CN117392857B - Truck type recognition system and recognition method based on Bluetooth network - Google Patents
Truck type recognition system and recognition method based on Bluetooth network Download PDFInfo
- Publication number
- CN117392857B CN117392857B CN202311327289.3A CN202311327289A CN117392857B CN 117392857 B CN117392857 B CN 117392857B CN 202311327289 A CN202311327289 A CN 202311327289A CN 117392857 B CN117392857 B CN 117392857B
- Authority
- CN
- China
- Prior art keywords
- sound
- model
- image
- accuracy
- recognition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000011156 evaluation Methods 0.000 claims abstract description 77
- 238000004088 simulation Methods 0.000 claims abstract description 29
- 238000012795 verification Methods 0.000 claims abstract description 22
- 238000012360 testing method Methods 0.000 claims abstract description 14
- 238000004364 calculation method Methods 0.000 claims abstract description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 18
- 238000007781 pre-processing Methods 0.000 claims description 15
- 230000005540 biological transmission Effects 0.000 claims description 14
- 230000010485 coping Effects 0.000 claims description 10
- 238000010801 machine learning Methods 0.000 claims description 10
- 238000007405 data analysis Methods 0.000 claims description 8
- 239000000284 extract Substances 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 abstract description 11
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000013528 artificial neural network Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 201000010099 disease Diseases 0.000 description 5
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 5
- 238000001228 spectrum Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000010191 image analysis Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000005236 sound signal Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/80—Services using short range communication, e.g. near-field communication [NFC], radio-frequency identification [RFID] or low energy communication
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Computational Linguistics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a truck type recognition system and a recognition method based on a Bluetooth network, and relates to the technical field of vehicle monitoring, wherein the recognition system comprises an audio-video acquisition module, an audio-video recognition module, a simulation module, a database contrast verification module and a system evaluation module, wherein the audio-video acquisition module is used for acquiring audio-video data, and comprises engine starting sound and a vehicle appearance image; establishing a corresponding recognition model in the audio-video recognition module; the technical key points are as follows: the accuracy of judging the truck type is improved by adopting a mode of analyzing and verifying through combination of image and sound collection, meanwhile, in order to ensure the running stability of the recognition system, a simulation test is required in advance, a system state evaluation value Xpgz obtained through analysis and calculation is compared with a preset threshold value, a result of whether the system is stable or not is obtained, and the recognition system is ensured to finally run stably through debugging or executing a strategy.
Description
Technical Field
The invention relates to the technical field of vehicle monitoring, in particular to a truck type recognition system and a recognition method based on a Bluetooth network.
Background
The vehicle monitoring technology is a technology for monitoring and analyzing the state, behavior and performance of a vehicle in real time by using various sensors, data acquisition devices and related algorithms and systems, for example: vehicle position monitoring: the method comprises the steps of acquiring the position information of a vehicle in real time through equipment such as a Global Positioning System (GPS), an Inertial Navigation System (INS) and the like, and matching the position data with a map by utilizing a Geographic Information System (GIS) to realize accurate positioning and track tracking of the vehicle; vehicle behavior monitoring: collecting relevant information such as acceleration, speed, angle and the like of a vehicle through equipment such as an acceleration sensor, a gyroscope and the like, and analyzing behaviors such as acceleration, braking, turning and the like of the vehicle, as well as the posture and the stability of the vehicle; vehicle fault monitoring: monitoring various systems of the vehicle, such as an engine, a transmission system, a braking system and the like, by utilizing an on-board diagnosis system (OBD), a sensor and a fault diagnosis algorithm, detecting fault codes and abnormal signals in real time, and early warning and diagnosing the faults of the vehicle in advance; vehicle type recognition: and (3) acquiring the vehicle appearance by utilizing an image recognition technology, comparing the vehicle appearance with more than 2000 vehicle appearances recorded in a database, and outputting a corresponding vehicle model if the vehicle appearances are consistent.
The prior application number is: 202010104820.0, the technical scheme of the Chinese invention patent named as a truck type and axle type identification method and system indicates that: the problem of identifying the truck type and the axle type of the truck by utilizing images or videos is solved by applying deep learning in target classification and identification, and the appearance characteristic data of the truck is intuitively structured; the method comprises the following steps: s1, acquiring side and front images or videos of an original vehicle; s2, after detecting the position of the positioned vehicle, inputting a trained deep learning model to position and identify the axle of the truck; s3, calculating parameters such as the axle number, the axle base, the tire number and the like of the vehicle according to the axle number and the position of the vehicle detected by the side image of the vehicle; s4, vehicle model identification and classification are carried out according to the front images of the vehicles; the technical scheme pointed out in the patent with the application number 202211298769.7 is as follows: (1) Selecting trucks of multiple vehicle types to shoot truck scene images, and constructing a truck disease data set; (2) acquiring a truck passing image to be detected; (3) Classifying the truck passing image to be detected based on a global multidimensional attention mechanism; (4) Detecting whether a truck passing image to be detected has diseases or not based on a Faster-RCNN deep learning neural network, and positioning the diseases; (5) identifying a disease; relates to application of a visual technology in identifying railway wagon diseases, and combines a deep neural network technology with an image detection and identification technology.
