picture into an input vector input into the initial unsupervised deep learning model and subsequent steps to reduce the output error such that the error between the output vector of the subsequent training and the input vector is minimized.
  • step S207 after repeatedly adjusting the model parameters of the unsupervised deep learning model, after performing multiple trainings, comparing the output error between each output vector and the input vector corresponding to the training group sample, if the output error satisfies the Pre-set conditions, such as an output error of less than 5%, can be determined that the unsupervised deep learning model training is completed.
  • the preset condition may be determined when training a specific unsupervised deep learning model, such as setting an output error to be less than a specific threshold, and the specific threshold may be a percentage value, and the smaller the specific threshold, the less the final training is completed. The more stable the supervised deep learning model is, the higher the recognition accuracy is.
  • the unsupervised deep learning model includes an input layer, a hidden layer, and an output layer
  • the model parameters of the unsupervised deep learning model include a weight matrix, a first offset vector between the input layer and the hidden layer, and a hidden layer.
  • a second partial vector to the output layer for inputting data, the hidden layer for encoding and decoding data, and the output layer for re-encoding the encoded and decoded data
  • Input to the input layer to open the iterative process of the next encoding and decoding of the hidden layer, wherein, at the beginning of the training, the weight matrix, the first bias vector and the second offset vector of the initial unsupervised deep learning model are first used.
  • the training process of the unsupervised deep learning model is specifically as follows: inputting an input vector of a sample picture into the unsupervised deep learning model through an input layer; the hidden layer in the unsupervised deep learning model is the input The vector is reconstructed to obtain the output vector, that is, the feature vector is generated by transforming the input vector and the weight matrix, and then the feature encoding is performed with the transposed matrix of the weight matrix to obtain an output vector; An output error between the output vector and the input vector, determining whether the output error reaches a preset minimum error value, and if so, determining that the unsupervised deep learning model training is completed, ie, the current weight matrix, An offset vector and a second offset vector are trained optimal model parameters; if not, the output error is inversely propagated to the hidden layer using a gradient descent method to update the weight matrix, the first offset vector And the second offset vector, while the output vector is re-introduced as input to the input layer, and the next training is started. Substituting process to reduce the output error.
  • the unsupervised learning mode is used to perform the training of the unsupervised deep learning model, and it is not necessary to pre-classify the training samples and do not need to know the classification labels of the training samples in advance, thereby reducing the difficulty of obtaining the training samples and improving the training. Efficiency, increasing the scope of application of the unsupervised deep learning model.
  • the recognition result in this embodiment may also be combined with the positioning information to provide early warning information for other vehicles.
  • the method for detecting and detecting snow and ice in front of the vehicle may further include:
  • Step S301 Acquire positioning information of the vehicle when the camera captures a first video of a scene in front of the vehicle;
  • Step S302 storing the positioning information in association with the identification result
  • Step S303 if the recognition result is that there is snow or ice on the road surface, and it is detected that the vehicle is going to travel through the location corresponding to the positioning information, the preset time or the preset distance is used to issue an early warning to the vehicle that is going to travel. information.
  • the server can combine the GPS positioning and the high-precision map, and determine the positioning information of the vehicle when the first video of the front scene is captured by the camera, that is, the identification is determined.
  • the result corresponds to the position of the road surface, and the positioning information is stored in association with the recognition result, so that the server can have snow or ice on which position of the corresponding position mark of the high-precision map, and which position does not have snow or ice.
  • the server is connected to a large number of vehicles, the more roads the vehicle has traveled, the more identified positions the server marks on the high-precision map, so that after a period of information accumulation, the server can theoretically know that The road surface condition of each road surface on the high-precision map.
  • the server Since the server stores the positioning information and the recognition result in association, when the server finds that the vehicle is about to travel past a certain location where there is snow or ice, the server may issue the vehicle a short time or distance. Early warning information so that the driver on the vehicle can make countermeasures or prepare for work in advance.
  • the method for detecting snow and ice in front of the vehicle may further include:
  • Step S401 if the recognition result is that there is snow or ice on the road surface, acquiring weather conditions of the area to which the location corresponding to the positioning information belongs;
  • Step S402 determining, according to the weather condition, a length of time required for snow or icing in the region to be ablated;
  • Step S403 determining, according to the determined duration, an ablation time of snow or ice on the location corresponding to the positioning information
  • Step S404 when the ablation time arrives, the state of the recognition result stored in association with the positioning information is modified to be invalid.
  • step S401 if the recognition result is that there is snow or ice on the road surface, if there is snow or ice on the road surface, the server has snow or ice on the location corresponding to the location information, and the server estimates the location.
  • the server may obtain weather information of the area from the website of the weather bureau, and the weather information includes information such as temperature, air humidity, and snowfall.
  • the length of time required for snow or icing to be ablated may be determined according to the weather condition.
  • the duration can be determined by the correspondence between the preset weather conditions and the ablation duration. For example, according to the empirical value, the ablation duration corresponding to "the temperature is between 10 and 20 degrees, and the snowfall is less than 1.0 mm in 12 hours" can be set to 1 hour.
  • step S403 it can be understood that after obtaining the length of time required for snow or icing ablation in the region to which it belongs, the ablation time of snow or ice on the ground corresponding to the positioning information can be calculated.
  • the duration required for the ablation is equal to the ablation time minus the shooting time point of the first video, and the server can know the shooting time point of the first video, so that the ablation time can be calculated to be equal to the shooting time of the first video plus The length of time required for the upper ablation. For example, if the shooting time of the first video is 9:00 am and the time required for ablation is 1 hour, the ablation time is 10:00 am.
  • step S404 when the ablation time arrives, it can be understood that if the recognition result is that there is snow or ice on the road surface, when the ablation time arrives, the positioning information may be considered to correspond to snow or knot on the ground.
  • the ice has been ablated, so the recognition result has failed for the warning function of the server, and the server can modify the state of the recognition result stored in association with the positioning information to be invalid, indicating that the location information corresponding to the positioning information is no longer There is snow or ice.
  • the present embodiment can maximize the warning of the snow or icing danger occurring in front of the driver's driving, and accurately determine the length of time of the danger signal at the alarm position.
  • the server can comprehensively and accurately provide the driver with a real-time warning of the danger of snow or ice in front of the vehicle to minimize the occurrence of traffic accidents.
  • the first video of the scene in front of the vehicle is captured by acquiring the camera, and a picture representing the condition of the road surface is intercepted therefrom, and the picture is input to an unsupervised deep learning model completed by the unsupervised deep learning pre-training to obtain the recognition result. Therefore, it is possible to obtain whether there is snow or ice on the front road surface of the vehicle, and if there is, an alarm message is issued, which realizes detection of snow and icing conditions of the road area and timely sends out alarm information, thereby greatly reducing snow and icing.
