CN110807493A - Optimization method and equipment of vehicle classification model - Google Patents
Optimization method and equipment of vehicle classification model Download PDFInfo
- Publication number
- CN110807493A CN110807493A CN201911077094.1A CN201911077094A CN110807493A CN 110807493 A CN110807493 A CN 110807493A CN 201911077094 A CN201911077094 A CN 201911077094A CN 110807493 A CN110807493 A CN 110807493A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- classification model
- classification
- model
- existing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- 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
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Traffic Control Systems (AREA)
Abstract
The method comprises the step of carrying out model training and testing on a known vehicle classification model through a newly added vehicle type and a vehicle image corresponding to the newly added vehicle type and a known vehicle type and a vehicle image corresponding to the known vehicle type to obtain an optimized classification model. The method and the device have the advantages that model iterative optimization is carried out on the existing classification model based on the classification result of each vehicle image acquired in the actual scene, the optimized classification model for classifying the vehicles is obtained, time cost and labor cost of the artificial optimized classification model are saved, accuracy of the optimized classification model for vehicle type prediction of the vehicle images is further improved, and therefore the method and the device are better suitable for users in the actual scene, and experience and satisfaction of the users are improved.
Description
Technical Field
The present application relates to the field of computers, and in particular, to a method and an apparatus for optimizing a vehicle classification model.
Background
In the prior art, in the field of intelligent traffic algorithms, vehicle or license plate classification is an important topic. The vehicle or license plate can be effectively classified, and the subsequent violation behavior judgment is a very important step. However, the actual situation is complex, the vehicle license plates in different regions are different, and some vehicle types or license plate types are rare, so that it is difficult to acquire data of all types in a short time, and it is difficult to train a model suitable for classifying all vehicle license plates.
When the adaptability of the model to new data types is poor, a large amount of data on the site are collected and then data screening and labeling are carried out by technicians, so that the model is trained and further updated to the site. The data transmission processing and the model training return, a large amount of time is needed, the field cannot be updated in time to obtain a better model, meanwhile, manual operation is needed for data labeling again, and therefore the labor cost is increased, the problem that data labeling is inaccurate probably occurs in manual labeling operation, and the model training is not facilitated to obtain the updated and optimized model.
Therefore, the iterative model training is continuously updated, so that the model is more quickly adapted to the classification of the newly added vehicle types, the accuracy of vehicle type prediction is improved, and the vehicle violation behaviors can be better judged.
Disclosure of Invention
An object of the present application is to provide an optimization method for a vehicle classification model, so as to solve the problems in the prior art that the time and labor cost required for the classification model optimization is high and the prediction of the vehicle type by the classification model is not accurate.
According to an aspect of the present application, there is provided a method for optimizing a vehicle classification model, comprising:
obtaining an existing classification model for classifying vehicles, wherein the existing classification model comprises at least two known vehicle types;
classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain a corresponding classification result, wherein the classification result comprises the vehicle image and a corresponding vehicle type;
and carrying out model training and testing again based on the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicles, wherein the optimized classification model comprises at least one newly added vehicle type and at least two known vehicle types.
Further, in the method for optimizing a vehicle classification model, the performing model training and testing again based on the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicle, where the optimized classification model includes at least one newly added vehicle type and the at least two known vehicle types, and includes:
randomly dividing the classification result into a training set and a test set according to a preset quantity proportion;
model training is carried out again based on the training set to obtain a new classification model for classifying the vehicles, wherein the new classification model comprises the at least one newly added vehicle type and the at least two known vehicle types;
and performing model test on the new classification model based on the test set and the existing classification model to obtain an optimized classification model for classifying the vehicles.
Further, in the method for optimizing a vehicle classification model, the performing a model test on the new classification model based on the test set and the existing classification model to obtain an optimized classification model for classifying the vehicle includes:
predicting the vehicle images in the classification results in the test set based on the existing classification model and the new classification model respectively, and counting to obtain a first accuracy of the existing classification model and a second accuracy of the new classification model;
judging whether the first accuracy is greater than the second accuracy; if so, determining a model obtained by performing arithmetic mean on the existing classification model and the new classification model as a model for vehicle classification on the at least two known vehicle types in the optimized classification model, and simultaneously determining the new classification model as a model for vehicle classification on the at least one newly added vehicle type in the optimized classification model to obtain the optimized classification model;
and if not, determining the new classification model as the optimized classification model.
Further, in the method for optimizing a vehicle classification model, predicting vehicle images in each classification result in the test set based on the existing classification model and the new classification model, and obtaining a first accuracy of the existing classification model and a second accuracy of the new classification model by statistics, includes:
predicting the vehicle images in the classification results in the test set respectively based on the existing classification model and the new classification model to obtain a first predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the existing classification model and a second predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the new classification model;
and obtaining a first accuracy of the existing classification model according to the classification results in the test set and a first predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the existing classification model, and obtaining a second accuracy of the new classification model according to the classification results in the test set and a second predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the new classification model.
Further, in the above method for optimizing a vehicle classification model, the classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain a corresponding classification result, where the classification result includes the vehicle image and a vehicle type corresponding to the vehicle image, includes:
classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain each known vehicle type of the at least two known vehicle types as a classification evaluation value of the type of the vehicle image;
determining whether a highest classification evaluation value of at least two of the classification evaluation values is greater than a preset classification evaluation threshold value,
if so, determining the known vehicle type corresponding to the highest classification evaluation value as the type of the vehicle image to obtain a classification result of the vehicle image;
if not, manually classifying the vehicle image again to obtain a classification result of the vehicle image.