However, for the prior art, conventionally, an electronic camera is usually equipped at a toll gate of a highway, in the process that each truck enters the toll gate to stop and pay, the electronic camera finishes shooting processing on the whole truck, usually, the recognition processing on the truck type is finished by image analysis or combining the image analysis with a deep neural network algorithm, but the truck type is recognized only based on the content of the image analysis, and the truck type cannot be verified in other ways after recognition, so that the accuracy of the recognition result of the truck type cannot be further improved, the stability of the system design in the conventional truck type recognition system is not verified, the condition that the system failure occurs after a period of working or just begins to work may occur, the effective proceeding of the subsequent truck type recognition work is affected, and then the overall efficiency of the truck type recognition work is affected.
Disclosure of Invention
(One) solving the technical problems
Aiming at the defects of the prior art, the invention provides a truck type recognition system and a recognition method based on a Bluetooth network, which are used for improving the accuracy of truck type judgment by adopting a mode of analyzing and verifying by combining image and sound collection, meanwhile, in order to ensure the stability of the operation of the recognition system, a simulation test is required in advance, a system state evaluation value Xpgz obtained by analysis and calculation is compared with a preset threshold value to obtain a result of whether the system is stable, and the final stable operation of the recognition system is ensured by debugging or executing a strategy, so that the problems that the recognition accuracy cannot be improved and the stability of the recognition system cannot be verified in advance in the conventional truck type recognition are solved.
(II) technical scheme
In order to achieve the above purpose, the invention is realized by the following technical scheme:
A bluetooth network-based wagon type recognition system, comprising:
The audio-video acquisition module acquires audio-video data, which comprises engine starting sound and vehicle appearance images;
The audio and video recognition module establishes a corresponding recognition model, processes audio and video data transmitted through a Bluetooth network, and extracts corresponding characteristics for vehicle type recognition;
The simulation module is used for establishing a simulation model, extracting relevant information of any given truck model in the simulation model, wherein the relevant information is sound characteristics and image characteristics of the corresponding truck model, and taking the sound characteristics and the image characteristics of the corresponding truck model as simulation signals;
the database comparison and verification module is used for building a vehicle type database and comprising sound features and image features corresponding to a plurality of truck types, comparing and matching the analog signals with data in the vehicle type database, mutually verifying the truck types obtained by matching, and acquiring the truck types to be output when verification results are consistent;
The system evaluation module is used for building a data analysis model after acquiring evaluation parameters, generating a system state evaluation value Xpgz according to the evaluation parameters, comparing the system state evaluation value Xpgz with a preset threshold, and normally outputting the corresponding truck model to be output when the system state evaluation value Xpgz exceeds the preset threshold.
Further, the audio-video collecting module comprises a sound collecting unit and an image collecting unit, the sound collecting unit is used for collecting engine starting sound, the image collecting unit is used for collecting vehicle appearance images, the audio-video identifying module comprises a sound identifying unit and an image identifying unit, a sound identifying model is built in the sound identifying unit, sound characteristics in engine starting sound transmitted through a Bluetooth network are extracted through a machine learning algorithm, an image identifying model is built in the image identifying unit, and image characteristics in vehicle appearance images transmitted through the Bluetooth network are extracted through an image identifying algorithm.
Further, when mutual verification is performed on the truck model obtained through matching, the condition that verification results are inconsistent is further included, early warning needs to be sent out, and a coping strategy is executed, wherein the coping strategy is to acquire the truck model, and the corresponding sound characteristics and image characteristics are input into a vehicle type database.
Further, the system evaluation module includes a preprocessing unit and an evaluation comparison unit, the preprocessing unit is used for obtaining evaluation parameters, the evaluation parameters include a sound accuracy coefficient Szx, an image accuracy coefficient Tzx and a bluetooth stability coefficient Lwx, a data analysis model is built in the evaluation comparison unit, a system state evaluation value Xpgz is generated according to the evaluation parameters, and the system state evaluation value Xpgz is compared with a preset threshold value.
Further, the step of acquiring the sound accuracy coefficient Szx in the preprocessing unit is as follows:
S101, calculating and acquiring in a voice recognition model, training and testing by inputting a group of voice samples with known labels and using a machine learning algorithm, training the voice recognition model by a training set, and evaluating the accuracy Ac1, the accuracy Pr1 and the recall rate Re1 of the voice recognition model by a testing set;
S102, when calculating the sound accuracy coefficient Szx according to the accuracy Ac1, the accuracy Pr1 and the recall rate Re1, the following formula is adopted:
Wherein a1, a2 and a3 are preset proportionality coefficients of accuracy Ac1, accuracy Pr1 and recall Re1 respectively, and a3 is larger than a1 and a2 is larger than 0, and a1+a2+a3=2.64.