  • the road surface condition has an adverse effect on driving.
  • the embodiment uses an unsupervised deep learning algorithm to detect the snow or icing condition of the road surface, and the accuracy of the detection can be continuously increased as the number of samples increases; and the vehicles can be interconnected by the server, provided by the vehicle.
  • the alarm information can provide early warning for other vehicles in the corresponding areas corresponding to the alarm within a certain period of time, thus forming an overall city (or other regional) alarm linkage network, providing a more comprehensive alarm strategy, and at the same time
  • the strategy is also more convenient for the operation and management of all vehicles.
  • the above mainly describes a method for detecting and warning of snow and ice in front of the vehicle, and a description will be given below of a detection and alarm device for snow and ice in front of the vehicle.
  • FIG. 5 is a structural diagram showing an embodiment of a detection and alarm device for snow and ice in front of a vehicle in an embodiment of the present invention.
  • a detection and alarm device for snow and ice in front of the vehicle includes:
  • the photographing video acquisition module 501 is configured to acquire a first video of the front camera scene captured by the camera on the vehicle;
  • a picture intercepting module 502 configured to intercept, from a video frame of the first video, a first picture that represents a road surface condition in the front scene;
  • An identification module 503 configured to input the first picture to an unsupervised deep learning model completed by using unsupervised deep learning pre-training, to obtain a recognition result output by the unsupervised deep learning model, where the recognition result is existing on the road surface Snow or ice, or no snow or ice on the road;
  • the alarm module 504 is configured to issue an alarm message if the recognition result is that there is snow or ice on the road surface.
  • the detecting and alarming device for snow and ice in front of the vehicle may further include:
  • a positioning information acquiring module configured to acquire positioning information of the vehicle when the camera captures a first video of a scene in front of the vehicle;
  • An association storage module configured to store the positioning information in association with the identification result
  • the early warning module is configured to: if the recognition result is that there is snow or ice on the road surface, and detecting that the vehicle is going to travel through the location corresponding to the positioning information, the preset time or the preset distance is to be traveled The vehicle issues an early warning message.
  • the detecting and alarming device for snow and ice in front of the vehicle may further include:
  • a weather condition obtaining module configured to obtain weather conditions of a region corresponding to the location corresponding to the positioning information if the recognition result is that snow or ice is present on the road surface
  • An ablation duration determining module configured to determine a length of time required for snow or icing ablation in the region to be located according to the weather condition
  • the ablation time determining module is configured to determine an ablation time of snow or ice on the location corresponding to the positioning information according to the determined duration;
  • the effectiveness modification module is configured to modify the state of the recognition result stored in association with the positioning information to be invalid when the ablation time arrives.
  • the unsupervised deep learning model can be pre-trained by the following steps:
  • a sample video collection module configured to pre-acquire a plurality of sample videos captured from a scene before the vehicle
  • a sample picture intercepting module configured to intercept, from each of the sample videos, a sample picture that characterizes a road surface condition
  • a sample picture input module configured to convert each of the sample pictures into an input vector input into an initial unsupervised deep learning model
  • a codec processing module configured to encode and decode the input vector by using the initial unsupervised deep learning model to obtain an output vector
  • An output error calculation module configured to calculate an output error between the output vector and the input vector
  • a model parameter adjustment module configured to adjust a model parameter of the unsupervised deep learning model if the output error does not meet a preset condition, and use an unsupervised deep learning model with model parameter adjustment as an initial unsupervised deep learning a model that returns to perform the step of converting each of the sample pictures into an input vector input into an initial unsupervised deep learning model and subsequent steps;
  • the training completion determining module is configured to determine that the unsupervised deep learning model training is completed until the error satisfies a preset condition.
  • the detecting and alarming device for snow and ice in front of the vehicle may further include:
  • An alarm list module configured to store the generated alarm information to a designated alarm list if the recognition result is that there is snow or ice on the road surface;
  • the alarm information querying module is configured to, when receiving the request for querying the alarm information, query the alarm list for the alarm information required by the request, and then feed back the alarm information required by the request to the requesting party.
  • FIG. 6 is a schematic diagram of a server according to an embodiment of the present invention.
  • the server 6 of this embodiment includes a processor 60, a memory 61, and a computer program 62 stored in the memory 61 and operable on the processor 60, for example, performing the above-described snow accumulation in front of the vehicle.
  • a program for detecting alarms with icing The processor 60 executes the computer program 62 to implement the steps in the foregoing embodiments of the method for detecting and warning the snow and ice in front of the vehicle, such as steps S101 to S104 shown in FIG.
  • the processor 60 when executing the computer program 62, implements the functions of the modules/units in the various apparatus embodiments described above, such as the functions of the modules 501 through 504 shown in FIG.
  • the computer program 62 can be partitioned into one or more modules/units that are stored in the memory 61 and executed by the processor 60 to complete this invention.
  • the one or more modules/units may be instruction segments of a series of computer programs capable of performing a particular function, the instruction segments being used to describe the execution of the computer program 62 in the server 6.
  • the server 6 can be a computing device such as a local server or a cloud server.
  • the server may include, but is not limited to, a processor 60, a memory 61. It will be understood by those skilled in the art that FIG. 6 is merely an example of the server 6, and does not constitute a limitation to the server 6, and may include more or less components than those illustrated, or some components may be combined, or different components, such as
  • the server may also include an input and output device, a network access device, a bus, and the like.
  • the processor 60 can be a central processing unit (Central Processing Unit, CPU), can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc.
  • the general purpose processor may be a microprocessor or the processor or any conventional processor or the like.
  • the memory 61 may be an internal storage unit of the server 6, such as a hard disk or a memory of the server 6.
  • the memory 61 may also be an external storage device of the server 6, such as a plug-in hard disk, a smart memory card (SMC), and a secure digital (SD) card provided on the server 6. Flash card (Flash Card) and so on.
  • the memory 61 may also include both an internal storage unit of the server 6 and an external storage device.
  • the memory 61 is used to store the computer program and other programs and data required by the server.
  • the memory 61 can also be used to temporarily store data that has been output or is about to be output.
  • modules, units, and/or method steps of the various embodiments described in connection with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods for implementing the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the present invention.
  • the disclosed system, apparatus, and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • there may be another division manner for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the integrated unit if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium.
  • the technical solution of the present invention which is essential or contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product stored in a storage medium.
  • a number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), and a random access memory (RAM, Random Access).