Further, the method for optimizing the vehicle classification model further includes:
acquiring a target vehicle image in an actual scene;
inputting the target vehicle image into the optimized classification model, classifying vehicles corresponding to the target vehicle image to obtain various vehicle types of the at least one newly added vehicle type and the at least two known vehicle types which are respectively used as target evaluation values of the types of the vehicles corresponding to the target vehicle image;
and determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value.
Further, in the method for optimizing a vehicle classification model, determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value includes:
determining whether the highest target evaluation value is greater than a preset classification evaluation threshold value,
if so, determining the vehicle type corresponding to the highest target evaluation value as the vehicle type corresponding to the target vehicle image;
if not, manually classifying the target vehicle image again to obtain the type of the vehicle corresponding to the target vehicle image.
Further, the method for optimizing the vehicle classification model further includes: and storing the target vehicle image and the type of the vehicle corresponding to the target vehicle image.
According to another aspect of the present application, there is also provided a computer readable medium having computer readable instructions stored thereon, which, when executed by a processor, cause the processor to implement the method of any one of the above.
According to another aspect of the present application, there is also provided an optimization apparatus of a vehicle classification model, the optimization apparatus including:
one or more processors;
a computer-readable medium for storing one or more computer-readable instructions,
when executed by the one or more processors, cause the one or more processors to implement a method as in any one of the above.
Compared with the prior art, the method and the device have the advantages that the existing classification model for classifying the vehicle is obtained, and the existing classification model comprises at least two known vehicle types; classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain a corresponding classification result, wherein the classification result comprises the vehicle image and a corresponding vehicle type; and carrying out model training and testing again on the basis of the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicles, wherein the optimized classification model comprises at least one newly added vehicle type and at least two known vehicle types, and the known vehicle classification model is subjected to model training and testing through the newly added vehicle type and a vehicle image corresponding to the newly added vehicle type and the known vehicle type and a vehicle image corresponding to the known vehicle type to obtain the optimized classification model. The method and the device have the advantages that model iterative optimization is carried out on the existing classification model based on the classification result of each vehicle image acquired in the actual scene, the optimized classification model for classifying the vehicles is obtained, time cost and labor cost of the artificial optimized classification model are saved, accuracy of the optimized classification model for vehicle type prediction of the vehicle images is further improved, and therefore the method and the device are better suitable for users in the actual scene, and experience and satisfaction of the users are improved.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:
FIG. 1 illustrates a flow chart of a method of optimizing a vehicle classification model according to one aspect of the present application;
FIG. 2 illustrates an optimized classification model training diagram in a method of optimizing a vehicle classification model according to an aspect of the subject application;
FIG. 3 illustrates a schematic diagram of optimized classification model usage in a method of optimizing a vehicle classification model according to an aspect of the subject application;
the same or similar reference numbers in the drawings identify the same or similar elements.
Detailed Description
The present application is described in further detail below with reference to the attached figures.
In a typical configuration of the present application, the terminal, the device serving the network, and the trusted party each include one or more processors (e.g., Central Processing Units (CPUs)), input/output interfaces, network interfaces, and memory.
The Memory may include volatile Memory in a computer readable medium, Random Access Memory (RAM), and/or nonvolatile Memory such as Read Only Memory (ROM) or flash Memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, Phase-Change RAM (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), flash Memory or other Memory technology, Compact Disc Read-Only Memory (CD-ROM), Digital Versatile Disc (DVD) or other optical storage, magnetic cassette tape, magnetic tape storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory computer readable media (transmyedia), such as modulated data signals and carrier waves.
Fig. 1 shows a flowchart of a method for optimizing a vehicle classification model according to an aspect of the present application, the method is applied to a vehicle type detection process, and the method includes steps S11, S12, and S13, where the method specifically includes:
step S11, an existing classification model for classifying the vehicle is obtained, the existing classification model including at least two known vehicle types. Here, the existing classification model may identify the vehicle picture of the existing vehicle type and obtain the type to which the vehicle corresponding to the vehicle picture belongs.
For example, an existing classification model M1 for classifying vehicles is obtained, where the existing classification model M1 includes five known vehicle types, i.e., a normal car, a police car, an ambulance, a bus, and a truck, and the sequence numbers of the known vehicle types are 0, 1, 2, 3, and 4, respectively, that is, "0" is used to indicate the vehicle type of the normal car.
Step S12, based on the existing classification model, vehicle classification is carried out on each vehicle image in the actual scene to obtain a corresponding classification result, and the classification result comprises the vehicle image and a corresponding vehicle type. For example, based on the existing classification model M1, vehicle classification is performed on each vehicle image P1, P2, and P3.... Pn in the actual scene, and classification results corresponding to each vehicle image P1, P2, and P3.... Pn in sequence are obtained: r1, R2, R3.. Rn, where n is a positive integer greater than or equal to 1 for indicating the number of vehicle images acquired in the actual scene.
And step S13, carrying out model training and testing again based on the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicles, wherein the optimized classification model comprises at least one newly added vehicle type and at least two known vehicle types. The optimized classification model can identify and predict the vehicle types of the vehicle images corresponding to the newly added vehicle types and the vehicle images corresponding to the known vehicle types, iterative optimization of the existing classification model is achieved, a large amount of time cost and labor cost are saved, and accuracy of the optimized classification model for predicting the vehicle types of the vehicle images is improved, so that the optimized classification model is better suitable for users, and satisfaction of the users on vehicle classification products is further improved.