Further, the step of acquiring the image accuracy factor Tzx in the preprocessing unit is as follows:
S201, calculating and acquiring an image accuracy coefficient Tzx by using an image recognition model, training and testing by using an image recognition algorithm by inputting a group of image samples with known labels, training the image recognition model by using a training set, and evaluating the accuracy Ac2, the accuracy Pr2 and the recall Re2 of the image recognition model by using a testing set;
S202, when calculating an image accuracy coefficient Tzx according to the accuracy Ac2, the accuracy Pr2 and the recall Re2, the following formula is adopted:
wherein b1, b2 and b3 are preset proportionality coefficients of an accuracy Ac1, an accuracy Pr1 and a recall Re1 respectively, and b3 is larger than b1 and b2 is larger than 0, and a1+a2+a3=2.57.
Further, the step of acquiring the bluetooth stability factor Lwx in the preprocessing unit is as follows:
S301, acquiring transmission rates Vr t under different time nodes in T time, wherein T represents the numbers of the transmission rates under the different time nodes in T time, and t= {1, 2,3, …, n }, wherein n is a positive integer;
s302, according to the average value of transmission rates in different time nodes in T time And a camera frame rate Vr t to calculate a bluetooth stability coefficient Lwx, where the calculation formula is as follows:
further, the formula according to which the system state evaluation value Xpgz is generated according to the evaluation parameter is as follows:
In the formula, alpha, beta and gamma are respectively preset proportional coefficients of sound accuracy coefficient Szx, image accuracy coefficient Tzx and Bluetooth stability coefficient Lwx, gamma is larger than beta than alpha is larger than 0, alpha+beta+gamma=3.54, and G is a constant correction coefficient.
Further, after comparing the system state evaluation value Xpgz with the preset threshold, if the system state evaluation value Xpgz does not exceed the preset threshold, an adjustment policy is required to be executed, and the adjustment policy is to use a grid search method to adjust the super parameters of the voice recognition model and the image recognition model, and overhaul the bluetooth network.
A truck type recognition method based on a Bluetooth network comprises the following steps:
Step one, acquiring audio-video data, which comprises engine starting sound and vehicle appearance images;
Step two, establishing a sound recognition model, extracting sound characteristics in engine starting sound transmitted through a Bluetooth network through a machine learning algorithm, synchronously establishing an image recognition model, and extracting image characteristics in a vehicle appearance image transmitted through the Bluetooth network through an image recognition algorithm;
Step three, a simulation model is established, relevant information of any given truck model is extracted from the simulation model, the relevant information is sound characteristics and image characteristics of the corresponding truck model, and the sound characteristics and the image characteristics of the corresponding truck model are used as simulation signals;
step four, a vehicle model database is built, the vehicle model database comprises sound features and image features corresponding to a plurality of truck models, the analog signals are compared and matched with data in the vehicle model database, and the truck models obtained through matching are mutually verified;
If the matching is successful, if the verification results are consistent, the identification is successful, and the corresponding truck model to be output is obtained; if the matching is successful, if the verification result is inconsistent, the identification is failed, an early warning is sent out, a coping strategy is executed, and the strategy is that the model of the truck is obtained, and the corresponding sound characteristic and image characteristic are input into a vehicle type database;
step five, building a data analysis model after acquiring the evaluation parameters, generating a system state evaluation value Xpgz according to the evaluation parameters, and comparing the system state evaluation value Xpgz with a preset threshold;
If the system state evaluation value Xpgz exceeds a preset threshold, the system state is stable, and the corresponding truck model to be output is normally output; if the system state evaluation value Xpgz does not exceed the preset threshold, the system state is unstable, and an adjustment strategy is needed to be executed, wherein the adjustment strategy is to adjust the super parameters of the voice recognition model and the image recognition model by using a grid search method, and overhaul the Bluetooth network.
(III) beneficial effects
The invention provides a truck type recognition system and a recognition method based on a Bluetooth network, which have the following beneficial effects:
Compared with the traditional method for recognizing the truck type only through image acquisition and analysis, the method combines image and sound acquisition, extracts corresponding image features and sound features to complete recognition work, wherein the sound recognition unit is matched with the image recognition unit, plays roles of verification and auxiliary judgment, increases the accuracy of the truck type judgment, and can be obviously improved compared with the traditional method;
According to the invention, audio and video data are transmitted through the Bluetooth network, so that real-time vehicle type recognition is realized, compared with the traditional method that time and manpower are required to collect and analyze data and lengthy data lines are arranged, the recognition system can timely acquire and process the data, a rapid recognition result is provided, and the potential problem of redundancy of the data lines is solved;
According to the invention, the simulation module is matched with the database contrast verification module, after matching and mutual verification of the simulation signals and the data in the vehicle type database are completed, the recognition processing of the truck type can be efficiently completed, a coping strategy exists for the truck type which is not recorded in the database, the recognition system is ensured to cope with various truck types, the recognition system is enabled to be continuously perfect, in order to ensure the stability and the effectiveness of the recognition system, the processing to be output is made in the first time of recognizing the truck type, and the subsequent completion of the inspection of the whole recognition system is facilitated;
The invention also adds a system evaluation module in the recognition system to ensure the stability of the recognition system in actual use, comprehensively considers the sound accuracy coefficient Szx, the image accuracy coefficient Tzx and the Bluetooth stability coefficient Lwx in the system evaluation module, ensures the accuracy of generating the system state evaluation value Xpgz, can rapidly judge whether the recognition system is stable in a simulated running state after comparing the system state evaluation value Xpgz with a preset threshold value, can output the recognized truck model result on the premise of ensuring the stability of the recognition system, and executes an adjustment strategy when the recognition system is unstable, thereby realizing the management and optimization of the whole recognition system.