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)

Abstract

一种车前积雪与结冰的检测报警方法，用于解决如何检测路面积雪与结冰情况并及时发出相应报警的问题。该方法包括：获取车辆上摄像头拍摄车前场景的第一视频（S101）；从第一视频的视频帧中截取出车前场景中表征路面状况的第一图片（S102）；将第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到无监督深度学习模型输出的识别结果（S103），识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；若识别结果为路面上存在积雪或结冰，则发出报警信息（S104）。

Description

一种车前积雪与结冰的检测报警方法、存储介质和服务器 技术领域
本发明涉及图像处理技术领域，尤其涉及一种车前积雪与结冰的检测报警方法、存储介质和服务器。
 背景技术
行车途中，路上若有积雪与结冰是令司机非常头疼的一件事，积雪与结冰是路面行车的不稳定因素，是造成很多交通事故的罪魁祸首，长期以来是司机师傅们的心头大患。
因此，如何检测路面积雪与结冰情况并及时发出相应报警成为本领域技术人员亟需解决的问题。
 技术问题
为了解决如何检测路面积雪与结冰情况并及时发出相应报警的问题，本发明实施例提供了一种车前积雪与结冰的检测报警方法、存储介质和服务器。
 技术解决方案
本发明第一方面，提供了一种车前积雪与结冰的检测报警方法，包括：
获取车辆上摄像头拍摄车前场景的第一视频；
从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
可选地，还包括：
获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
将所述定位信息与所述识别结果关联存储；
若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
可选地，在将所述定位信息与所述识别结果关联存储之后，还包括：
若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
可选地，所述无监督深度学习模型通过以下步骤预先训练得到：
预先采集多个拍摄自车前场景的样本视频；
从各个所述样本视频中截取出表征路面状况的样本图片；
将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中；
通过所述初始的无监督深度学习模型对所述输入向量进行编码、解码处理，得到输出向量；
计算所述输出向量与所述输入向量之间的输出误差；
若所述输出误差不符合预设条件，则调整所述无监督深度学习模型的模型参数，并将模型参数调整后的无监督深度学习模型作为初始的无监督深度学习模型，返回执行将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中的步骤以及后续步骤；
直到所述输出误差满足所述预设条件，确定所述无监督深度学习模型训练完成。
可选地，还包括：
若所述识别结果为路面上存在积雪或结冰，则将生成的报警信息存储至指定的报警列表；
当接收到查询报警信息的请求时，从所述报警列表中查询所述请求所需的报警信息，然后将所述请求所需的报警信息反馈给请求方。
本发明第二方面，提供了一种车前积雪与结冰的检测报警装置，包括：
拍摄视频获取模块，用于获取车辆上摄像头拍摄车前场景的第一视频；
图片截取模块，用于从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
识别模块，用于将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
报警模块，用于若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
可选地，所述车前积雪与结冰的检测报警装置还包括：
定位信息获取模块，用于获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
关联存储模块，用于将所述定位信息与所述识别结果关联存储；
提前预警模块，用于若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
可选地，所述车前积雪与结冰的检测报警装置还包括：
天气情况获取模块，用于若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
消融时长确定模块，用于根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
消融时间确定模块，用于根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
效力修改模块，用于当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
本发明第三方面，提供了一种服务器，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现上述车前积雪与结冰的检测报警方法的步骤。
本发明第四方面，提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行时实现上述车前积雪与结冰的检测报警方法的步骤。
 有益效果
本发明实施例中，首先，获取车辆上摄像头拍摄车前场景的第一视频；然后，从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；接着，将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；若所述识别结果为路面上存在积雪或结冰，则发出报警信息。在本发明实施例中，通过获取摄像头拍摄车前场景的第一视频，并从中截取表征路面状况的图片，将该图片输入至利用无监督深度学习预训练完成的无监督深度学习模型得到识别结果，从而可以得到该车辆当前车前路面上是否存在积雪或结冰，若存在，则发出报警信息，实现了检测路面积雪与结冰情况并及时发出报警信息，大大减轻了积雪与结冰的路面情况对行车带来的不良影响。
 附图说明
图1为本发明实施例中一种车前积雪与结冰的检测报警方法一个实施例流程图；
图2为本发明实施例中一种车前积雪与结冰的检测报警方法在一个应用场景下预先训练无监督深度学习模型的流程示意图；
图3为本发明实施例中一种车前积雪与结冰的检测报警方法在一个应用场景下为其它车辆提供预警信息的流程示意图；
图4为本发明实施例中一种车前积雪与结冰的检测报警方法在一个应用场景下估算冰雪消融时间的流程示意图；
图5为本发明实施例中一种车前积雪与结冰的检测报警装置一个实施例结构图；
图6为本发明一实施例提供的服务器的示意图。
 本发明的实施方式
本发明实施例提供了一种车前积雪与结冰的检测报警方法、存储介质和服务器，用于解决如何检测路面积雪与结冰情况并及时发出相应报警的问题。
本发明采用基于无监督深度学***台连网，形成一个城市网，在每个城市内的所有车辆间能够信息共享，在一辆车检测出某地存在路面积雪与结冰后能在一定时间内对经过此地的其他车辆进行提前提醒，最大限度上对危险情况进行提前预警，防止不必要的事故发生。
为使得本发明的发明目的、特征、优点能够更加的明显和易懂，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，下面所描述的实施例仅仅是本发明一部分实施例，而非全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都属于本发明保护的范围。
请参阅图1，本发明实施例中一种车前积雪与结冰的检测报警方法一个实施例包括：
步骤S101、获取车辆上摄像头拍摄车前场景的第一视频；
本实施例的执行主体可以是终端设备或者服务器，优选地，本实施例的执行主体为一服务器，比如云端服务器平台。
该车辆可以在车辆前窗玻璃的适当位置上安装ADAS（高级驾驶辅助***）摄像头，用于拍摄车前场景的视频、录像，从而，服务器可以与车辆的ADAS***通信获取到该第一视频。该摄像头具体可以安装在车辆前窗玻璃的竖直中心线上，以避免雨刮对其工作造成影响或者对其造成损坏。
需要说明的是车辆上与服务器通信的通信模块具体可以连接在MDVR（Mobile Digital Video Recorders）上，并可适用于2G/3G/4G/5G网络带宽，其中，通信模块一般通过相应天线连接设备实现，而室内测试环境下亦可用有线网络通讯。
步骤S102、从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
可以理解的是，在获取到第一视频之后，可以获取第一视频中各个视频帧，并从视频帧中截取出包含有车前路面状况的图片，即上述的第一图片。
可以理解的是，摄像头通常是固定安装在一个车辆上的，且安装后摄像头的拍摄角度不变，比如可以将该摄像头安装在车辆前窗玻璃的竖直中心线上，摄像头拍摄角度正对车辆前方的路面。这样，摄像头拍摄出来的第一视频中包含的路面状况图像占据视频帧中的区域是固定的。因此，在截取出所述车前场景中表征路面状况的第一图片时，可以截取第一视频的视频帧中的固定区域图像，截取出的固定区域图像即为该第一图片。
具体地，该第一图片的截取可以包括以下步骤：
（1）预先获取车辆上摄像头拍摄车前场景的测试视频；
（2）选取出所述测试视频的视频帧中包含车前场景的路面状况的截取区域，并存储所述截取区域；
（3）在执行步骤S102时，从所述第一视频的视频帧中截取出所述截取区域内的图像，得到所述第一图片。
