WO2023123818A1 - Vehicle retrofitting detection method and apparatus, electronic device, computer-readable storage medium and computer program product - Google Patents

Abstract

Embodiments of the present disclosure provide a vehicle retrofitting detection method and apparatus, an electronic device, a computer-readable storage medium and a computer program product. The method comprises: obtaining a to-be-detected vehicle image corresponding to a target vehicle; determining a re-identification feature similarity of the target vehicle in the to-be-detected vehicle image, wherein the re-identification feature similarity is determined on the basis of a first re-identification feature extracted from the to-be-detected vehicle image and a second re-identification feature extracted from an original vehicle image corresponding to the target vehicle; determining a vehicle attribute confidence of the target vehicle in the to-be-detected vehicle image; and determining a retrofitting detection result of the target vehicle on the basis of the re-identification feature similarity and the vehicle attribute confidence.

Description

Vehicle modification detection method, device, electronic device, computer-readable storage medium, and computer program product

Cross References to Related Applications

The embodiment of the present disclosure is based on the application number 202111658296.2, the application date is December 30, 2021, the applicant is Beijing SenseTime Technology Development Co., Ltd., and the application name is "vehicle modification detection method, device, electronic equipment and storage medium". A patent application is filed and priority is claimed to this Chinese patent application, the entire contents of which are hereby incorporated by reference into this disclosure.

technical field

Embodiments of the present disclosure relate to but are not limited to the technical field of artificial intelligence, and particularly relate to a vehicle modification detection method, device, electronic equipment, computer-readable storage medium, and computer program product.

Background technique

In the detection and identification of vehicle violations, it is generally judged whether the vehicle is illegal by manual observation; The identification of whether the vehicle is modified or whether the vehicle in the video is modified by the staff can be determined by the vehicle information observed by human eyes to determine whether the appearance of the vehicle has changed, so as to determine whether the vehicle has been modified.

Contents of the invention

Embodiments of the present disclosure provide a vehicle modification detection method, device, electronic equipment, computer-readable storage medium, and computer program product.

An embodiment of the present disclosure provides a vehicle modification detection method, the method is executed by an electronic device, including:

Obtain the image of the vehicle to be detected corresponding to the target vehicle;

determining the re-identification feature similarity of the target vehicle in the vehicle image to be detected; wherein the re-identification feature similarity is based on the first re-identification feature extracted from the vehicle image to be detected, and from the target vehicle Determination of the second identification feature extracted from the original vehicle image corresponding to the vehicle;

determining the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected;

Based on the similarity of the re-identified features and the confidence of the vehicle attribute, a modification detection result of the target vehicle is determined.

Compared with the method of manually identifying vehicle modifications in the related art, the efficiency is low, and due to the limitation of human eye observation, the accuracy of the recognition results of vehicle modification is not high; in the above technical solution, the electronic device is based on the weight of the target vehicle. Identifying feature similarity and vehicle attribute confidence of the target vehicle to determine the modification detection result of the target vehicle can improve the efficiency and accuracy of vehicle modification detection and save labor costs. Moreover, the above technical solution detects the modification of the target vehicle from the two different aspects of the re-identification feature similarity of the target vehicle and the confidence of the vehicle attribute of the target vehicle, which can improve the accuracy and robustness of the target vehicle modification detection.

For the description of the effects of the following devices, electronic equipment, etc., please refer to the description of the above-mentioned vehicle modification detection method.

An embodiment of the present disclosure also provides a vehicle modification detection device, including: an image acquisition part of a vehicle to be detected, configured to obtain an image of a vehicle to be detected corresponding to a target vehicle; a re-identification feature similarity determination part, configured to determine the In detecting the vehicle image, the re-identification feature similarity of the target vehicle; wherein, the re-identification feature similarity is based on the first re-identification feature extracted from the vehicle image to be detected, and the corresponding original The second identification feature determination of vehicle image extraction; the vehicle attribute confidence determination part is configured to determine the vehicle attribute confidence of the target vehicle in the vehicle image to be detected; the refit detection result determination part is configured based on The re-identification feature similarity and the vehicle attribute confidence determine the refit detection result of the target vehicle.

An embodiment of the present disclosure also provides an electronic device, including: a processor; a memory configured to store instructions executable by the processor; wherein the processor is configured to execute the instructions to implement any of the above method described in the item.

The embodiment of the present disclosure also provides a computer-readable storage medium, when the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device can execute any one of the above-mentioned instructions in the embodiments of the present disclosure. vehicle modification detection method.

An embodiment of the present disclosure also provides a computer program product containing instructions, and when the computer program or instruction is run on an electronic device, the electronic device is made to execute the vehicle modification detection method as described in any of the above-mentioned embodiments. .

Embodiments of the present disclosure at least provide a vehicle modification detection method, device, electronic equipment, computer-readable storage medium, and computer program product; Due to the limitation of observation, the accuracy of the recognition result of vehicle modification is not high; in the above technical solution, the electronic device determines the detection result of the modification of the target vehicle based on the similarity of the re-identification features of the target vehicle and the confidence of the vehicle attribute of the target vehicle. The efficiency and accuracy of vehicle modification detection can be improved, and labor costs can be saved. Moreover, the above technical solution detects the modification of the target vehicle from the two different aspects of the re-identification feature similarity of the target vehicle and the confidence of the vehicle attribute of the target vehicle, which can improve the accuracy and robustness of the target vehicle modification detection.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings that need to be used in the embodiments of the present disclosure will be described below.

The accompanying drawings here are incorporated into the description and constitute a part of the present description. These drawings show embodiments consistent with the present disclosure, and are used together with the description to explain the technical solution of the present disclosure.

FIG. 1 is a flow chart of a vehicle modification detection method provided by an embodiment of the present disclosure;

FIG. 2 is a flowchart of a method for determining an original vehicle image provided by an embodiment of the present disclosure;

FIG. 3 is a flow chart of a method for determining a vehicle identification information recognition result provided by an embodiment of the present disclosure;

FIG. 4 is a flow chart of a method for identifying license plate information based on license plate area detection provided by an embodiment of the present disclosure;

FIG. 5 is a flow chart of a method for identifying license plate information based on character area detection provided by an embodiment of the present disclosure;

FIG. 6 is a flowchart of a method for determining similarity of re-identification features provided by an embodiment of the present disclosure;

FIG. 7 is a flow chart of a method for determining the confidence level of a vehicle attribute provided by an embodiment of the present disclosure;

FIG. 8 is a flow chart of a method for identifying vehicle attributes of an image of a vehicle to be detected according to an embodiment of the present disclosure;

FIG. 9 is a flow chart of a method for determining a modification detection result provided by an embodiment of the present disclosure;

FIG. 10 is a flow chart of a method for fusing re-identification feature similarity and prediction confidence provided by an embodiment of the present disclosure;

Fig. 11 is a schematic diagram of a vehicle modification detection device provided by an embodiment of the present disclosure;

FIG. 12 is a schematic structural diagram of an electronic device for vehicle modification detection provided by an embodiment of the present disclosure;

FIG. 13 is a schematic structural diagram of an electronic device for vehicle modification detection provided by an embodiment of the present disclosure.

Detailed ways

In order to enable ordinary persons in the art to better understand the solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Apparently, the described embodiments are some of the embodiments of the present application, but not all of them. Based on the embodiments in the present disclosure, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present disclosure.

It should be noted that the terms "first" and "second" in the specification and claims of the present disclosure and the above drawings are used to distinguish similar objects, but not necessarily used to describe a specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "include" and "have", as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include for these processes, Other steps or elements inherent in a method, product, or apparatus. The implementations described in the following exemplary examples do not represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present disclosure as recited in the appended claims.

It should be understood that in this application, "at least one (item)" means one or more, "multiple" means two or more, and "at least two (items)" means two or three And three or more, "and/or", is used to describe the association relationship of associated objects, indicating that there can be three types of relationships, for example, "A and/or B" can mean: only A exists, only B exists, and A exists at the same time and B, where A and B can be singular or plural. The character "/" can indicate that the contextual objects are an "or" relationship, which refers to any combination of these items, including any combination of single items (items) or plural items (items). For example, at least one item (piece) of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c ", where a, b, c can be single or multiple. The character "/" can also represent a division sign in mathematical operations, for example, a/b=a divided by b; 6/3=2. "At least one of the following" or similar expressions.

An embodiment of the present disclosure provides a vehicle modification detection method. FIG. 1 is a flow chart of a vehicle modification detection method provided by an embodiment of the present disclosure. Please refer to FIG. 1. This method can be used in electronic devices such as terminals, servers, and edge computing nodes. , may include the following steps:

S110: Obtain an image of the vehicle to be detected corresponding to the target vehicle.

In some embodiments, the image of the vehicle to be detected is a vehicle image corresponding to the target vehicle, that is, the image of the vehicle to be detected contains the vehicle image of the target vehicle, and the image of the vehicle to be detected can be obtained through a pre-trained vehicle detection model; for example, The vehicle detection model can carry out vehicle recognition on the input image, determine the target vehicle area in the image; crop the vehicle image in the target vehicle area, and obtain the vehicle image to be detected corresponding to the target vehicle. Among them, the structure of the vehicle detection model can be a two-stage network structure, such as a regional convolutional neural network (Faster Regions with CNN features, Faster RCNN), and a single-stage detection network structure (RetinaNet). One or more vehicles may be contained in one image, so the vehicle images to be detected of one or more target vehicles contained in the image can be respectively obtained through the vehicle detection model, and each target vehicle corresponds to a vehicle to be detected image.

In some embodiments, the image can be collected by an image acquisition device. The image acquisition device can be a device installed at a fixed collection point, for example, it can be installed at a fixed collection point such as a road side, a parking lot, a gantry, or an overpass. It may be a mobile image acquisition device installed on a vehicle.

In some embodiments, the image may be an image captured by an image acquisition device, or may be a video frame image extracted from a video captured by the image acquisition device.

S120: Determine the re-identification feature similarity of the target vehicle in the image of the vehicle to be detected.

Wherein, the re-identification feature similarity is determined based on the first re-identification feature extracted from the image of the vehicle to be detected and the second re-identification feature extracted from the original vehicle image corresponding to the target vehicle, the re-identification feature The degree of similarity can be used to characterize the degree of similarity between the first re-identification feature and the second re-identification feature.

The first recognition feature can be used to represent the feature information of the target vehicle in the vehicle image to be detected, and the second recognition feature can be used to represent the feature information of the target vehicle in the original vehicle image, where the original vehicle image can be the target vehicle before it is modified The vehicle image can be stored in the preset vehicle information library; thus, according to the comparison result of the first re-identification feature and the second re-identification feature, it can be determined whether the characteristics of the target vehicle have changed, and the re-identification feature can be understood as the same Multiple feature representations of the target vehicle.

In some implementations, the re-identification feature extraction model can be used to perform re-identification feature extraction on the image of the vehicle to be detected and the original vehicle image. The first recognition feature extraction of the image of the vehicle to be detected can be performed after the image of the vehicle to be detected of the target vehicle is determined. For the second recognition feature extraction of the original vehicle image, it can be extracted through the re-identification feature extraction model in advance, and the extracted second recognition feature is stored in the preset vehicle information library, so that the second recognition feature needs to be used. It can be obtained directly when re-identifying features.

In some embodiments, since the re-identification feature extraction model may be updated, when the second re-identification feature is needed, the current re-identification feature extraction model can be called to perform the second re-identification feature extraction, so as to improve the second Accuracy of re-identification feature extraction.

S130: Determine the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected.

The vehicle attribute confidence can be used to represent the vehicle attribute information of the target vehicle in the vehicle image to be detected, and the predicted matching degree of the vehicle attribute corresponding to the target vehicle in the original vehicle image, so that the vehicle attribute confidence can be used to subsequently determine whether the target vehicle is modified A factor.

S140: Determine a modification detection result of the target vehicle based on the re-identification feature similarity and the vehicle attribute confidence.

Compared with the method of manually identifying vehicle modifications in the related art, the efficiency is low, and due to the limitation of human eye observation, the accuracy of the recognition results of vehicle modification is not high; in the above technical solution, the electronic device is based on the weight of the target vehicle. Identifying feature similarity and vehicle attribute confidence of the target vehicle to determine the modification detection result of the target vehicle can improve the efficiency and accuracy of vehicle modification detection and save labor costs. Moreover, the above technical solution detects the modification of the target vehicle from the two different aspects of the re-identification feature similarity of the target vehicle and the confidence of the vehicle attribute of the target vehicle, which can improve the accuracy and robustness of the target vehicle modification detection.

The vehicle image to be detected corresponding to the target vehicle and the original vehicle image have a corresponding relationship. The target vehicle in the vehicle image to be detected is the same vehicle as the target vehicle in the original vehicle image, so they have the same vehicle identification information; please refer to Figure 2, Figure 2 It is a flowchart of a method for determining an original vehicle image provided by an embodiment of the present disclosure, the method is executed by an electronic device, and the method may include the following steps:

S210: Perform vehicle identification information identification on the image of the vehicle to be detected to obtain a vehicle identification information identification result of the target vehicle.

S220: Based on the identification result of the vehicle identification information, match the original vehicle image and the vehicle attributes corresponding to the original vehicle image from a preset vehicle information database.

The electronic device can perform image recognition on the image of the vehicle to be detected to obtain the recognition result of the vehicle identification information, and the recognition result is the vehicle identification information. Using the preset vehicle identification information stored in the vehicle information database and the original vehicle image corresponding to the preset vehicle identification information, the electronic device matches the identified vehicle identification information with the preset vehicle identification information. After the preset vehicle identification information, the original vehicle image corresponding to the target vehicle can be determined.

In some embodiments, the vehicle attribute corresponding to the original vehicle image can be the vehicle attribute before the target vehicle is refitted, so that the vehicle attribute corresponding to the original vehicle image can also be identified in advance and stored in the vehicle information database; The vehicle attributes corresponding to the original vehicle image can be matched.

In the above technical solution, the vehicle identification information can be used to uniquely identify the vehicle. By identifying the vehicle identification information of the target vehicle, the electronic device can conveniently obtain the original vehicle image corresponding to the vehicle identification information, thereby improving the efficiency of obtaining the original vehicle image. Efficiency, and then improve the efficiency of vehicle modification detection.

In some implementations, the vehicle identification information includes the license plate information of the target vehicle, so that the identification result of the vehicle identification information can be the identification result of the license plate information; A flowchart of a method for determining a vehicle identification information recognition result, the method is executed by an electronic device, and the method may include the following steps:

S310: Perform license plate information recognition on the image of the vehicle to be detected to obtain license plate information of the target vehicle.

S320: Determine the license plate information of the target vehicle as a recognition result of the vehicle identification information of the target vehicle.

In the above technical solution, the license plate information is identification information that is easily obtained by the target vehicle. By identifying the license plate information on the image of the vehicle to be detected, the electronic device can obtain the vehicle identification information, which improves the convenience of vehicle identification information identification. Moreover, in a variety of application scenarios, such as fixed collection points such as gantry frames and overpasses, or mobile image collection equipment installed on vehicles, electronic equipment can obtain vehicle identification information by performing license plate information recognition on images of vehicles to be detected , the vehicle modification detection method can be applied to a variety of application scenarios, improving the universality of the vehicle modification detection method.

In some implementations, the license plate information area is a smaller area than the image of the vehicle to be detected, so that the license plate information can be identified by the method of area detection; please refer to FIG. A flow chart of a method for license plate information recognition based on license plate area detection, the method is executed by an electronic device, and the method may include the following steps:

S410: Perform license plate area detection on the image of the vehicle to be detected, and determine a target license plate area in the image of the vehicle to be detected.

S420: Perform text recognition on the target license plate area to obtain license plate information of the target vehicle.

In some embodiments, the recognition of the license plate information by the electronic device can be realized by optical character recognition (Optical Character Recognition, OCR). The image containing the text is cut into independently identifiable units, and then the algorithm is used to analyze the text in each image unit. morphological characteristics. By comparing the data in the standard feature database, the standard encoding of the character in the computer is judged and output and saved in a text file in a common format.

In some embodiments, the identification of the license plate information by the electronic device can also be realized by the license plate recognition model. By inputting the image of the vehicle to be detected into the license plate recognition model, the license plate recognition model first detects the license plate area, and then detects the detected target license plate area to perform text recognition, and output the recognition result of the license plate information of the target vehicle.

In the above technical solution, firstly, the license plate area is detected in the image of the vehicle to be detected, and then text recognition is performed based on the target license plate area, which can reduce the area requiring text recognition, reduce the data processing capacity of electronic equipment in the text recognition process, and improve the license plate information recognition. s efficiency.

In some implementations, please refer to FIG. 5. FIG. 5 is a flow chart of a method for identifying license plate information based on text area detection provided by an embodiment of the present disclosure. The method is executed by an electronic device, and the method may include the following steps:

S510: Perform license plate text area detection on the target license plate area to obtain a target license plate text area in the target license plate area.

S520: Perform character recognition on the license plate text image in the target license plate text area to obtain license plate information of the target vehicle.

When the license plate recognition model performs text recognition on the target license plate area, it can first detect the license plate text area on the target license plate area, and then perform text recognition on the license plate text area, which reduces the scope of text recognition and reduces the amount of data processing. The efficiency of license plate information recognition is improved.

In some implementations, please refer to FIG. 6. FIG. 6 is a flow chart of a method for determining the similarity of re-identification features provided by an embodiment of the present disclosure. The method is executed by an electronic device, and the method may include the following steps:

S610: Determine a re-identification feature distance between the first re-identification feature and the second re-identification feature.

S620: Determine the re-identification feature similarity based on the re-identification feature distance.

Wherein, the first re-identification feature is used to characterize the feature description information of the target vehicle in the vehicle image to be detected, and the second re-identification feature is used to characterize the feature of the target vehicle in the original vehicle image Description.

The re-identification feature similarity can be used to characterize the similarity between the first re-identification feature and the second re-identification feature; the higher the re-identification feature similarity, the closer the first re-identification feature is to the second re-identification feature, and the re-identification feature The lower the similarity, the greater the difference between the first recognition feature and the second recognition feature.

The first re-identification feature and the second re-identification feature can be feature vectors, and the re-identification feature distance calculation can be realized by a method that can perform feature distance calculation in related technologies, such as Euclidean distance (Euclidean Distance), cosine distance (Cosine Distance) )wait.

The larger the re-identification feature distance between the first re-identification feature and the second re-identification feature, the smaller the similarity of the corresponding re-identification features, and the smaller the re-identification feature distance between the first re-identification feature and the second re-identification feature, the corresponding re-identification The greater the feature similarity.

In some implementations, if the normalized re-identification feature distance is a (0<a<1), the corresponding re-identification feature similarity may be 1-a, or 1/a.

In some embodiments, the first re-identification feature can also be used to characterize the feature description information of the target vehicle at the first angle, and the second re-identification feature can be used to characterize the target vehicle at the second angle The feature description information; the first angle and the second angle can be the same angle or different angles; when the first angle and the second angle are the same angle, the corresponding vehicle image to be detected and the original vehicle image can be Images collected from the same angle lens; when the first angle and the second angle are different, the corresponding image of the vehicle to be detected and the original vehicle image can be images of different angle lenses, that is, images collected across lenses. In practical application scenarios, the original vehicle image can generally be obtained by collecting the image acquisition device at the registration point when the target vehicle information is registered into the warehouse; for the vehicle image to be detected, it is generally installed on the side of the road, in the parking lot, Image acquisition equipment at fixed acquisition points such as gantry frames and overpasses are collected, so that the image of the vehicle to be detected and the original vehicle image are generally cross-lens images.

In the above technical solution, for the image information of the target vehicle collected at the same or different angles, the electronic device can determine the corresponding re-identification feature based on the re-identification feature distance between the first re-identification feature and the second re-identification feature The similarity can improve the efficiency of re-identification feature similarity calculation, thereby improving the efficiency of vehicle modification detection.

In some implementations, please refer to FIG. 7. FIG. 7 is a flow chart of a method for determining the confidence level of a vehicle attribute provided by an embodiment of the present disclosure. The method is executed by an electronic device, and the method may include the following steps:

S710: Carry out vehicle attribute recognition on the image of the vehicle to be detected, and obtain multiple predicted categories of vehicle attributes corresponding to the target vehicle, and prediction confidence levels corresponding to each of the multiple predicted categories.

S720: From the plurality of predicted categories, determine a matching predicted category that matches the vehicle attribute corresponding to the original vehicle image.

S730: Determine the prediction confidence of the matching prediction category as the vehicle attribute confidence.

In some embodiments, the vehicle attribute may include multiple attribute categories, such as attribute category 1, attribute category 2, attribute category 3, etc., so that when the vehicle attribute is identified on the image of the vehicle to be detected, multiple attributes of the target vehicle can be obtained. prediction categories, and the prediction confidence corresponding to each prediction category, the sum of the prediction confidence corresponding to multiple prediction categories is 1. Taking the vehicle attribute including three attribute categories as an example, the prediction confidence corresponding to attribute category 1 can be 0.1, the prediction confidence corresponding to attribute category 2 is 0.8, and the prediction confidence corresponding to attribute category 3 is 0.1; each prediction category The corresponding prediction confidence can be used to characterize the possibility that the vehicle attribute of the predicted target vehicle is the predicted category.

According to the vehicle attribute corresponding to the original vehicle image, one or two or more predicted categories matching the vehicle attribute are determined from multiple predicted categories, and the prediction confidence of the matched predicted category is determined as the vehicle attribute confidence. Here, the matching predicted category may be the same predicted category as the vehicle attribute corresponding to the original vehicle image, or a predicted category whose vehicle attribute corresponding to the original vehicle image satisfies a preset condition, for example, the corresponding vehicle attribute of the original vehicle image Predicted color classes for which vehicle attributes differ in lightness and/or saturation by less than a preset threshold.

In the above technical solution, for each predicted category, there is a prediction confidence corresponding to the predicted category, which improves the contingency caused by predicting a predicted category, thereby improving the accuracy of vehicle attribute category prediction, In turn, the accuracy of determining the confidence degree of the vehicle attribute can be improved.

In some implementations, the vehicle attribute information can be information obtained through visual observation, for example, the vehicle attribute can include the color of the target vehicle, the type of the target vehicle, etc.; each vehicle attribute can include multiple vehicle attribute categories, for example The color of the target vehicle includes color categories such as black, white, red, and gray, and the type of the target vehicle includes types and categories such as cars, trucks, and vans.

Therefore, the vehicle attributes may include vehicle color attributes and vehicle type attributes; please refer to FIG. 8 , which is a flow chart of a method for identifying vehicle attributes in an image of a vehicle to be detected according to an embodiment of the present disclosure. The method is executed by an electronic device. The method may include the steps of:

S810: Carry out vehicle color attribute identification on the image of the vehicle to be detected, and obtain multiple color prediction categories corresponding to the vehicle color attributes, and prediction confidence levels corresponding to the multiple color prediction categories.

S820: Carry out vehicle type attribute recognition on the image of the vehicle to be detected, and obtain multiple type prediction categories corresponding to the vehicle type attributes, and prediction confidence levels corresponding to the multiple type prediction categories.

In some embodiments, the identification of vehicle attributes can be carried out through the vehicle attribute recognition model, and the vehicle attribute recognition model can use a classification network such as a residual network (ResNet), VGG (Visual Geometry Group), etc., to input the image of the vehicle to be detected into the vehicle In the attribute recognition model, the prediction confidence of the target vehicle under each vehicle attribute category will be obtained. Assume that for the vehicle attribute of the target vehicle color, there are three color categories: black, white, and red. For the image of the vehicle to be detected, the vehicle attribute recognition model will predict a confidence score for each of the three color categories, representing each color category. The predicted confidence value of , where the sum of the confidence scores corresponding to the three color categories is 1. Similarly, for the vehicle attribute of vehicle type, the prediction confidence corresponding to each type category is predicted separately.

In the above technical solution, the vehicle attributes may include vehicle color attributes and vehicle type attributes. For each of the various vehicle attributes, corresponding multiple prediction categories can be obtained, thereby improving the comprehensiveness of the vehicle attribute description, and further Determining the confidence level of the vehicle attribute based on multiple prediction categories corresponding to multiple vehicle attributes can improve the accuracy of determining the confidence level of the vehicle attribute.

In some implementations, the matching prediction categories include matching color prediction categories and matching type prediction categories; correspondingly, please refer to FIG. The method is performed by an electronic device, and the method may include the following steps:

S910: Determine the prediction confidence of the matching color prediction category and the prediction confidence of the matching type prediction category as the vehicle attribute confidence.

S920: Perform information fusion on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category, to obtain a modification detection result of the target vehicle.

From the above content of this embodiment, it can be known that the corresponding original vehicle attribute can be determined through the identified vehicle identification information of the target vehicle. The vehicle attribute is the vehicle color as an example for illustration. For the target vehicle, if the original vehicle attribute category is black, From the vehicle attribute prediction results corresponding to the target vehicle, the prediction confidence corresponding to the vehicle color being black is determined; the prediction confidence corresponding to the vehicle color being black is fused with the re-identification feature similarity to determine the target vehicle. Modified test results.

In the above technical solution, the confidence of the vehicle attribute can be determined based on the prediction confidence of the matching color prediction category and the prediction confidence of the matching type prediction category, that is to say, the vehicle attribute confidence is determined based on the prediction categories of various vehicle attributes , so that the modification detection result can be determined based on the information fusion result of the vehicle attribute confidence and the re-identification feature similarity, which can improve the accuracy and robustness of the target vehicle modification detection.

In some implementations, please refer to FIG. 10. FIG. 10 is a flow chart of a method for fusing re-identification feature similarity and prediction confidence provided by an embodiment of the present disclosure. The method is executed by an electronic device, and the method can be Include the following steps:

S1010: Perform a weighted summation of the similarity of the re-identified features, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the vehicle modification confidence.

S1020: In a case where the reliability of the vehicle modification device is less than a preset reliability threshold, determine that the target vehicle has been modified in a modification detection result of the target vehicle.

S1030: If the vehicle modification reliability is greater than or equal to the preset reliability threshold, determine that the target vehicle modification detection result is that the target vehicle has not been modified.

For example, the prediction confidence of the vehicle color attribute is s ₁ , the prediction confidence of the vehicle type attribute is s ₂ , the similarity between the first recognition feature and the second recognition feature is s ₃ , and the preset reliability threshold is thr, Then s ₁ , s ₂ and s ₃ can be weighted and summed, as shown in formula (1):

s=w ₁ *s ₁ +w ₂ *s ₂ +w ₃ *s ₃ (1);

Among them, s is the vehicle modification reliability obtained after the weighted summation, where w ₁ , w ₂ , and w ₃ can be flexibly adjusted according to needs. For example, it is necessary to give greater weight to the similarity of re-identification features, and w ₃ can be set greater than w ₁ and w ₂ , eg set w ₁ =0.25, w ₂ =0.25, w ₃ =0.5. In the case of s<thr, it is judged that the target vehicle has been modified, and in the case of s≥thr, it is judged that the target vehicle has not been modified.

In the above technical solution, weighted summation is performed on the re-identification feature similarity, the prediction confidence of the matching color prediction category and the prediction confidence of the matching type prediction category, and the corresponding vehicle modification equipment confidence can be obtained; in addition, in order to adapt to different scenarios , the corresponding weights can be flexibly adjusted to obtain the reliability of the vehicle modification device in the corresponding scene; in addition, the modification detection result of the target vehicle can be determined according to the vehicle modification device reliability and the preset reliability threshold, thereby improving the determination of the modification detection result. convenience.

In some implementations, when the modification detection result of the target vehicle is that the target vehicle has been modified, output vehicle modification prompt information. Compared with the related technology that needs to manually judge whether the vehicle is modified, the vehicle modification result is determined through heavy feature comparison, and when the detection result is modified, the modification prompt information is output to improve the detection efficiency of vehicle modification.

It can be seen from the technical solutions provided by the above-mentioned embodiments of the present disclosure that in the embodiments of the present disclosure, the electronic device determines the modification detection result of the target vehicle based on the re-identification feature similarity of the target vehicle and the vehicle attribute confidence of the target vehicle. Based on the two different aspects of the re-identification feature similarity of the vehicle and the vehicle attribute confidence of the target vehicle, the modification detection of the target vehicle is relatively inefficient compared to the method of manually performing vehicle modification recognition in the related art, and due to human Due to limitations of eye observation, the accuracy of vehicle modification recognition results is not high; the vehicle modification detection method provided in the embodiments of the present disclosure can improve the efficiency, robustness and accuracy of vehicle modification detection, and can save labor costs.

Those skilled in the art can understand that in the above-mentioned method of specific implementation, the writing order of each step does not imply a strict execution order and constitutes any limitation on the implementation process, and the execution order of each step should be based on its function and possible internal Logically OK. The serial numbers of the above embodiments of the present application are for description only, and do not represent the advantages and disadvantages of the embodiments.

Based on the same inventive concept, the embodiment of the present disclosure also provides a vehicle modification detection device corresponding to the vehicle modification detection method. Since the problem-solving principle of the device in the embodiment of the disclosure is similar to the above-mentioned vehicle modification detection method of the embodiment of the disclosure, therefore The implementation of the device can refer to the implementation of the method. Figure 11 is a vehicle modification detection device provided by an embodiment of the present disclosure, including:

The image acquisition part 1110 of the vehicle to be detected is configured to acquire the image of the vehicle to be detected corresponding to the target vehicle;

The re-identification feature similarity determining part 1120 is configured to determine the re-identification feature similarity of the target vehicle in the image of the vehicle to be detected; wherein the re-identification feature similarity is based on the The first identification feature, and the determination of the second identification feature extracted from the original vehicle image corresponding to the target vehicle;

The vehicle attribute confidence determination part 1130 is configured to determine the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected;

The modification detection result determining part 1140 is configured to determine the modification detection result of the target vehicle based on the re-identification feature similarity and the vehicle attribute confidence.

In some embodiments, the device also includes:

The vehicle identification information identification part is configured to perform vehicle identification information identification on the image of the vehicle to be detected to obtain a vehicle identification information identification result of the target vehicle;

The vehicle identification information matching part is configured to match the original vehicle image and the vehicle attribute corresponding to the original vehicle image from a preset vehicle information database based on the identification result of the vehicle identification information.

In some implementations, the vehicle identification information includes license plate information of the target vehicle;

The identification part of the vehicle identification information includes:

The license plate information recognition part is configured to perform license plate information recognition on the image of the vehicle to be detected to obtain the license plate information of the target vehicle;

The first determining part is configured to determine the license plate information of the target vehicle as the identification result of the vehicle identification information of the target vehicle.

In some implementations, the re-identification feature similarity determination part 1120 includes:

A feature distance calculation part configured to determine a re-identification feature distance between the first re-identification feature and the second re-identification feature;

The second determination part is configured to determine the re-identification feature similarity based on the re-identification feature distance;

Wherein, the first re-identification feature is used to characterize the feature description information of the target vehicle in the vehicle image to be detected, and the second re-identification feature is used to characterize the feature of the target vehicle in the original vehicle image Description.

In some implementations, the vehicle attribute confidence determination part 1130 includes:

The vehicle attribute identification part is configured to perform vehicle attribute identification on the image of the vehicle to be detected, to obtain a plurality of predicted categories of vehicle attributes corresponding to the target vehicle, and respective prediction confidence levels corresponding to the plurality of predicted categories;

a matching prediction category determination section configured to determine a matching prediction category matching the vehicle attribute corresponding to the original vehicle image from the plurality of prediction categories;

A third determination section configured to determine the prediction confidence level of the matching predicted category as the vehicle attribute confidence level.

In some embodiments, the vehicle attributes include vehicle color attributes and vehicle type attributes;

The vehicle attribute identification part includes:

The vehicle color attribute identification part is configured to perform vehicle color attribute identification on the vehicle image to be detected, and obtain a plurality of color prediction categories corresponding to vehicle color attributes, and prediction confidence levels corresponding to the plurality of color prediction categories;

The vehicle type attribute identification part is configured to perform vehicle type attribute identification on the image of the vehicle to be detected to obtain a plurality of type prediction categories corresponding to the vehicle type attributes, and prediction confidence levels corresponding to the plurality of type prediction categories.

In some embodiments, the match prediction categories include match color prediction categories and match type prediction categories;

The third determination part includes:

a fourth determination section configured to determine the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category as the vehicle attribute confidence;

The modification detection result determination part includes:

The information fusion part is configured to perform information fusion on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the modification detection of the target vehicle result.

In some embodiments, the information fusion part includes:

The weighted summing part is configured to perform weighted summation on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the vehicle modification equipment confidence;

The modification determination part is configured to determine that the modification detection result of the target vehicle is that the target vehicle has been modified when the reliability of the vehicle modification is less than a preset reliability threshold; or,

The non-modification determining part is configured to determine that the target vehicle is not modified in the modification detection result of the target vehicle when the vehicle modification reliability is greater than or equal to the preset reliability threshold.

In some embodiments, the device also includes:

The output part is configured to output vehicle modification prompt information when the modification detection result of the target vehicle is that the target vehicle has been modified.

With regard to the apparatus in the above embodiments, the specific manner in which each part executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

In the embodiments of the present disclosure and other embodiments, a "part" may be a part of a circuit, a part of a processor, a part of a program or software, etc., of course it may also be a unit, a module or a non-modular one.

FIG. 12 is a schematic structural diagram of an electronic device for vehicle modification detection provided by an embodiment of the present disclosure. The electronic device may be a terminal, and its internal structure may be as shown in FIG. 12 . The electronic device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Wherein, the processor of the electronic device is configured to provide calculation and control capabilities. The storage of the electronic device includes a non-volatile storage medium, a volatile storage medium or an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the electronic device is configured to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a vehicle modification detection method is realized. The display screen of the electronic device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device may be a touch layer covered on the display screen, or a button, a trackball or a touch pad provided on the housing of the electronic device , and can also be an external keyboard, touchpad, or mouse.

FIG. 13 is a schematic structural diagram of an electronic device for vehicle modification detection provided by an embodiment of the present disclosure. The electronic device may be a server, and its internal structure may be as shown in FIG. 13 . The electronic device includes a processor, memory and network interface connected by a system bus. Wherein, the processor of the electronic device is configured to provide calculation and control capabilities. The storage of the electronic device includes a non-volatile storage medium, a volatile storage medium or an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the electronic device is configured to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a vehicle modification detection method is realized.

Those skilled in the art can understand that the structures shown in FIG. 12 and FIG. 13 are structural schematic diagrams of partial structures related to the disclosed solution, and do not constitute a limitation on the electronic equipment that the disclosed solution is configured on. The electronic equipment may include There may be more or fewer components than shown in the figures, or certain components may be combined, or have different component arrangements.

In some embodiments, there is also provided an electronic device, including: a processor; a memory configured to store instructions executable by the processor; wherein, the processor is configured to execute the instructions, so as to realize the implementation of the present disclosure. The vehicle modification detection method in the example.

In some embodiments, a computer-readable storage medium is also provided. When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device can perform the vehicle modification detection in the embodiments of the present disclosure. method. A computer readable storage medium may be a tangible device that holds and stores instructions for use by an instruction execution device, and may be a volatile storage medium or a nonvolatile storage medium.

In some embodiments, a computer program product containing instructions is also provided, and when the computer program or instructions are run on the electronic device, the electronic device is made to execute the vehicle as described in any one of the above embodiments. Modify the detection method.

Wherein, the above-mentioned computer program product may be implemented by means of hardware, software or a combination thereof. In some embodiments, the computer program product may be embodied as a computer storage medium, and in other embodiments, the computer program product may be embodied as a software product, such as a software development kit (Software Development Kit, SDK) and the like.

The device involved in the embodiments of the present disclosure may be at least one of a system, a method, and a computer program product. A computer program product may include a computer readable storage medium having computer readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.

A computer readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer readable storage medium may be, for example, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. Examples of computer-readable storage media (a non-exhaustive list) include: portable computer disks, hard disks, Random Access Memory (RAM), Read-Only Memory (ROM), erasable Electrical Programmable Read Only Memory (EPROM) or flash memory, Static Random-Access Memory (Static Random-Access Memory, SRAM), Portable Compact Disc Read-Only Memory (CD-ROM), Digital Video Discs (DVDs), memory sticks, floppy disks, mechanically encoded devices such as punched cards or raised structures in grooves with instructions stored thereon, and any suitable combination of the foregoing. As used herein, computer-readable storage media are not to be construed as transient signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., pulses of light through fiber optic cables), or transmitted electrical signals.

The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device over at least one of a network, such as the Internet, a local area network, a wide area network, and a wireless network. . The network may include at least one of copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and edge servers. A network adapter card or a network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing/processing device .

Computer program instructions for performing the operations of the present disclosure may be assembly instructions, Industry Standard Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or in one or more source or object code written in any combination of programming languages, including object-oriented programming languages—such as Smalltalk, C++, etc., and conventional procedural programming languages, such as the “C” language or similar programming languages. Computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server implement. In cases involving a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or it may be connected to an external computer (for example, using Internet Service Provider to connect via the Internet). In some embodiments, electronic circuits, such as programmable logic circuits, FPGAs, or programmable logic arrays (Programmable Logic Arrays, PLAs), can be customized by using state information of computer-readable program instructions, which can execute computer-readable Read program instructions, thereby implementing various aspects of the present disclosure.

Other implementations of the disclosed embodiments will be readily apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. The embodiments of the present disclosure are intended to cover any modification, use or adaptation of the present disclosure. These modifications, uses or adaptations follow the general principles of the present disclosure and include common knowledge in the technical field not disclosed by the embodiments of the present disclosure. or customary technical means. It is intended that the specification and examples be considered exemplary only, with a true scope and spirit of the disclosed embodiments being indicated by the following claims.

It should be understood that the embodiments of the present disclosure are not limited to the precise structures that have been described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the disclosed embodiments is limited only by the appended claims.

Industrial Applicability

Embodiments of the present disclosure provide a vehicle modification detection method, device, electronic equipment, computer-readable storage medium, and computer program product, wherein the vehicle modification detection method includes: acquiring an image of the vehicle to be detected corresponding to the target vehicle; determining the vehicle to be detected In the image, the re-identification feature similarity of the target vehicle; wherein, the re-identification feature similarity is determined based on the first re-identification feature extracted from the image of the vehicle to be detected and the second re-identification feature extracted from the original vehicle image corresponding to the target vehicle ; Determine the vehicle attribute confidence of the target vehicle in the vehicle image to be detected; determine the modification detection result of the target vehicle based on the re-identification feature similarity and the vehicle attribute confidence. Compared with the method of manually identifying vehicle modifications in the related art, the efficiency is low, and due to the limitation of human eye observation, the accuracy of the recognition results of vehicle modification is not high; in the above technical solution, the electronic device is based on the weight of the target vehicle. Identifying feature similarity and vehicle attribute confidence of the target vehicle to determine the modification detection result of the target vehicle can improve the efficiency and accuracy of vehicle modification detection and save labor costs. Moreover, the above technical solution detects the modification of the target vehicle from the two different aspects of the re-identification feature similarity of the target vehicle and the confidence of the vehicle attribute of the target vehicle, which can improve the accuracy and robustness of the target vehicle modification detection.

Claims (20)

A vehicle modification detection method, comprising: Obtain the image of the vehicle to be detected corresponding to the target vehicle; determining the re-identification feature similarity of the target vehicle in the vehicle image to be detected; wherein the re-identification feature similarity is based on the first re-identification feature extracted from the vehicle image to be detected, and from the target vehicle Determination of the second identification feature extracted from the original vehicle image corresponding to the vehicle; determining the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected; Based on the similarity of the re-identified features and the confidence of the vehicle attribute, a modification detection result of the target vehicle is determined. The method according to claim 1, wherein said obtaining the image of the vehicle to be detected corresponding to the target vehicle further comprises: Carrying out vehicle identification information identification on the image of the vehicle to be detected to obtain a vehicle identification information identification result of the target vehicle; Based on the identification result of the vehicle identification information, the original vehicle image and the vehicle attribute corresponding to the original vehicle image are matched from a preset vehicle information database. The method according to claim 2, wherein the vehicle identification information includes license plate information of the target vehicle; The identifying the vehicle identification information on the image of the vehicle to be detected to obtain the identification result of the vehicle identification information of the target vehicle includes: Carrying out license plate information recognition on the image of the vehicle to be detected to obtain the license plate information of the target vehicle; Determining the license plate information of the target vehicle as the identification result of the vehicle identification information of the target vehicle. The method according to any one of claims 1 to 3, wherein the determining the re-identification feature similarity of the target vehicle in the image of the vehicle to be detected comprises: determining a re-identification feature distance between the first re-identification feature and the second re-identification feature; determining the re-identification feature similarity based on the re-identification feature distance; Wherein, the first re-identification feature is used to characterize the feature description information of the target vehicle in the vehicle image to be detected, and the second re-identification feature is used to characterize the feature of the target vehicle in the original vehicle image Description. The method according to any one of claims 1 to 3, wherein said determining the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected comprises: Carrying out vehicle attribute recognition on the image of the vehicle to be detected, to obtain a plurality of predicted categories of vehicle attributes corresponding to the target vehicle, and prediction confidence degrees corresponding to each of the plurality of predicted categories; From the plurality of predicted categories, determining a matching predicted category that matches the vehicle attribute corresponding to the original vehicle image; The prediction confidence degree of the matching prediction category is determined as the vehicle attribute confidence degree. The method of claim 5, wherein the vehicle attributes include vehicle color attributes and vehicle type attributes; The identifying the vehicle attribute on the image of the vehicle to be detected to obtain a plurality of predicted categories of the vehicle attribute corresponding to the target vehicle, and respective prediction confidence levels corresponding to the plurality of predicted categories includes: Carrying out vehicle color attribute recognition on the vehicle image to be detected to obtain a plurality of color prediction categories corresponding to vehicle color attributes, and prediction confidence levels corresponding to the plurality of color prediction categories; Carrying out vehicle type attribute recognition on the image of the vehicle to be detected to obtain multiple type prediction categories corresponding to the vehicle type attributes, and prediction confidence levels corresponding to the multiple type prediction categories. The method of claim 6, wherein the matching prediction categories include a matching color prediction category and a matching type prediction category; The determining the prediction confidence degree of the matching prediction category as the vehicle attribute confidence degree includes: determining the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category as the vehicle attribute confidence; The determining the modification detection result of the target vehicle based on the re-identification feature similarity and the vehicle attribute confidence includes: Information fusion is performed on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain a modification detection result of the target vehicle. The method according to claim 7, wherein the information fusion is performed on the similarity of the re-identification feature, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the The modification detection results of the target vehicle, including: Carrying out weighted summation of the similarity of the re-identification feature, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the vehicle modification equipment confidence; If the reliability of the vehicle modification device is less than a preset reliability threshold, it is determined that the modification detection result of the target vehicle is that the target vehicle has been modified; or, In a case where the vehicle refitting reliability is greater than or equal to the preset reliability threshold, it is determined that the target vehicle is not refitted as a refitting detection result. The method according to any one of claims 1 to 8, wherein the method further comprises: If the modification detection result of the target vehicle is that the target vehicle has been modified, output vehicle modification prompt information. A vehicle modification detection device, comprising: The image acquisition part of the vehicle to be detected is configured to acquire the image of the vehicle to be detected corresponding to the target vehicle; The re-identification feature similarity determining part is configured to determine the re-identification feature similarity of the target vehicle in the image of the vehicle to be detected; wherein the re-identification feature similarity is based on the first extracted from the image of the vehicle to be detected One identification feature, and the determination of the second identification feature extracted from the original vehicle image corresponding to the target vehicle; a vehicle attribute confidence determination part configured to determine the vehicle attribute confidence of the target vehicle in the image of the vehicle to be detected; The modification detection result determination part is configured to determine the modification detection result of the target vehicle based on the re-identification feature similarity and the vehicle attribute confidence. The device according to claim 10, wherein the device further comprises: The vehicle identification information identification part is configured to perform vehicle identification information identification on the image of the vehicle to be detected to obtain a vehicle identification information identification result of the target vehicle; The vehicle identification information matching part is configured to match the original vehicle image and the vehicle attribute corresponding to the original vehicle image from a preset vehicle information database based on the identification result of the vehicle identification information. The apparatus according to claim 11, wherein the vehicle identification information includes license plate information of the target vehicle; The identification part of the vehicle identification information includes: The license plate information recognition part is configured to perform license plate information recognition on the image of the vehicle to be detected to obtain the license plate information of the target vehicle; The first determining part is configured to determine the license plate information of the target vehicle as the identification result of the vehicle identification information of the target vehicle. The device according to any one of claims 10 to 12, wherein the re-identification feature similarity determining part comprises: A feature distance calculation part configured to determine a re-identification feature distance based on the first re-identification feature and the second re-identification feature; The second determination part is configured to determine the re-identification feature similarity based on the re-identification feature distance; Wherein, the first re-identification feature is used to characterize the feature description information of the target vehicle in the vehicle image to be detected, and the second re-identification feature is used to characterize the feature of the target vehicle in the original vehicle image Description. The apparatus according to any one of claims 10 to 12, wherein the vehicle attribute confidence determination part comprises: The vehicle attribute identification part is configured to perform vehicle attribute identification on the image of the vehicle to be detected, to obtain a plurality of predicted categories of vehicle attributes corresponding to the target vehicle, and respective prediction confidence levels corresponding to the plurality of predicted categories; a matching prediction category determination section configured to determine a matching prediction category matching the vehicle attribute corresponding to the original vehicle image from the plurality of prediction categories; A third determination section configured to determine the prediction confidence level of the matching predicted category as the vehicle attribute confidence level. The apparatus of claim 14, wherein the vehicle attributes include a vehicle color attribute and a vehicle type attribute; The vehicle attribute identification part includes: The vehicle color attribute identification part is configured to perform vehicle color attribute identification on the vehicle image to be detected, and obtain a plurality of color prediction categories corresponding to vehicle color attributes, and prediction confidence levels corresponding to the plurality of color prediction categories; The vehicle type attribute identification part is configured to perform vehicle type attribute identification on the image of the vehicle to be detected to obtain a plurality of type prediction categories corresponding to the vehicle type attributes, and prediction confidence levels corresponding to the plurality of type prediction categories. The apparatus of claim 15, wherein, The matching prediction categories include matching color prediction categories and matching type prediction categories; The third determination part includes: a fourth determination section configured to determine the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category as the vehicle attribute confidence; The modification detection result determination part includes: The information fusion part is configured to perform information fusion on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the modification detection of the target vehicle result. The apparatus according to claim 16, wherein the information fusion part comprises: The weighted summing part is configured to perform weighted summation on the re-identification feature similarity, the prediction confidence of the matching color prediction category, and the prediction confidence of the matching type prediction category to obtain the vehicle modification equipment confidence; The modification determination part is configured to determine that the modification detection result of the target vehicle is that the target vehicle has been modified when the reliability of the vehicle modification is less than a preset reliability threshold; or, The non-modification determining part is configured to determine that the target vehicle is not modified in the modification detection result of the target vehicle when the vehicle modification reliability is greater than or equal to the preset reliability threshold. An electronic device comprising: processor; a memory configured to store said processor-executable instructions; Wherein, the processor is configured to execute the instructions, so as to realize the vehicle modification detection method according to any one of claims 1-9. A computer-readable storage medium, when the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device can perform the vehicle modification according to any one of claims 1 to 9 Detection method. A computer program product, the computer program product comprising a computer program or an instruction, when the computer program or instruction is run on an electronic device, the electronic device is made to execute any one of claims 1 to 9 vehicle modification detection method.

Images