Drawings
Fig. 1 is an overall modular schematic block diagram of a truck type recognition system based on a bluetooth network according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1:
Referring to fig. 1, the embodiment specifically describes a truck type recognition system based on a bluetooth network, including:
The audio-video acquisition module acquires audio-video data, which comprises engine starting sound and vehicle appearance images;
The sound and image acquisition module comprises a sound acquisition unit and an image acquisition unit, wherein the sound acquisition unit is used for acquiring engine starting sound, a sound sensor arranged at the position 1 m+/-0.5 m above a horizontal plane is particularly adopted, the position can acquire cleaned engine starting sound to the greatest extent, and the image acquisition unit is used for acquiring vehicle appearance images, and a high-definition camera arranged at the position 5 m-7 m above the horizontal plane is particularly adopted;
The integral truck type recognition system is applied to a scene of parking payment or inspection in a high-speed intersection gate-crossing toll gate, and an audio-video acquisition module can acquire information of a relatively stationary vehicle, and can freely disassemble and assemble a sound sensor and a high-definition camera applied during information acquisition, so that required information can be acquired, and the accuracy and the effectiveness of the information are preliminarily ensured.
The audio and video recognition module is established, processes audio and video data transmitted through a Bluetooth network, extracts corresponding characteristics and is used for vehicle type recognition;
The audio-video recognition module comprises a sound recognition unit and an image recognition unit;
In the voice recognition unit, a voice recognition model is built, and voice features in engine starting voice transmitted through a Bluetooth network are extracted through a machine learning algorithm, wherein the machine learning algorithm is specifically a Deep Neural Network (DNN): DNN is a deep learning algorithm based on a neural network, and through a multi-layer neural network structure, DNN can automatically learn abstract representation of sound characteristics and realize accurate classification;
the above-mentioned sound features include frequency spectrum features, sound spectrum envelope, sound duration, frequency features and sound intensity, which can be used to describe the frequency domain and time domain features of engine starting sound, so as to distinguish different truck models, each truck model having corresponding sound features;
Specifically, spectral characteristics: the frequency domain analysis can be carried out on the sound through Fourier transformation, and the frequency spectrum characteristics including the intensity and distribution conditions of different frequency components in the sound are extracted; the sound spectrum envelope: the sound signal can obtain the energy change rule, namely the sound spectrum envelope, by an envelope extraction technology, and reflects the time domain characteristics and the frequency domain characteristics of the sound; duration of sound: the duration of a sound refers to the time interval between the beginning and the end of the sound; frequency characteristics: the sound signal may acquire its frequency characteristics, such as dominant frequency and frequency variation, by a Fast Fourier Transform (FFT); sound intensity: the amplitude of the sound signal may reflect the intensity of the sound and may be obtained by measuring the amplitude of the sound.
In an image recognition unit, an image recognition model is established, and image features in an appearance image of a vehicle transmitted through a Bluetooth network are extracted through an image recognition algorithm, wherein the image recognition algorithm is a Convolutional Neural Network (CNN): CNN is a deep learning algorithm suitable for image processing and computer vision tasks, which extracts image features through multi-layer convolution and pooling layers and classifies using fully connected layers;
The image features include the appearance, size and number of wheels of the vehicle, wherein the appearance, size and number of wheels are used for distinguishing different truck types, and each truck type has corresponding image features.
Specifically, the information acquired by the audio-video acquisition module is transmitted to the audio-video identification module by using the Bluetooth network, and the Bluetooth modules are carried in the audio-video acquisition module and the audio-video identification module, so that the information to be transmitted can be transmitted wirelessly through the Bluetooth network, and the condition that excessive lines are disordered is avoided.
By adopting the technical scheme: the method combines the image and sound collection, extracts the corresponding image features and sound features to finish the recognition work, wherein the sound recognition unit is matched with the image recognition unit, and plays roles of verification and auxiliary judgment, so that the accuracy of the judgment of the truck type is improved, and compared with the traditional method, the accuracy can be remarkably improved.
The simulation module is used for establishing a simulation model, extracting relevant information of any given truck model in the simulation model, wherein the relevant information is sound characteristics and image characteristics of the corresponding truck model, and taking the sound characteristics and the image characteristics of the corresponding truck model as simulation signals; specifically, the purpose of designing the simulation module is to ensure that the whole recognition system can be practically used after no obvious problem is found after the whole recognition system is tested or tested, so that the whole recognition system has higher practicability.
The database comparison and verification module is used for building a vehicle model database and comprising sound features and image features corresponding to a plurality of truck models, comparing and matching the analog signals with data in the vehicle model database, and mutually verifying the truck models obtained by matching;
If the matching is successful, if the verification results are consistent, the identification is successful, and the corresponding truck model to be output is obtained; if the matching is successful, if the verification results are inconsistent, the recognition failure is indicated, an early warning is sent out, a coping strategy is executed, the strategy is used for manually acquiring the model of the truck, and the corresponding sound characteristics and image characteristics are input into a vehicle type database.
