Vehicle Event Recognition Method, Device, Equipment and Storage Medium

Abstract

The present invention discloses a vehicle event recognition method, device, equipment and storage medium. The method includes dividing an original image data set into a construction image data set and a training image data set; extracting semantic words of frame images in the construction image data set and constructing a word list by using the semantic words; determining a semantic word sequence corresponding to a frame image in the training image data set based on the word list and converting the semantic word sequence into a semantic word matrix; inputting the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model; and when receiving a test image data set, using the vehicle event recognition model to recognize the test image data set to obtain a vehicle event classification result. By acquiring the semantic words of vehicle driving images, constructing a semantic word matrix and further training to obtain a vehicle event recognition model, the present invention performs semantic recognition by using vehicle driving images during vehicle driving, improves the efficiency and accuracy of vehicle event recognition, and avoids the occurrence of traffic accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and particularly to a vehicle event recognition method, device, equipment and storage medium. Background Art

[0002] Currently, cars have almost become one of the essential tools for every household to travel. Coupled with the continuous development of electronic technology and the popularization of intelligent electronic devices, concepts such as the Internet of Things, big data, and vehicle networking have taken root in people's hearts. The intelligent vehicle industry shows a prosperous scene, bringing more user-friendly experiences and services to car owners. However, frequent traffic accidents are hard to guard against, and currently, there is a lack of safety detection devices in cars.

[0003] Although some high-end cars on the current market are installed with driving recorders, automotive assistant systems ADAS, facial image recognition systems, and the original vehicle sensors are deeply integrated with the vehicle yard, etc., which can effectively improve driving safety and reduce the accident rate. However, most of these devices or systems are used for driving data collection, and the efficiency and accuracy of real-time vehicle event recognition are not high. Therefore, how to improve the efficiency and accuracy of vehicle event recognition to avoid traffic accidents is a technical problem that urgently needs to be solved.

[0004] The above content is only used to assist in understanding the technical solution of the present invention, and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of the present invention is to provide a vehicle event recognition method, device, equipment and storage medium, aiming to solve the technical problem that the current vehicle safety monitoring device has low efficiency and accuracy in real-time vehicle event recognition.

[0006] To achieve the above purpose, the present invention provides a vehicle event recognition method, and the method includes the following steps:

[0007] Obtain the original image data set of vehicle driving, and divide the original image data set into a construction image data set and a training image data set;

[0008] Extract the semantic words of the frame images in the construction image data set, and construct a word list using the semantic words;

[0009] Based on the word list, determine the semantic word sequence corresponding to the frame images in the training image data set, and convert the semantic word sequence into a semantic word matrix;

[0010] Input the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model;

[0011] When receiving the test image dataset, use the vehicle event recognition model to recognize the test image dataset to obtain the vehicle event classification result.

[0012] In the present invention, by obtaining the semantic words of vehicle driving images, constructing a semantic word matrix, and further training to obtain a vehicle event recognition model, during vehicle driving, semantic recognition is performed using vehicle driving images, improving the efficiency and accuracy of vehicle event recognition to avoid traffic accidents.

[0013] Optionally, the step of extracting the semantic words of the frame images in the constructed image dataset and constructing a word list using the semantic words specifically includes:

[0014] Extract the feature information of each frame image in the constructed image dataset;

[0015] Using the feature information, match the corresponding vehicle behavior, and convert the feature information of each frame image into the semantic words of the corresponding vehicle behavior in the database;

[0016] Construct a word list using the semantic words.

[0017] In the present invention, by extracting the feature information of each frame image in the constructed image dataset to match the corresponding vehicle behavior, the extraction of the semantic words of the frame images is realized.

[0018] Optionally, before the step of constructing a word list using the semantic words, the method further includes:

[0019] Query the call frequency of the semantic words stored in the database;

[0020] Obtain a call frequency threshold, and based on the call frequency threshold, determine the infrequently used semantic words in the database;

[0021] Update the semantic words corresponding to each frame image to eliminate the infrequently used semantic words in the database.