It should be noted that the new vehicle type needs to be classified by a user, and the vehicle image classified by the user and the new vehicle type corresponding to the vehicle image are stored correspondingly, so as to be used for training and testing during optimization of the existing classification model in the following. The data is directly marked when the user examines the data (such as the vehicle image and the corresponding vehicle type) with poor classification results predicted by the existing classification model, so that the labor cost for marking the data again is saved, the accuracy of controlling the marking of the data by the user is improved, and a training set is cleaner during subsequent model optimization.
And in the steps from the step S11 to the step S13, model training and testing are carried out on the known vehicle classification model through the newly added vehicle type and the vehicle image corresponding to the newly added vehicle type and the known vehicle type and the vehicle image corresponding to the newly added vehicle type, so as to obtain the optimized classification model. The method and the device have the advantages that model iterative optimization is carried out on the existing classification model based on the classification result of each vehicle image acquired in the actual scene, the optimized classification model for classifying the vehicles is obtained, time cost and labor cost of the artificial optimized classification model are saved, accuracy of the optimized classification model for vehicle type prediction of the vehicle images is further improved, and therefore the method and the device are better suitable for users in the actual scene, and experience and satisfaction of the users are improved.
For example, an existing classification model M1 for classifying vehicles is obtained, where the existing classification model M1 includes five known vehicle types, i.e., a common car, a police car, an ambulance, a bus, and a truck, and the corresponding serial numbers in sequence are 0, 1, 2, 3, and 4, respectively. Based on the existing classification model M1, vehicle classification is performed on each vehicle image P1, P2, and P3.... No. Pn in the actual scene, and classification results R1, R2, and R3.. No. Rn corresponding to each vehicle image P1, P2, and P3.. No. Pn are obtained. And (4) training and testing the model again based on the classification results R1, R2 and R3. Here, the classification results R1, R2, and R3.. to Rn include not only the known vehicle types and vehicle images thereof, but also newly added vehicle types and vehicle images thereof, which are classified and predicted by the existing classification model M1, and then are determined and stored by manual inspection. In a preferred embodiment of the present application, if the optimized classification model M (optimal) includes two newly added vehicle types and five known vehicle types, where the two newly added vehicle types are a motorcycle and a three-wheeled vehicle, respectively, the corresponding serial numbers are respectively: 5 and 6, so that in the actual scene, the optimized classification model M (optimal) can not only identify and predict the corresponding vehicle types of the vehicle images respectively corresponding to the common cars, police cars, ambulances, buses and trucks, but also identify and predict the corresponding vehicle types of the vehicle images respectively corresponding to the motorcycles and the tricycles, thereby realizing the iterative optimization of the existing classification model, saving a large amount of time and labor cost for model optimization training and testing, and improving the accuracy of the optimized classification model for predicting the vehicle types of the vehicle images, thereby better adapting to the use of users and improving the satisfaction degree of the users on the vehicle classification products of the application.
Following the above embodiment of the present application, the step S12, based on the existing classification model, performs vehicle classification on each vehicle image in an actual scene to obtain a corresponding classification result, where the classification result includes the vehicle image and a vehicle type corresponding to the vehicle image, and includes:
classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain each known vehicle type of the at least two known vehicle types as a classification evaluation value of the type of the vehicle image; here, the classification evaluation value indicates the possibility that each known vehicle type is a type of a vehicle corresponding to a vehicle image acquired in an actual scene, and the possibility may be represented by a score system or a probability value such as a scale.
Determining whether a highest classification evaluation value of at least two of the classification evaluation values is greater than a preset classification evaluation threshold value,
if so, determining the known vehicle type corresponding to the highest classification evaluation value as the type of the vehicle image to obtain a classification result of the vehicle image;
if not, manually classifying the vehicle image again to obtain a classification result of the vehicle image.
For example, as shown in fig. 2, in the process of classifying the vehicle types of the vehicle images in the actual scene, if the vehicle image is the vehicle image a, the vehicle image a to be recognized acquired from the actual scene is classified according to the existing classification model M1, and each known vehicle type of the known vehicle types with the vehicle type number of 0-4 is obtained as the classification evaluation values score0, score1, score2, score3 and score4 of the vehicle type corresponding to the vehicle image a in sequence: 60. 70, 80, 93, and 50, wherein the classification evaluation score1 is used to indicate a known vehicle type: the possibility of the police car as the type of the vehicle to which the vehicle image a corresponds, and the classification evaluation value score4 is used to indicate the known vehicle type: the possibility of a truck as the type of vehicle corresponding to the vehicle image a. These five classification evaluation values are judged: score0, score1, score2, socre3, and score4 correspond in sequence to: 60. if the highest classification evaluation value socre3 ═ 93 in 70, 80, 93 and 50 is greater than a preset classification evaluation threshold value T ═ 90 (in a preferred embodiment of the present application, the preset classification evaluation threshold value T is preferably 90, although any other value may be included in the protection range of the present application), that is, socre3 > T, the known vehicle type corresponding to the highest classification evaluation value socre3 is determined: the bus is determined as the type of the vehicle corresponding to the vehicle image A, so as to obtain a classification result of the vehicle image A, that is, the vehicle corresponding to the vehicle image A belongs to the bus, and the classification result is that the vehicle image A and the type of the vehicle corresponding to the vehicle image A are: a bus. For another example, in another preferred embodiment of the present application, if the preset classification evaluation threshold T is preferably 95, the five classification evaluation values are determined as follows: score0, score1, score2, socre3, and score4 correspond in sequence to: 60. if the highest classification evaluation value socre3 ═ 93 in 70, 80, 93 and 50 is greater than the preset classification evaluation threshold value T ═ 95, and the highest classification evaluation value socre3 ═ 93 is less than the preset classification evaluation threshold value T ═ 95, that is socre3 < T, then manually re-classifying the vehicle image a to obtain the classification result of the vehicle image a, and if the result of manually classifying the vehicle image a by the vehicle is: and the motorcycle, the obtained classification result of the vehicle image A is as follows: vehicle image a and its corresponding vehicle type: a motorcycle.
Next to the above embodiment of the present application, the step S13 performs model training and testing again based on the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicle, where the optimized classification model includes at least one new vehicle type and at least two known vehicle types, and specifically includes:
step S131 (not shown), randomly dividing the classification result into a training set and a testing set according to a preset quantity ratio. Here, the preset quantity ratio may be any ratio, and in a preferred embodiment of the present application, the preset quantity ratio is preferably 10: 1, randomly dividing the classification result into 11 parts, wherein 10 parts constitute a training set, and the remaining 1 part constitutes a testing set, and certainly, the serial numbers of the vehicle image and the corresponding vehicle type in each classification result are still unchanged.
Step S132 (not shown), performing model training again based on the training set to obtain a new classification model for classifying the vehicle, where the new classification model includes the at least one new vehicle type and the at least two known vehicle types. The new classification model of the classification recognition of the vehicle images respectively corresponding to the newly added vehicle type and the known vehicle type is obtained through model training, the existing classification model is optimized, a large amount of time and cost are saved, and the classification recognition of the vehicle type of the vehicle corresponding to the vehicle image corresponding to the newly added vehicle type is completed.
Step S133 (not shown), performing model testing on the new classification model based on the test set and the existing classification model to obtain an optimized classification model for classifying the vehicle. The method further completes the test of the new classification model so as to obtain a more optimized vehicle classification model, namely the optimized classification model, realizes the iterative optimization of the existing classification model, saves a large amount of time and cost, and simultaneously improves the accuracy of the classification model in vehicle type prediction of the vehicle image, thereby better adapting to the use of users and improving the satisfaction degree of the users on products.
For example, after the classification results with the sequence numbers of 0-6 are stored to a certain data volume, all the classification results are classified according to the following relation of 10: 1, randomly dividing into a training set U1 and a testing set U2, then taking an existing classification model M1 as a basic training model, and slightly modifying a network structure corresponding to the existing classification model M1; then training an existing classification model M1 based on the training set U1 to obtain a trained existing classification model M2; and then performing model fine-turning on the trained existing classification model M2 according to a test set U2, and finally finding out the best trained optimized classification model M (optimal), thereby realizing iterative optimization of the classification model, saving a large amount of time and cost, and simultaneously improving the accuracy of the classification model for predicting the vehicle type of the vehicle image, so that the method is better suitable for the use of a user, and improves the satisfaction degree of the user on the product.
Next to the above embodiment of the present application, the step S133 performs model testing on the new classification model based on the test set and the existing classification model to obtain an optimized classification model for classifying the vehicle, including:
step S1331, predicting the vehicle images in the classification results in the test set based on the existing classification model and the new classification model, and counting to obtain a first accuracy of the existing classification model and a second accuracy of the new classification model. The method synthesizes the prediction results of the existing classification model and the new classification model to obtain the first accuracy of the existing classification model and the second accuracy of the new classification model, so that the prediction results can be judged to obtain the most accurate prediction result, and the high-accuracy optimized classification model can be obtained.
Step S1332, determining whether the first accuracy is greater than the second accuracy; if so, determining a model obtained by performing arithmetic mean on the existing classification model and the new classification model as a model for vehicle classification on the at least two known vehicle types in the optimized classification model, and simultaneously determining the new classification model as a model for vehicle classification on the at least one newly added vehicle type in the optimized classification model to obtain the optimized classification model;
and if not, determining the new classification model as the optimized classification model. The method realizes the optimization of the classification model, and simultaneously improves the accuracy of the classification model for predicting the vehicle type of the vehicle image, thereby better adapting to the use of users and improving the satisfaction degree of the users to products.
It should be noted that the existing classification model M1 can well identify and classify the vehicle types with the sequence numbers of 0-4, but the newly added vehicle types with the sequence numbers of 5-6 do not participate in the training of the existing classification model M1, so the existing classification model M1 cannot classify the vehicle types with the sequence numbers of 5-6. The training set of the new classification model M2 contains vehicle type data with serial numbers of 0-6, so the classification result of the new classification model M2 on the vehicle types with serial numbers of 0-6 is better than that of the existing classification model M1 on the whole, but the classification result of the new classification model M2 on the vehicle types with serial numbers of 0-4 is not necessarily better than that of the existing classification model M1. And combining the results of the existing classification model M1 and the new classification model M2, so that the models are more important for the classification results of the vehicle types with the sequence numbers of 0-6 respectively.
For example, as shown in fig. 3, vehicle images with vehicle type numbers of 0 to 4 in a test set are taken, vehicle images in each classification result in the test set U2 are predicted based on an existing classification model M1 and a new classification model M2, and a first accuracy1 of the existing classification model and a second accuracy2 of the new classification model are obtained through statistics. Then, it is determined whether the first accuracy accurve 1 is greater than the second accuracy accurve 2.
If yes, namely the accuracy1 is larger than the accuracy2, the result of the existing classification model M1 is better than that of the new classification model M2 for the classification result with the vehicle type serial number of 0-4, but the result of the new classification model M2 for the classification result with the serial number of 5-6 is higher than that of the existing classification model M1, and the results of the two models are integrated to judge the result. Determining a model obtained by performing arithmetic mean on the existing classification model M1 and the new classification model M2 as a model used for vehicle classification on the at least two known vehicle types in the optimized classification model M (optimal), and determining the new classification model M2 as a model used for vehicle classification on the at least one newly added vehicle type in the optimized classification model M (optimal) to obtain the optimized classification model M (optimal), namely, when the serial number of the vehicle types is 0-4, the optimized classification model M (optimal) is: (M1+ M2)/2, when the serial number of the vehicle type is 5-6, the optimized classification model M (optimal) is as follows: m2. Namely, when the same vehicle image B is input, two models are respectively used: when the existing classification model M1 and the new classification model M2 are judged, a one-dimensional array { (score01, 0), (score11, 1), (score21, 2), (score31, 3), (score41, 4) } containing 5 elements is obtained by using the existing classification model M1, wherein the 5 elements sequentially correspond to the classification evaluation values of the vehicle types with the sequence numbers of 0 to 4, and a one-dimensional array { (score02, 0), (score12, 1), (score22, 2), (score32, 3), (score42, 4), (score52, 5), (score62, 6) } containing 7 elements sequentially correspond to the evaluation values of the vehicle types with the sequence numbers of 0 to 6, and the first 5 elements of the two arrays have the same meaning and respectively represent the evaluation values of the vehicle types with the sequence numbers of 0 to 4. And correspondingly adding the first 5 classification evaluated values of the two arrays and dividing the result by 2 to be used as the first 5 classification evaluated values of the new array, and using the last two values of the array obtained by the new classification model M2 as the last classification evaluated values of the new array to obtain a new one-dimensional array containing 7 elements { ((score01+ score02)/2, 0), ((score11+ score12)/2, 1), ((score21+ score22)/2, 2), ((score31+ score32)/2, 3), ((score41+ score42)/2, 4), (score52, 5), (score62, 6) }, so as to obtain the optimal classification prediction result. If not, the new classification model M2 is determined to be the optimized classification model M (optimal). The method and the device realize the optimization of the classification model and improve the accuracy of the classification model in predicting the vehicle type of the vehicle image, thereby better adapting to the use of users and improving the satisfaction degree of the users to products.
Next to the above embodiment of the present application, the step S1331 of predicting the vehicle images in the classification results in the test set based on the existing classification model and the new classification model, and obtaining a first accuracy of the existing classification model and a second accuracy of the new classification model through statistics, including:
and predicting the vehicle images in the classification results in the test set respectively based on the existing classification model and the new classification model to obtain a first predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the existing classification model and a second predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the new classification model, so that the accuracy of the existing classification model and the accuracy of the new classification model can be obtained in the next step, the classification model is optimized, and the accuracy of the classification model prediction results is improved.
And obtaining a first accuracy of the existing classification model according to the classification results in the test set and a first predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the existing classification model, and obtaining a second accuracy of the new classification model according to the classification results in the test set and a second predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the new classification model. In this case, further optimization of the classification model is facilitated.
For example, 10 classification results, namely R101, R102, … … and R110, are contained in the test set U2, and the vehicle types of the vehicle images in the 10 classification results are predicted by using the existing classification model M1 and the new classification model M2 respectively; the result of the first prediction of the existing classification model M1: m101, M102, …, M110, the result of the first prediction comprising a first predicted vehicle type corresponding to a vehicle image; second predicted outcome of new classification model M2: m201, M202, … and M210, wherein the result of the second prediction comprises a second predicted vehicle type corresponding to the vehicle image; and respectively comparing the results with the predicted vehicle types in the real classification results R101, R102, … … and R110 to obtain a first predicted accuracy accurve 1 corresponding to the existing classification model M1 and a second predicted accuracy accurve 2 corresponding to the new classification model M2 so as to test the classification models subsequently to obtain an optimized classification model M (optimal) with higher accuracy.
Following the above embodiments of the present application, a method for optimizing a vehicle classification model further includes:
acquiring a target vehicle image in an actual scene;
inputting the target vehicle image into the optimized classification model, classifying vehicles corresponding to the target vehicle image to obtain various vehicle types of the at least one newly added vehicle type and the at least two known vehicle types which are respectively used as target evaluation values of the types of the vehicles corresponding to the target vehicle image; here, the target evaluation value represents a possibility that each vehicle type is a type of the target image vehicle, and may be represented as a score or a proportional equal probability in the present application.
And determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value.
For example, target vehicle images P1, P2, P3. The target vehicle images P, P3.... Pn are input into the optimized classification model M (excellent), and the vehicles corresponding to the target vehicle images P, P3.... Pn are classified, so that target evaluation values P { score ', score', score ', sock', score ', score', score ', P { score', score ', sock', score ', score' }, P { score ', score', sock. If the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P1 is score1 ', the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P2 is score4 ', and the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P3 is score5 '. And determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value, so that the accuracy of predicting the vehicle type of the classification model is improved.
Further, determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value includes:
determining whether the highest target evaluation value is greater than a preset classification evaluation threshold value,
if so, determining the vehicle type corresponding to the highest target evaluation value as the vehicle type corresponding to the target vehicle image;
if not, manually classifying the target vehicle image again to obtain the type of the vehicle corresponding to the target vehicle image. Here, the target vehicle image and the type of the vehicle corresponding thereto are stored. The user can classify the target vehicle image again, so that a large amount of labor and time are saved, and cleaner data can be obtained so as to continuously optimize the classification model. When the manually classified data is accumulated to a certain amount, all the data are put together to be randomly disordered (the images and the corresponding serial numbers are unchanged), and are divided into a training set and a testing set according to a certain proportion. And taking an optimized classification model M (optimal) as a basic model, slightly modifying a corresponding network structure, performing model fine-turning by using a prepared training set and a test set, and finally obtaining a more optimized classification model, thereby realizing iterative optimization of the classification model, saving a large amount of time and cost, and simultaneously improving the accuracy of the classification model for predicting the vehicle type of the vehicle image, so that the method is better suitable for the use of a user and improves the satisfaction degree of the user on the product.
For example, the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P1 is score1 ' 87, the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P2 is score4 ' 95, and the highest target evaluation value of the type of vehicle corresponding to the target vehicle image P3 is score5 ' 93. Then, it is determined whether or not each of the above-described highest target evaluation values is greater than a preset classification evaluation threshold Q of 90. If the P1(score1 ═ 87) < Q ═ 90, manually re-classifying the target vehicle image P1 to obtain the type of the vehicle corresponding to the target vehicle image P1; p2(score4 '═ 95) > Q ═ 90, the vehicle type corresponding to the highest target evaluation value score 4' ═ 95 is determined as the type of the vehicle corresponding to the target vehicle image; p3(score5 '═ 93) > Q ═ 90, the vehicle type corresponding to the highest target evaluation value score 5' is determined as the type of the vehicle corresponding to the target vehicle image P3; .... Pn (score 1' ═ 50) < Q ═ 90, then manually re-classifying the target vehicle image Pn to obtain the type of the vehicle corresponding to the target vehicle image Pn, and sequentially judging and comparing according to the method to obtain the types of the vehicles corresponding to the target vehicle images P1, P2 and P3.... Pn; and then, the target vehicle image and the type of the vehicle corresponding to the target vehicle image are stored, so that a large amount of labor and time are saved, cleaner data can be obtained so as to continuously optimize the classification model subsequently, and the accuracy of the classification model for predicting the vehicle type of the vehicle image is improved.
Following the above embodiments of the present application, a method for optimizing a vehicle classification model further includes:
and storing the target vehicle image and the type of the vehicle corresponding to the target vehicle image so as to continuously optimize the classification model in the following process and improve the accuracy of the classification model for predicting the vehicle type of the vehicle image.
For example, the target vehicle image P1 obtained by manually re-classifying the target vehicle image P1, the.
According to another aspect of the present application, there is also provided a computer readable medium having stored thereon computer readable instructions, which, when executed by a processor, cause the processor to implement the method of controlling user base alignment as described above.
According to another aspect of the present application, there is also provided an optimization apparatus of a vehicle classification model, characterized by comprising:
one or more processors;
a computer-readable medium for storing one or more computer-readable instructions,
when executed by the one or more processors, cause the one or more processors to implement a method of controlling user base station on a device as described above.
Here, for details of each embodiment of the device, reference may be specifically made to corresponding parts of the embodiment of the method for controlling user base pairing at the device side, and details are not described here.
In summary, the optimization classification model is obtained by performing model training and testing on the known vehicle classification model through the newly added vehicle type and the vehicle image corresponding to the newly added vehicle type and the known vehicle type and the vehicle image corresponding to the newly added vehicle type. The method and the device have the advantages that model iterative optimization is carried out on the existing classification model based on the classification result of each vehicle image acquired in the actual scene, the optimized classification model for classifying the vehicles is obtained, time cost and labor cost of the artificial optimized classification model are saved, accuracy of the optimized classification model for vehicle type prediction of the vehicle images is further improved, and therefore the method and the device are better suitable for users in the actual scene, and experience and satisfaction of the users are improved.
It should be noted that the present application may be implemented in software and/or a combination of software and hardware, for example, implemented using Application Specific Integrated Circuits (ASICs), general purpose computers or any other similar hardware devices. In one embodiment, the software programs of the present application may be executed by a processor to implement the steps or functions described above. Likewise, the software programs (including associated data structures) of the present application may be stored in a computer readable recording medium, such as RAM memory, magnetic or optical drive or diskette and the like. Additionally, some of the steps or functions of the present application may be implemented in hardware, for example, as circuitry that cooperates with the processor to perform various steps or functions.
In addition, some of the present application may be implemented as a computer program product, such as computer program instructions, which when executed by a computer, may invoke or provide methods and/or techniques in accordance with the present application through the operation of the computer. Program instructions which invoke the methods of the present application may be stored on a fixed or removable recording medium and/or transmitted via a data stream on a broadcast or other signal-bearing medium and/or stored within a working memory of a computer device operating in accordance with the program instructions. An embodiment according to the present application comprises an apparatus comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein the computer program instructions, when executed by the processor, trigger the apparatus to perform a method and/or a solution according to the aforementioned embodiments of the present application.
It will be evident to those skilled in the art that the present application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned. Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units or means recited in the apparatus claims may also be implemented by one unit or means in software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.
Claims (10)
1. A method of optimizing a vehicle classification model, the method comprising:
obtaining an existing classification model for classifying vehicles, wherein the existing classification model comprises at least two known vehicle types;
classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain a corresponding classification result, wherein the classification result comprises the vehicle image and a corresponding vehicle type;
and carrying out model training and testing again based on the classification result and the existing classification model to obtain an optimized classification model for classifying the vehicles, wherein the optimized classification model comprises at least one newly added vehicle type and at least two known vehicle types.
2. The method of claim 1, wherein the model training and testing are repeated based on the classification result and the existing classification model to obtain an optimized classification model for classifying vehicles, wherein the optimized classification model comprises at least one new vehicle type and the at least two known vehicle types, and comprises:
randomly dividing the classification result into a training set and a test set according to a preset quantity proportion;
model training is carried out again based on the training set to obtain a new classification model for classifying the vehicles, wherein the new classification model comprises the at least one newly added vehicle type and the at least two known vehicle types;
and performing model test on the new classification model based on the test set and the existing classification model to obtain an optimized classification model for classifying the vehicles.
3. The method of claim 2, wherein said model testing said new classification model based on said test set and said existing classification models results in an optimized classification model for classifying vehicles comprising:
predicting the vehicle images in the classification results in the test set based on the existing classification model and the new classification model respectively, and counting to obtain a first accuracy of the existing classification model and a second accuracy of the new classification model;
judging whether the first accuracy is greater than the second accuracy; if so, determining a model obtained by performing arithmetic mean on the existing classification model and the new classification model as a model for vehicle classification on the at least two known vehicle types in the optimized classification model, and simultaneously determining the new classification model as a model for vehicle classification on the at least one newly added vehicle type in the optimized classification model to obtain the optimized classification model;
and if not, determining the new classification model as the optimized classification model.
4. The method of claim 3, wherein the predicting the vehicle images in the classification results of the test set based on the existing classification model and the new classification model, respectively, and statistically obtaining a first accuracy of the existing classification model and a second accuracy of the new classification model comprises:
predicting the vehicle images in the classification results in the test set respectively based on the existing classification model and the new classification model to obtain a first predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the existing classification model and a second predicted vehicle type corresponding to the vehicle images in the classification results in the test set predicted by the new classification model;
and obtaining a first accuracy of the existing classification model according to the classification results in the test set and a first predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the existing classification model, and obtaining a second accuracy of the new classification model according to the classification results in the test set and a second predicted vehicle type corresponding to the vehicle image in the classification results in the test set predicted by the new classification model.
5. The method according to any one of claims 1 to 4, wherein the vehicle classification of each vehicle image in an actual scene based on the existing classification model to obtain a corresponding classification result, wherein the classification result includes the vehicle image and a corresponding vehicle type thereof, and comprises:
classifying vehicles of each vehicle image in an actual scene based on the existing classification model to obtain each known vehicle type of the at least two known vehicle types as a classification evaluation value of the type of the vehicle image;
determining whether a highest classification evaluation value of at least two of the classification evaluation values is greater than a preset classification evaluation threshold value,
if so, determining the known vehicle type corresponding to the highest classification evaluation value as the type of the vehicle image to obtain a classification result of the vehicle image;
if not, manually classifying the vehicle image again to obtain a classification result of the vehicle image.
6. The method of any of claims 1-5, wherein the method further comprises:
acquiring a target vehicle image in an actual scene;
inputting the target vehicle image into the optimized classification model, classifying vehicles corresponding to the target vehicle image to obtain various vehicle types of the at least one newly added vehicle type and the at least two known vehicle types which are respectively used as target evaluation values of the types of the vehicles corresponding to the target vehicle image;
and determining the type of the vehicle corresponding to the target vehicle image according to the highest target evaluation value.
7. The method of claim 6, wherein the determining the type of the vehicle to which the target vehicle image corresponds from the highest target evaluation value comprises:
determining whether the highest target evaluation value is greater than a preset classification evaluation threshold value,
if so, determining the vehicle type corresponding to the highest target evaluation value as the vehicle type corresponding to the target vehicle image;
if not, manually classifying the target vehicle image again to obtain the type of the vehicle corresponding to the target vehicle image.
8. The method of claim 7, wherein the method further comprises:
and storing the target vehicle image and the type of the vehicle corresponding to the target vehicle image.
9. A computer readable medium having computer readable instructions stored thereon, which, when executed by a processor, cause the processor to implement the method of any one of claims 1 to 8.
10. An apparatus for optimizing a vehicle classification model, the apparatus comprising:
one or more processors;
a computer-readable medium for storing one or more computer-readable instructions,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911077094.1A CN110807493A (en) | 2019-11-06 | 2019-11-06 | Optimization method and equipment of vehicle classification model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911077094.1A CN110807493A (en) | 2019-11-06 | 2019-11-06 | Optimization method and equipment of vehicle classification model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110807493A true CN110807493A (en) | 2020-02-18 |
Family
ID=69501413
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911077094.1A Pending CN110807493A (en) | 2019-11-06 | 2019-11-06 | Optimization method and equipment of vehicle classification model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110807493A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111709371A (en) * | 2020-06-17 | 2020-09-25 | 腾讯科技(深圳)有限公司 | Artificial intelligence based classification method, device, server and storage medium |
CN112180913A (en) * | 2020-09-01 | 2021-01-05 | 芜湖酷哇机器人产业技术研究院有限公司 | Special vehicle identification method |
CN112448868A (en) * | 2020-12-02 | 2021-03-05 | 新华三人工智能科技有限公司 | Network traffic data identification method, device and equipment |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050100209A1 (en) * | 2003-07-02 | 2005-05-12 | Lockheed Martin Corporation | Self-optimizing classifier |
CN104281679A (en) * | 2014-09-30 | 2015-01-14 | 东软集团股份有限公司 | Goods classification method and goods classification device both based on image features |
CN106203330A (en) * | 2016-07-08 | 2016-12-07 | 西安理工大学 | A kind of vehicle classification method based on convolutional neural networks |
CN107358257A (en) * | 2017-07-07 | 2017-11-17 | 华南理工大学 | Under a kind of big data scene can incremental learning image classification training method |
CN108256550A (en) * | 2017-12-14 | 2018-07-06 | 北京木业邦科技有限公司 | A kind of timber classification update method and device |
CN108830332A (en) * | 2018-06-22 | 2018-11-16 | 安徽江淮汽车集团股份有限公司 | A kind of vision vehicle checking method and system |
CN109948643A (en) * | 2019-01-21 | 2019-06-28 | 东南大学 | A kind of type of vehicle classification method based on deep layer network integration model |
US20190204834A1 (en) * | 2018-01-04 | 2019-07-04 | Metawave Corporation | Method and apparatus for object detection using convolutional neural network systems |
CN110210560A (en) * | 2019-05-31 | 2019-09-06 | 北京市商汤科技开发有限公司 | Increment training method, classification method and the device of sorter network, equipment and medium |
-
2019
- 2019-11-06 CN CN201911077094.1A patent/CN110807493A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050100209A1 (en) * | 2003-07-02 | 2005-05-12 | Lockheed Martin Corporation | Self-optimizing classifier |
CN104281679A (en) * | 2014-09-30 | 2015-01-14 | 东软集团股份有限公司 | Goods classification method and goods classification device both based on image features |
CN106203330A (en) * | 2016-07-08 | 2016-12-07 | 西安理工大学 | A kind of vehicle classification method based on convolutional neural networks |
CN107358257A (en) * | 2017-07-07 | 2017-11-17 | 华南理工大学 | Under a kind of big data scene can incremental learning image classification training method |
CN108256550A (en) * | 2017-12-14 | 2018-07-06 | 北京木业邦科技有限公司 | A kind of timber classification update method and device |
US20190204834A1 (en) * | 2018-01-04 | 2019-07-04 | Metawave Corporation | Method and apparatus for object detection using convolutional neural network systems |
CN108830332A (en) * | 2018-06-22 | 2018-11-16 | 安徽江淮汽车集团股份有限公司 | A kind of vision vehicle checking method and system |
CN109948643A (en) * | 2019-01-21 | 2019-06-28 | 东南大学 | A kind of type of vehicle classification method based on deep layer network integration model |
CN110210560A (en) * | 2019-05-31 | 2019-09-06 | 北京市商汤科技开发有限公司 | Increment training method, classification method and the device of sorter network, equipment and medium |
Non-Patent Citations (2)
Title |
---|
XUEZHI WEN ET AL.: "A rapid learning algorithm for vehicle classification", 《INFORMATION SCIENCES》 * |
薛峰 等: "结合改进的SVM和随机森林算法车标分类识别", 《计算机工程与设计》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111709371A (en) * | 2020-06-17 | 2020-09-25 | 腾讯科技(深圳)有限公司 | Artificial intelligence based classification method, device, server and storage medium |
CN111709371B (en) * | 2020-06-17 | 2023-12-22 | 腾讯科技(深圳)有限公司 | Classification method, device, server and storage medium based on artificial intelligence |
CN112180913A (en) * | 2020-09-01 | 2021-01-05 | 芜湖酷哇机器人产业技术研究院有限公司 | Special vehicle identification method |
CN112448868A (en) * | 2020-12-02 | 2021-03-05 | 新华三人工智能科技有限公司 | Network traffic data identification method, device and equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
van Schagen et al. | The potential benefits of naturalistic driving for road safety research: Theoretical and empirical considerations and challenges for the future | |
CN110807493A (en) | Optimization method and equipment of vehicle classification model | |
CN110378258B (en) | Image-based vehicle seat information detection method and device | |
CN110473211B (en) | Method and equipment for detecting number of spring pieces | |
CN113287120A (en) | Vehicle driving environment abnormity monitoring method and device, electronic equipment and storage medium | |
CN110751012A (en) | Target detection evaluation method and device, electronic equipment and storage medium | |
CN114764774A (en) | Defect detection method, device, electronic equipment and computer readable storage medium | |
CN114626477A (en) | Target detection method, device and equipment | |
CN112434717B (en) | Model training method and device | |
CN112270319B (en) | Event labeling method and device and electronic equipment | |
CN111369790B (en) | Vehicle passing record correction method, device, equipment and storage medium | |
CN112686125A (en) | Vehicle type determination method and device, storage medium and electronic device | |
CN116662667A (en) | Vehicle-mounted message pushing method based on user tag and related equipment | |
CN114724107B (en) | Image detection method, device, equipment and medium | |
CN111832599A (en) | Gas station prediction method based on machine learning random forest | |
CN116152637A (en) | Evaluation method of automatic driving perception model, computer equipment and storage medium | |
CN116052429A (en) | Road section congestion identification method and device, electronic equipment and medium | |
CN112597835B (en) | Driving behavior evaluation method and device, electronic equipment and readable storage medium | |
CN112289026B (en) | Vehicle route restoration method, equipment and medium | |
CN114119953A (en) | Method for quickly positioning and correcting license plate, storage medium and equipment | |
CN114169247A (en) | Method, device and equipment for generating simulated traffic flow and computer readable storage medium | |
CN113159640A (en) | Method, device, equipment and medium for determining evaluation index weight | |
CN112533208A (en) | Model training method, false terminal identification method and device, and electronic device | |
US20240135679A1 (en) | Method for classifying images and electronic device | |
CN118331865A (en) | Display effect automatic evaluation method and device, electronic equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
AD01 | Patent right deemed abandoned | ||
AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20230203 |