By adopting the technical scheme: the design simulation module is matched with the database contrast verification module, after the simulation signal is matched with the data in the vehicle type database and mutually verified, the recognition processing of the truck type can be efficiently completed, a coping strategy also exists for the truck type which is not recorded in the database, the recognition system is ensured to cope with various truck types, the recognition system can be continuously perfected, the stability and the effectiveness of the recognition system are ensured, the processing to be output is carried out in the first time of recognizing the truck type, and the subsequent completion of the inspection of the whole recognition system is facilitated.
The system evaluation module comprises a preprocessing unit and an evaluation comparison unit;
Acquiring evaluation parameters including a sound accuracy coefficient Szx, an image accuracy coefficient Tzx and a Bluetooth stability coefficient Lwx through a preprocessing unit, constructing a data analysis model in an evaluation comparison unit, generating a system state evaluation value Xpgz according to the evaluation parameters, and comparing the system state evaluation value Xpgz with a preset threshold;
The step of obtaining the sound accuracy factor Szx in the preprocessing unit is as follows:
s101, calculating and acquiring in a voice recognition model, training and testing by using a machine learning algorithm (deep neural network) through inputting a group of voice samples with known labels, training the voice recognition model through a training set, and evaluating the accuracy Ac1, the accuracy Pr1 and the recall Re1 of the voice recognition model by using a testing set;
S102, when calculating the sound accuracy coefficient Szx according to the accuracy Ac1, the accuracy Pr1 and the recall rate Re1, the following formula is adopted:
Wherein a1, a2 and a3 are preset proportionality coefficients of accuracy Ac1, accuracy Pr1 and recall Re1 respectively, and a3 is larger than a1 and a2 is larger than 0, and a1+a2+a3=2.64.
The step of acquiring the image accuracy factor Tzx in the preprocessing unit is as follows:
S201, calculating and acquiring an image accuracy coefficient Tzx by using an image recognition model, training and testing by using an image recognition algorithm (such as a deep learning model) by inputting a group of image samples with known labels, training the image recognition model by using a training set, and evaluating the accuracy Ac2, the accuracy Pr2 and the recall Re2 of the image recognition model by using a testing set;
S202, when calculating an image accuracy coefficient Tzx according to the accuracy Ac2, the accuracy Pr2 and the recall Re2, the following formula is adopted:
Wherein b1, b2 and b3 are preset proportionality coefficients of an accuracy Ac1, an accuracy Pr1 and a recall Re1 respectively, and b3 is more than b1 and more than b2 is more than 0, and a1+a2+a3=2.57; from the above steps, it can be seen that the principle is the same for acquiring the image accuracy factor Tzx and the sound accuracy factor Szx.
The step of acquiring the bluetooth stability factor Lwx in the preprocessing unit is as follows:
s301, acquiring transmission rates Vr t under different time nodes in T time, wherein T represents the numbers of the transmission rates under the different time nodes in T time, and t= {1, 2, 3, …, n }, wherein n is a positive integer; for example, when the T time is 1h, each time node in the T time is 1min, 2min, … min, 60min, where n is 60;
units of transmission rate are typically expressed in terms of bit rate (bit/s), but multiples thereof such as gigabits per second (Gbps) or megabits per second (Mbps) may also be used; to obtain the transmission rate at each minute time node, this can be typically achieved using a network monitoring tool or device, such as a network traffic monitoring tool: traffic monitoring software or hardware devices to monitor network traffic in real time typically provide real time monitoring and recording of network bandwidth and transmission rate, and can obtain real time transmission rate data at nodes every 1 minute.
S302, according to the average value of transmission rates in different time nodes in T timeAnd a camera frame rate Vr t to calculate a bluetooth stability coefficient Lwx, where the calculation formula is as follows:
the formula according to which the system state evaluation value Xpgz is generated according to the evaluation parameters is as follows:
In the formula, alpha, beta and gamma are respectively preset proportional coefficients of sound accuracy coefficient Szx, image accuracy coefficient Tzx and Bluetooth stability coefficient Lwx, gamma is larger than beta and larger than alpha is larger than 0, alpha+beta+gamma=3.54, G is a constant correction coefficient, a specific value of the constant correction coefficient can be adjusted and set by a user or is generated by fitting an analysis function, and the specific value of G in the invention is 1.08.
It should be noted that: a person skilled in the art collects a plurality of groups of sample data and sets a corresponding preset scaling factor for each group of sample data; substituting the preset proportionality coefficient, which can be the preset proportionality coefficient and the acquired sample data, into a formula, forming a ternary once equation set by any three formulas, screening the calculated coefficient, taking an average value, and obtaining a value;
The magnitude of the coefficient is a specific numerical value obtained by quantizing each parameter, so that the subsequent comparison is convenient, the magnitude of the coefficient depends on the number of sample data and the corresponding preset proportional coefficient preliminarily set by a person skilled in the art for each group of sample data, that is, the coefficient is preset according to the actual practice, so long as the proportional relation between the parameter and the quantized numerical value is not influenced, and the above description is also adopted for the preset proportional coefficient and the constant correction coefficient described in other formulas.