步骤S103、将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
在截取出第一图片之后，可以将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，所述无监督深度学习模型对所述第一图片进行识别判断，得到输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰。
可以理解的是，所述无监督深度学习模型是预先经过大量的训练样本训练完成得到的，可以对第一图片中的路面状况进行识别判断，从而得知该第一图片中的路面是否存在积雪或结冰。
其中，上述无监督深度学习模型的预训练过程将在下述内容中进行详细描述。
步骤S104、若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
在本实施例中，若所述识别结果为路面上存在积雪或结冰，则可以认为当前的路面状况存在不稳定因素，因此发出报警信息。具体地，该报警信息可以由车辆上的LED或LCD提示器发出，比如可以向车辆的驾驶员发出语音提示或者画面提示，使得驾驶员能第一时间察觉危险并及时做出减速慢行的处理。
进一步地，为了便于报警信息的管理，若所述识别结果为路面上存在积雪或结冰，则将生成的报警信息存储至指定的报警列表；当接收到查询报警信息的请求时，从所述报警列表中查询所述请求所需的报警信息，然后将所述请求所需的报警信息反馈给请求方。在本实施例中，在报警信息生成后，该报警信息可以传到服务器上，服务器将报警信息添加至指定的报警列表中，可以选择按照时间或者按照报警类型对报警列表中的报警信息分类、排列，以方便查询。另外，服务器还可以根据需要存储第一图片和/或第一视频，其中图片的张数或视频的长度均可以根据需要设置，以留作证据使用，便于后续根据需要来调用这些报警图片或报警视频。
下面，将对上述卷积神经网络的预训练过程进行详细介绍。如图2所示，所述无监督深度学习模型可以通过以下步骤预先训练得到：
步骤S201、预先采集多个拍摄自车前场景的样本视频；
步骤S202、从各个所述样本视频中截取出表征路面状况的样本图片；
步骤S203、将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中；
步骤S204、通过所述初始的无监督深度学习模型对所述输入向量进行编码、解码处理，得到输出向量；
步骤S205、计算所述输出向量与所述输入向量之间的输出误差；
步骤S206、若所述输出误差不符合预设条件，则调整所述无监督深度学习模型的模型参数，并将模型参数调整后的无监督深度学习模型作为初始的无监督深度学习模型，返回执行将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中的步骤以及后续步骤；
步骤S207、直到所述输出误差满足预设条件时，确定所述无监督深度学习模型训练完成。
对于上述步骤S201，在训练无监督深度学习模型之前，需要预先采集用于训练的多个样本视频，这些样本视频均拍摄自车前场景中，包含了车前路面存在积雪或结冰，以及车前路面不存在积雪或结冰的情况。这些样本视频的数据量越大，对无监督深度学习模型的训练效果就越好。
上述步骤S202与上述步骤S102的内容相似，原理基本相同，此处不再赘述。
对于上述步骤S203，在采集、截取到这些训练用的样本图片之后，将这些样本图片转换成向量的形式输入至所述无监督深度学习模型中，即将这些样本图片转换成输入向量输入至所述无监督深度学习模型中，以方便后续的计算与处理。
在此，所述无监督深度学习模型采用一种自动编码方式，即在上述步骤S204中，所述初始的无监督深度学习模型获取这些样本图片的输入向量之后，通过自动编码器将这些输入向量转化为特征编码，以完成编码处理；随后再通过解码器将这些特征编码转换成无监督深度学习模型能够识别的数据形式，即通过解码器的解码处理得到输出向量，以使得无监督深度学习模型能对所要提取的特征进行学习，即学习到对样本图片的特殊表达方式。
对于上述步骤S205，在所述初始的无监督深度学习模型得到输出向量之后，可以计算所述输出向量与样本图片对应的输入向量之间的输出误差，并判断该输出误差是否符合预设条件。
对于上述步骤S206，若该输出误差不符合所述预设条件的话，则调整所述无监督深度学习模型的模型参数，并将模型参数调整后的无监督深度学习模型作为初始的无监督深度学习模型，返回执行将各个样本图片转换成输入向量输入至初始的无监督深度学习模型中的步骤以及后续步骤，以减小输出误差，使得后续训练的输出向量与输入向量之间的误差最小化。
对于上述步骤S207，在反复调整无监督深度学习模型的模型参数，进行多次训练之后，对比每次的输出向量与训练组样本对应的输入向量之间的输出误差，如果该输出误差满足所述预设条件，比如输出误差小于5%，则可以确定所述无监督深度学习模型训练完成。其中，所述预设条件可以在训练具体的无监督深度学习模型时确定，比如设定输出误差小于特定阈值，该特定阈值可以是一个百分比数值，特定阈值越小，则最后训练完成得到的无监督深度学习模型越稳定，识别精度越高。
具体地，所述无监督深度学习模型包括输入层、隐藏层及输出层，所述无监督深度学习模型的模型参数包括权值矩阵、输入层到隐藏层间的第一偏置向量以及隐藏层到输出层间的第二偏向量，所述输入层用于进行数据的输入，所述隐藏层用于对数据进行编码、解码处理，所述输出层用于将编码、解码处理后的数据重新输入至输入层，以开启隐藏层下一次编码、解码的迭代处理，其中，在训练开始时，首先对初始的无监督深度学习模型的权值矩阵、第一偏置向量以及第二偏置向量进行初始化。
因而，所述无监督深度学习模型的训练过程具体如下：将样本图片的输入向量通过输入层输入至所述无监督深度学习模型中；所述无监督深度学习模型中的隐藏层则对该输入向量进行向量重构，得到输出向量，即通过对该输入向量与权值矩阵进行变换生成特征编码，随后再将该特征编码与该权值矩阵的转置矩阵进行运算，得到输出向量；计算此时的输出向量与输入向量之间的输出误差，判断该输出误差是否达到预设的最小误差值，如果是的话，则确定所述无监督深度学习模型训练完成，即当前的权值矩阵、第一偏置向量及第二偏置向量为训练所得的最优模型参数；如果不是的话，则使用梯度下降法将所述输出误差逆传播至隐藏层，以更新权值矩阵、第一偏置向量和第二偏置向量，同时将所述输出向量作为输入重新投入至输入层，开启下一次训练的迭代处理，以减小输出误差。当输出误差满足预设条件，即达到预设的最小误差值时，则终止迭代，完成所述无监督深度学习模型的训练。
在此，采用无监督的学习方式来进行所述无监督深度学习模型的训练，不需要预先进行训练样本的分类以及不需要预先知道训练样本的分类标签，可降低训练样本的获取难度，提高训练效率，增大所述无监督深度学习模型的适用范围。
进一步地，本实施例中的识别结果还可以与定位信息结合起来，为其它车辆提供预警信息。如图3所示，该车前积雪与结冰的检测报警方法还可以包括：
步骤S301、获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
步骤S302、将所述定位信息与所述识别结果关联存储；
步骤S303、若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
对于上述步骤S301~步骤S303，可以理解的是，服务器可以将GPS定位和高精度地图结合起来，通过获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息，即确定出识别结果对应的路面位置，并将该定位信息与识别结果关联存储，从而服务器可以在高精度地图的对应位置标记上哪个位置存在积雪或结冰、哪个位置不存在积雪或结冰。当服务器与大量的车辆通信连接，车辆行驶过的路面越多时，服务器在高精度地图上标记的已识别的位置也就越多，从而经过一段时间的信息积累之后，服务器理论上可以得知该高精度地图上每一路面位置的路面状况。由于服务器是关联存储所述定位信息与所述识别结果的，因此，当服务器发现有车辆即将行驶经过某个存在积雪或结冰的位置时，服务器可以提前一小段时间或距离对该车辆发出预警信息，以便于车辆上的驾驶员可以提前做出应对措施或准备工作。
进一步地，在将所述定位信息与所述识别结果关联存储之后，如图4所示，该车前积雪与结冰的检测报警方法还可以包括：
步骤S401、若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
步骤S402、根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
步骤S403、根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
步骤S404、当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
对于步骤S401，对于某个定位信息所关联的识别结果，如果该识别结果为路面上存在积雪或结冰，则代表该定位信息对应的地点存在积雪或结冰，服务器为了估算这个地点的积雪或结冰什么时候消融，需要获取该定位信息对应的地点所属地区的天气情况。具体地，服务器可以从气象局的网站获取到该地区的天气信息，该天气信息包括温度、空气湿度、降雪量等信息。
对于步骤S402，在获取到该定位信息对应的地面所属地区的天气情况之后，可以根据该天气情况确定出积雪或结冰消融所需的时长。该时长可以由预先设置的天气情况与消融时长的对应关系确定得到。比如，可以根据经验值设置“温度在10~20度之间，降雪量为12小时内小于1.0mm”对应的消融时长为1小时。
