CN116258919A

CN116258919A - Method, device, medium and program product for acquiring data

Info

Publication number: CN116258919A
Application number: CN202111497797.7A
Authority: CN
Inventors: 姚人杰; 段宁; 张超; 张行剑
Original assignee: Mobility Asia Smart Technology Co Ltd
Current assignee: Mobility Asia Smart Technology Co Ltd
Priority date: 2021-12-09
Filing date: 2021-12-09
Publication date: 2023-06-13

Abstract

According to example embodiments of the present disclosure, methods, devices, media, and program products for collecting data are provided. In the method, data is received from an input device of a vehicle. Reference data is obtained from a database stored on the vehicle, the reference data being associated with a model deployed on the vehicle for assisting driving. In response to similarity between the data and the reference data being above a threshold, the data is added to a training set for training the model. According to the scheme, the data similar to the reference data can be acquired in a targeted manner to train a model for assisting driving, so that the efficiency of data acquisition is improved.

Description

Method, device, medium and program product for acquiring data

Technical Field

Embodiments of the present disclosure relate generally to the field of computers and, more particularly, relate to methods, apparatuses, computer-readable storage media, and computer program products for collecting data.

Background

Currently, more and more vehicles deploy machine learning models to assist users in driving vehicles to enhance user experience. Typically, these machine learning models may be trained and updated with data collected by the vehicle. For example, various data may be collected using sensors such as cameras, microphones, lidars, etc. deployed on the vehicle as training data for the training model. However, taking into account the limitations of network bandwidth and computing resources, it is not practical to collect all the data received by the sensors in real time as training data. Thus, a solution for collecting data is needed to efficiently collect data that is valuable for model training.

Disclosure of Invention

According to an example embodiment of the present disclosure, a scheme for collecting data is provided to efficiently collect data similar to reference data that is valuable for model training.

In a first aspect of the present disclosure, a method for acquiring data is provided. The method includes receiving data from an input device of a vehicle; obtaining reference data from a database stored on the vehicle, the reference data being associated with a model deployed on the vehicle for assisting driving; in response to similarity between the data and the reference data being above a threshold, the data is added to a training set for training the model.

In some embodiments, the method further comprises: the similarity is determined based on the data and the reference data using a comparison model deployed on the vehicle, the comparison model being trained based on a training set associated with the reference data.

In some embodiments, determining the similarity using a comparison model deployed on the vehicle includes: determining a feature vector characterizing the data using the comparison model based on the data; determining a reference feature vector characterizing the reference data using the comparison model based on the reference data; and determining similarity using a comparison model based on the feature vector and the reference feature vector.

In some embodiments, the method further comprises: based on the data, feature vectors characterizing the data are determined using at least a portion of a model or a first model deployed on the vehicle, the first model being different from the at least a portion of the model.

In some embodiments, the method further comprises: based on the reference data, a reference feature vector characterizing the reference data is determined using one of: at least a portion of the model, the first model, or a second model deployed on the vehicle, the second model being different from at least a portion of the model and the first model.

In some embodiments, the method further comprises: similarity between the data and the reference data is determined based on distances between the feature vectors and the reference feature vectors in the vector space.

In some embodiments, the method further comprises: the similarity is determined based on the feature vector and the reference feature vector using a third model deployed on the vehicle, the third model being trained based on a training set associated with the reference feature vector.

In some embodiments, obtaining the reference data includes: determining a type of reference data for at least one of a performance of the plurality of types of data based on the function of the model and the model; and selecting a type of reference data from a plurality of types of reference data in the database.

In a second aspect of the present disclosure, an electronic device is provided. The electronic device includes: at least one processing unit; and at least one memory coupled to the at least one processing unit and storing instructions for execution by the at least one processing unit, the instructions when executed by the at least one processing unit causing the device to perform actions. The actions include: receiving data from an input device of a vehicle; obtaining reference data from a database stored on the vehicle, the reference data being associated with a model deployed on the vehicle for assisting driving; in response to similarity between the data and the reference data being above a threshold, the data is added to a training set for training the model.

In a third aspect of the present disclosure, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a device, causes the device to perform a method according to the first aspect of the present disclosure.

According to a fourth aspect of the present disclosure, a computer program product is provided. The computer program product comprises a computer program which, when executed by a processor, implements the method of the first aspect.

It should be understood that what is described in this summary is not intended to limit the critical or essential features of the embodiments of the disclosure nor to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following description.

Drawings

The above and other features, advantages and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. In the drawings, wherein like or similar reference numerals denote like or similar elements, in which:

FIG. 1 illustrates a schematic diagram of an example environment, according to an embodiment of the present disclosure;

FIG. 2 illustrates a flow chart of a process for acquiring data according to some embodiments of the present disclosure;

FIG. 3 illustrates an example of a process of determining similarity according to some embodiments of the present disclosure;

FIG. 4 illustrates an example of another process of determining similarity according to some embodiments of the present disclosure; and

fig. 5 illustrates a block diagram of an electronic device capable of implementing various embodiments of the present disclosure.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure have been illustrated in the accompanying drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather, these embodiments are provided so that this disclosure will be more thorough and complete. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.

In describing embodiments of the present disclosure, the term "comprising" and its like should be taken to be open-ended, i.e., including, but not limited to. The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions are also possible below.

As mentioned above, a solution to collect data is needed to efficiently collect data that is valuable for model training. Currently, in view of the limitations of networks and computing resources, data acquisition schemes applied to vehicles typically utilize a random acquisition approach to acquire data received by sensors, such as data acquired at intervals, thereby reducing the consumption of resources by the acquisition process. However, with this approach, the acquired data is completely random. There is also a need to manually screen and annotate data that is valuable to training models.

Still other schemes collect data based on performance of the model. By observing the performance of the models for the different data received by the sensor, it can be learned which data models perform poorly. Based on this, these data may be collected as training data to update the model, thereby improving the performance of the model for such data.

For example, if the model is observed to be less accurate in identifying the received pictures containing signal lights, these pictures containing signal lights may be collected as training data for further training the model, i.e. updating the model.

However, data for which the model has poor performance is not necessarily data that is valuable for training the model. For example, for models used to identify signal lights in pictures, the models would be less effective at identifying pictures containing pedestrians rather than signal lights, but the ability of the models to identify signal lights cannot be improved using these pictures containing pedestrians as training data. Thus, such data acquisition schemes do not accurately acquire data that is valuable for model training.

According to an embodiment of the present disclosure, a scheme for collecting data is provided. In this scenario, data is received from an input device of a vehicle. Reference data is obtained from a database stored on the vehicle, the reference data being associated with a model deployed on the vehicle for assisting driving. In response to similarity between the data and the reference data being above a threshold, the data is added to a training set for training the model. According to the scheme, the data similar to the reference data can be acquired in a targeted manner to train a model for assisting driving, so that the efficiency of data acquisition is improved.

Example embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

Fig. 1 illustrates a schematic diagram of an example environment 100, according to an embodiment of the disclosure. In this example environment 100, the vehicle 110 may include an input device 120 for receiving data. The input device 120 may include various sensors such as a camera, microphone, and the like. Input device 120 is illustrated in fig. 1 as a camera by way of example and not limitation. The vehicle 110 also includes a computing device 130. The computing device 130 may include or be implemented on any electronic device having computing capabilities, such as an electronic device implemented as a central control system of the vehicle 110. The vehicle 110 further includes a database 140 for storing reference data. The database 140 may be stored in a storage device in the vehicle 110. Although the data 140 is shown separately in fig. 1, the database 140 may also be integrated in a memory unit of the computing device 130. The computing device 130 may be in communication with the input device 120 and the database 140.

As used herein, a vehicle refers to any type of tool that is capable of carrying a person and/or object and that is movable. In fig. 1, as well as other figures and descriptions herein, the vehicle 110 is illustrated as a vehicle. The vehicle may be a motor vehicle or a non-motor vehicle, examples of which include, but are not limited to, a car, a sedan, a truck, a bus, an electric vehicle, a motorcycle, a bicycle, and the like. However, it should be understood that a vehicle is only one example of a vehicle. Embodiments of the present disclosure are equally applicable to vehicles other than vehicles, such as boats, trains, planes, and the like.

It should be understood that fig. 1 only schematically illustrates objects, units, elements, or components related to embodiments of the present disclosure. In practice, the example environment 100 may also include other objects, units, elements, or components, among others. In addition, the particular number of objects, units, elements, or components shown in fig. 1 is merely illustrative and is not intended to limit the scope of the disclosure in any way. In other embodiments, the example environment 100 may include any suitable number of objects, units, elements, or components, etc. For example, the vehicle 110 may include a plurality of different types of input devices 120. Embodiments of the present disclosure are not limited to the specific scenario depicted in fig. 1, but rather are generally applicable to any technical environment in which a vehicle carries onboard electronic equipment.

Some embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings.

Fig. 2 illustrates a flow chart of a process 200 for collecting data according to some embodiments of the present disclosure. Process 200 may be implemented by computing device 130 of fig. 1. For ease of discussion, the description will be described with reference to environment 100 of FIG. 1.

At block 210, the computing device 130 receives data from the input device 120 of the vehicle 110. The data may be data acquired from the outside via the input device 120. The data may have any suitable form depending on the type of input device 120. For example, the received data may be image data acquired from a camera, sound data acquired from a microphone, or the like.

At block 220, the computing device 130 obtains reference data from the database 140 stored on the vehicle 110, the reference data being associated with a model deployed on the vehicle 110 for assisting driving. The model deployed on the vehicle 110 for assisting driving may also be referred to as a business model hereinafter. The business model may be any suitable machine learning model for assisting a user in driving the vehicle 110 to enhance the user experience. For example, the business model may have functions of image recognition, object detection, path planning, etc.

At block 230, the computing device 130 adds the received data to a training set for training the business model in response to the similarity between the received data and the reference data being above a threshold. The reference data may be standard data associated with the business model for comparison with the received data. In other words, the reference data is data that is desired to be collected for training the business model.

For example, when a business model is used to identify different signal lights, the data that is desired to be collected for training the business model includes image data that includes the various signal lights. In this case, the reference data may include image data including a signal lamp. In this way, when the similarity between the received data and the reference data is above a threshold, i.e., the received data is similar to the reference data, the data may be collected for training the traffic model, thereby improving the ability of the traffic model to identify the signal.

In another example, when the business model is used to detect pedestrians and vehicles in a field of view, the reference data may include image data including pedestrians and/or vehicles. Thus, when the received data is not related to a pedestrian or a vehicle, for example, the received data is image data containing only roads, the received data will not be collected for training the traffic model because the received data is dissimilar to the reference data.

In this way, by collecting only received data that is similar to the reference data, data that is valuable for training the business model can be efficiently collected.

In some embodiments, database 140 may be pre-stored with multiple types of reference data. The plurality of types of reference data pre-stored in the database 140 may be determined according to functions of the business model. For example, image data containing pedestrians, vehicles, and/or signals may be stored in advance for a traffic model for identifying pedestrians, vehicles, and/or signals.

In this case, the computing device 130 may obtain target reference data from multiple types of reference data stored in the database 140 as needed for comparison with the received data. In some embodiments, the computing device 130 may determine a type of reference data to obtain based on the functionality of the business model and retrieve the type of reference data from a plurality of types of reference data in the database 140. For example, when the business model is used only to detect pedestrians, image data containing pedestrians may be acquired from the database 140 as target reference data. In this way, the computing device 130 may collect only pedestrian-related data in the received data as training data, thereby improving the efficiency of collecting the data.

Alternatively or additionally, the computing device 130 may determine a type of reference data to obtain based on the performance of the business model for the plurality of types of data, and retrieve the type of reference data from the plurality of types of reference data in the database 140.

Computing device 130 may observe the performance of the business model for different types of data. The computing device 130 may monitor and analyze the inference results of the business model to determine the type of data for which the business model has poor performance. For example, the computing device 130 may find that the accuracy of the business model's detection of pedestrians is lower than the detection of vehicles.

In this case, the computing device 130 may regard the type of data for which the business model has poor performance as the type of reference data. For example, the computing device 130 may obtain image data containing pedestrians from the database 140 as target reference data. In this way, the computing device 130 may collect only pedestrian-related data in the received data as training data, thereby improving the efficiency of collecting the data.

In some embodiments, the computing device 130 may obtain a plurality of reference data from the database 140 for comparison with the received data. In this case, the computing device 130 may determine a similarity between each of the plurality of reference data and the received data, respectively. If the similarity between the received data and any of the plurality of reference data is above a threshold, the computing device 130 may collect the received data for training the business model.

Details of comparing the received data with the reference data will be described in detail below in conjunction with the figures. Fig. 3 illustrates an example of a process 300 for determining similarity between received data and reference data.

In some embodiments, the computing device 130 may utilize a comparison model deployed on the vehicle 110 to determine similarity between the received data and the reference data based on the received data and the acquired reference data.

As shown in fig. 3, based on the received data 310 and the reference data 320 obtained from the database 140, the comparison model 330 may determine a similarity 340 between the data 310 and the reference data 320. The comparison model 330 may be any suitable model for comparing data similarities. Examples of the comparison model 330 may include a model for comparing image similarity, a model for comparing sound similarity, and so forth.

In some embodiments, the comparison model 330 may be a model based on a non-machine learning algorithm. For example, the comparison model 330 may be a model that determines image similarity based on a hashing algorithm or a Scale Invariant Feature Transform (SIFT) algorithm. In this case, the comparison model 330 may not need to be trained.

In some embodiments, the comparison model 330 may be a model based on a machine learning algorithm. For example, the comparison model 330 may be a neural network model. The comparison model 330 may be trained with a dataset associated with the reference data.

The reference data can be used to construct positive sample pairs (e.g., two images containing signal lights) and negative sample pairs (one image containing signal lights and one image not containing signal lights). The positive and negative pairs of samples may be manually labeled "1" and "0" respectively, thereby constructing a labeled training set.

It should be appreciated that the positive and negative pairs of samples may be constructed according to a particular implementation. Positive sample pairs may refer to pairs of samples of the same type, or peers of samples containing the same object. Negative sample pairs may refer to pairs of uncorrelated samples.

For example, for a traffic model identifying a signal, a positive pair of samples may consist of two images containing a signal, while a negative pair of samples may consist of one image containing a signal and one image containing only pedestrians.

Using the training set, the comparison model 330 may be trained to determine whether the input data pairs are identical or to determine the probability that the input data pairs are identical. The similarity 340 may be shown as a binary classification result of 0 or 1 or in the form of a probability. For example, the trained comparison model 330 may determine similarity between the data 310 and the reference data 320 based on the input data 310 and the reference data 320.

The use of the machine learning algorithm based comparison model 330 will be described in detail below with reference to fig. 4. Fig. 4 illustrates an example of a process 400 for determining similarity between received data and reference data.

As shown in fig. 4, the comparison model 330 may include sub-models 410, 420 for feature extraction and sub-model 430 for comparing extracted features. The sub-model 410 may be used to extract features of the data 310 and output feature vectors 440 that characterize the data 310. The submodel 420 may be used to extract features of the reference data 320 and output a reference feature vector 450 for characterizing the reference data 320. Sub-model 430 may determine similarity 340 based on feature vector 440 and reference feature vector 450.

In some embodiments, sub-models 410 and 420 may be feature extraction models based on non-machine learning algorithms, such as SIFT algorithm based models. Alternatively, sub-model 410 and sub-model 420 may be models based on machine learning algorithms. For example, the output of the penultimate layer (depending on the specific layer structure of the network) of a neural network model (e.g., VGG or GoogleNet) for image classification can be taken as the extracted feature vector for characterizing the data.

In some embodiments, sub-model 410 and sub-model 420 may be identical models, such as neural network models of identical structure and weight. Alternatively, sub-model 410 and sub-model 420 may be different models.

In some embodiments, sub-model 410 and sub-model 420 may be at least part of a business model. The traffic model for assisting driving may comprise a plurality of sub-models with different functions, such as a sub-model for image classification, a sub-model for path planning, etc. In this case, the output of the penultimate layer of the submodel for image classification may be taken as a feature vector for characterizing the data. In this way, at least a portion of the business model may be multiplexed, thereby increasing the utilization of computing resources.

In some embodiments, sub-model 410 and sub-model 420 may be other models that are different from at least a portion of the business model. For example, sub-model 410 and sub-model 420 may be simpler models than the portions of the business model used for feature extraction, thereby reducing consumption of resources.

In some embodiments, sub-model 410 may be at least a portion of a business model and sub-model 420 may be a different other model. Since the business model needs to process the received data 310, the processing result of the data 310 by at least a portion of the business model may be output as the feature vector 440. In this way, the utilization of resources can be improved.

In some embodiments, the submodel 430 for comparing feature vectors may determine the similarity 340 based on vector distance. Sub-model 430 may determine similarity 340 between data 310 and reference data 320 based on the distance between feature vector 440 and reference feature vector 450 in the vector space. For example, sub-model 430 may calculate a cosine distance or euclidean distance between feature vector 440 and reference feature vector 450 to determine similarity 340.

In some embodiments, sub-model 430 may be a model based on a machine learning algorithm, such as a neural network model. The sub-models 430 may be trained separately. The sub-model 430 may be trained based on a training set constructed from the reference data 320. For example, positive and negative pairs of samples may be constructed based on the reference feature vector 450 as a training set.

Alternatively, sub-model 430 may be trained with

sub-models

410 and 420. It should be appreciated that when sub-model 410 and sub-model 420 are at least a portion of a business model, sub-model 410 and sub-model 420 are not trained so as not to affect the normal functioning of the business model.

The process of collecting data 310 and the process of determining similarity based on similarity between received data 310 and reference data 320 is described above with reference to fig. 2-4. It should be understood that the

processes

200, 300, 400 are merely exemplary and not limiting. For example, the reference feature vector 450 may be stored directly in the database 140 or cached in the computing device 130, such that the reference feature vector 450 may be obtained directly from the database 140 or computing device 130 for comparison. In this way, the consumption of resources can be reduced.

Fig. 5 shows a schematic block diagram of an electronic device 500 that may be used to implement embodiments of the present disclosure. Electronic device 500 may be used to implement computing device 130 as shown in fig. 1. For example, computing device 130 may be implemented as or included in electronic device 500.

As shown in fig. 5, the electronic device 500 includes a Central Processing Unit (CPU) 501 that can perform various suitable actions and processes in accordance with computer program instructions stored in a Read Only Memory (ROM) 502 or loaded from a storage unit 508 into a Random Access Memory (RAM) 503. In the RAM503, various programs and data required for the operation of the electronic device 500 may also be stored. The CPU 501, ROM 502, and RAM503 are connected to each other through a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.

A number of components in electronic device 500 are connected to I/O interface 505, including: an input unit 506 such as a keyboard, a mouse, etc.; an output unit 507 such as various types of displays, speakers, and the like; a storage unit 508 such as a magnetic disk, an optical disk, or the like; and a communication unit 509 such as a network card, modem, wireless communication transceiver, etc. The communication unit 509 allows the electronic device 500 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The processing unit 501 performs the various methods and processes described above, such as

processes

200, 300, 400. For example, in some embodiments, the

processes

200, 300, 400 may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 500 via the ROM 502 and/or the communication unit 509. When the computer program is loaded into RAM503 and executed by CPU 501, one or more of the steps of

processes

200, 300, 400 described above may be performed. Alternatively, in other embodiments, CPU 501 may be configured to perform

processes

200, 300, 400 in any other suitable manner (e.g., by means of firmware).

In some embodiments, a computer program product may also be provided. The computer program product may comprise a computer program which, when executed by the processor 501, implements the

processes

200, 300, 400.

The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a load programmable logic device (CPLD), and the like.

Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

Moreover, although operations are depicted in a particular order, this should be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single implementation. Conversely, various features that are described in the context of a single implementation can also be implemented in multiple implementations separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are example forms of implementing the claims.

Claims

1. A method of collecting data, comprising:

receiving data from an input device of a vehicle;

obtaining reference data from a database stored on the vehicle, the reference data being associated with a model deployed on the vehicle for assisting driving; and

in response to similarity between the data and the reference data being above a threshold, the data is added to a training set used to train the model.

2. The method of claim 1, further comprising:

based on the data and the reference data, the similarity is determined using a comparison model deployed on the vehicle, the comparison model being trained based on a training set associated with the reference data.

3. The method of claim 2, wherein determining the similarity using a comparison model deployed on the vehicle comprises:

determining, based on the data, a feature vector characterizing the data using the comparison model;

determining, based on the reference data, a reference feature vector characterizing the reference data using the comparison model; and

the similarity is determined using the comparison model based on the feature vector and the reference feature vector.

4. The method of claim 1, further comprising:

based on the data, a feature vector characterizing the data is determined using at least a portion of the model or a first model deployed on the vehicle, the first model being different from at least a portion of the model.

5. The method of claim 4, further comprising:

based on the reference data, a reference feature vector characterizing the reference data is determined using one of: at least a portion of the model, the first model, or a second model deployed on the vehicle, the second model being different from at least a portion of the model and the first model.

6. The method of claim 5, further comprising:

the similarity between the data and the reference data is determined based on a distance between the feature vector and the reference feature vector in a vector space.

7. The method of claim 5, further comprising:

the similarity is determined based on the feature vector and the reference feature vector using a third model deployed on the vehicle, the third model trained based on a training set associated with the reference feature vector.

8. The method of claim 1, wherein obtaining the reference data comprises:

determining a type of the reference data based on at least one of a function of the model and a performance of the model for a plurality of types of data; and

the type of reference data is selected from a plurality of types of reference data in the database.

9. An electronic device, comprising:

at least one processing unit; and

at least one memory coupled to the at least one processing unit and storing instructions for execution by the at least one processing unit, the instructions when executed by the at least one processing unit cause the apparatus to perform actions comprising:

receiving data from an input device of a vehicle;

10. The apparatus of claim 9, the acts further comprising:

11. The apparatus of claim 10, wherein utilizing a comparison model deployed on the vehicle to determine the similarity comprises:

12. The apparatus of claim 9, the acts further comprising:

13. The apparatus of claim 12, the acts further comprising:

14. The apparatus of claim 13, the acts further comprising:

15. The apparatus of claim 13, the acts further comprising:

16. The apparatus of claim 9, wherein obtaining the reference data comprises:

17. A computer readable storage medium having stored thereon a computer program which when executed by a processor implements the method according to any of claims 1 to 8.

18. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 8.