[0022] In the present invention, by eliminating the extracted semantic words before constructing the word list, eliminating the semantic words with a low call frequency in the database, making the constructed word list more effective, and thus improving the recognition accuracy.

[0023] Optionally, the step of constructing a word list using the semantic words specifically includes:

[0024] Construct a word list for the semantic words corresponding to all the frame images in the constructed image dataset according to a preset sorting rule; wherein, the preset sorting rule includes the call frequency order and / or the word part-of-speech order.

[0025] In the present invention, semantic words in the frame images are sorted according to the rules of call frequency or word part of speech, and then a word list is constructed to facilitate the search for the semantic words corresponding to the features during training, thereby improving the recognition efficiency.

[0026] Optionally, the step of determining the semantic word sequence corresponding to the frame image in the training image dataset based on the word list specifically includes:

[0027] Extract the feature information of each frame image in the training image dataset;

[0028] Using the feature information, match the corresponding vehicle behavior, and convert the feature information of each frame image into the semantic word of the corresponding vehicle behavior in the database;

[0029] Use the semantic words of the vehicle behavior to match the semantic words in the word list to determine the semantic word sequence corresponding to the frame image in the training image dataset.

[0030] In the present invention, when determining the semantic word sequence of the frame image, by extracting the feature information of the frame image, using the feature information to match the vehicle behavior, and then obtaining its semantic word and semantic word sequence according to the vehicle behavior, semantic recognition is performed using the vehicle driving image, thereby improving the efficiency and accuracy of vehicle event recognition.

[0031] Optionally, the step of converting the semantic word sequence into a semantic word matrix specifically includes:

[0032] For the semantic word sequence corresponding to each frame image in the training image dataset, generate a semantic word binary sequence according to a preset transformation rule;

[0033] Generate a group of semantic word binary sequences according to the semantic word binary sequences of all frame images in the training image dataset, and transform the group of semantic word binary sequences into a semantic word matrix.

[0034] In the present invention, in order to train the semantic words of the obtained frame image to obtain a vehicle event recognition model, the semantic word sequence is converted into a semantic word matrix to improve the recognition accuracy of the vehicle event recognition model.

[0035] Optionally, the step of recognizing the test image dataset using the vehicle event recognition model specifically includes:

[0036] Extract the feature information of each frame image in the test image dataset;

[0037] Using the feature information, match the corresponding vehicle behavior, and convert the feature information of each frame image into the semantic word of the corresponding vehicle behavior in the database;

[0038] Apply the preset transformation rules to the semantic words and convert them into a binary sequence of semantic words;

[0039] Based on all the binary sequences of semantic words in the test image dataset, determine the sequence group of semantic words and convert it into a semantic word matrix;

[0040] Input the semantic word matrix into the vehicle event recognition model.

[0041] In the present invention, after obtaining the test image dataset, in order for the vehicle event recognition model to recognize the test image dataset, it is necessary to extract the feature information in the test image dataset, construct a semantic word matrix using the feature information, and perform semantic analysis using the semantic word matrix to improve the efficiency and accuracy of the safety monitoring device in recognizing immediate vehicle events.

[0042] In addition, to achieve the above object, the present invention also provides a vehicle event recognition device, which includes:

[0043] A partitioning module for obtaining the original image dataset of vehicle driving and partitioning the original image dataset into a construction image dataset and a training image dataset;

[0044] An extraction module for extracting the semantic words of the frame images in the construction image dataset and constructing a word list using the semantic words;

[0045] A determination module for determining the semantic word sequence corresponding to the frame images in the training image dataset based on the word list and converting the semantic word sequence into a semantic word matrix;

[0046] A training module for inputting the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model;

[0047] A recognition module for, when receiving a test image dataset, using the vehicle event recognition model to recognize the test image dataset and obtaining a vehicle event classification result.

[0048] In addition, to achieve the above object, the present invention also provides a vehicle event recognition device, which includes: a memory, a processor, and a vehicle event recognition program stored on the memory and executable on the processor. When the vehicle event recognition program is executed by the processor, the steps of the vehicle event recognition method described above are implemented.

[0049] In addition, to achieve the above object, the present invention also provides a storage medium on which a vehicle event recognition program is stored. When the vehicle event recognition program is executed by a processor, the steps of the vehicle event recognition method described above are implemented.

[0050] A vehicle event recognition method, device, equipment and storage medium proposed by an embodiment of the present invention. The method includes obtaining an original image data set of vehicle driving, dividing the original image data set into a construction image data set and a training image data set; extracting semantic words of frame images in the construction image data set, and constructing a word list by using the semantic words; based on the word list, determining a semantic word sequence corresponding to a frame image in the training image data set, and converting the semantic word sequence into a semantic word matrix; inputting the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model; when receiving a test image data set, using the vehicle event recognition model to recognize the test image data set to obtain a vehicle event classification result. By obtaining the semantic words of vehicle driving images, constructing a semantic word matrix, and further training to obtain a vehicle event recognition model, the present invention performs semantic recognition by using vehicle driving images when the vehicle is driving, improves the efficiency and accuracy of vehicle event recognition, and avoids the occurrence of traffic accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic structural diagram of a vehicle event recognition device in an embodiment of the present invention;

[0052] Figure 2 It is a schematic flowchart of an embodiment of the vehicle event recognition method of the present invention;

[0053] Figure 3 It is a structural block diagram of a vehicle event recognition device in an embodiment of the present invention.

[0054] The realization, functional characteristics and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] At present, cars have almost become one of the essential tools for every family to travel. Coupled with the continuous development of electronic technology and the popularization of intelligent electronic devices, concepts such as the Internet of Things, big data, and the Internet of Vehicles have taken root in people's hearts, and the intelligent vehicle industry shows a prosperous scene, bringing more user-friendly experiences and services to car owners. However, frequent traffic accidents are beyond our control, and there is a lack of safety detection devices in current cars.

[0057] Although some high - end cars on the current market are installed with driving recorders, automotive assistant systems ADAS, facial image recognition systems, and original vehicle sensors that are deeply integrated with the vehicle yard, etc., which can effectively improve driving safety and reduce the accident rate. However, most of these devices or systems are used for driving data collection, and the efficiency and accuracy of real - time vehicle event recognition are not high. Therefore, how to improve the efficiency and accuracy of vehicle event recognition to avoid traffic accidents is a technical problem that urgently needs to be solved.

[0058] To solve this problem, various embodiments of the vehicle event recognition method of the present invention are proposed. The vehicle event recognition method provided by the present invention obtains the semantic words of vehicle driving images, constructs a semantic word matrix, and further trains to obtain a vehicle event recognition model. When the vehicle is driving, semantic recognition is performed using the vehicle driving images to improve the efficiency and accuracy of vehicle event recognition and avoid traffic accidents.

[0059] Refer to Figure 1 , Figure 1 which is a schematic structural diagram of the vehicle event recognition device involved in the embodiment solution of the present invention.

[0060] The device can be a user equipment (UE) such as a mobile phone, a smart phone, a laptop computer, a digital broadcast receiver, a personal digital assistant (PDA), a tablet computer (PAD), a handheld device, an in - vehicle device, a wearable device, a computing device or other processing devices connected to a wireless modem, a mobile station (MS), etc. The device may be referred to as a user terminal, a portable terminal, a desktop terminal, etc.

[0061] Generally, the device includes: at least one processor 301, a memory 302, and a vehicle event recognition program stored on the memory and executable on the processor. The vehicle event recognition program is configured to implement the steps of the vehicle event recognition method as described above.

[0062] The processor 301 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. The processor 301 may also include an AI (Artificial Intelligence) processor, which is used to process vehicle event recognition operations, enabling the vehicle event recognition model to autonomously train and learn, improving efficiency and accuracy.

[0063] The memory 302 may include one or more computer-readable storage media, and the computer-readable storage media may be non-transitory. The memory 302 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 302 is used to store at least one instruction, and the at least one instruction is used to be executed by the processor 301 to implement the vehicle event recognition method provided in the method embodiments of the present application.

[0064] In some embodiments, the terminal may also optionally include: a communication interface 303 and at least one peripheral device. The processor 301, the memory 302, and the communication interface 303 may be connected through a bus or signal lines. Each peripheral device may be connected to the communication interface 303 through a bus, signal lines, or a circuit board. Specifically, the peripheral devices include at least one of a radio frequency circuit 304, a display screen 305, and a power supply 306.

[0065] The communication interface 303 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302. The communication interface 303 is used by the peripheral device to receive the movement trajectories and other data of multiple mobile terminals uploaded by the user. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 can be implemented on a separate chip or circuit board, and this embodiment does not limit this.

[0066] The radio frequency circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 304 communicates with the communication network and other communication devices through electromagnetic signals, so as to obtain the movement trajectories and other data of multiple mobile terminals. The radio frequency circuit 304 converts an electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 304 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and so on. The radio frequency circuit 304 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to: metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area network, and / or WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 304 may further include a circuit related to NFC (Near Field Communication), and this application does not limit this.

[0067] The display screen 305 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 305 is a touch display screen, the display screen 305 also has the ability to collect touch signals on or above the surface of the display screen 305. The touch signals can be input to the processor 301 for processing as control signals. At this time, the display screen 305 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 305 can be one, the front panel of the electronic device; in other embodiments, the display screen 305 can be at least two, respectively arranged on different surfaces of the electronic device or in a folding design; in still other embodiments, the display screen 305 can be a flexible display screen, arranged on the curved surface or folding surface of the electronic device. Even, the display screen 305 can also be set to an irregular shape that is not rectangular, that is, a special-shaped screen. The display screen 305 can be prepared from materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0068] The power supply 306 is used to supply power to each component in the electronic device. The power supply 306 can be alternating current, direct current, a disposable battery, or a rechargeable battery. When the power supply 306 includes a rechargeable battery, the rechargeable battery can support wired charging or wireless charging. The rechargeable battery can also be used to support fast charging technology.

[0069] Those skilled in the art can understand that Figure 1 the structure shown in does not constitute a limitation on the vehicle event recognition device, and may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0070] An embodiment of the present invention provides a vehicle event recognition method. Refer to Figure 2 , Figure 2 which is a schematic flowchart of the embodiment of the vehicle event recognition method of the present invention.

[0071] In this embodiment, the vehicle event recognition method includes the following steps:

[0072] Step S100, obtain the original image data set of the vehicle driving, and divide the original image data set into a construction image data set and a training image data set.

[0073] Specifically, in order to achieve vehicle event recognition, this embodiment adopts a semantic-based image recognition strategy. By constructing a word list, the image is semantically processed, and then vehicle events in the image are recognized. Among them, in the original image dataset of vehicle driving, in order to construct a word list, a part of the image data in the original image dataset is divided as the construction image dataset. At the same time, in order to train and obtain a vehicle recognition model, a part of the image data in the original image dataset is divided as the training image dataset.

[0074] Step S200: Extract the semantic words of the frame images in the construction image dataset, and construct a word list using the semantic words.

[0075] Specifically, for the obtained construction image dataset, the semantic words of the frame images in the construction image dataset can be extracted to construct a word list.

[0076] In practical applications, extract the feature information of each frame image in the construction image dataset; use the feature information to match the corresponding vehicle behavior, convert the feature information of each frame image into the semantic words of the corresponding vehicle behavior in the database; construct a word list using the semantic words.

[0077] It is easy to understand that when constructing a word list, for each frame image in the construction image dataset, match the semantic words corresponding to the vehicle behavior in the database, and summarize the semantic words corresponding to all frame images to construct a word list.

[0078] It should be noted that before constructing a word list using semantic words, the call frequency of the semantic words stored in the database can also be queried; obtain the call frequency threshold, and based on the call frequency threshold, determine the infrequently used semantic words in the database; update the semantic words corresponding to each frame image to eliminate the infrequently used semantic words in the database.

[0079] Furthermore, when constructing a word list using semantic words, the semantic words corresponding to all the frame images in the construction image dataset can be sorted according to a preset sorting rule to construct a word list; where the preset sorting rule includes the call frequency order and / or the word part-of-speech order.

[0080] Step S300: Based on the word list, determine the semantic word sequence corresponding to the frame images in the training image dataset, and convert the semantic word sequence into a semantic word matrix.

[0081] Specifically, after the word list is constructed, when using the training image dataset for training, the semantic word sequence corresponding to the frame images in the training image dataset needs to be extracted first, and the semantic word sequence is converted into a semantic word matrix for model training.

[0082] In practical applications, extract the feature information of each frame of image in the training image dataset; use the feature information to match the corresponding vehicle behavior, convert the feature information of each frame of image into the semantic words of the corresponding vehicle behavior in the database; use the semantic words of the vehicle behavior to match the semantic words in the word list to determine the semantic word sequence corresponding to the frame image in the training image dataset.

[0083] Further, it is easy to understand that for the semantic word sequence corresponding to each frame of image in the training image dataset, generate a binary semantic word sequence according to a preset transformation rule; generate a binary semantic word sequence group according to the binary semantic word sequences of all frames of images in the training image dataset, and transform the binary semantic word sequence group into a semantic word matrix.

[0084] Step S400, input the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model.

[0085] Specifically, after obtaining the semantic word matrix, the semantic word matrix can be input into a convolutional network for training to further obtain a vehicle event recognition model.

[0086] Step S500, when receiving a test image dataset, use the vehicle event recognition model to recognize the test image dataset to obtain a vehicle event classification result.

[0087] Specifically, after obtaining the vehicle event recognition model, if a test image dataset is received, it is necessary to perform the steps of extracting the semantic word sequence from the test image dataset and converting the semantic word sequence into a semantic word matrix in the same way;

[0088] In practical applications, extract the feature information of each frame of image in the test image dataset; use the feature information to match the corresponding vehicle behavior, convert the feature information of each frame of image into the semantic words of the corresponding vehicle behavior in the database; convert the semantic words into a binary semantic word sequence according to a preset transformation rule; determine a binary semantic word sequence group based on all the binary semantic word sequences in the test image dataset, and convert it into a semantic word matrix; input the semantic word matrix into the vehicle event recognition model.

[0089] In this embodiment, by obtaining the semantic words of vehicle driving images, constructing a semantic word matrix, and further training to obtain a vehicle event recognition model, when the vehicle is driving, semantic recognition is performed using vehicle driving images to improve the efficiency and accuracy of vehicle event recognition to avoid traffic accidents.

[0090] Refer to Figure 3 , Figure 3 which is the structural block diagram of the embodiment of the vehicle event recognition device of the present invention.

[0091] As shown in Figure 3 the vehicle event recognition device proposed in the embodiment of the present invention includes:

[0092] A partitioning module 10, configured to obtain an original image dataset of vehicle driving, and partition the original image dataset into a constructed image dataset and a training image dataset;

[0093] An extraction module 20, configured to extract semantic words of frame images in the constructed image dataset, and construct a word list using the semantic words;

[0094] A determination module 30, configured to determine a semantic word sequence corresponding to a frame image in the training image dataset based on the word list, and convert the semantic word sequence into a semantic word matrix;

[0095] A training module 40, configured to input the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model;

[0096] An identification module 50, configured to, when receiving a test image dataset, use the vehicle event recognition model to identify the test image dataset to obtain a vehicle event classification result.

[0097] As an implementation manner, the extraction module 50 is further configured to extract feature information of each frame image in the constructed image dataset; use the feature information to match corresponding vehicle behaviors, and convert the feature information of each frame image into semantic words of corresponding vehicle behaviors in the database; construct a word list using the semantic words.

[0098] As an implementation manner, the extraction module 50 is further configured to query the call frequency of the semantic words stored in the database; obtain a call frequency threshold, and based on the call frequency threshold, determine the semantic words that are not commonly used in the database; update the semantic words corresponding to each frame image to eliminate the semantic words that are not commonly used in the database.

[0099] As an implementation manner, the extraction module 50 is further configured to construct a word list for the semantic words corresponding to all the frame images in the constructed image dataset according to a preset sorting rule; wherein, the preset sorting rule includes the call frequency order and / or the word part-of-speech order.

[0100] As an implementation manner, the determination module 30 is further configured to extract feature information of each frame image in the training image dataset; use the feature information to match corresponding vehicle behaviors, and convert the feature information of each frame image into semantic words of corresponding vehicle behaviors in the database; match the semantic words of the vehicle behaviors with the semantic words in the word list to determine the semantic word sequence corresponding to the frame image in the training image dataset.

[0101] As an implementation manner, the determination module 30 is further configured to generate a semantic word binary sequence for the semantic word sequence corresponding to each frame of image in the training image dataset according to a preset transformation rule; generate a semantic word binary sequence group according to the semantic word binary sequences of all frames of images in the training image dataset, and transform the semantic word binary sequence group into a semantic word matrix.

[0102] As an implementation manner, the recognition module 50 is further configured to extract the feature information of each frame of image in the test image dataset; use the feature information to match the corresponding vehicle behavior, convert the feature information of each frame of image into the semantic word of the corresponding vehicle behavior in the database; convert the semantic word into a semantic word binary sequence according to a preset transformation rule; determine a semantic word binary sequence group based on all the semantic word binary sequences in the test image dataset, and convert it into a semantic word matrix; input the semantic word matrix into the vehicle event recognition model.

[0103] The vehicle event recognition device provided in this embodiment obtains the semantic words of the vehicle driving images, constructs a semantic word matrix, further trains to obtain a vehicle event recognition model, and uses the vehicle driving images for semantic recognition when the vehicle is driving, improving the efficiency and accuracy of vehicle event recognition to avoid traffic accidents.

[0104] For other embodiments or specific implementation manners of the vehicle event recognition device of the present invention, reference may be made to the above method embodiments, and details will not be described here again.

[0105] In addition, an embodiment of the present invention further provides a storage medium, on which a vehicle event recognition program is stored. When the vehicle event recognition program is executed by a processor, it implements the steps of the vehicle event recognition method as described above. Therefore, details will not be described here again. In addition, the beneficial effects of using the same method will not be described again. For the technical details not disclosed in the embodiment of the computer-readable storage medium involved in this application, please refer to the description of the method embodiment of this application. By way of example, the program instructions can be deployed to be executed on a computing device, or on multiple computing devices located at one location, or on multiple computing devices distributed at multiple locations and interconnected through a communication network.

[0106] Those of ordinary skill in the art can understand that all or part of the processes of implementing the above method embodiments can be completed by instructing related hardware through a computer program. The above program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. Among them, the above storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0107] It should be further noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement this without creative efforts.

[0108] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by dedicated hardware including application-specific integrated circuits, dedicated CPUs, dedicated memories, dedicated components, etc. Generally, functions accomplished by computer programs can be easily implemented by corresponding hardware, and the specific hardware structures for implementing the same function can also be various, such as analog circuits, digital circuits or dedicated circuits. However, for the present invention, in more cases, software program implementation is a better implementation manner. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a floppy disk, a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disc of a computer, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

Claims

1. A vehicle event recognition method, characterized in that, The method includes the following steps: Obtain the original image dataset of vehicle driving, and divide the original image dataset into a construction image dataset and a training image dataset; Extract the semantic words of the frame images in the construction image dataset, and construct a word list using the semantic words; Based on the word list, determine the semantic word sequence corresponding to the frame images in the training image dataset, and convert the semantic word sequence into a semantic word matrix; Input the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model; When receiving a test image dataset, use the vehicle event recognition model to recognize the test image dataset to obtain a vehicle event classification result.

2. The vehicle event recognition method according to claim 1, characterized in that The step of extracting the semantic words of the frame images in the construction image dataset and constructing a word list using the semantic words specifically includes: Extract the feature information of each frame image in the construction image dataset; Using the feature information, match the corresponding vehicle behaviors, and convert the feature information of each frame image into the semantic words of the corresponding vehicle behaviors in the database; Construct a word list using the semantic words.

3. The vehicle event recognition method according to claim 2, wherein Before the step of constructing a word list using the semantic words, the method further includes: Query the call frequency of the semantic words stored in the database; Obtain a call frequency threshold, and based on the call frequency threshold, determine the less frequently used semantic words in the database; Update the semantic words corresponding to each frame image to eliminate the less frequently used semantic words in the database.

4. The vehicle event recognition method according to claim 3, wherein The step of constructing a word list using the semantic words specifically includes: For the semantic words corresponding to all the frame images in the construction image dataset, construct a word list according to a preset sorting rule; wherein, the preset sorting rule includes the call frequency order and / or the word part-of-speech order.

5. The vehicle event recognition method according to claim 1, wherein The step of determining the semantic word sequence corresponding to the frame images in the training image dataset based on the word list specifically includes: Extract the feature information of each frame image in the training image dataset; Using the feature information, match the corresponding vehicle behaviors, and convert the feature information of each frame image into the semantic words of the corresponding vehicle behaviors in the database; Match the semantic words of the vehicle behaviors with the semantic words in the word list to determine the semantic word sequence corresponding to the frame images in the training image dataset.

6. The vehicle event recognition method according to claim 5, characterized in that, The step of converting the semantic word sequence into a semantic word matrix specifically includes: For the semantic word sequence corresponding to each frame image in the training image dataset, generate a semantic word binary sequence according to a preset transformation rule; According to the semantic word binary sequences of all the frame images in the training image dataset, generate a group of semantic word binary sequences, and transform the group of semantic word binary sequences into a semantic word matrix.

7. The vehicle event recognition method according to claim 6, wherein, The step of using the vehicle event recognition model to recognize the test image dataset specifically includes: Extract the feature information of each frame image in the test image dataset; Using the feature information, match the corresponding vehicle behaviors, and convert the feature information of each frame image into the semantic words of the corresponding vehicle behaviors in the database; Convert the semantic words into a semantic word binary sequence according to a preset transformation rule; Based on all semantic word binary sequences in the test image dataset, determine the semantic word sequence group and convert it into a semantic word matrix; Input the semantic word matrix into the vehicle event recognition model.

8. A vehicle event recognition device, characterized in that, The vehicle event recognition device includes: A partitioning module, configured to obtain the original image dataset of vehicle driving, and partition the original image dataset into a construction image dataset and a training image dataset; An extraction module, configured to extract the semantic words of the frame images in the construction image dataset, and construct a word list using the semantic words; A determination module, configured to determine the semantic word sequence corresponding to the frame images in the training image dataset based on the word list, and convert the semantic word sequence into a semantic word matrix; A training module, configured to input the semantic word matrix into a convolutional network for training to obtain a vehicle event recognition model; An identification module, configured to, when receiving a test image dataset, use the vehicle event recognition model to identify the test image dataset and obtain a vehicle event classification result.

9. A vehicle event recognition device, characterized in that, The vehicle event recognition device includes: a memory, a processor, and a vehicle event recognition program stored on the memory and executable on the processor. When the vehicle event recognition program is executed by the processor, the steps of the vehicle event recognition method according to any one of claims 1 to 7 are implemented.

10. A storage medium, characterized in that, A vehicle event recognition program is stored on the storage medium. When the vehicle event recognition program is executed by a processor, the steps of the vehicle event recognition method according to any one of claims 1 to 7 are implemented.

Images