对于步骤S403，可以理解的是，在得到该所属地区中积雪或结冰消融所需的时长之后，可以计算出该定位信息对应的地面上积雪或结冰的消融时间。具体可以是，消融所需的时长等于消融时间减去第一视频的拍摄时间点，服务器可以获知到第一视频的拍摄时间点，从而可以计算得出消融时间等于第一视频的拍摄时间点加上消融所需的时长。举例说明，假设第一视频的拍摄时间点为上午9点，消融所需的时长为1小时，则消融时间为上午10点。
对于步骤S404，当所述消融时间到达时，可以理解的是，如果该识别结果为路面上存在积雪或结冰，当消融时间到达时，可以认为该定位信息对应地面上的积雪或结冰已经消融，因此该识别结果对于服务器的预警功能来说已经失效了，服务器即可以将与所述定位信息关联存储的识别结果的状态修改为失效，以表明该定位信息对应的路面上已不存在积雪或结冰。
由上述步骤S401~步骤S404的内容可知，本实施例可以最大限度做好对驾驶员行车中的前方出现的积雪或结冰危险进行预警提醒，以及准确判断报警位置处危险信号存在时间长短。这样，服务器便能全面而准确得为驾驶员做好车前积雪或结冰危险实时预警，最大限度地避免交通事故的发生。
本实施例中，首先，获取车辆上摄像头拍摄车前场景的第一视频；然后，从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；接着，将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；若所述识别结果为路面上存在积雪或结冰，则发出报警信息。在本实施例中，通过获取摄像头拍摄车前场景的第一视频，并从中截取表征路面状况的图片，将该图片输入至利用无监督深度学习预训练完成的无监督深度学习模型得到识别结果，从而可以得到该车辆当前车前路面上是否存在积雪或结冰，若存在，则发出报警信息，实现了检测路面积雪与结冰情况并及时发出报警信息，大大减轻了积雪与结冰的路面情况对行车带来的不良影响。
另外，本实施例采用无监督深度学习算法检测路面的积雪或结冰状况，可随样本数量的不断增大不断提高检测的准确率；并且，车辆之间可以通过服务器实现互联，车辆提供的报警信息在对本车实时预警之后，一定时间内可持续为其他经过报警对应地区的其他车辆提供预警，从而形成整体城市（或其他区域）性报警联动网络，提供更加全面的报警策略，同时车联网策略也更加方便全部车辆的运营管理。
应理解，上述实施例中各步骤的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本发明实施例的实施过程构成任何限定。
上面主要描述了一种车前积雪与结冰的检测报警方法，下面将对一种车前积雪与结冰的检测报警装置进行详细描述。
图5示出了本发明实施例中一种车前积雪与结冰的检测报警装置一个实施例结构图。
本实施例中，一种车前积雪与结冰的检测报警装置包括：
拍摄视频获取模块501，用于获取车辆上摄像头拍摄车前场景的第一视频；
图片截取模块502，用于从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
识别模块503，用于将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
报警模块504，用于若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
进一步地，所述车前积雪与结冰的检测报警装置还可以包括：
定位信息获取模块，用于获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
关联存储模块，用于将所述定位信息与所述识别结果关联存储；
提前预警模块，用于若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
进一步地，所述车前积雪与结冰的检测报警装置还可以包括：
天气情况获取模块，用于若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
消融时长确定模块，用于根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
消融时间确定模块，用于根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
效力修改模块，用于当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
进一步地，所述无监督深度学习模型可以通过以下步骤预先训练得到：
样本视频收集模块，用于预先采集多个拍摄自车前场景的样本视频；
样本图片截取模块，用于从各个所述样本视频中截取出表征路面状况的样本图片；
样本图片输入模块，用于将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中；
编解码处理模块，用于通过所述初始的无监督深度学习模型对所述输入向量进行编码、解码处理，得到输出向量；
输出误差计算模块，用于计算所述输出向量与所述输入向量之间的输出误差；
模型参数调整模块，用于若所述输出误差不符合预设条件，则调整所述无监督深度学习模型的模型参数，并将模型参数调整后的无监督深度学习模型作为初始的无监督深度学习模型，返回执行将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中的步骤以及后续步骤；
训练完成确定模块，用于直到所述误差满足预设条件时，确定所述无监督深度学习模型训练完成。
进一步地，所述车前积雪与结冰的检测报警装置还可以包括：
报警列表模块，用于若所述识别结果为路面上存在积雪或结冰，则将生成的报警信息存储至指定的报警列表；
报警信息查询模块，用于当接收到查询报警信息的请求时，从所述报警列表中查询所述请求所需的报警信息，然后将所述请求所需的报警信息反馈给请求方。
图6是本发明一实施例提供的服务器的示意图。如图6所示，该实施例的服务器6包括：处理器60、存储器61以及存储在所述存储器61中并可在所述处理器60上运行的计算机程序62，例如执行上述车前积雪与结冰的检测报警方法的程序。所述处理器60执行所述计算机程序62时实现上述各个车前积雪与结冰的检测报警方法实施例中的步骤，例如图1所示的步骤S101至步骤S104。或者，所述处理器60执行所述计算机程序62时实现上述各装置实施例中各模块/单元的功能，例如图5所示模块501至504的功能。
示例性的，所述计算机程序62可以被分割成一个或多个模块/单元，所述一个或者多个模块/单元被存储在所述存储器61中，并由所述处理器60执行，以完成本发明。所述一个或多个模块/单元可以是能够完成特定功能的一系列计算机程序的指令段，该指令段用于描述所述计算机程序62在所述服务器6中的执行过程。
所述服务器6可以是本地服务器、云端服务器等计算设备。所述服务器可包括，但不仅限于，处理器60、存储器61。本领域技术人员可以理解，图6仅仅是服务器6的示例，并不构成对服务器6的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件，例如所述服务器还可以包括输入输出设备、网络接入设备、总线等。
所述处理器60可以是中央处理单元(Central Processing Unit，CPU)，还可以是其他通用处理器、数字信号处理器 (Digital Signal Processor，DSP)、专用集成电路 (Application Specific Integrated Circuit，ASIC)、现成可编程门阵列 (Field-Programmable Gate Array，FPGA) 或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
所述存储器61可以是所述服务器6的内部存储单元，例如服务器6的硬盘或内存。所述存储器61也可以是所述服务器6的外部存储设备，例如所述服务器6上配备的插接式硬盘，智能存储卡（Smart Media Card, SMC），安全数字（Secure Digital, SD）卡，闪存卡（Flash Card）等。进一步地，所述存储器61还可以既包括所述服务器6的内部存储单元也包括外部存储设备。所述存储器61用于存储所述计算机程序以及所述服务器所需的其他程序和数据。所述存储器61还可以用于暂时地存储已经输出或者将要输出的数据。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的***，装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述或记载的部分，可以参见其它实施例的相关描述。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各实施例的模块、单元和/或方法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本发明的范围。
在本申请所提供的几个实施例中，应该理解到，所揭露的***，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个***，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备（可以是个人计算机，服务器，或者网络设备等）执行本发明各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器（ROM，Read-Only Memory）、随机存取存储器（RAM，Random Access Memory）、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述，以上实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。

Claims (10)

  1. 一种车前积雪与结冰的检测报警方法，其特征在于，包括：
    获取车辆上摄像头拍摄车前场景的第一视频；
    从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
    将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
    若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
  2. 根据权利要求1所述的车前积雪与结冰的检测报警方法，其特征在于，还包括：
    获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
    将所述定位信息与所述识别结果关联存储；
    若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
  3. 根据权利要求2所述的车前积雪与结冰的检测报警方法，其特征在于，在将所述定位信息与所述识别结果关联存储之后，还包括：
    若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
    根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
    根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
    当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
  4. 根据权利要求1所述的车前积雪与结冰的检测报警方法，其特征在于，所述无监督深度学习模型通过以下步骤预先训练得到：
    预先采集多个拍摄自车前场景的样本视频；
    从各个所述样本视频中截取出表征路面状况的样本图片；
    将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中；
    通过所述初始的无监督深度学习模型对所述输入向量进行编码、解码处理，得到输出向量；
    计算所述输出向量与所述输入向量之间的输出误差；
    若所述输出误差不符合预设条件，则调整所述无监督深度学习模型的模型参数，并将模型参数调整后的无监督深度学习模型作为初始的无监督深度学习模型，返回执行将各个所述样本图片转换成输入向量输入至初始的无监督深度学习模型中的步骤以及后续步骤；
    直到所述输出误差满足预设条件时，确定所述无监督深度学习模型训练完成。
  5. 根据权利要求1至4中任一项所述的车前积雪与结冰的检测报警方法，其特征在于，还包括：
    若所述识别结果为路面上存在积雪或结冰，则将生成的报警信息存储至指定的报警列表；
    当接收到查询报警信息的请求时，从所述报警列表中查询所述请求所需的报警信息，然后将所述请求所需的报警信息反馈给请求方。
  6. 一种车前积雪与结冰的检测报警装置，其特征在于，包括：
    拍摄视频获取模块，用于获取车辆上摄像头拍摄车前场景的第一视频；
    图片截取模块，用于从所述第一视频的视频帧中截取出所述车前场景中表征路面状况的第一图片；
    识别模块，用于将所述第一图片输入至利用无监督深度学习预训练完成的无监督深度学习模型，得到所述无监督深度学习模型输出的识别结果，所述识别结果为路面上存在积雪或结冰，或者路面上不存在积雪或结冰；
    报警模块，用于若所述识别结果为路面上存在积雪或结冰，则发出报警信息。
  7. 根据权利要求6所述的车前积雪与结冰的检测报警装置，其特征在于，所述车前积雪与结冰的检测报警装置还包括：
    定位信息获取模块，用于获取所述摄像头拍摄车前场景的第一视频时所述车辆的定位信息；
    关联存储模块，用于将所述定位信息与所述识别结果关联存储；
    提前预警模块，用于若所述识别结果为路面上存在积雪或结冰，且检测到车辆将要行驶经过所述定位信息对应的地点，则提前预设时间或者预设距离对将要行驶经过的车辆发出预警信息。
  8. 根据权利要求7所述的车前积雪与结冰的检测报警装置，其特征在于，所述车前积雪与结冰的检测报警装置还包括：
    天气情况获取模块，用于若所述识别结果为路面上存在积雪或结冰，获取所述定位信息对应的地点所属地区的天气情况；
    消融时长确定模块，用于根据所述天气情况确定所述所属地区中积雪或结冰消融所需的时长；
    消融时间确定模块，用于根据确定出的所述时长确定所述定位信息对应的地点上积雪或结冰的消融时间；
    效力修改模块，用于当所述消融时间到达时，将与所述定位信息关联存储的识别结果的状态修改为失效。
  9. 一种服务器，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，其特征在于，所述处理器执行所述计算机程序时实现如权利要求1至5中任一项所述车前积雪与结冰的检测报警方法的步骤。
  10. 一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现如权利要求1至5中任一项所述车前积雪与结冰的检测报警方法的步骤。
PCT/CN2018/080477 2018-03-26 2018-03-26 一种车前积雪与结冰的检测报警方法、存储介质和服务器 WO2019183752A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201880000213.0A CN108701396B (zh) 2018-03-26 2018-03-26 一种车前积雪与结冰的检测报警方法、存储介质和服务器
PCT/CN2018/080477 WO2019183752A1 (zh) 2018-03-26 2018-03-26 一种车前积雪与结冰的检测报警方法、存储介质和服务器

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/080477 WO2019183752A1 (zh) 2018-03-26 2018-03-26 一种车前积雪与结冰的检测报警方法、存储介质和服务器

Publications (1)

Publication Number Publication Date
WO2019183752A1 true WO2019183752A1 (zh) 2019-10-03

Family

ID=63841486

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/080477 WO2019183752A1 (zh) 2018-03-26 2018-03-26 一种车前积雪与结冰的检测报警方法、存储介质和服务器

Country Status (2)

Country Link
CN (1) CN108701396B (zh)
WO (1) WO2019183752A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114022765A (zh) * 2021-11-03 2022-02-08 应急管理部国家自然灾害防治研究院 一种采用图像识别滑坡崩塌落石智能监测预警方法及***
CN114596315A (zh) * 2022-05-10 2022-06-07 中国民航大学 一种飞机地面检测积冰方法、装置、***及计算机设备
CN114913477A (zh) * 2022-05-06 2022-08-16 广州市城市规划勘测设计研究院 城市管线防挖掘预警方法、装置、设备及介质
CN116469013A (zh) * 2023-06-20 2023-07-21 云途信息科技(杭州)有限公司 道路积水预测方法、装置、计算机设备及存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108701396B (zh) * 2018-03-26 2020-06-16 深圳市锐明技术股份有限公司 一种车前积雪与结冰的检测报警方法、存储介质和服务器
CN110186471A (zh) * 2019-05-06 2019-08-30 平安科技（深圳）有限公司 基于历史视频的导航方法、装置、计算机设备及存储介质
CN110363070A (zh) * 2019-05-31 2019-10-22 合刃科技（武汉）有限公司 智能识别道路状态的方法、装置及计算机程序产品
CN110705370B (zh) * 2019-09-06 2023-08-18 中国平安财产保险股份有限公司 基于深度学习的路况识别方法、装置、设备及存储介质
CN114999117B (zh) * 2022-06-29 2023-09-01 洛阳腾飞建设工程集团有限公司 一种机场道面冰雪状况监测预警方法、***、终端及介质
CN115649455B (zh) * 2022-12-07 2023-03-28 中国空气动力研究与发展中心低速空气动力研究所 一种基于电热除冰信号判别结冰的方法及装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030026483A1 (en) * 2001-02-01 2003-02-06 Pietro Perona Unsupervised learning of object categories from cluttered images
CN104103182A (zh) * 2014-07-28 2014-10-15 张蕾 道路及通行条件信息采集装置
CN204154308U (zh) * 2014-07-28 2015-02-11 张蕾 图像与激光复合式遥感路面监测自动报警***
CN104652341A (zh) * 2015-03-24 2015-05-27 淄博高华新能源科技有限公司 一种路面冰情检测及自动融冰***
CN105184271A (zh) * 2015-09-18 2015-12-23 苏州派瑞雷尔智能科技有限公司 一种基于深度学习的车辆自动检测方法
EP3104284A1 (en) * 2015-06-12 2016-12-14 Volkswagen Aktiengesellschaft Automatic labeling and learning of driver yield intention
CN108701396A (zh) * 2018-03-26 2018-10-23 深圳市锐明技术股份有限公司 一种车前积雪与结冰的检测报警方法、存储介质和服务器

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS61138193A (ja) * 1984-12-10 1986-06-25 Matsushita Seiko Co Ltd 路面凍結予知装置
JPS63191991A (ja) * 1987-02-05 1988-08-09 Matsushita Electric Ind Co Ltd 路面凍結予知装置
JPH0786208B2 (ja) * 1991-12-13 1995-09-20 古河電気工業株式会社 融雪設備の温度制御方法
JP3742810B2 (ja) * 2003-08-29 2006-02-08 クラリオン株式会社 路面状況検出システム
JP2009013703A (ja) * 2007-07-06 2009-01-22 Msk Corp 融雪方法
CN101235621A (zh) * 2007-12-25 2008-08-06 大连理工大学 水泥混凝土桥面融雪化冰的电热法
CN101621190A (zh) * 2009-06-17 2010-01-06 西安工程大学 输电线路覆冰及融冰过程在线监测***
CN103208205B (zh) * 2013-03-20 2015-05-13 北京航空航天大学 一种基于车联网的车辆安全行驶预警方法
CN105788313A (zh) * 2016-05-18 2016-07-20 陕西科技大学 一种高速公路路面结冰的监测识别装置及方法
CN106504590B (zh) * 2016-11-21 2019-06-25 温州云航信息科技有限公司 一种交通调度指令确认***和方法
CN107554420A (zh) * 2017-09-11 2018-01-09 安徽实运信息科技有限责任公司 一种基于道路环境的安全车距报警***
CN107609602A (zh) * 2017-09-28 2018-01-19 吉林大学 一种基于卷积神经网络的驾驶场景分类方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030026483A1 (en) * 2001-02-01 2003-02-06 Pietro Perona Unsupervised learning of object categories from cluttered images
CN104103182A (zh) * 2014-07-28 2014-10-15 张蕾 道路及通行条件信息采集装置
CN204154308U (zh) * 2014-07-28 2015-02-11 张蕾 图像与激光复合式遥感路面监测自动报警***
CN104652341A (zh) * 2015-03-24 2015-05-27 淄博高华新能源科技有限公司 一种路面冰情检测及自动融冰***
EP3104284A1 (en) * 2015-06-12 2016-12-14 Volkswagen Aktiengesellschaft Automatic labeling and learning of driver yield intention
CN105184271A (zh) * 2015-09-18 2015-12-23 苏州派瑞雷尔智能科技有限公司 一种基于深度学习的车辆自动检测方法
CN108701396A (zh) * 2018-03-26 2018-10-23 深圳市锐明技术股份有限公司 一种车前积雪与结冰的检测报警方法、存储介质和服务器

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114022765A (zh) * 2021-11-03 2022-02-08 应急管理部国家自然灾害防治研究院 一种采用图像识别滑坡崩塌落石智能监测预警方法及***
CN114022765B (zh) * 2021-11-03 2022-07-08 应急管理部国家自然灾害防治研究院 一种采用图像识别滑坡崩塌落石智能监测预警方法及***
CN114913477A (zh) * 2022-05-06 2022-08-16 广州市城市规划勘测设计研究院 城市管线防挖掘预警方法、装置、设备及介质
CN114596315A (zh) * 2022-05-10 2022-06-07 中国民航大学 一种飞机地面检测积冰方法、装置、***及计算机设备
CN116469013A (zh) * 2023-06-20 2023-07-21 云途信息科技(杭州)有限公司 道路积水预测方法、装置、计算机设备及存储介质
CN116469013B (zh) * 2023-06-20 2023-09-08 云途信息科技(杭州)有限公司 道路积水预测方法、装置、计算机设备及存储介质

Also Published As

Publication number Publication date
CN108701396A (zh) 2018-10-23
CN108701396B (zh) 2020-06-16

Similar Documents

Publication Publication Date Title
WO2019183752A1 (zh) 一种车前积雪与结冰的检测报警方法、存储介质和服务器
WO2019183751A1 (zh) 一种车前积雪与结冰的检测报警方法、存储介质和服务器
WO2022083402A1 (zh) 障碍物检测方法、装置、计算机设备和存储介质
CN109460699B (zh) 一种基于深度学习的驾驶员安全带佩戴识别方法
US9154741B2 (en) Apparatus and method for processing data of heterogeneous sensors in integrated manner to classify objects on road and detect locations of objects
WO2018153211A1 (zh) 获取交通路况信息的方法及装置、计算机存储介质
CN108764042B (zh) 一种异常路况信息识别方法、装置及终端设备
WO2021135879A1 (zh) 监控车辆数据方法、装置、计算机设备及存储介质
CN109191830B (zh) 一种基于视频图像处理的道路拥堵检测方法
WO2020042984A1 (zh) 一种车辆行为检测方法及装置
WO2021155685A1 (zh) 一种更新地图的方法、装置和设备
WO2022227766A1 (zh) 交通异常检测的方法和装置
KR20220047732A (ko) 차량 감시 방법 및 장치, 전자 기기, 저장 매체 및 컴퓨터 프로그램, 클라우드 제어 플랫폼 및 차량 도로 협조 시스템
WO2022227490A1 (zh) 行为识别方法、装置、设备、存储介质、计算机程序及程序产品
WO2022134387A1 (zh) 车辆逆行检测方法、装置、设备、计算机可读存储介质、计算机程序产品
WO2012024976A1 (zh) 一种交通信息处理方法及装置
CN103236181A (zh) 一种基于机器视觉的交通信号灯状态监控***及方法
CN110838230B (zh) 一种移动视频监控方法、监控中心及***
WO2021195955A1 (zh) 检测车辆行驶场景的复杂度的方法和装置
JP2018106243A5 (zh)
WO2022198897A1 (zh) 一种路侧停车的管理方法及装置
CN111898491A (zh) 一种车辆逆向行驶的识别方法、装置及电子设备
TWI774034B (zh) 基於車聯網的行車示警方法、系統及設備
CN107170267A (zh) 一种信号灯的提示方法及提示装置
WO2018209470A1 (zh) 一种车牌辨识方法和***

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18911698

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18911698

Country of ref document: EP

Kind code of ref document: A1