The result of comparing the system state evaluation value Xpgz with the preset threshold is as follows:
if the system state evaluation value Xpgz exceeds the preset threshold, the system state is stable, and the corresponding truck model to be output can be normally output;
If the system state evaluation value Xpgz does not exceed the preset threshold, the system state is unstable, and an adjustment strategy needs to be executed, wherein the adjustment strategy is as follows: adjusting super parameters of the voice recognition model and the image recognition model to optimize the effect and performance of the model; the method used in adjusting the hyper-parameters of the voice recognition model and the image recognition model is GRID SEARCH, GRID SEARCH to find the best hyper-parameter setting by traversing all possible combinations of hyper-parameters, trying different hyper-parameter values, such as learning rate, batch size and hidden layer size, to find the best model performance using a mesh search method;
In addition, it is also necessary to repair or replace a bluetooth module specifically used in the bluetooth network.
In sound and image recognition models, common super-parameters include:
Learning rate (LEARNING RATE): controlling the step size of the model updating weight at each iteration, a higher learning rate may lead to model divergence, while a lower learning rate may lead to model convergence slower; batch Size (Batch Size): specifying the number of samples used to update the model weights in each iteration, a larger batch size may increase computational efficiency, but may increase memory requirements. Smaller batch sizes may result in unstable direction of updates; hidden layer Size (HIDDEN LAYER Size): the number of neurons of the hidden layer in the neural network model, the choice of hidden layer size depends on the complexity of the data set and the learning capabilities of the model. Smaller hidden layer sizes may result in a under fit, while larger hidden layer sizes may result in an over fit.
By adopting the technical scheme: the system evaluation module is added into the recognition system, so that the stability of the recognition system in actual use is ensured, the sound accuracy coefficient Szx, the image accuracy coefficient Tzx and the Bluetooth stability coefficient Lwx are comprehensively considered in the system evaluation module, the accuracy of generating the system state evaluation value Xpgz is ensured, whether the recognition system is stable in a simulated running state can be rapidly judged after the system state evaluation value Xpgz is compared with a preset threshold value, the result of the recognized truck model can be output on the premise of ensuring the stability of the recognition system, and an adjustment strategy is executed when the recognition system is unstable, so that the management and optimization of the whole recognition system are realized.
Example 2:
based on embodiment 1, the embodiment specifically describes a truck type identification method based on a bluetooth network, which comprises the following steps:
Step one, acquiring audio-video data, which comprises engine starting sound and vehicle appearance images;
Step two, establishing a sound recognition model, extracting sound characteristics in engine starting sound transmitted through a Bluetooth network through a machine learning algorithm, synchronously establishing an image recognition model, and extracting image characteristics in a vehicle appearance image transmitted through the Bluetooth network through an image recognition algorithm;
Step three, a simulation model is established, relevant information of any given truck model is extracted from the simulation model, the relevant information is sound characteristics and image characteristics of the corresponding truck model, and the sound characteristics and the image characteristics of the corresponding truck model are used as simulation signals;
step four, a vehicle model database is built, the vehicle model database comprises sound features and image features corresponding to a plurality of truck models, the analog signals are compared and matched with data in the vehicle model database, and the truck models obtained through matching are mutually verified;
If the matching is successful, if the verification results are consistent, the identification is successful, and the corresponding truck model to be output is obtained; if the matching is successful, if the verification result is inconsistent, the identification is failed, an early warning is sent out, a coping strategy is executed, and the strategy is that the model of the truck is obtained, and the corresponding sound characteristic and image characteristic are input into a vehicle type database;
step five, building a data analysis model after acquiring the evaluation parameters, generating a system state evaluation value Xpgz according to the evaluation parameters, and comparing the system state evaluation value Xpgz with a preset threshold;
If the system state evaluation value Xpgz exceeds a preset threshold, the system state is stable, and the corresponding truck model to be output is normally output; if the system state evaluation value Xpgz does not exceed the preset threshold, the system state is unstable, and an adjustment strategy is needed to be executed, wherein the adjustment strategy is to adjust the super parameters of the voice recognition model and the image recognition model by using a grid search method, and overhaul the Bluetooth network.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application.
Claims (3)
1. A wagon type recognition system based on a bluetooth network, comprising:
The audio-video acquisition module acquires audio-video data, which comprises engine starting sound and vehicle appearance images;
The sound-image acquisition module comprises a sound acquisition unit and an image acquisition unit, the sound acquisition unit is used for acquiring engine starting sound, the image acquisition unit is used for acquiring vehicle appearance images, the sound-image recognition module comprises a sound recognition unit and an image recognition unit, a sound recognition model is built in the sound recognition unit, sound characteristics in engine starting sound transmitted through a Bluetooth network are extracted through a machine learning algorithm, an image recognition model is built in the image recognition unit, and image characteristics in vehicle appearance images transmitted through the Bluetooth network are extracted through an image recognition algorithm;
The audio and video recognition module establishes a corresponding recognition model, processes audio and video data transmitted through a Bluetooth network, and extracts corresponding characteristics for vehicle type recognition;
The simulation module is used for establishing a simulation model, extracting relevant information of any given truck model in the simulation model, wherein the relevant information is sound characteristics and image characteristics of the corresponding truck model, and taking the sound characteristics and the image characteristics of the corresponding truck model as simulation signals;
The database comparison and verification module is used for building a vehicle type database and comprising sound features and image features corresponding to a plurality of truck types, comparing and matching the analog signals with data in the vehicle type database, mutually verifying the truck types obtained by matching, and acquiring the truck types to be output when verification results are consistent; when the matched truck models are mutually verified, the method also comprises the condition that verification results are inconsistent, early warning is required to be sent out, a coping strategy is required to be executed, and the coping strategy is that the truck model is obtained, and corresponding sound characteristics and image characteristics are input into a vehicle type database;
The system evaluation module is used for building a data analysis model after acquiring evaluation parameters, generating a system state evaluation value Xpgz according to the evaluation parameters, comparing the system state evaluation value Xpgz with a preset threshold, and normally outputting the corresponding truck model to be output when the system state evaluation value Xpgz exceeds the preset threshold;
The system evaluation module comprises a preprocessing unit and an evaluation comparison unit, wherein the preprocessing unit is used for acquiring evaluation parameters, the evaluation parameters comprise sound accuracy coefficients Szx, image accuracy coefficients Tzx and Bluetooth stability coefficients Lwx, a data analysis model is built in the evaluation comparison unit, a system state evaluation value Xpgz is generated according to the evaluation parameters, and the system state evaluation value Xpgz is compared with a preset threshold value;
The step of obtaining the sound accuracy factor Szx in the preprocessing unit is as follows:
S101, calculating and acquiring in a voice recognition model, training and testing by inputting a group of voice samples with known labels and using a machine learning algorithm, training the voice recognition model by a training set, and evaluating the accuracy Ac1, the accuracy Pr1 and the recall rate Re1 of the voice recognition model by a testing set;
S102, when calculating the sound accuracy coefficient Szx according to the accuracy Ac1, the accuracy Pr1 and the recall rate Re1, the following formula is adopted:
;
Wherein a1, a2 and a3 are preset proportionality coefficients of an accuracy Ac1, an accuracy Pr1 and a recall Re1 respectively, and a3 is more than a1 and more than a2 is more than 0, and a1+a2+a3=2.64;
The step of acquiring the image accuracy factor Tzx in the preprocessing unit is as follows:
S201, calculating and acquiring an image accuracy coefficient Tzx by using an image recognition model, training and testing by using an image recognition algorithm by inputting a group of image samples with known labels, training the image recognition model by using a training set, and evaluating the accuracy Ac2, the accuracy Pr2 and the recall rate Re2 of the image recognition model by using a testing set;
S202, when calculating an image accuracy coefficient Tzx according to the accuracy Ac2, the accuracy Pr2 and the recall Re2, the following formula is adopted:
;
wherein b1, b2 and b3 are preset proportionality coefficients of an accuracy Ac2, an accuracy Pr2 and a recall Re2 respectively, and b3 is more than b1 and more than b2 is more than 0, a1+a2+a3=2.57;
the step of acquiring the bluetooth stability factor Lwx in the preprocessing unit is as follows:
s301, acquiring transmission rates of different time nodes in T time T represents the number of transmission rates at different time nodes within T time, t=/>Wherein n is a positive integer;
s302, according to the average value of transmission rates in different time nodes in T time Camera frame rate/>To calculate the bluetooth stability factor Lwx, the calculation formula is as follows:
;
the formula according to which the system state evaluation value Xpgz is generated according to the evaluation parameter is as follows:
;
In the method, in the process of the invention, Preset scaling coefficients of sound accuracy factor Szx, image accuracy factor Tzx, and bluetooth stability factor Lwx, respectively, and/>>/>>/>>0,/>=3.54, G is a constant correction coefficient.
2. A bluetooth network based wagon type identification system according to claim 1, wherein: after comparing the system state evaluation value Xpgz with a preset threshold, if the system state evaluation value Xpgz does not exceed the preset threshold, executing an adjustment strategy, wherein the adjustment strategy is to use a grid search method to adjust the super parameters of the voice recognition model and the image recognition model, and overhauling the Bluetooth network.
3. A method for identifying a wagon model based on a bluetooth network, using the system according to any one of claims 1 to 2, characterized in that: the method comprises the following steps:
Step one, acquiring audio-video data, which comprises engine starting sound and vehicle appearance images;
Step two, establishing a sound recognition model, extracting sound characteristics in engine starting sound transmitted through a Bluetooth network through a machine learning algorithm, synchronously establishing an image recognition model, and extracting image characteristics in a vehicle appearance image transmitted through the Bluetooth network through an image recognition algorithm;
Step three, a simulation model is established, relevant information of any given truck model is extracted from the simulation model, the relevant information is sound characteristics and image characteristics of the corresponding truck model, and the sound characteristics and the image characteristics of the corresponding truck model are used as simulation signals;
step four, a vehicle model database is built, the vehicle model database comprises sound features and image features corresponding to a plurality of truck models, the analog signals are compared and matched with data in the vehicle model database, and the truck models obtained through matching are mutually verified;
If the matching is successful, if the verification results are consistent, the identification is successful, and the corresponding truck model to be output is obtained; if the matching is successful, if the verification result is inconsistent, the identification is failed, an early warning is sent out, a coping strategy is executed, and the strategy is that the model of the truck is obtained, and the corresponding sound characteristic and image characteristic are input into a vehicle type database;
step five, building a data analysis model after acquiring the evaluation parameters, generating a system state evaluation value Xpgz according to the evaluation parameters, and comparing the system state evaluation value Xpgz with a preset threshold;
If the system state evaluation value Xpgz exceeds a preset threshold, the system state is stable, and the corresponding truck model to be output is normally output; if the system state evaluation value Xpgz does not exceed the preset threshold, the system state is unstable, and an adjustment strategy is needed to be executed, wherein the adjustment strategy is to adjust the super parameters of the voice recognition model and the image recognition model by using a grid search method, and overhaul the Bluetooth network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311327289.3A CN117392857B (en) | 2023-10-13 | 2023-10-13 | Truck type recognition system and recognition method based on Bluetooth network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311327289.3A CN117392857B (en) | 2023-10-13 | 2023-10-13 | Truck type recognition system and recognition method based on Bluetooth network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117392857A CN117392857A (en) | 2024-01-12 |
CN117392857B true CN117392857B (en) | 2024-06-18 |
Family
ID=89438420
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311327289.3A Active CN117392857B (en) | 2023-10-13 | 2023-10-13 | Truck type recognition system and recognition method based on Bluetooth network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117392857B (en) |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105489019B (en) * | 2015-12-18 | 2018-04-24 | 中山大学 | A kind of traffic throughput monitor system for dividing vehicle based on double-audio signal collection |
CN110168625A (en) * | 2017-05-03 | 2019-08-23 | 华为技术有限公司 | A kind of recognition methods of emergency vehicles and device |
CN107730895B (en) * | 2017-10-25 | 2024-01-30 | 同济大学 | Buried road vehicle type identification system and method |
CN109740478B (en) * | 2018-12-26 | 2023-04-28 | 杨先明 | Vehicle detection and identification method, device, computer equipment and readable storage medium |
CN113614732A (en) * | 2019-03-29 | 2021-11-05 | 索尼集团公司 | Information processing apparatus and information processing method |
CN111680556B (en) * | 2020-04-29 | 2024-06-07 | 平安国际智慧城市科技股份有限公司 | Method, device, equipment and storage medium for identifying traffic gate vehicle type |
US20210103747A1 (en) * | 2020-12-17 | 2021-04-08 | Hassnaa Moustafa | Audio-visual and cooperative recognition of vehicles |
CN115641470A (en) * | 2022-10-17 | 2023-01-24 | 阿波罗智联(北京)科技有限公司 | Method, device and equipment for training classification model and vehicle image classification model |
-
2023
- 2023-10-13 CN CN202311327289.3A patent/CN117392857B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN117392857A (en) | 2024-01-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110164128B (en) | City-level intelligent traffic simulation system | |
US10520947B2 (en) | Machine learning for event detection and classification in autonomous vehicles | |
CN103235933B (en) | A kind of vehicle abnormality behavioral value method based on HMM | |
CN112015153B (en) | System and method for detecting abnormity of sterile filling production line | |
CN107657237A (en) | Car crass detection method and system based on deep learning | |
CN110006672A (en) | Rail vehicle fault monitoring method based on acoustic imaging technology | |
CN114157513B (en) | Vehicle networking intrusion detection method and equipment based on improved convolutional neural network | |
CN112418649A (en) | Building stair pedestrian flow estimation system based on multi-dimensional MEMS inertial sensor | |
CN113932758B (en) | Road surface flatness prediction method and device | |
CN108197731B (en) | Motor vehicle exhaust remote measurement and vehicle inspection result consistency method based on co-training | |
Yu et al. | Pavement roughness index estimation and anomaly detection using smartphones | |
CA3040961A1 (en) | Airport noise classification method and system | |
CN114926825A (en) | Vehicle driving behavior detection method based on space-time feature fusion | |
KR102383051B1 (en) | Method for providing deep learning model based vehicle part testing service | |
US11941986B2 (en) | Methods for characterizing a low-impact vehicle collision using high-rate acceleration data | |
CN113221759A (en) | Road scattering identification method and device based on anomaly detection model | |
CN116758762A (en) | Control method based on big data | |
CN110263836B (en) | Bad driving state identification method based on multi-feature convolutional neural network | |
CN111967308A (en) | Online road surface unevenness identification method and system | |
CN115546742A (en) | Rail foreign matter identification method and system based on monocular thermal infrared camera | |
CN117392857B (en) | Truck type recognition system and recognition method based on Bluetooth network | |
CN109720353A (en) | A kind of driving behavior detection method based on smart phone | |
CN116665155B (en) | Underground trackless man-vehicle getting-off early warning safety management system based on image processing technology | |
CN107539038A (en) | A kind of vehicle tyre pressure method for monitoring state and device | |
CN113064940A (en) | Highway intelligence real-time charging analytic system based on big data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |