[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

WO2022001091A1 - 一种危险驾驶行为识别方法、装置、电子设备及存储介质 - Google Patents

一种危险驾驶行为识别方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2022001091A1
WO2022001091A1 PCT/CN2021/073483 CN2021073483W WO2022001091A1 WO 2022001091 A1 WO2022001091 A1 WO 2022001091A1 CN 2021073483 W CN2021073483 W CN 2021073483W WO 2022001091 A1 WO2022001091 A1 WO 2022001091A1
Authority
WO
WIPO (PCT)
Prior art keywords
face detection
detection frame
driving behavior
dangerous driving
face
Prior art date
Application number
PCT/CN2021/073483
Other languages
English (en)
French (fr)
Inventor
王珂尧
冯浩城
岳海潇
Original Assignee
北京百度网讯科技有限公司
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 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Priority to KR1020217032212A priority Critical patent/KR20210128491A/ko
Priority to US17/599,901 priority patent/US20230116040A1/en
Priority to JP2021559129A priority patent/JP2022544635A/ja
Priority to EP21777920.6A priority patent/EP3961498A4/en
Publication of WO2022001091A1 publication Critical patent/WO2022001091A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
    • G06V20/597—Recognising the driver's state or behaviour, e.g. attention or drowsiness
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification

Definitions

  • the embodiments of the present application relate to the field of computer technology, for example, to the field of artificial intelligence, deep learning, and image recognition, and can be applied to the field of automatic driving. Specifically, it relates to a dangerous driving behavior identification method, device, electronic device and storage medium.
  • the traditional method usually adopts random inspection of surveillance video, and then makes judgment with the naked eye; in recent years, with the rise of Convolutional Neural Networks (CNN), some methods have introduced artificial intelligence-assisted identification. , but these methods usually only perform direct binary classification on the entire monitoring image or the driver's body area to make judgments.
  • the artificial naked eye method has disadvantages such as slow speed, large error, high time and labor cost; the direct classification method based on CNN, because the objects such as smoking, calling, drinking water are small in the image, it can be extracted.
  • the features of the vehicle are sparse, and there is a lot of interference information around, which leads to a low recognition accuracy and an unsatisfactory recognition effect in the real vehicle scene.
  • the present application provides a method, device, electronic device and storage medium for identifying dangerous driving behaviors, which can greatly improve the accuracy of identifying dangerous driving behaviors of drivers, and at the same time, can greatly reduce the calculation cost and obtain high-accuracy dangerous driving behaviors. Real-time recognition of driving behavior.
  • a method for identifying dangerous driving behavior including:
  • a dangerous driving behavior recognition device comprising: a face detection module and a behavior recognition module; wherein,
  • the face detection module is used to input the to-be-recognized image into a pre-trained face detection model, and perform face detection on the to-be-recognized image through the face detection model to obtain the face of the to-be-recognized image detection frame;
  • the behavior recognition module is used to input the face detection frame into a pre-trained dangerous driving behavior recognition model, and perform dangerous driving behavior recognition on the face detection frame through the dangerous driving behavior recognition model, and obtain the The recognition result of dangerous driving behavior corresponding to the face detection frame.
  • an electronic device comprising:
  • processors one or more processors
  • memory for storing one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors implement the method for recognizing dangerous driving behavior described in any embodiment of the present application.
  • a storage medium on which a computer program is stored, and when the program is executed by a processor, implements the method for recognizing dangerous driving behavior described in any embodiment of the present application.
  • the technical solution according to the present application solves the problem of directly recognizing the image to be recognized based on CNN in the prior art. Since the objects such as smoking, calling, drinking water are small in the image, the features that can be extracted are few, and there are a lot of interference around. information, resulting in low recognition accuracy and unsatisfactory recognition effect in real vehicle scenarios.
  • the technical solution provided in this application can greatly improve the accuracy of identifying drivers' dangerous driving behavior, and can also greatly reduce Calculate the cost to obtain high-accuracy real-time identification of dangerous driving behaviors.
  • FIG. 1 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application
  • FIG. 4 is a first structural schematic diagram of a dangerous driving behavior identification device provided by an embodiment of the present application.
  • FIG. 5 is a second schematic structural diagram of the device for identifying dangerous driving behavior provided by an embodiment of the present application.
  • FIG. 6 is a schematic structural diagram of a preprocessing module provided by an embodiment of the present application.
  • FIG. 7 is a block diagram of an electronic device used to implement the method for identifying a dangerous driving behavior according to an embodiment of the present application.
  • FIG. 1 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application.
  • the method may be executed by a dangerous driving behavior identification device or an electronic device, and the device or electronic device may be implemented by software and/or hardware. , the device or electronic device can be integrated into any smart device with network communication function.
  • the method for identifying dangerous driving behavior may include the following steps:
  • S101 Input the to-be-recognized image into a pre-trained face detection model, and perform face detection on the to-be-recognized image by using the face detection model to obtain a face detection frame of the to-be-recognized image.
  • the electronic device can input the image to be recognized into a pre-trained face detection model, and perform face detection on the image to be recognized by the face detection model to obtain a face detection frame of the image to be recognized.
  • the coordinates of the four vertices of the face detection frame can be obtained through the face detection model, and the face detection frame can be obtained based on the coordinates of the four vertices.
  • the electronic device may first use the first layer of the convolutional neural network of the face detection model as the current layer of the convolutional neural network; use the image to be recognized as the detection object of the current layer of the convolutional neural network;
  • the convolutional neural network performs image downsampling on the detection object of the current layer of the convolutional neural network to obtain the face feature extraction result corresponding to the current layer of the convolutional neural network;
  • the face feature extraction result corresponding to the Nth layer convolutional neural network is extracted from the detection object of the convolutional neural network; among them, N is a natural number greater than 1; finally, based on the first layer of the convolutional neural network to the Nth layer of the convolutional neural network
  • the face feature extraction result corresponding to each layer of convolutional neural network in the above is obtained, and the face detection frame of the image to be recognized is obtained.
  • the electronic device can perform image downsampling through the six-layer convolutional neural network of the face detection model, and obtain the face feature extraction result corresponding to the six-layer convolutional neural network; based on the last three layers of the convolutional neural network, preset fixed The number of face anchor boxes of different sizes is used for face detection frame regression, and finally the face detection result is obtained, that is, the coordinates of the four vertices of the face detection frame.
  • the electronic device can input the face detection frame into a pre-trained dangerous driving behavior recognition model, and use the dangerous driving behavior recognition model to perform dangerous driving behavior recognition on the face detection frame to obtain the face detection frame.
  • the corresponding dangerous driving behavior recognition results may first input the face detection frame into the convolution layer in the dangerous driving behavior recognition model, and perform a convolution operation on the face detection frame through the convolution layer to obtain the face corresponding to the convolution layer.
  • Feature extraction results then input the face feature extraction results corresponding to the convolutional layer to the pooling layer in the dangerous driving behavior recognition model, and perform the pooling operation on the face detection frame corresponding to the convolutional layer through the pooling layer.
  • the face feature extraction results corresponding to the pooling layer are then input to the fully connected layer in the dangerous driving behavior model, and the face feature extraction features corresponding to the pooling layer are extracted through the fully connected layer.
  • the classification operation is performed to obtain the identification result of dangerous driving behavior corresponding to the face detection frame.
  • the electronic device can perform feature extraction on the face detection frame through a dangerous driving behavior recognition model composed of eight convolution layers and five pooling layers, and finally output the dangerous driving behavior identification results through a fully connected layer.
  • driving behaviors can be defined into five categories, namely: non-dangerous behaviors, making phone calls, smoking, eating, drinking water, and numbers 0-4 are used to represent labels of various driving behaviors.
  • the image to be recognized is first input into a pre-trained face detection model, and the face detection model is used to perform face detection on the image to be recognized, so as to obtain a face detection frame of the image to be recognized; Then, the face detection frame is input into the pre-trained dangerous driving behavior recognition model, and the dangerous driving behavior recognition is performed on the face detection frame through the dangerous driving behavior recognition model, and the dangerous driving behavior recognition result corresponding to the face detection frame is obtained. That is to say, the present application can first extract a face detection frame from the image to be recognized, and then perform dangerous driving behavior recognition based on the face detection frame. In the existing dangerous driving behavior recognition methods, the image to be recognized is directly recognized based on CNN.
  • the present application adopts the technical means of first extracting the face detection frame from the image to be recognized, and then identifying the dangerous driving behavior based on the face detection frame, it overcomes the direct recognition of the image to be recognized based on CNN in the prior art, Because the targets such as smoking, calling, drinking water are small in the image, the features that can be extracted are scarce, and there is a lot of interference information around, resulting in low recognition accuracy and unsatisfactory recognition in real vehicle scenes.
  • the technical solution provided by this application can greatly improve the accuracy of identifying the driver's dangerous driving behavior, and at the same time can also greatly reduce the calculation cost, and obtain a high-accuracy real-time identification ability of dangerous driving behavior;
  • the technical solution of the example is simple and convenient to implement, easy to popularize, and has a wider application range.
  • FIG. 2 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application. This embodiment is an optional solution proposed on the basis of the foregoing embodiment. As shown in Figure 2, the method for identifying dangerous driving behavior may include the following steps:
  • S201 Input the to-be-recognized image into a pre-trained face detection model, and perform face detection on the to-be-recognized image through the face detection model to obtain a face detection frame of the to-be-recognized image.
  • S202 Perform image preprocessing on the face detection frame to obtain a face detection frame after image preprocessing.
  • the electronic device may perform image preprocessing on the face detection frame to obtain the image preprocessed face detection frame; input the image preprocessed face detection frame into the dangerous driving behavior recognition model .
  • the electronic device may first perform an enlarging process on the face detection frame to obtain an enlarged face detection frame; and then perform clipping processing on the enlarged face detection frame to obtain a clipped face detection frame; then normalize the cropped face detection frame to obtain the normalized face detection frame; use the normalized face detection frame as the face detection frame after image preprocessing frame.
  • S203 Input the preprocessed face detection frame into the dangerous driving behavior recognition model, and perform the dangerous driving behavior recognition on the preprocessed face detection frame through the dangerous driving behavior recognition model, so as to obtain the dangerous driving corresponding to the face detection frame Behavioral recognition results.
  • the electronic device may input the face detection frame after image preprocessing into the dangerous driving behavior recognition model, and perform the dangerous driving behavior recognition on the preprocessed face detection frame through the dangerous driving behavior recognition model. , to obtain the recognition result of dangerous driving behavior corresponding to the face detection frame.
  • the electronic device can first input the preprocessed face detection frame to the convolution layer in the dangerous driving behavior recognition model, and perform a convolution operation on the preprocessed face detection frame through the convolution layer, Obtain the face feature extraction result corresponding to the convolution layer; then input the face feature extraction result corresponding to the convolution layer to the pooling layer in the dangerous driving behavior recognition model, and detect the face corresponding to the convolution layer through the pooling layer.
  • the frame is pooled to obtain the face feature extraction results corresponding to the pooling layer; then the face feature extraction results corresponding to the pooling layer are input into the fully connected layer in the dangerous driving behavior model, and the pooling layer is processed by the fully connected layer.
  • the corresponding face feature extraction features are used for classification operation, and the identification result of dangerous driving behavior corresponding to the face detection frame is obtained.
  • the image to be recognized is first input into a pre-trained face detection model, and the face detection model is used to perform face detection on the image to be recognized, and a face detection frame of the image to be recognized is obtained; Then, the face detection frame is input into the pre-trained dangerous driving behavior recognition model, and the dangerous driving behavior recognition is performed on the face detection frame through the dangerous driving behavior recognition model, and the dangerous driving behavior recognition result corresponding to the face detection frame is obtained. That is to say, the present application can first extract a face detection frame from the image to be recognized, and then perform dangerous driving behavior recognition based on the face detection frame. In the existing dangerous driving behavior recognition methods, the image to be recognized is directly recognized based on CNN.
  • this application adopts the technical means of first extracting the face detection frame from the image to be recognized, and then identifying the dangerous driving behavior based on the face detection frame, it overcomes the direct recognition of the image to be recognized based on CNN in the prior art.
  • Targets such as smoking, calling, drinking water are small in the image, and the features that can be extracted are scarce.
  • the technical solution provided by the present application can greatly improve the accuracy of identifying the driver's dangerous driving behavior, and at the same time, can also greatly reduce the calculation cost, and obtain a high-accuracy real-time identification capability of dangerous driving behavior; and, the embodiment of the present application
  • the technical solution is simple and convenient to implement, easy to popularize, and has a wider application range.
  • FIG. 3 is a schematic flowchart of a method for identifying a dangerous driving behavior provided by an embodiment of the present application. This embodiment is an optional solution proposed on the basis of the foregoing embodiment. As shown in Figure 3, the method for identifying dangerous driving behavior may include the following steps:
  • S301 Input the to-be-recognized image into a pre-trained face detection model, and perform face detection on the to-be-recognized image through the face detection model to obtain a face detection frame of the to-be-recognized image.
  • the electronic device may perform an enlarging process on the face detection frame to obtain an enlarged face detection frame. In this step, the electronic device can enlarge the face detection frame by two times.
  • image scaling refers to the process of adjusting the size of digital images. Image scaling requires a trade-off between processing efficiency and the smoothness and sharpness of the result. As the size of an image increases, the pixels that make up the image become more visible, making the image appear "soft". Conversely, scaling down an image will enhance its smoothness and sharpness.
  • enlarging an image or referred to as upsampling or image interpolation, mainly aims to enlarge the original image so that it can be displayed on a display device with a higher resolution.
  • the electronic device may perform cropping processing on the enlarged face detection frame to obtain the cropped face detection frame.
  • the electronic device may transform the cropped face detection frame into an image of a predetermined size, for example, transform the cropped face detection frame into a 140 ⁇ 140 image.
  • the electronic device can normalize the cropped face detection frame to obtain the normalized face detection frame; the normalized face detection frame is used as The face detection frame after image preprocessing.
  • the pixel value of each pixel in the normalized face detection frame is within a predetermined range, for example, the pixel value of each pixel is between [-0.5, 0.5].
  • Image normalization refers to the process of performing a series of standard processing and transformation on the image to transform it into a fixed standard form, and the standard image is called a normalized image.
  • Image normalization is to convert the original image to be processed into a corresponding unique standard form through a series of transformations (that is, using the invariant moment of the image to find a set of parameters that can eliminate the influence of other transformation functions on image transformation).
  • Images in standard form are invariant to affine transformations such as translation, rotation, scaling, etc.
  • the face detection model may also be trained first. Specifically, the electronic device may first use the pre-acquired first face image sample as the current face image sample; if the face detection model does not meet the convergence conditions corresponding to the preset face detection model, the current face image sample Input to the face detection model, and use the current face image sample to train the face detection model; take the next face image sample of the current face image sample as the current face image sample, and repeat the above operations until the person The face detection model satisfies the convergence conditions corresponding to the face detection model.
  • the electronic device before inputting the face detection frame into the pre-trained dangerous driving behavior recognition model, can also first train the dangerous driving behavior recognition model. Specifically, the electronic device can first use the pre-acquired first face detection frame sample as the current face detection frame sample; if the dangerous driving behavior recognition model does not meet the preset convergence conditions corresponding to the dangerous driving behavior recognition model, the current The face detection frame sample is input to the dangerous driving behavior recognition model, and the current face detection frame sample is used to train the dangerous driving behavior recognition model; the next face detection frame sample of the current face detection frame sample is used as the current face detection frame sample, and repeat the above operations until the driving behavior recognition model satisfies the convergence conditions corresponding to the dangerous driving behavior recognition model.
  • the image to be recognized is first input into a pre-trained face detection model, and the face detection model is used to perform face detection on the image to be recognized, so as to obtain a face detection frame of the image to be recognized; Then, the face detection frame is input into the pre-trained dangerous driving behavior recognition model, and the dangerous driving behavior recognition is performed on the face detection frame through the dangerous driving behavior recognition model, and the dangerous driving behavior recognition result corresponding to the face detection frame is obtained. That is to say, the present application can first extract a face detection frame from the image to be recognized, and then perform dangerous driving behavior recognition based on the face detection frame. In the existing dangerous driving behavior recognition methods, the image to be recognized is directly recognized based on CNN.
  • this application adopts the technical means of first extracting the face detection frame from the image to be recognized, and then identifying the dangerous driving behavior based on the face detection frame, it overcomes the direct recognition of the image to be recognized based on CNN in the prior art.
  • Targets such as smoking, calling, drinking water are small in the image, and the features that can be extracted are scarce.
  • the technical solution provided by the present application can greatly improve the accuracy of identifying the driver's dangerous driving behavior, and at the same time, can also greatly reduce the calculation cost, and obtain a high-accuracy real-time identification capability of dangerous driving behavior; and, the embodiment of the present application
  • the technical solution is simple and convenient to implement, easy to popularize, and has a wider application range.
  • FIG. 4 is a first structural schematic diagram of a dangerous driving behavior identification device provided by an embodiment of the present application.
  • the device 400 includes: a face detection module 401 and a behavior recognition module 402; wherein,
  • the face detection module 401 is used to input the image to be recognized into a pre-trained face detection model, and perform face detection on the image to be recognized through the face detection model to obtain the person of the image to be recognized. face detection frame;
  • the behavior recognition module 402 is used to input the face detection frame into a pre-trained dangerous driving behavior recognition model, and perform dangerous driving behavior recognition on the face detection frame through the dangerous driving behavior recognition model, and obtain the result.
  • FIG. 5 is a second schematic structural diagram of the device for recognizing dangerous driving behavior provided by an embodiment of the present application.
  • the apparatus 400 further includes: a preprocessing module 403, configured to perform image preprocessing on the face detection frame to obtain a preprocessed face detection frame;
  • the face detection frame is input to the dangerous driving behavior recognition model.
  • FIG. 6 is a schematic structural diagram of a preprocessing module provided in Embodiment 4 of the present application.
  • the preprocessing module 403 includes: amplifying sub-module 4031, cropping sub-module 4032 and normalizing sub-module 4033; wherein,
  • the amplifying sub-module 4031 is used to amplify the face detection frame to obtain an enlarged face detection frame;
  • the cropping sub-module 4032 is configured to perform cropping processing on the enlarged face detection frame to obtain a cropped face detection frame;
  • the normalization sub-module 4033 is used to normalize the cropped face detection frame to obtain the normalized face detection frame;
  • the detection frame is used as the face detection frame after the image is preprocessed.
  • the face detection module 401 is specifically configured to use the first-layer convolutional neural network of the face detection model as the current-layer convolutional neural network; use the to-be-recognized image as the current-layer convolutional neural network.
  • N is a natural number greater than 1; based on the face feature extraction result corresponding to each layer of convolutional neural network in the first layer of convolutional neural network to the Nth layer of convolutional neural network, the to-be-recognized image is obtained. face detection frame.
  • the behavior recognition module 402 is specifically configured to input the face detection frame into the convolution layer in the dangerous driving behavior recognition model, and roll the face detection frame through the convolution layer. product operation to obtain the face feature extraction result corresponding to the convolutional layer; input the face feature extraction result corresponding to the convolutional layer to the pooling layer in the dangerous driving behavior recognition model, through the pooling layer layer performs pooling operation on the face detection frame corresponding to the convolutional layer to obtain the face feature extraction result corresponding to the pooling layer; input the face feature extraction result corresponding to the pooling layer to the danger
  • the fully-connected layer in the driving behavior model performs classification operations on the facial feature extraction features corresponding to the pooling layer through the fully-connected layer to obtain the identification result of dangerous driving behavior corresponding to the face detection frame.
  • the device further includes: a face detection training module 404 (not shown in the figure), used for taking the pre-acquired first face image sample as the current face image sample; if the face detection model does not Meet the preset convergence conditions corresponding to the face detection model, input the current face image sample into the face detection model, and use the current face image sample to train the face detection model; Taking the next face image sample of the current face image sample as the current face image sample, and repeating the above operations until the face detection model satisfies the convergence condition corresponding to the face detection model.
  • a face detection training module 404 (not shown in the figure), used for taking the pre-acquired first face image sample as the current face image sample; if the face detection model does not Meet the preset convergence conditions corresponding to the face detection model, input the current face image sample into the face detection model, and use the current face image sample to train the face detection model; Taking the next face image sample of the current face image sample as the current face image sample, and repeating the above operations until the face detection model
  • the device further includes: a behavior recognition training module 405 (not shown in the figure), used for taking the pre-acquired first face detection frame sample as the current face detection frame sample; if the dangerous driving behavior The recognition model does not meet the preset convergence conditions corresponding to the dangerous driving behavior recognition model, the current face detection frame sample is input into the dangerous driving behavior recognition model, and the current face detection frame sample is used for the The dangerous driving behavior recognition model is trained; the next face detection frame sample of the current face detection frame sample is used as the current face detection frame sample, and the above operations are repeated until the driving behavior recognition model satisfies the Convergence conditions corresponding to the dangerous driving behavior recognition model.
  • a behavior recognition training module 405 (not shown in the figure), used for taking the pre-acquired first face detection frame sample as the current face detection frame sample; if the dangerous driving behavior The recognition model does not meet the preset convergence conditions corresponding to the dangerous driving behavior recognition model, the current face detection frame sample is input into the dangerous driving behavior recognition model, and the current face detection frame sample is used
  • the dangerous driving behavior identification device of the above verification processor can execute the method provided by any embodiment of the present application, and has functional modules and beneficial effects corresponding to the execution method.
  • the dangerous driving behavior identification device of the above verification processor can execute the method provided by any embodiment of the present application, and has functional modules and beneficial effects corresponding to the execution method.
  • the present application further provides an electronic device and a readable storage medium.
  • FIG. 7 it is a block diagram of an electronic device of the method for identifying dangerous driving behavior according to an embodiment of the present application.
  • Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.
  • Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices.
  • the components shown herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the application described and/or claimed herein.
  • the electronic device includes: one or more processors 701 , a memory 702 , and interfaces for connecting various components, including a high-speed interface and a low-speed interface.
  • the various components are interconnected using different buses and may be mounted on a common motherboard or otherwise as desired.
  • the processor may process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of the GUI on an external input/output device, such as a display device coupled to the interface.
  • multiple processors and/or multiple buses may be used with multiple memories and multiple memories, if desired.
  • multiple electronic devices may be connected, each providing some of the necessary operations (eg, as a server array, a group of blade servers, or a multiprocessor system).
  • a processor 701 is taken as an example in FIG. 7 .
  • the memory 702 is the non-transitory computer-readable storage medium provided by the present application.
  • the memory stores instructions executable by at least one processor, so that the at least one processor executes the dangerous driving behavior identification method provided by the present application.
  • the non-transitory computer-readable storage medium of the present application stores computer instructions, and the computer instructions are used to cause the computer to execute the dangerous driving behavior identification method provided by the present application.
  • the memory 702 can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as program instructions/modules corresponding to the dangerous driving behavior identification method in the embodiments of the present application (for example, , the face detection module 401 and the behavior recognition module 402 shown in FIG. 4 ).
  • the processor 701 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions and modules stored in the memory 702, that is, to implement the dangerous driving behavior identification method in the above method embodiments.
  • the memory 702 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the electronic device of the dangerous driving behavior identification method. data etc. Additionally, memory 502 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 702 may optionally include memory located remotely relative to the processor 701, and these remote memories may be connected to the electronic device of the method for identifying dangerous driving behaviors through a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
  • the electronic device of the dangerous driving behavior identification method may further include: an input device 703 and an output device 704 .
  • the processor 701 , the memory 702 , the input device 703 and the output device 704 may be connected by a bus or in other ways, and the connection by a bus is taken as an example in FIG. 7 .
  • the input device 703 can receive input numerical or character information, and generate key signal input related to user settings and function control of the electronic device of the dangerous driving behavior recognition method, such as a touch screen, a keypad, a mouse, a trackpad, a touchpad, an indication A stick, one or more mouse buttons, a trackball, a joystick, and other input devices.
  • Output devices 704 may include display devices, auxiliary lighting devices (eg, LEDs), haptic feedback devices (eg, vibration motors), and the like.
  • the display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some implementations, the display device may be a touch screen.
  • Various implementations of the systems and techniques described herein can be implemented in digital electronic circuitry, integrated circuit systems, application specific ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs executable and/or interpretable on a programmable system including at least one programmable processor that The processor, which may be a special purpose or general-purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device an output device.
  • the processor which may be a special purpose or general-purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device an output device.
  • machine-readable medium and “computer-readable medium” refer to any computer program product, apparatus, and/or apparatus for providing machine instructions and/or data to a programmable processor ( For example, magnetic disks, optical disks, memories, programmable logic devices (PLDs), including machine-readable media that receive machine instructions as machine-readable signals.
  • machine-readable signal refers to any signal used to provide machine instructions and/or data to a programmable processor.
  • the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or trackball) through which a user can provide input to the computer.
  • a display device eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor
  • a keyboard and pointing device eg, a mouse or trackball
  • Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); and can be in any form (including acoustic input, voice input, or tactile input) to receive input from the user.
  • the systems and techniques described herein may be implemented on a computing system that includes back-end components (eg, as a data server), or a computing system that includes middleware components (eg, an application server), or a computing system that includes front-end components (eg, a user's computer having a graphical user interface or web browser through which a user may interact with implementations of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system.
  • the components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include: Local Area Networks (LANs), Wide Area Networks (WANs), the Internet, and blockchain networks.
  • a computer system can include clients and servers.
  • Clients and servers are generally remote from each other and usually interact through a communication network.
  • the relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other.
  • the image to be recognized is first input into a pre-trained face detection model, and the face detection model is used to perform face detection on the image to be recognized to obtain a face detection frame of the image to be recognized;
  • the face detection frame is input into the pre-trained dangerous driving behavior recognition model, and the dangerous driving behavior recognition model is used to identify the dangerous driving behavior of the face detection frame, and the dangerous driving behavior recognition result corresponding to the face detection frame is obtained.
  • the present application can first extract a face detection frame from the image to be recognized, and then perform dangerous driving behavior recognition based on the face detection frame.
  • the image to be recognized is directly recognized based on CNN.
  • this application adopts the technical means of first extracting the face detection frame from the image to be recognized, and then identifying the dangerous driving behavior based on the face detection frame, it overcomes the direct recognition of the image to be recognized based on CNN in the prior art.
  • Targets such as smoking, calling, drinking water are small in the image, and the features that can be extracted are scarce.
  • the technical solution provided by the present application can greatly improve the accuracy of identifying the driver's dangerous driving behavior, and at the same time, can also greatly reduce the calculation cost, and obtain a high-accuracy real-time identification capability of dangerous driving behavior; and, the embodiment of the present application
  • the technical solution is simple and convenient to implement, easy to popularize, and has a wider application range.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Human Computer Interaction (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Analysis (AREA)

Abstract

本申请公开了一种危险驾驶行为识别方法、装置、电子设备及存储介质，涉及人工智能、深度学习以及图像识别领域，可应用于自动驾驶领域。具体方案为：将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框；将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。本申请实施例可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力。

Description

一种危险驾驶行为识别方法、装置、电子设备及存储介质
本申请要求在2020年06月29日提交中国专利局、申请号为202010611370.4的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及计算机技术领域，例如涉及人工智能、深度学习以及图像识别领域，可应用于自动驾驶领域。具体地，涉及一种危险驾驶行为识别方法、装置、电子设备及存储介质。
背景技术
随着互联网和人工智能技术的不断发展，越来越多的领域开始涉及自动化计算与分析，其中监控安防领域是最为重要的场景之一。
对于公共运营车辆，如出租车、公交车、长途大巴车等，由于涉及众多乘客安全，驾驶员的驾驶安全显得尤为重要。因此，许多公共运营车辆已安装车载监控摄像头，方便对应公司或监管部门对驾驶员驾驶行为进行监控。对于驾驶员经常出现的一些危险驾驶行为，如抽烟、打电话、未系安全带等，需要及时发现并进行警告，最大限度保证车辆行驶安全。
针对驾驶员安全带判断，传统方法通常采用对监控视频进行抽查，之后人工肉眼进行判断；近年来，随着卷积神经网络(Convolutional Neural Networks，CNN)的兴起，一些方法已引入人工智能辅助识别，不过这些方法通常只是对整张监控图片或者司机身体区域进行直接二分类来进行判断。在已有方案中，人工肉眼方式存在速度慢，误差大，时间、人力成本高昂等缺点；基于CNN的直接分类法，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准确率较低，识别效果不理想。
发明内容
以下是对本文详细描述的主题的概述。本概述并非是为了限制权利要求的保护范围。
本申请提供了一种危险驾驶行为识别方法、装置、电子设备及存储介质，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力。
根据本公开的一方面，提供了一种危险驾驶行为识别方法，包括：
将待识别图像输入至预先训练的人脸检测模型，通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框；
将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果。
根据本公开的一方面，提供了一种危险驾驶行为识别装置，包括：人脸检测模块和行为识别模块；其中，
所述人脸检测模块，用于将待识别图像输入至预先训练的人脸检测模型，通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框；
所述行为识别模块，用于将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果。
根据本公开的一方面，提供了一种电子设备，包括：
一个或多个处理器；
存储器，用于存储一个或多个程序，
当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现本申请任意实施例所述的危险驾驶行为识别方法。
根据本公开的一方面，提供了一种存储介质，其上存储有计算机程序，该程序被处理器执行时实现本申请任意实施例所述的危险驾驶行为识别方法。
根据本申请的技术方案解决了现有技术中基于CNN直接对待识别图像进行识别，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准 确率较低，识别效果不理想的技术问题，本申请提供的技术方案，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力。
应当理解，本部分所描述的内容并非旨在标识本公开的实施例的关键或重要特征，也不用于限制本公开的范围。本公开的其它特征将通过以下的说明书而变得容易理解。在阅读并理解了附图和详细描述后，可以明白其他方面。
附图说明
附图用于更好地理解本方案，不构成对本申请的限定。其中：
图1是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图；
图2是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图；
图3是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图；
图4是本申请实施例提供的危险驾驶行为识别装置的第一结构示意图；
图5是本申请实施例提供的危险驾驶行为识别装置的第二结构示意图；
图6是本申请实施例提供的预处理模块的结构示意图；
图7是用来实现本申请实施例的危险驾驶行为识别方法的电子设备的框图。
具体实施方式
以下结合附图对本申请的示范性实施例做出说明，其中包括本申请实施例的各种细节以助于理解，应当将它们认为仅仅是示范性的。因此，本领域普通技术人员应当认识到，可以对这里描述的实施例做出各种改变和修改，而不会背离本申请的范围和精神。同样，为了清楚和简明，以下的描述中省略了对公知功能和结构的描述。
实施例一
图1是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图，该方法可以由危险驾驶行为识别装置或者电子设备来执行，该装置或者电子设备可以由软件和/或硬件的方式实现，该装置或者电子设备可以集成在 任何具有网络通信功能的智能设备中。如图1所示，危险驾驶行为识别方法可以包括以下步骤：
S101、将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框。
在本申请的具体实施例中，电子设备可以将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框。具体地，通过人脸检测模型可以得到人脸检测框的四个顶点坐标，基于这四个顶点的坐标可以得到人脸检测框。在一个实施例中，电子设备可以先将人脸检测模型的第一层卷积神经网络作为当前层卷积神经网络；将待识别图像作为当前层卷积神经网络的检测对象；然后通过当前层卷积神经网络对当前层卷积神经网络的检测对象进行图像下采样，得到当前层卷积神经网络对应的人脸特征提取结果；再将当前层卷积神经网络对应的人脸特征提取结果作为当前层卷积神经网络的下一层卷积神经网络的检测对象；将下一层卷积神经网络作为当前层卷积神经网络，重复执行上述操作，直到在人脸检测模型的第N层卷积神经网络的检测对象中提取出第N层卷积神经网络对应的人脸特征提取结果；其中，N为大于1的自然数；最后基于第一层卷积神经网络至第N层卷积神经网络中的每一层卷积神经网络对应的人脸特征提取结果，得到待识别图像的人脸检测框。具体地，电子设备可以通过人脸检测模型的六层卷积神经网络进行图像下采样，得到六层卷积神经网络对应的人脸特征提取结果；基于最后三层卷积神经网络分别预设置固定数目的不同尺寸人脸锚点框进行人脸检测框回归，最终获得人脸检测结果，即人脸检测框的四个顶点的坐标。
S102、将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。
在本申请的具体实施例中，电子设备可以将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。在一个实施例中，电子设备可以先将人脸检测框输入至危险驾驶行为识别模 型中的卷积层，通过卷积层对人脸检测框进行卷积操作，得到卷积层对应的人脸特征提取结果；然后将卷积层对应的人脸特征提取结果输入至危险驾驶行为识别模型中的池化层，通过池化层对卷积层对应的人脸检测框进行池化操作，得到池化层对应的人脸特征提取结果；再将池化层对应的人脸特征提取结果输入至危险驾驶行为模型中的全连接层，通过全连接层对池化层对应的人脸特征提取特征进行分类操作，得到人脸检测框对应的危险驾驶行为识别结果。具体地，电子设备可以通过由八层卷积层和五层池化层组成的危险驾驶行为识别模型对人脸检测框进行特征提取，最后通过全连接层输出危险驾驶行为识别结果。
在本申请的具体实施例中，驾驶行为可以定义为五类，分别为：无危险行为、打电话、抽烟、吃东西、喝水，分别用数字0-4表示各类驾驶行为的标签。
本申请实施例提出的危险驾驶行为识别方法，先将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框；然后将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。也就是说，本申请可以先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别。而在现有的危险驾驶行为识别方法中，基于CNN直接对待识别图像进行识别。因为本申请采用了先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别的技术手段，克服了现有技术中基于CNN的直接对待识别图像进行识别，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准确率较低，识别效果不理想的技术问题，本申请提供的技术方案，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力；并且，本申请实施例的技术方案实现简单方便、便于普及，适用范围更广。
实施例二
图2是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图。本实施例是在上述实施例的基础上提出的一种可选方案。如图2所示，危险驾驶行为识别方法可以包括以下步骤：
S201、将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框。
S202、对人脸检测框进行图像预处理，得到图像预处理后的人脸检测框。
在本申请的具体实施例中，电子设备可以对人脸检测框进行图像预处理，得到图像预处理后的人脸检测框；将图像预处理后的人脸检测框输入至危险驾驶行为识别模型。在一个实施例中，电子设备可以先将人脸检测框进行放大处理，得到放大处理后的人脸检测框；然后将放大处理后的人脸检测框进行裁剪处理，得到裁剪处理后的人脸检测框；再将裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将归一化后的人脸检测框作为图像预处理后的人脸检测框。
S203、将图像预处理后的人脸检测框输入至危险驾驶行为识别模型，通过危险驾驶行为识别模型对预处理后的人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。
在本申请的具体实施例中，电子设备可以将图像预处理后的人脸检测框输入至危险驾驶行为识别模型，通过危险驾驶行为识别模型对预处理后的人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。在一个实施例中，电子设备可以先将预处理后的人脸检测框输入至危险驾驶行为识别模型中的卷积层，通过卷积层对预处理后的人脸检测框进行卷积操作，得到卷积层对应的人脸特征提取结果；然后将卷积层对应的人脸特征提取结果输入至危险驾驶行为识别模型中的池化层，通过池化层对卷积层对应的人脸检测框进行池化操作，得到池化层对应的人脸特征提取结果；再将池化层对应的人脸特征提取结果输入至危险驾驶行为模型中的全连接层，通过全连接层对池化层对应的人脸特征提取特征进行分类操作，得到人脸检测框对应的危险驾驶行为识别结果。
本申请实施例提出的危险驾驶行为识别方法，先将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测， 得到待识别图像的人脸检测框；然后将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。也就是说，本申请可以先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别。而在现有的危险驾驶行为识别方法中，基于CNN的直接对待识别图像进行识别。因为本申请采用了先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别的技术手段，克服了现有技术中基于CNN直接对待识别图像进行识别，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准确率较低，识别效果不理想的技术问题，本申请提供的技术方案，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力；并且，本申请实施例的技术方案实现简单方便、便于普及，适用范围更广。
实施例三
图3是本申请实施例提供的一种危险驾驶行为识别方法的流程示意图。本实施例是在上述实施例的基础上提出的一种可选方案。如图3所示，危险驾驶行为识别方法可以包括以下步骤：
S301、将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框。
S302、将人脸检测框进行放大处理，得到放大处理后的人脸检测框。
在本申请的具体实施例中，电子设备可以将人脸检测框进行放大处理，得到放大处理后的人脸检测框。在本步骤中，电子设备可以将人脸检测框放大两倍。在计算机图像处理和计算机图形学中，图像缩放(image scaling)是指对数字图像的大小进行调整的过程。图像缩放需要在处理效率以及结果的平滑度(smoothness)和清晰度(sharpness)上做一个权衡。当一个图像的大小增加之后，组成图像的像素的可见度将会变得更高，从而使得图像表现得“软”。相反地，缩小一个图像将会增强它的平滑度和清晰度。具体地，放大图像，或称为上采样(upsampling)或图像插值(interpolating)， 主要目的是放大原图像，从而可以显示在更高分辨率的显示设备上。
S303、将放大处理后的人脸检测框进行裁剪处理，得到裁剪处理后的人脸检测框。
在本申请的具体实施例中，电子设备可以将放大处理后的人脸检测框进行裁剪处理，得到裁剪处理后的人脸检测框。在本步骤中，电子设备可以将裁剪后的人脸检测框变换为预定尺寸的图像，例如，将裁剪后的人脸检测框变换为140×140的图像。
S304、将裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将归一化后的人脸检测框作为图像预处理后的人脸检测框。
在本申请的具体实施例中，电子设备可以将裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将归一化后的人脸检测框作为图像预处理后的人脸检测框。在本步骤中，归一化处理后的人脸检测框中每个像素的像素值在预定范围内，例如，每个像素的像素值在[-0.5，0.5]之间。图像归一化是指对图像进行了一系列标准的处理变换，使之变换为一固定标准形式的过程，该标准图像称作归一化图像。图像归一化就是通过一系列变换(即利用图像的不变矩寻找一组参数使其能够消除其他变换函数对图像变换的影响)，将待处理的原始图像转换成相应的唯一标准形式，该标准形式的图像对平移、旋转、缩放等仿射变换具有不变特性。
S305、将图像预处理后的人脸检测框输入至危险驾驶行为识别模型，通过危险驾驶行为识别模型对预处理后的人脸检测框进行危险驾驶行为识别，得到预处理后的人脸检测框对应的危险驾驶行为识别结果。
较佳地，在本申请的具体实施例中，电子设备在将待识别图像输入至预先训练的人脸检测模型之前，还可以先对人脸检测模型进行训练。具体地，电子设备可以先将预先获取的第一个人脸图像样本作为当前人脸图像样本；若人脸检测模型不满足预先设置的人脸检测模型对应的收敛条件，将当前人脸图像样本输入至所述人脸检测模型，使用当前人脸图像样本对人脸检测模型进行训练；将当前人脸图像样本的下一个人脸图像样本作为当前人脸图像样本，重复执行上述操作，直到人脸检测模型满足人脸检测 模型对应的收敛条件。
较佳地，在本申请的具体实施例中，电子设备在将人脸检测框输入至预先训练的危险驾驶行为识别模型之前，还可以先对危险驾驶行为识别模型进行训练。具体地，电子设备可以先将预先获取的第一个人脸检测框样本作为当前人脸检测框样本；若危险驾驶行为识别模型不满足预先设置的危险驾驶行为识别模型对应的收敛条件，将当前人脸检测框样本输入至危险驾驶行为识别模型，使用当前人脸检测框样本对危险驾驶行为识别模型进行训练；将当前人脸检测框样本的下一个人脸检测框样本作为当前人脸检测框样本，重复执行上述操作，直到驾驶行为识别模型满足危险驾驶行为识别模型对应的收敛条件。
本申请实施例提出的危险驾驶行为识别方法，先将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别图像的人脸检测框；然后将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。也就是说，本申请可以先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别。而在现有的危险驾驶行为识别方法中，基于CNN的直接对待识别图像进行识别。因为本申请采用了先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别的技术手段，克服了现有技术中基于CNN直接对待识别图像进行识别，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准确率较低，识别效果不理想的技术问题，本申请提供的技术方案，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力；并且，本申请实施例的技术方案实现简单方便、便于普及，适用范围更广。
实施例四
图4是本申请实施例提供的危险驾驶行为识别装置的第一结构示意图。如图4所示，所述装置400包括：人脸检测模块401和行为识别模块402； 其中，
所述人脸检测模块401，用于将待识别图像输入至预先训练的人脸检测模型，通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框；
所述行为识别模块402，用于将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果。
图5是本申请实施例提供的危险驾驶行为识别装置的第二结构示意图。如图5所示，所述装置400还包括：预处理模块403，用于对所述人脸检测框进行图像预处理，得到图像预处理后的人脸检测框；将所述图像预处理后的人脸检测框输入至所述危险驾驶行为识别模型。
图6是本申请实施例四提供的预处理模块的结构示意图。如图6所示，所述预处理模块403包括：放大子模块4031、裁剪子模块4032和归一化子模块4033；其中，
所述放大子模块4031，用于将所述人脸检测框进行放大处理，得到放大处理后的人脸检测框；
所述裁剪子模块4032，用于将所述放大处理后的人脸检测框进行裁剪处理，得到裁剪处理后的人脸检测框；
所述归一化子模块4033，用于将所述裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将所述归一化后的人脸检测框作为所述图像预处理后的人脸检测框。
进一步的，所述人脸检测模块401，具体用于将所述人脸检测模型的第一层卷积神经网络作为当前层卷积神经网络；将所述待识别图像作为所述当前层卷积神经网络的检测对象；通过所述当前层卷积神经网络对所述当前层卷积神经网络的检测对象进行图像下采样，得到所述当前层卷积神经网络对应的人脸特征提取结果；将所述当前层卷积神经网络对应的人脸特征提取结果作为所述当前层卷积神经网络的下一层卷积神经网络的检测对象；将所述下一层卷积神经网络作为所述当前层卷积神经网络，重复执行上述操作，直到在所述人脸检测模型的第N层卷积神经网络的检测对 象中提取出第N层卷积神经网络对应的人脸特征提取结果；其中，N为大于1的自然数；基于所述第一层卷积神经网络至所述第N层卷积神经网络中的每一层卷积神经网络对应的人脸特征提取结果，得到所述待识别图像的人脸检测框。
进一步的，所述行为识别模块402，具体用于将所述人脸检测框输入至所述危险驾驶行为识别模型中的卷积层，通过所述卷积层对所述人脸检测框进行卷积操作，得到所述卷积层对应的人脸特征提取结果；将所述卷积层对应的人脸特征提取结果输入至所述危险驾驶行为识别模型中的池化层，通过所述池化层对所述卷积层对应的人脸检测框进行池化操作，得到所述池化层对应的人脸特征提取结果；将所述池化层对应的人脸特征提取结果输入至所述危险驾驶行为模型中的全连接层，通过所述全连接层对所述池化层对应的人脸特征提取特征进行分类操作，得到所述人脸检测框对应的危险驾驶行为识别结果。
进一步，所述装置还包括：人脸检测训练模块404(图中未示出)，用于将预先获取的第一个人脸图像样本作为当前人脸图像样本；若所述人脸检测模型不满足预先设置的所述人脸检测模型对应的收敛条件，将所述当前人脸图像样本输入至所述人脸检测模型，使用所述当前人脸图像样本对所述人脸检测模型进行训练；将所述当前人脸图像样本的下一个人脸图像样本作为所述当前人脸图像样本，重复执行上述操作，直到所述人脸检测模型满足所述人脸检测模型对应的收敛条件。
进一步的，所述装置还包括：行为识别训练模块405(图中未示出)，用于将预先获取的第一个人脸检测框样本作为当前人脸检测框样本；若所述危险驾驶行为识别模型不满足预先设置的所述危险驾驶行为识别模型对应的收敛条件，将所述当前人脸检测框样本输入至所述危险驾驶行为识别模型，使用所述当前人脸检测框样本对所述危险驾驶行为识别模型进行训练；将所述当前人脸检测框样本的下一个人脸检测框样本作为所述当前人脸检测框样本，重复执行上述操作，直到所述驾驶行为识别模型满足所述危险驾驶行为识别模型对应的收敛条件。
上述验证处理器的危险驾驶行为识别装置可执行本申请任意实施例所提供的方法，具备执行方法相应的功能模块和有益效果。未在本实施例 中详尽描述的技术细节，可参见本申请任意实施例提供的验证处理器的危险驾驶行为识别方法。
实施例五
根据本申请的实施例，本申请还提供了一种电子设备和一种可读存储介质。
如图7所示，是根据本申请实施例的危险驾驶行为识别方法的电子设备的框图。电子设备旨在表示各种形式的数字计算机，诸如，膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算机。电子设备还可以表示各种形式的移动装置，诸如，个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例，并且不意在限制本文中描述的和/或者要求的本申请的实现。
如图7所示，该电子设备包括：一个或多个处理器701、存储器702，以及用于连接各部件的接口，包括高速接口和低速接口。各个部件利用不同的总线互相连接，并且可以被安装在公共主板上或者根据需要以其它方式安装。处理器可以对在电子设备内执行的指令进行处理，包括存储在存储器中或者存储器上以在外部输入/输出装置(诸如，耦合至接口的显示设备)上显示GUI的图形信息的指令。在其它实施方式中，若需要，可以将多个处理器和/或多条总线与多个存储器和多个存储器一起使用。同样，可以连接多个电子设备，各个设备提供部分必要的操作(例如，作为服务器阵列、一组刀片式服务器、或者多处理器系统)。图7中以一个处理器701为例。
存储器702即为本申请所提供的非瞬时计算机可读存储介质。其中，所述存储器存储有可由至少一个处理器执行的指令，以使所述至少一个处理器执行本申请所提供的危险驾驶行为识别方法。本申请的非瞬时计算机可读存储介质存储计算机指令，该计算机指令用于使计算机执行本申请所提供的危险驾驶行为识别方法。
存储器702作为一种非瞬时计算机可读存储介质，可用于存储非瞬时软件程序、非瞬时计算机可执行程序以及模块，如本申请实施例中的危险 驾驶行为识别方法对应的程序指令/模块(例如，附图4所示的人脸检测模块401和行为识别模块402)。处理器701通过运行存储在存储器702中的非瞬时软件程序、指令以及模块，从而执行服务器的各种功能应用以及数据处理，即实现上述方法实施例中的危险驾驶行为识别方法。
存储器702可以包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需要的应用程序；存储数据区可存储根据危险驾驶行为识别方法的电子设备的使用所创建的数据等。此外，存储器502可以包括高速随机存取存储器，还可以包括非瞬时存储器，例如至少一个磁盘存储器件、闪存器件、或其他非瞬时固态存储器件。在一些实施例中，存储器702可选包括相对于处理器701远程设置的存储器，这些远程存储器可以通过网络连接至危险驾驶行为识别方法的电子设备。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
危险驾驶行为识别方法的电子设备还可以包括：输入装置703和输出装置704。处理器701、存储器702、输入装置703和输出装置704可以通过总线或者其他方式连接，图7中以通过总线连接为例。
输入装置703可接收输入的数字或字符信息，以及产生与危险驾驶行为识别方法的电子设备的用户设置以及功能控制有关的键信号输入，例如触摸屏、小键盘、鼠标、轨迹板、触摸板、指示杆、一个或者多个鼠标按钮、轨迹球、操纵杆等输入装置。输出装置704可以包括显示设备、辅助照明装置(例如，LED)和触觉反馈装置(例如，振动电机)等。该显示设备可以包括但不限于，液晶显示器(LCD)、发光二极管(LED)显示器和等离子体显示器。在一些实施方式中，显示设备可以是触摸屏。
此处描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、专用ASIC(专用集成电路)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括：实施在一个或者多个计算机程序中，该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释，该可编程处理器可以是专用或者通用可编程处理器，可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令，并且将数据和指令传输至该存储系统、该至少一个输 入装置、和该至少一个输出装置。
这些计算程序(也称作程序、软件、软件应用、或者代码)包括可编程处理器的机器指令，并且可以利用高级过程和/或面向对象的编程语言、和/或汇编/机器语言来实施这些计算程序。如本文使用的，术语“机器可读介质”和“计算机可读介质”指的是用于将机器指令和/或数据提供给可编程处理器的任何计算机程序产品、设备、和/或装置(例如，磁盘、光盘、存储器、可编程逻辑装置(PLD))，包括，接收作为机器可读信号的机器指令的机器可读介质。术语“机器可读信号”指的是用于将机器指令和/或数据提供给可编程处理器的任何信号。
为了提供与用户的交互，可以在计算机上实施此处描述的系统和技术，该计算机具有：用于向用户显示信息的显示装置(例如，CRT(阴极射线管)或者LCD(液晶显示器)监视器)；以及键盘和指向装置(例如，鼠标或者轨迹球)，用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互；例如，提供给用户的反馈可以是任何形式的传感反馈(例如，视觉反馈、听觉反馈、或者触觉反馈)；并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据服务器)、或者包括中间件部件的计算系统(例如，应用服务器)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(LAN)、广域网(WAN)、互联网和区块链网络。
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。
根据本申请实施例的技术方案，先将待识别图像输入至预先训练的人脸检测模型，通过人脸检测模型对待识别图像进行人脸检测，得到待识别 图像的人脸检测框；然后将人脸检测框输入至预先训练的危险驾驶行为识别模型，通过危险驾驶行为识别模型对人脸检测框进行危险驾驶行为识别，得到人脸检测框对应的危险驾驶行为识别结果。也就是说，本申请可以先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别。而在现有的危险驾驶行为识别方法中，基于CNN的直接对待识别图像进行识别。因为本申请采用了先在待识别图像中提取出人脸检测框，然后基于该人脸检测框进行危险驾驶行为识别的技术手段，克服了现有技术中基于CNN直接对待识别图像进行识别，由于抽烟、打电话、喝水等目标在图像中较小，可提取出的特征稀少，同时周围又存在大量干扰信息，导致在真实车载场景中识别准确率较低，识别效果不理想的技术问题，本申请提供的技术方案，可以极大地提高识别驾驶员危险驾驶行为的准确度，同时还可以极大地减小计算成本，获得高准确度的危险驾驶行为的实时识别能力；并且，本申请实施例的技术方案实现简单方便、便于普及，适用范围更广。
应该理解，可以使用上面所示的各种形式的流程，重新排序、增加或删除步骤。例如，本申请中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行，只要能够实现本申请公开的技术方案所期望的结果，本文在此不进行限制。
上述具体实施方式，并不构成对本申请保护范围的限制。本领域技术人员应该明白的是，根据设计要求和其他因素，可以进行各种修改、组合、子组合和替代。

Claims (16)

  1. 一种危险驾驶行为识别方法，包括：
    将待识别图像输入至预先训练的人脸检测模型，通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框；
    将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果。
  2. 根据权利要求1所述的方法，在所述将所述人脸检测框输入至预先训练的危险驾驶行为识别模型之前，所述方法还包括：
    对所述人脸检测框进行图像预处理，得到图像预处理后的人脸检测框；将所述图像预处理后的人脸检测框输入至所述危险驾驶行为识别模型。
  3. 根据权利要求2所述的方法，其中，所述对所述人脸检测框进行图像预处理，得到图像预处理后的人脸检测框，包括：
    将所述人脸检测框进行放大处理，得到放大处理后的人脸检测框；
    将所述放大处理后的人脸检测框进行裁剪处理，得到裁剪处理后的人脸检测框；
    将所述裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将所述归一化后的人脸检测框作为所述图像预处理后的人脸检测框。
  4. 根据权利要求1所述的方法，其中，所述通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框，包括：
    将所述人脸检测模型的第一层卷积神经网络作为当前层卷积神经网络；将所述待识别图像作为所述当前层卷积神经网络的检测对象；
    通过所述当前层卷积神经网络对所述当前层卷积神经网络的检测对象进行图像下采样，得到所述当前层卷积神经网络对应的人脸 特征提取结果；将所述当前层卷积神经网络对应的人脸特征提取结果作为所述当前层卷积神经网络的下一层卷积神经网络的检测对象；将所述下一层卷积神经网络作为所述当前层卷积神经网络，重复执行上述操作，直到在所述人脸检测模型的第N层卷积神经网络的检测对象中提取出第N层卷积神经网络对应的人脸特征提取结果；其中，N为大于1的自然数；
    基于所述第一层卷积神经网络至所述第N层卷积神经网络中的每一层卷积神经网络对应的人脸特征提取结果，得到所述待识别图像的人脸检测框。
  5. 根据权利要求1所述的方法，其中，所述将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果，包括：
    将所述人脸检测框输入至所述危险驾驶行为识别模型中的卷积层，通过所述卷积层对所述人脸检测框进行卷积操作，得到所述卷积层对应的人脸特征提取结果；
    将所述卷积层对应的人脸特征提取结果输入至所述危险驾驶行为识别模型中的池化层，通过所述池化层对所述卷积层对应的人脸检测框进行池化操作，得到所述池化层对应的人脸特征提取结果；
    将所述池化层对应的人脸特征提取结果输入至所述危险驾驶行为模型中的全连接层，通过所述全连接层对所述池化层对应的人脸特征提取特征进行分类操作，得到所述人脸检测框对应的危险驾驶行为识别结果。
  6. 根据权利要求1所述的方法，在所述将待识别图像输入至预先训练的人脸检测模型之前，所述方法还包括：
    将预先获取的第一个人脸图像样本作为当前人脸图像样本；
    若所述人脸检测模型不满足预先设置的所述人脸检测模型对应的收敛条件，将所述当前人脸图像样本输入至所述人脸检测模型，使用所述当前人脸图像样本对所述人脸检测模型进行训练；将所述当前人脸图像样本的下一个人脸图像样本作为所述当前人脸图像样 本，重复执行上述操作，直到所述人脸检测模型满足所述人脸检测模型对应的收敛条件。
  7. 根据权利要求1所述的方法，在所述将所述人脸检测框输入至预先训练的危险驾驶行为识别模型之前，所述方法还包括：
    将预先获取的第一个人脸检测框样本作为当前人脸检测框样本；
    若所述危险驾驶行为识别模型不满足预先设置的所述危险驾驶行为识别模型对应的收敛条件，将所述当前人脸检测框样本输入至所述危险驾驶行为识别模型，使用所述当前人脸检测框样本对所述危险驾驶行为识别模型进行训练；将所述当前人脸检测框样本的下一个人脸检测框样本作为所述当前人脸检测框样本，重复执行上述操作，直到所述驾驶行为识别模型满足所述危险驾驶行为识别模型对应的收敛条件。
  8. 一种危险驾驶行为识别装置，包括：人脸检测模块和行为识别模块；其中，
    所述人脸检测模块，用于将待识别图像输入至预先训练的人脸检测模型，通过所述人脸检测模型对所述待识别图像进行人脸检测，得到所述待识别图像的人脸检测框；
    所述行为识别模块，用于将所述人脸检测框输入至预先训练的危险驾驶行为识别模型，通过所述危险驾驶行为识别模型对所述人脸检测框进行危险驾驶行为识别，得到所述人脸检测框对应的危险驾驶行为识别结果。
  9. 根据权利要求8所述的装置，其中，所述装置还包括：预处理模块，用于对所述人脸检测框进行图像预处理，得到图像预处理后的人脸检测框；将所述图像预处理后的人脸检测框输入至所述危险驾驶行为识别模型。
  10. 根据权利要求9所述的装置，其中，所述预处理模块包括：放大子模块、裁剪子模块和归一化子模块；其中，
    所述放大子模块，用于将所述人脸检测框进行放大处理，得到放大处理后的人脸检测框；
    所述裁剪子模块，用于将所述放大处理后的人脸检测框进行裁 剪处理，得到裁剪处理后的人脸检测框；
    所述归一化子模块，用于将所述裁剪处理后的人脸检测框进行归一化处理，得到归一化处理后的人脸检测框；将所述归一化后的人脸检测框作为所述图像预处理后的人脸检测框。
  11. 根据权利要求8所述的装置，其中，所述人脸检测模块，具体用于将所述人脸检测模型的第一层卷积神经网络作为当前层卷积神经网络；将所述待识别图像作为所述当前层卷积神经网络的检测对象；
    通过所述当前层卷积神经网络对所述当前层卷积神经网络的检测对象进行图像下采样，得到所述当前层卷积神经网络对应的人脸特征提取结果；将所述当前层卷积神经网络对应的人脸特征提取结果作为所述当前层卷积神经网络的下一层卷积神经网络的检测对象；将所述下一层卷积神经网络作为所述当前层卷积神经网络，重复执行上述操作，直到在所述人脸检测模型的第N层卷积神经网络的检测对象中提取出第N层卷积神经网络对应的人脸特征提取结果；其中，N为大于1的自然数；
    基于所述第一层卷积神经网络至所述第N层卷积神经网络中的每一层卷积神经网络对应的人脸特征提取结果，得到所述待识别图像的人脸检测框。
  12. 根据权利要求8所述的装置，其中，所述行为识别模块，具体用于将所述人脸检测框输入至所述危险驾驶行为识别模型中的卷积层，通过所述卷积层对所述人脸检测框进行卷积操作，得到所述卷积层对应的人脸特征提取结果；将所述卷积层对应的人脸特征提取结果输入至所述危险驾驶行为识别模型中的池化层，通过所述池化层对所述卷积层对应的人脸检测框进行池化操作，得到所述池化层对应的人脸特征提取结果；将所述池化层对应的人脸特征提取结果输入至所述危险驾驶行为模型中的全连接层，通过所述全连接层对所述池化层对应的人脸特征提取特征进行分类操作，得到所述人脸检测框对应的危险驾驶行为识别结果。
  13. 根据权利要求8所述的装置，其中，所述装置还包括：人 脸检测训练模块，用于将预先获取的第一个人脸图像样本作为当前人脸图像样本；若所述人脸检测模型不满足预先设置的所述人脸检测模型对应的收敛条件，将所述当前人脸图像样本输入至所述人脸检测模型，使用所述当前人脸图像样本对所述人脸检测模型进行训练；将所述当前人脸图像样本的下一个人脸图像样本作为所述当前人脸图像样本，重复执行上述操作，直到所述人脸检测模型满足所述人脸检测模型对应的收敛条件。
  14. 根据权利要求8所述的装置，其中，所述装置还包括：行为识别训练模块，用于将预先获取的第一个人脸检测框样本作为当前人脸检测框样本；若所述危险驾驶行为识别模型不满足预先设置的所述危险驾驶行为识别模型对应的收敛条件，将所述当前人脸检测框样本输入至所述危险驾驶行为识别模型，使用所述当前人脸检测框样本对所述危险驾驶行为识别模型进行训练；将所述当前人脸检测框样本的下一个人脸检测框样本作为所述当前人脸检测框样本，重复执行上述操作，直到所述驾驶行为识别模型满足所述危险驾驶行为识别模型对应的收敛条件。
  15. 一种电子设备，包括：
    至少一个处理器；以及
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行权利要求1-7中任一项所述的方法。
  16. 一种存储有计算机指令的非瞬时计算机可读存储介质，所述计算机指令用于使所述计算机执行权利要求1-7中任一项所述的方法。
PCT/CN2021/073483 2020-06-29 2021-01-25 一种危险驾驶行为识别方法、装置、电子设备及存储介质 WO2022001091A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
KR1020217032212A KR20210128491A (ko) 2020-06-29 2021-01-25 위험 주행 행위 식별 방법, 장치, 전자 설비 및 저장매체
US17/599,901 US20230116040A1 (en) 2020-06-29 2021-01-25 Method for recognizing dangerous driving behavior, electronic device and storage medium
JP2021559129A JP2022544635A (ja) 2020-06-29 2021-01-25 危険運転行動認識方法、装置、電子機器および記憶媒体
EP21777920.6A EP3961498A4 (en) 2020-06-29 2021-01-25 DANGEROUS DRIVING BEHAVIOR RECOGNITION METHOD AND APPARATUS, AND ELECTRONIC DEVICE AND STORAGE MEDIA

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010611370.4A CN111814637A (zh) 2020-06-29 2020-06-29 一种危险驾驶行为识别方法、装置、电子设备及存储介质
CN202010611370.4 2020-06-29

Publications (1)

Publication Number Publication Date
WO2022001091A1 true WO2022001091A1 (zh) 2022-01-06

Family

ID=72856282

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/073483 WO2022001091A1 (zh) 2020-06-29 2021-01-25 一种危险驾驶行为识别方法、装置、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN111814637A (zh)
WO (1) WO2022001091A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114724246A (zh) * 2022-04-11 2022-07-08 中国人民解放军东部战区总医院 危险行为识别方法及装置
CN114944009A (zh) * 2022-05-10 2022-08-26 联通(广东)产业互联网有限公司 一种抽烟识别方法、系统、设备及存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111814637A (zh) * 2020-06-29 2020-10-23 北京百度网讯科技有限公司 一种危险驾驶行为识别方法、装置、电子设备及存储介质
CN112633057B (zh) * 2020-11-04 2024-01-30 北方工业大学 一种公交车内异常行为智能监控方法
CN113033374A (zh) * 2021-03-22 2021-06-25 开放智能机器（上海）有限公司 人工智能识别危险行为方法、装置、电子设备和存储介质
CN113011347B (zh) * 2021-03-23 2022-01-07 深圳信可通讯技术有限公司 一种基于人工智能的智能驾驶方法、装置及相关产品
CN114663863A (zh) * 2022-02-24 2022-06-24 北京百度网讯科技有限公司 图像处理方法、装置、电子设备和计算机存储介质

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10035514B1 (en) * 2014-11-21 2018-07-31 Lytx, Inc. Dynamic configuration of event recorder content capture
CN108960065A (zh) * 2018-06-01 2018-12-07 浙江零跑科技有限公司 一种基于视觉的驾驶行为检测方法
CN109740477A (zh) * 2018-12-26 2019-05-10 联创汽车电子有限公司 驾驶员疲劳检测系统及其疲劳检测方法
CN109948450A (zh) * 2019-02-22 2019-06-28 深兰科技（上海）有限公司 一种基于图像的用户行为检测方法、装置和存储介质
CN110674701A (zh) * 2019-09-02 2020-01-10 东南大学 一种基于深度学习的驾驶员疲劳状态快速检测方法
CN110837815A (zh) * 2019-11-15 2020-02-25 济宁学院 一种基于卷积神经网络的驾驶员状态监测方法
CN111626272A (zh) * 2020-07-28 2020-09-04 长沙超创电子科技有限公司 一种基于深度学习的驾驶员疲劳监测系统
CN111814637A (zh) * 2020-06-29 2020-10-23 北京百度网讯科技有限公司 一种危险驾驶行为识别方法、装置、电子设备及存储介质
CN112183482A (zh) * 2020-10-29 2021-01-05 苏州市职业大学 危险驾驶行为识别方法、装置、系统及可读存储介质

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110969130B (zh) * 2019-12-03 2023-04-18 厦门瑞为信息技术有限公司 一种基于yolov3司机危险动作识别方法及系统

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10035514B1 (en) * 2014-11-21 2018-07-31 Lytx, Inc. Dynamic configuration of event recorder content capture
CN108960065A (zh) * 2018-06-01 2018-12-07 浙江零跑科技有限公司 一种基于视觉的驾驶行为检测方法
CN109740477A (zh) * 2018-12-26 2019-05-10 联创汽车电子有限公司 驾驶员疲劳检测系统及其疲劳检测方法
CN109948450A (zh) * 2019-02-22 2019-06-28 深兰科技（上海）有限公司 一种基于图像的用户行为检测方法、装置和存储介质
CN110674701A (zh) * 2019-09-02 2020-01-10 东南大学 一种基于深度学习的驾驶员疲劳状态快速检测方法
CN110837815A (zh) * 2019-11-15 2020-02-25 济宁学院 一种基于卷积神经网络的驾驶员状态监测方法
CN111814637A (zh) * 2020-06-29 2020-10-23 北京百度网讯科技有限公司 一种危险驾驶行为识别方法、装置、电子设备及存储介质
CN111626272A (zh) * 2020-07-28 2020-09-04 长沙超创电子科技有限公司 一种基于深度学习的驾驶员疲劳监测系统
CN112183482A (zh) * 2020-10-29 2021-01-05 苏州市职业大学 危险驾驶行为识别方法、装置、系统及可读存储介质

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114724246A (zh) * 2022-04-11 2022-07-08 中国人民解放军东部战区总医院 危险行为识别方法及装置
CN114724246B (zh) * 2022-04-11 2024-01-30 中国人民解放军东部战区总医院 危险行为识别方法及装置
CN114944009A (zh) * 2022-05-10 2022-08-26 联通(广东)产业互联网有限公司 一种抽烟识别方法、系统、设备及存储介质

Also Published As

Publication number Publication date
CN111814637A (zh) 2020-10-23

Similar Documents

Publication Publication Date Title
WO2022001091A1 (zh) 一种危险驾驶行为识别方法、装置、电子设备及存储介质
US11643076B2 (en) Forward collision control method and apparatus, electronic device, program, and medium
KR102447352B1 (ko) 교통 신호등 검출 및 지능형 주행을 위한 방법 및 디바이스, 차량, 및 전자 디바이스
US11170210B2 (en) Gesture identification, control, and neural network training methods and apparatuses, and electronic devices
KR102597377B1 (ko) 이미지 인식방법, 장치, 기기, 컴퓨터 저장매체 및 컴퓨터 프로그램
WO2022033095A1 (zh) 一种文本区域的定位方法及装置
WO2019136623A1 (en) Apparatus and method for semantic segmentation with convolutional neural network
US20230267735A1 (en) Method for structuring pedestrian information, device, apparatus and storage medium
US20210209385A1 (en) Method and apparatus for recognizing wearing state of safety belt
US20210295472A1 (en) Method and apparatus for recognizing abnormal license plate, device and readable storage medium
CN111914698A (zh) 图像中人体的分割方法、分割系统、电子设备及存储介质
EP3846122B1 (en) Method and apparatus for generating background-free image, device, and medium
EP4080470A2 (en) Method and apparatus for detecting living face
Harini et al. Sign language translation
US20230116040A1 (en) Method for recognizing dangerous driving behavior, electronic device and storage medium
US11881050B2 (en) Method for detecting face synthetic image, electronic device, and storage medium
US20230030431A1 (en) Method and apparatus for extracting feature, device, and storage medium
US20210201501A1 (en) Motion-based object detection method, object detection apparatus and electronic device
CN115358392A (zh) 深度学习网络的训练方法、文本检测方法及装置
CN111862031B (zh) 一种人脸合成图检测方法、装置、电子设备及存储介质
CN111814636A (zh) 一种安全带检测方法、装置、电子设备及存储介质
WO2023020176A1 (zh) 图像识别方法和装置
US20230245429A1 (en) Method and apparatus for training lane line detection model, electronic device and storage medium
CN111932530B (zh) 三维对象检测方法、装置、设备和可读存储介质
CN113361536A (zh) 图像语义分割模型训练、图像语义分割方法及相关装置

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2021559129

Country of ref document: JP

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 20217032212

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2021777920

Country of ref document: EP

Effective date: 20211007

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

Ref document number: 21777920

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE