Target object recognition method, device, equipment and storage medium

By obtaining and utilizing the perspective information and features of the vehicle and combining the perspective information to perform feature distance calculation, the problem of misjudgment caused by the perspective difference in the vehicle recognition is solved, and the recognition accuracy is improved.

CN111767940BActive Publication Date: 2024-11-26YUANLI TUXIN (CHONGQING) TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010406727.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-05-14
Publication Date
2024-11-26
Estimated Expiration
2040-05-14

AI Technical Summary

Technical Problem

The prior art has the problem of inaccurate identification results in vehicle recognition, especially because the similarity of the same target object at different shooting angles is smaller than the similarity of different target objects at the same shooting angles, resulting in misjudgment of the recognition results. .

Method used

By obtaining the viewing angle information and features of the image to be identified and the reference image, and combining the viewing angle information to perform feature distance calculation, determine whether the image to be identified and the reference image to be identified contains the same target object, avoiding a simple recognition method based on similarity, and improving the recognition accuracy.

Benefits of technology

It effectively avoids misjudgment of identification caused by perspective differences, improves the accuracy of vehicle identification, and ensures the reliability of identification results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111767940B_ABST
    Figure CN111767940B_ABST
Patent Text Reader

Abstract

The present application relates to a target object recognition method, device, equipment and storage medium, which obtains the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image, and then determines whether the image to be recognized and the corresponding reference image contain the same target object based on the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image. In the above method, the computer device makes full use of the viewing angle information of the image to be recognized and the viewing angle information of the reference image to achieve the purpose of identifying the target object in the image to be recognized. The method avoids the situation that the recognition result of the target object to be recognized is easily misjudged due to the similarity of the same target object at different shooting angles being less than the similarity of different target objects at the same shooting angle, thereby improving the accuracy of the recognition result of the target object to be recognized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of image recognition technology, and in particular to a method, device, equipment and storage medium for identifying a target object. Background Art

[0002] As cities expand and the number of cars increases, there are scenarios where vehicles need to be identified in the fields of public safety and intelligent transportation.

[0003] When identifying a vehicle, a picture of the vehicle is often input into a preset recognition model, a feature vector of a fixed dimension is obtained through the preset recognition model, and a determination is made based on the similarity between the picture of the vehicle to be identified and the reference picture of the vehicle to be identified whether the vehicle in the picture of the vehicle to be identified and the vehicle in the reference picture of the vehicle are the same vehicle, thereby obtaining a recognition result of the picture of the vehicle to be identified.

[0004] However, the recognition result of the vehicle recognition using the above method is inaccurate. Summary of the invention

[0005] Based on this, it is necessary to provide a target object recognition method, device, equipment and storage medium to address the problem of inaccurate recognition results for vehicles.

[0006] In a first aspect, a method for identifying a vehicle comprises:

[0007] Obtaining viewing angle information of at least one image to be identified and viewing angle information of a reference image;

[0008] Acquire features of the image to be identified and features of the reference image;

[0009] According to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image, it is determined whether the image to be identified and the corresponding reference image contain the same target object.

[0010] In a second aspect, a target object recognition device is provided, the device comprising:

[0011] A first acquisition module, used to acquire viewing angle information of at least one image to be recognized and viewing angle information of a reference image;

[0012] A second acquisition module is used to acquire the features of the image to be identified and the features of the reference image;

[0013] The recognition module is used to determine whether the image to be recognized and the corresponding reference image contain the same target object based on the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image.

[0014] In a third aspect, a computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned target object recognition method when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium stores a computer program, which implements the steps of the above-mentioned target object recognition method when executed by a processor.

[0016] The above-mentioned target object recognition method, device, equipment and storage medium first obtain the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image, and then determine whether the image to be recognized and the corresponding reference image contain the same target object according to the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image. In the above-mentioned recognition method, the computer device makes full use of the viewing angle information of the image to be recognized and the viewing angle information of the reference image to achieve the purpose of recognizing the target object in the image to be recognized. This method avoids the situation that the recognition result of the target object to be recognized is easily misjudged due to the similarity of the same target object under different shooting angles being less than the similarity of different target objects under the same shooting angle, thereby improving the accuracy of the recognition result of the target object to be recognized. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of an application environment of a target object recognition method provided by an embodiment;

[0018] Figure 2 A schematic diagram of a flow chart of a target object recognition method provided by an embodiment;

[0019] Figure 3 for Figure 2 A flowchart of an implementation method of Example S103;

[0020] Figure 4 for Figure 3 A flowchart of an implementation method of Example S201;

[0021] Figure 5 for Figure 4 A flowchart of an implementation method of embodiment S301;

[0022] Figure 6 for Figure 3 A flowchart of an implementation method of Example S202;

[0023] Figure 7 A schematic diagram of the structure of an image recognition network provided in one embodiment;

[0024] Figure 8 A schematic diagram of the structure of an image recognition network provided in another embodiment;

[0025] Fig. 9 for Figure 2 A flowchart of an implementation method of Example S101;

[0026] Fig.10 A schematic diagram of the structure of an image recognition network provided in another embodiment;

[0027] Fig.11 for Figure 2 A flowchart of an implementation method of Example S102;

[0028] Fig.12 A flowchart of a training method provided by an embodiment;

[0029] Fig.13 for Fig.12 A flowchart of an implementation method of embodiment S802;

[0030] Fig.14 for Fig.12 A flowchart of another implementation method of embodiment S802;

[0031] Fig.15 A schematic diagram of the structure of a training network provided by an embodiment;

[0032] Fig.16 A schematic diagram of the structure of a target object recognition device provided by an embodiment;

[0033] Fig.17 An internal structural diagram of a computer device provided for one embodiment. DETAILED DESCRIPTION

[0034] The vehicle identification method, device, equipment and storage medium provided by the present application are intended to solve the problem of inaccurate identification results of vehicles by traditional methods. The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems will be described in detail below through embodiments and in conjunction with the accompanying drawings. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0035] The target object recognition method provided in this embodiment can be applied to Figure 1In the application environment shown. Among them, the computer device 102 communicates with the image acquisition device 104 through a wired or wireless network. The computer device 102 is used to identify and process the target object in the image acquired by the image acquisition device 104. The computer device 102 can be, but is not limited to, an electronic device with a data processing function such as a smart phone, a tablet computer, a laptop computer, a desktop computer or a personal digital assistant. This embodiment does not limit the specific form of the computer device 102. The image acquisition device 104 can be, but is not limited to, a camera, a camera and an electronic device with a shooting function. It should be noted that the image acquisition device 104 and the computer device 102 can be the same electronic device integrated together, Figure 1 The application environment shown is only an example of the connection relationship between the image acquisition device 104 and the computer device 102. The embodiment of the present application does not limit whether the image acquisition device 104 and the computer device 102 are integrated into the same electronic device. When the image acquisition device 104 and the computer device 102 are integrated together to form an electronic device, the electronic device can be installed on any carrier, for example, a vehicle.

[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0037] Figure 2 A flow chart of a target object recognition method provided by an embodiment, the method is applied to Figure 1 The computer device shown in . This embodiment relates to a specific process in which a computer device identifies a target object in an image to be identified acquired by an image acquisition device. The method includes the following steps:

[0038] S101: Obtain viewing angle information of at least one image to be recognized and viewing angle information of a reference image.

[0039] Among them, the image to be identified may include any type of target object. For example, the image to be identified may include people, scenery, animals, vehicles, etc. The type of target object contained in the image to be identified can be determined according to the actual user identification needs, and this embodiment does not limit this. In addition, the image to be identified may include at least one target object of the same type, and may also include at least one target object of a different type, which is not limited in this embodiment. In one application, if the image to be identified includes a vehicle, the vehicle may include but is not limited to vehicles, ships, aircraft, bicycles, and electric vehicles, and the image to be identified may include vehicles from any shooting angle. In actual applications, the image to be identified can be Figure 1The image to be recognized can be acquired by the image acquisition device shown in , wherein the image acquisition device can be a camera, a webcam or an electronic device with a shooting function; optionally, the image to be recognized can also be acquired by Figure 1 The computer device in the embodiment is obtained from a database or downloaded from the Internet. This embodiment does not limit the method of obtaining the image to be identified. It should be noted that the image to be identified can be multiple images. For example, when the image acquisition device is a camera, it can acquire video data, and then several frames of images can be selected from the video data as the image to be identified.

[0040] The reference image includes objects of the same type as the target object to be identified. For example, if the target object to be identified is a vehicle, the reference image includes the vehicle. If the target object to be identified is a face, the reference image includes the face. The computer device can pre-set the reference image. Specifically, the reference image can be determined by receiving a command input by the user, or the reference image can be automatically determined according to preset conditions. The embodiments of the present application do not limit this.

[0041] The viewing angle information of the image to be identified is used to indicate the shooting viewing angle of the image to be identified, and the shooting viewing angle can be determined according to the parts of the target object. For example, if the target object includes parts such as the front, left front, right front, left side, rear, right side, left rear or right rear, correspondingly, the shooting viewing angle specifically indicates the front, left front, right front, left side, rear, right side, left rear or right rear of the target object in the image to be identified. Of course, the present application can also divide the target object into other parts, so that the shooting viewing angle may not be limited to the left front, right front, left side, rear, right side, left rear or right rear, but may be other directions, and the embodiments of the present application do not limit this. The viewing angle information of the reference image is used to indicate the shooting viewing angle of the reference image, and the shooting viewing angle may specifically indicate the front, left front, right front, left side, rear, right side, left rear or right rear of the object in the reference image, and the embodiments of the present application do not limit this. It should be noted that the viewing angle information of the image to be identified and the viewing angle information of the reference image may be the same or different.

[0042] In this embodiment, the computer device may first obtain the image to be identified through the image acquisition device, and then further obtain the perspective information of the image to be identified by analyzing the shooting perspective of the target object in the identified image; at the same time, the computer device may obtain the perspective information of the reference image by analyzing the shooting perspective of the object in the predetermined reference image. It should be noted that when the computer device identifies multiple images to be identified, and the multiple images to be identified all contain one type of target object, the corresponding reference image may be one reference image; when the computer device identifies multiple images to be identified, and the multiple images to be identified contain multiple types of target objects, the reference image may be multiple images corresponding to the multiple images to be identified.

[0043] S102: Acquire features of the image to be identified and features of the reference image.

[0044] When the computer device obtains the image to be identified and the reference image, it can further perform feature extraction operations on the image to be identified and the reference image, thereby obtaining the features of the image to be identified and the reference image for subsequent analysis and use. The computer device can use an existing neural network for feature extraction to extract features of the image to be identified and the reference image simultaneously or separately, or it can obtain the features of the image to be identified and the reference image through other means, which is not limited in this embodiment.

[0045] S103 , determining whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image.

[0046] When the computer device obtains the viewing angle information of the image to be identified and the viewing angle information of the reference image, it can determine whether the shooting angle of the target object in the image to be identified is the same as the shooting angle of the object in the reference image. If they are the same, the computer device can first determine the features of the image to be identified with the same shooting angle, and then further based on the image recognition method in the same shooting angle application environment, by comparing and analyzing the features of the image to be identified with the same shooting angle and the features of the corresponding reference image, to determine whether the image to be identified and the corresponding reference image contain the same target object. If they are not the same, the computer device can first determine the features of the image to be identified with different shooting angles, and then further based on the image recognition method in the different shooting angle application environment, by comparing and analyzing the features of the image to be identified with different shooting angles and the features of the corresponding reference image, to determine whether the image to be identified and the corresponding reference image contain the same target object. The above-mentioned image recognition method based on the application environment of the same shooting perspective is used to identify the target object in the image to be identified according to the characteristics of the image to be identified with the same shooting perspective and the characteristics of the corresponding reference image, and obtain the recognition result; the above-mentioned image recognition method based on the application environment of different shooting perspectives is used to identify the target object in the image to be identified according to the characteristics of the image to be identified with different shooting perspectives and the characteristics of the corresponding reference image, and obtain the recognition result.

[0047] The target object recognition method provided in the above embodiment first obtains the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image, and then determines whether the image to be recognized and the corresponding reference image contain the same target object according to the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image. In the above recognition method, the computer device makes full use of the viewing angle information of the image to be recognized and the viewing angle information of the reference image to achieve the purpose of recognizing the target object in the image to be recognized. This method avoids the situation that the recognition result of the target object to be recognized is easily misjudged due to the similarity of the same target object under different shooting angles being less than the similarity of different target objects under the same shooting angle, thereby improving the accuracy of the recognition result of the target object to be recognized.

[0048] Figure 3 for Figure 2 A flow chart of an implementation method of Example S103 is as follows: Figure 3 As shown, the above S103 "determining whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image" includes:

[0049] S201. Determine a viewing angle feature distance between the image to be identified and the corresponding reference image according to viewing angle information of the image to be identified, viewing angle information of the reference image, features of the image to be identified and features of the reference image.

[0050] Among them, the viewing angle feature distance represents the distance between the feature vector of the image to be identified and the feature vector of the corresponding reference image. When the computer device obtains the viewing angle information of the image to be identified and the viewing angle information of the reference image, it can determine whether the shooting angle of the target object in the image to be identified is the same as the shooting angle of the object in the reference image. If they are the same, the computer device can first determine the feature vector of the image to be identified with the same shooting angle according to the features of the image to be identified, and determine the feature vector of the corresponding reference image according to the features of the corresponding reference image, and then further adopt the corresponding feature distance calculation method to calculate the distance (for example, Euclidean distance) between the feature vector of the image to be identified with the same shooting angle and the feature vector of the corresponding reference image, and use the distance as the viewing angle feature distance between the image to be identified with the same shooting angle and the corresponding reference image, so as to determine whether the image to be identified and the reference image with the same shooting angle contain the same target object according to the viewing angle feature distance. Similarly, if they are not the same, the computer device may first determine the feature vector of the image to be identified with a different shooting angle according to the features of the image to be identified, and determine the feature vector of the corresponding reference image according to the features of the corresponding reference image, and then further adopt the corresponding feature distance calculation method to calculate the distance between the feature vector of the image to be identified with a different shooting angle and the feature vector of the corresponding reference image, and use the distance as the perspective feature distance between the image to be identified with a different shooting angle and the corresponding reference image, so as to determine whether the image to be identified with a different shooting angle and the reference image contain the same target object based on the perspective feature distance. The above two feature distance calculation methods can adopt any existing distance calculation method, and the two feature distance calculation methods can be the same or different, which is not limited in this embodiment.

[0051] S202: Determine whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle feature distance between the image to be identified and the corresponding reference image.

[0052] When the computer device obtains the viewing angle characteristic distance between the image to be identified (the image to be identified with the same shooting angle or the image to be identified with different shooting angles) and the corresponding reference image based on the aforementioned step S201, it can further determine whether the image to be identified and the corresponding reference image contain the same target object by analyzing the size of the viewing angle characteristic distance. For example, if the size of the viewing angle characteristic distance can meet the preset recognition accuracy, the computer device determines that the image to be identified and the corresponding reference image contain the same target object; if the size of the viewing angle characteristic distance does not meet the preset recognition accuracy, the computer device determines that the image to be identified and the corresponding reference image do not contain the same target object.

[0053] In practical applications, when the perspective information of multiple images to be identified acquired by the computer device is the same as the perspective information of the corresponding reference image, the perspective feature distance is also the same perspective feature distance; when the perspective information of multiple images to be identified acquired by the computer device is different from the perspective information of the corresponding reference image, the perspective feature distance is also the different perspective feature distance. When the perspective information of some of the multiple images to be identified acquired by the computer device is the same as the perspective information of the corresponding reference image, and the perspective information of some of the images to be identified is different from the perspective information of the corresponding reference image, the perspective feature distance includes the same perspective feature distance and the different perspective feature distance.

[0054] In the above application scenario, the present application also provides a specific implementation method of S201, such as Figure 4 As shown, the above S201 "determine the viewing angle feature distance between the image to be identified and the corresponding reference image according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the feature of the image to be identified and the feature of the reference image" includes:

[0055] S301. Obtain same-viewing angle features and / or different-viewing angle features of the image to be identified according to viewing angle information of the image to be identified, viewing angle information of a reference image, features of the image to be identified, and features of the reference image.

[0056] When the computer device obtains the perspective information of multiple images to be identified and the perspective information of the corresponding reference images, the computer device compares whether the perspective information of each image to be identified is the same as the perspective information of the corresponding reference image. If the perspective information of each image to be identified is the same as the perspective information of the corresponding reference image, the computer device uses a corresponding feature extraction method to extract features in each image to be identified, and uses the extracted features as the same-perspective features corresponding to each image to be identified. If the perspective information of each image to be identified is different from the perspective information of the corresponding reference image, the computer device uses a corresponding feature extraction method to extract features in each image to be identified, and uses the extracted features as the different-perspective features corresponding to each image to be identified. If the perspective information of some of the multiple images to be identified is the same as the perspective information of the corresponding reference image, and the perspective information of some of the images is different from the perspective information of the corresponding reference image, the computer device uses a corresponding feature extraction method to extract features in some of the images to be identified as the same-perspective features corresponding to some of the images to be identified, and uses a corresponding feature extraction method to extract features in another part of the images to be identified as the different-perspective features corresponding to another part of the images to be identified.

[0057] S302: Calculate the distance between the same-viewing angle feature of the image to be identified and the feature of the corresponding reference image to obtain the same-viewing angle feature distance between the image to be identified and the corresponding reference image.

[0058] In this embodiment, when the computer device obtains the same-perspective feature corresponding to the image to be identified, it can further adopt a corresponding distance calculation method to calculate the distance between the same-perspective feature of the image to be identified and the feature of the corresponding reference image (for example, Euclidean distance), and use the calculated distance as the same-perspective feature distance between the image to be identified and the corresponding reference image.

[0059] S303, and / or, calculating the distance between the different viewing angle features of the image to be identified and the features of the corresponding reference image, to obtain the different viewing angle feature distance between the image to be identified and the corresponding reference image.

[0060] In this embodiment, when the computer device obtains the different-perspective features corresponding to the image to be identified, it can further use the corresponding distance calculation method to calculate the distance between the different-perspective features of the image to be identified and the features of the corresponding reference image (for example, Euclidean distance), and use the calculated distance as the different-perspective feature distance between the image to be identified and the corresponding reference image. It can be understood that when the computer device simultaneously obtains the different-perspective features and the same-perspective features corresponding to the image to be identified, the corresponding distance calculation method can be used to calculate the distance between the different-perspective features and the features of the corresponding reference image, and the distance between the same-perspective features and the features of the corresponding reference image, and the different-perspective feature distance and the same-perspective feature distance can be obtained at the same time.

[0061] Optionally, an implementation of the above S301 "obtaining same-viewing angle features and / or different-viewing angle features of the image to be identified according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image" is as follows: Figure 5 As shown, including:

[0062] S401. According to the perspective information of the image to be identified and the perspective information of the corresponding reference image, determine whether the image to be identified and the corresponding reference image belong to the same perspective image. If so, execute step S402; if not, execute step S403.

[0063] This embodiment involves a step in which a computer device determines whether the viewing angle information of an image to be identified is the same as the viewing angle information of a corresponding reference image. If they are the same, it means that the shooting angle of the target object in the image to be identified is the same as the shooting angle of the object in the corresponding reference image. For example, the vehicle in the image to be identified is shot from the front, and the vehicle in the corresponding reference image is also shot from the front. If they are not the same, it means that the shooting angle of the target object in the image to be identified is different from the shooting angle of the object in the corresponding reference image. For example, the vehicle in the image to be identified is shot from the front, and the vehicle in the corresponding reference image is also shot from the side. The computer device will execute different steps according to different judgment results.

[0064] S402: extracting same-viewing angle features of the image to be identified from the features of the image to be identified.

[0065] This embodiment involves the specific operation of a computer device when determining that the viewing angle information of an image to be identified is the same as the viewing angle information of a corresponding reference image. Specifically, the computer device extracts features of an image to be identified that has the same shooting angle from features of one or more images to be identified, and directly uses them as the same viewing angle features of the image to be identified for subsequent use.

[0066] S403: extracting different-viewing angle features of the image to be identified from the features of the image to be identified.

[0067] This embodiment involves the specific operation of a computer device when determining that the viewing angle information of an image to be identified is different from the viewing angle information of a corresponding reference image. Specifically, the computer device extracts features of an image to be identified that has a different shooting angle from features of one or more images to be identified, and directly uses them as different viewing angle features of the image to be identified for subsequent use.

[0068] When the above application scenario is that the viewing angle characteristic distance calculated by the computer device includes both the same viewing angle characteristic distance and the different viewing angle characteristic distance, the present application also provides a specific implementation method of S202, such as Figure 6 As shown, the above S202 "determining whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle feature distance between the image to be identified and the corresponding reference image" includes:

[0069] S501: Fusing the same-viewing angle feature distance and the different-viewing angle feature distance to obtain a fused viewing angle feature distance between the image to be identified and the corresponding reference image.

[0070] When the computer device obtains the same-perspective feature distances of some images to be identified and the different-perspective feature distances of some images to be identified based on the above steps, a distance fusion method can be further adopted. For example, weighted accumulation and other operations are performed on the values ​​representing the distances to obtain the fused perspective feature distances between multiple images to be identified and the corresponding reference images.

[0071] S502: Determine whether the image to be identified and the corresponding reference image contain the same target object according to the fusion viewing angle feature distance.

[0072] When the computer device calculates the fusion perspective feature distance between multiple images to be identified and the corresponding reference images, it can be determined according to the fusion perspective feature distance which of the multiple images to be identified contain the same target object as the corresponding reference image and which images to be identified do not contain the same target object as the corresponding reference image.

[0073] The target recognition method provided in the above embodiment determines whether the image to be recognized contains the same target object as that in the reference image by respectively calculating the same-viewing angle feature distance and the different-viewing angle feature distance. This method not only determines the recognition result according to the feature distance, but also makes full use of the viewing angle information of the image to be recognized and the viewing angle information of the reference image, and selects the corresponding recognition method according to the same viewing angle information, and selects the corresponding recognition method according to different viewing angle information, thereby refining the recognition method, and thus greatly improving the recognition accuracy of the computer device for the target object in the image to be recognized.

[0074] In practical applications, computer devices usually use a trained image recognition network to identify target objects. Accordingly, the present application also provides a trained image recognition network. The computer device uses the image recognition network to determine whether the image to be identified and the corresponding reference image contain the same target object based on the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified, and the features of the reference image. The following embodiments describe in detail how a computer device uses the image recognition network provided by the present application to identify a target object, that is, the image recognition network can be applied to the target object recognition method described in any of the above embodiments.

[0075] Optionally, the above S103 "determine whether the image to be identified and the corresponding reference image contain the same target object based on the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image" specifically includes: inputting the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image into a preset image recognition network to identify the target object, and determine whether the image to be identified and the corresponding reference image contain the same target object.

[0076] The above image recognition network includes a same-view recognition network and a different-view recognition network, such as Figure 7 As shown, the same-view recognition network and the different-view recognition network are connected in parallel, and the parameters of the same-view recognition network and the different-view recognition network are not shared.

[0077] The above-mentioned same-viewing angle recognition network is used to extract the same-viewing angle features of the image to be recognized from the features of the image to be recognized according to the viewing angle information of the image to be recognized and the viewing angle information of the reference image; then, according to the same-viewing angle features of the image to be recognized and the features of the corresponding reference image, the same-viewing angle feature distance between the image to be recognized and the corresponding reference image is calculated; when the same-viewing angle feature distance is less than the first preset distance, it is determined that the image to be recognized and the corresponding reference image contain the same target object (the first recognition result in the figure); when the same-viewing angle feature distance is greater than or equal to the first preset distance, it is determined that the image to be recognized and the corresponding reference image do not contain the same target object (the first recognition result in the figure). The size of the first preset distance can be predetermined by the computer device according to the actual recognition accuracy, and this embodiment does not limit this. It is noted that the concepts of same-viewing angle features, same-viewing angle feature distances, etc. involved in this embodiment have been explained in the aforementioned description. Please refer to the aforementioned description for details, and the redundant description is not repeated here.

[0078] The above-mentioned different-viewing angle recognition network is used to extract the different-viewing angle features of the image to be recognized from the features of the image to be recognized according to the viewing angle information of the image to be recognized and the viewing angle information of the reference image; then, according to the different-viewing angle features of the image to be recognized and the features of the corresponding reference image, the different-viewing angle feature distance between the image to be recognized and the corresponding reference image is calculated; when the different-viewing angle feature distance of the image to be recognized is less than the second preset distance, it is determined that the image to be recognized and the corresponding reference image contain the same target object (the second recognition result in the figure); when the different-viewing angle feature distance is greater than or equal to the second preset distance, it is determined that the image to be recognized and the corresponding reference image do not contain the same target object (the second recognition result in the figure). The size of the second preset distance can be predetermined by the computer device according to the actual recognition accuracy, and this embodiment does not limit this. The second preset distance can be equal to the above-mentioned first preset distance, or it can be different from the above-mentioned first preset distance.

[0079] The above-mentioned same-viewing angle recognition network and different-viewing angle recognition network can both be a neural network or other types of machine learning networks. Specifically, they can be a multi-layer convolutional neural network or a deep convolutional neural network, etc., which is not limited in this embodiment.

[0080] In this embodiment, when the computer device obtains the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized, and the features of the reference image, these data or information can be input into a preset image recognition network, and the same-viewing angle recognition network in the image recognition network extracts the same-viewing angle features of the image to be recognized and the features of the corresponding reference image from the features of the image to be recognized according to the viewing angle information of the image to be recognized and the viewing angle information of the reference image, and then calculates the same-viewing angle feature distance between the image to be recognized and the corresponding reference image according to the same-viewing angle features of the image to be recognized and the features of the corresponding reference image, and then determines that the image to be recognized and the corresponding reference image do not contain the same target object according to the calculated same-viewing angle feature distance; correspondingly, the different-viewing angle recognition network in the image recognition network extracts the different-viewing angle features of the image to be recognized and the features of the corresponding reference image from the features of the image to be recognized according to the viewing angle information of the image to be recognized and the viewing angle information of the reference image, and then calculates the different-viewing angle feature distance between the image to be recognized and the corresponding reference image according to the different-viewing angle features of the image to be recognized and the features of the corresponding reference image, and then determines that the image to be recognized and the corresponding reference image do not contain the same target object according to the calculated different-viewing angle feature distance.

[0081] In an application scenario, when a computer device obtains the viewing angle information of an image to be recognized and the viewing angle information of a reference image, it can also obtain the information through a trained visual discrimination network. In the above application scenario, the present application provides another structure of an image recognition network (such as Figure 8 As shown), and the specific implementation of the above S101 applied to the network structure, for example, the above S101 "obtains viewing angle information of at least one image to be identified and viewing angle information of a reference image", such as Fig. 9 As shown, including:

[0082] S601: Input the image to be identified into a preset visual discrimination network to obtain the viewing angle information of the image to be identified.

[0083] The visual discrimination network is used to discriminate the shooting angle of the input image, and is a trained neural network or other types of networks. When the computer device obtains the image to be identified, the computer device inputs the image to be identified into the trained visual discrimination network to determine the shooting angle of the target object in the image to be identified, so as to determine the viewing angle information of the image to be identified.

[0084] S602: Input the reference image into the visual discrimination network to obtain the viewing angle information of the reference image.

[0085] When the computer device acquires the reference image, the computer device inputs the reference image into the trained visual discrimination network, determines the shooting angle of the object in the reference image, and determines the viewing angle information of the reference image. The visual discrimination network in this step is the same as the visual discrimination network in step S601. It can be understood that the computer device can simultaneously input the image to be identified and the reference image into the trained visual discrimination network to obtain the viewing angle information of the image to be identified and the viewing angle information of the reference image, respectively. When the above-mentioned visual discrimination network obtains the viewing angle information of the image to be identified and the viewing angle information of the reference image, it can also further determine whether the viewing angle information of the image to be identified and the viewing angle information of the reference image are the same, thereby obtaining a discrimination result, and inputting the judgment result into the same-viewing angle discrimination network, so that the same-viewing angle discrimination network extracts the features of the image to be identified with the same viewing angle from the features of the image to be identified, and inputting the judgment result into the different-viewing angle discrimination network, so that the different-viewing angle discrimination network extracts the features of the image to be identified with different viewing angles from the features to be identified.

[0086] In one application scenario, when the computer device obtains the features of the image to be recognized and the features of the reference image, it can also obtain them through a trained image feature extraction network. In the above application scenario, the present application provides another structure of an image recognition network (such as Fig.10 As shown), and the specific implementation of the above S102 applied to the network structure, for example, the above S102 "obtaining the features of the image to be identified and the features of the reference image", such as Fig.11 As shown, including:

[0087] S701, using a preset image feature extraction network to perform shallow feature extraction on the image to be identified, to obtain features of the image to be identified.

[0088] S702: Use an image feature extraction network to perform shallow feature extraction on the reference image to obtain features of the reference image.

[0089] The preset image feature extraction network is used for shallow feature extraction, which is a trained neural network or other types of networks. When the computer device obtains the image to be identified and the reference image, the computer device inputs the image to be identified and the reference image into the trained image feature extraction network for shallow feature extraction, and obtains the features of the image to be identified and the reference image respectively.

[0090] It should be noted that the above Fig.10The visual discrimination network and the image feature extraction network in the image recognition network shown are connected in parallel, the same-view recognition network and the different-view recognition network are connected in parallel, and the output ends of the visual discrimination network and the image feature extraction network are connected to the input ends of the same-view recognition network and the different-view recognition network. It is particularly noted that the output results of the visual discrimination network are different, and the corresponding same-view discrimination network and the different-view discrimination network are also different. For example, when the visual discrimination network outputs the view information of the image to be recognized and the view information of the reference image, the computer device inputs the view information of the image to be recognized, the view information of the reference image, the features of the image to be recognized and the features of the reference image into the input ends of the same-view recognition network and the different-view recognition network respectively. At this time, the same-view recognition network needs to first determine whether it belongs to the same view image according to the view information of the image to be recognized and the view information of the reference image. If so, the features of the image to be recognized belonging to the same view are further extracted from the features of the image to be recognized, and then the features of the image to be recognized at the same view are further extracted to obtain the same-view features of the image to be recognized. Similarly, the different-viewing angle recognition network also needs to first determine whether it belongs to the same viewing angle image according to the viewing angle information of the image to be recognized and the viewing angle information of the reference image. If they are not the same, the features of the image to be recognized that do not belong to the same viewing angle are further extracted from the features of the image to be recognized, and then the features of the images to be recognized with different viewing angles are further extracted to obtain the different viewing angle features of the image to be recognized. For another example, when the visual discrimination network outputs the discrimination result (described above), the computer device can further input the features of the image to be recognized with the same viewing angle and the features of the corresponding reference image into the same viewing angle recognition network according to the discrimination result, so that the same viewing angle recognition network further extracts the features of the image to be recognized with the same viewing angle, and obtains the same viewing angle features of the image to be recognized. Similarly, the computer device can also further input the features of the images to be recognized with different viewing angles and the features of the corresponding reference image into the different viewing angle recognition network according to the discrimination result, so that the different viewing angle recognition network further extracts the features of the images to be recognized with different viewing angles, and obtains the different viewing angle features of the image to be recognized. As for the other functional descriptions of each network in the above-mentioned image recognition network, please refer to the specific statements of the above-mentioned embodiments, and the redundant descriptions are not repeated here.

[0091] Based on the network architecture of the image recognition network provided above, the present application also provides three methods for training the image recognition network. The following embodiments describe these three training methods in detail.

[0092] The first training method, such as Fig.12 As shown, including:

[0093] S801. Obtain viewing angle information of a sample image, viewing angle information of a sample reference image, features of the sample image, and features of the sample reference image.

[0094] This embodiment relates to a method for a computer device to obtain the viewing angle information of a sample image, the viewing angle information of a sample reference image, the features of the sample image, and the features of the sample reference image. Optionally, this application provides a method for obtaining the above information, such as Fig.13 As shown, the method specifically includes:

[0095] S8011. Obtain a sample image and a sample reference image.

[0096] S8012: Input the sample image and the sample reference image into a trained visual discrimination network to obtain viewing angle information of the sample image and the viewing angle information of the sample reference image.

[0097] S8013. Input the sample image and the sample reference image into a trained feature extraction network to obtain features of the sample image and features of the sample reference image.

[0098] The specific description of the above method is explained in detail in the aforementioned embodiments. The specific content can be found in the contents of the aforementioned embodiments, and the redundant description will not be repeated here.

[0099] S802, input the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into an initial same-view recognition network and an initial different-view recognition network respectively to obtain a target loss value.

[0100] The initial same-perspective recognition network is the recognition network to be trained, and the initial different-perspective recognition network is the recognition network to be trained. When the initial same-perspective recognition network obtains a recognition result based on the perspective information of the sample image, the perspective information of the sample reference image, the features of the sample image, and the features of the sample reference image input by the computer device, and the initial different-perspective recognition network obtains a recognition result based on the perspective information of the sample image, the perspective information of the sample reference image, the features of the sample image, and the features of the sample reference image input by the computer device, the computer device can further obtain a target loss value based on the two recognition results, so as to be used in the subsequent training of the parameters of the initial different-perspective recognition network and the same-perspective recognition network.

[0101] S803. Adjust the parameters of the initial same-view recognition network and the parameters of the different-view recognition network simultaneously according to the target loss value until the target loss value meets the first preset training condition, thereby obtaining the trained same-view recognition network and the different-view recognition network.

[0102] Among them, the first preset training condition can be determined according to actual training needs. When the computer device obtains the target loss value, the parameters of the initial same-view recognition network and the parameters of the different-view recognition network can be adjusted simultaneously by judging whether the size of the target loss value meets the first preset training condition. After the adjustment is completed, the training is completed. At this time, the parameters of the same-view recognition network and the parameters of the different-view recognition network obtained are the parameters of the finally trained same-view recognition network and the parameters of the different-view recognition network. The parameters of the trained same-view recognition network and the parameters of the different-view recognition network are not shared.

[0103] Furthermore, a specific implementation of the above S802, that is, a specific implementation of the computer device obtaining the target loss value, is as follows: Fig.13 As shown, the above S802 "inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-view recognition network and the initial different-view recognition network, respectively, to obtain the target loss value" includes:

[0104] S901, input the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into an initial same-viewing angle recognition network, and calculate the same-viewing angle feature distance between the sample image and the corresponding reference image.

[0105] S902: Input the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into an initial different viewing angle recognition network, and calculate the different viewing angle feature distance between the sample image and the corresponding reference image.

[0106] S903: Fusing the same-viewing angle feature distance and the different-viewing angle feature distance to obtain a fused viewing angle feature distance.

[0107] S904: Perform weighted accumulation operation on the same-viewing angle feature distance, the different-viewing angle feature distance and the fused-viewing angle feature distance to calculate a target loss value.

[0108] In this embodiment, the same-view feature distance, different-view feature distance, and fused-view feature distance calculated by the above method can be specifically represented by a matrix. For example, the target loss value L obtained according to step S904 can be represented by the following relationship (1):

[0109] L = a*Ls+b*Ld+c*Lcross (1);

[0110] The above Ls represents the matrix of the same-view feature distance, Ld represents the matrix of the different-view feature distance, Lcross represents the matrix of the fused-view feature distance, and a, b, and c are weight coefficients respectively.

[0111] The above training method applies metric learning to the matrix of same-perspective feature distance, the matrix of different-perspective feature distance, and the matrix of fused-perspective feature distance respectively to train the same-perspective recognition network and the different-perspective recognition network, so that the same-perspective recognition network can make full use of the same-perspective features for recognition training, and the different-perspective recognition network can make full use of the different-perspective features for recognition training. Such targeted training can not only reduce the difficulty of training the recognition network, but also improve the recognition accuracy of the same-perspective recognition network and the different-perspective recognition network after training.

[0112] The second training method, for specific methods, please refer to the above Fig.12 The training method described in the embodiment is only that step S802 thereof is the same as the aforementioned Fig.13 The implementation methods described in the embodiments are different, that is, the computer device obtains the target loss value in different ways, such as Fig.14 As shown, another implementation of the above S802 "inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-view recognition network and the initial different-view recognition network, respectively, to obtain the target loss value" includes:

[0113] S1001. Input the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into an initial same-viewing angle recognition network, and calculate the same-viewing angle feature distance between the sample image and the corresponding reference image.

[0114] S1002: Input the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into an initial different viewing angle recognition network, and calculate the different viewing angle feature distance between the sample image and the corresponding reference image.

[0115] S1003: Fuse the same-viewing angle feature distance and the different-viewing angle feature distance to obtain a target loss value.

[0116] The target loss value L1 obtained by the above method can be expressed using the following equation (2):

[0117] L1=d*Ls+e*Ld (2);

[0118] The above Ls represents the matrix of the same-view feature distance, Ld represents the matrix of the different-view feature distance, Lcross represents the matrix of the fused-view feature distance, and d and e are weight coefficients respectively.

[0119] The third training method, such as Fig.15 As shown, including:

[0120] S2001. Obtain viewing angle information of a sample image, viewing angle information of a sample reference image, features of the sample image, and features of the sample reference image.

[0121] S2002. Input the perspective information of the sample image, the perspective information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-perspective recognition network to obtain a same-perspective loss value; and adjust the parameters of the initial same-perspective recognition network according to the same-perspective loss value until the same-perspective loss value meets the second preset training condition, thereby obtaining a trained same-perspective recognition network.

[0122] S2003. Input the perspective information of the sample image, the perspective information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial different-perspective recognition network to obtain a different-perspective loss value; and adjust the parameters of the initial different-perspective recognition network according to the different-perspective loss value until the different-perspective loss value meets the third preset training condition, thereby obtaining a trained different-perspective recognition network.

[0123] S2004. Connect the trained same-view recognition network and the trained different-view recognition network in parallel to obtain an image recognition network.

[0124] The third training method mentioned above is that the computer equipment trains the same-view recognition network and the different-view recognition network separately, and when training the same-view recognition network, the same-view loss value determined by the output result of the same-view recognition network is used to adjust the parameters of the initial same-view recognition network; when training the different-view recognition network, the different-view loss value determined by the output result of the different-view recognition network is used to adjust the parameters of the initial same-view recognition network. The same-view recognition network and the different-view recognition network can be trained sequentially or simultaneously, but the training processes of each are independent of each other. This training method is simple, fast and practical.

[0125] Based on the training method described in the above embodiment, Fig.15 As shown, the present application also provides a training network, which includes a trained visual discrimination network, a trained image feature extraction network, an initial same-perspective recognition network, and an initial different-perspective recognition network. The trained visual discrimination network and the trained image feature extraction network are connected in parallel, and the output ends of the trained visual discrimination network and the trained image feature extraction network are connected to the input ends of the initial same-perspective recognition network and the initial different-perspective recognition network, and the initial same-perspective recognition network and the initial different-perspective recognition network are connected in parallel. The process of training the initial same-perspective recognition network and the initial different-perspective recognition network using the above-mentioned training network can be specifically referred to the description of the aforementioned training method, and the cumbersome description will not be repeated here. It should be noted that the training network can apply any of the above-mentioned training methods.

[0126] It should be understood that although Figure 2-15 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2-15 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0127] Fig.16 FIG. 1 is a schematic diagram of the structure of a target object recognition device provided in an embodiment. Fig.16 As shown, the device includes: a first acquisition module 11, a second acquisition module 12 and an identification module 13, wherein:

[0128] A first acquisition module 11 is used to acquire viewing angle information of at least one image to be recognized and viewing angle information of a reference image;

[0129] A second acquisition module 12, used to acquire features of the image to be identified and features of the reference image;

[0130] The recognition module 13 is used to determine whether the image to be recognized and the corresponding reference image contain the same target object according to the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the features of the image to be recognized and the features of the reference image.

[0131] For the specific definition of a target object recognition device, please refer to the definition of the target object recognition method above, which will not be repeated here. Each module in the above-mentioned target object recognition device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0132] In one embodiment, a computer device is provided. The computer device may be a terminal device, and its internal structure diagram may be as follows: Fig.17As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a target object recognition method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0133] Those skilled in the art will understand that Fig.17 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0134] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0135] Obtaining viewing angle information of at least one image to be identified and viewing angle information of a reference image;

[0136] Acquire features of the image to be identified and features of the reference image;

[0137] According to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image, it is determined whether the image to be identified and the corresponding reference image contain the same target object. The computer device provided in this embodiment has similar implementation principles and technical effects as those of the above method embodiments, which will not be described in detail here.

[0138] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0139] Obtaining viewing angle information of at least one image to be identified and viewing angle information of a reference image;

[0140] Acquire features of the image to be identified and features of the reference image;

[0141] According to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image, it is determined whether the image to be identified and the corresponding reference image contain the same target object. The computer-readable storage medium provided in this embodiment has similar implementation principles and technical effects as those of the above method embodiments, and will not be described in detail here.

[0142] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0143] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0144] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A target object recognition method, It is characterized in that The method comprises: Obtaining viewing angle information of at least one image to be identified and viewing angle information of a reference image; Acquire features of the image to be identified and features of the reference image; Determining whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image; The determining, based on the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the feature of the image to be recognized, and the feature of the reference image, whether the image to be recognized and the corresponding reference image contain the same target object comprises: Determine a viewing angle feature distance between the image to be identified and the corresponding reference image according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the feature of the image to be identified and the feature of the reference image; the viewing angle feature distance represents the distance between the feature vector of the image to be identified and the feature vector of the corresponding reference image, and the viewing angle feature distance includes a same-viewing angle feature distance and / or a different-viewing angle feature distance; Determining whether the image to be identified and the corresponding reference image contain the same target object according to the visual feature distance between the image to be identified and the corresponding reference image; The determining, according to the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the feature of the image to be recognized and the feature of the reference image, of the viewing angle feature distance between the image to be recognized and the corresponding reference image comprises: Obtaining same-viewing angle features and / or different-viewing angle features of the image to be identified according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified, and the features of the reference image; Calculating the distance between the same-viewing angle feature of the image to be identified and the corresponding feature of the reference image to obtain the same-viewing angle feature distance between the image to be identified and the corresponding reference image; And / or, calculating the distance between the different-viewing angle features of the image to be identified and the corresponding features of the reference image to obtain the different-viewing angle feature distance between the image to be identified and the corresponding reference image.

2. The method according to claim 1, It is characterized in that The obtaining the same-viewing angle feature and / or different-viewing angle feature of the image to be identified according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the feature of the image to be identified and the feature of the reference image comprises: According to the viewing angle information of the image to be identified and the viewing angle information of the corresponding reference image, determining whether the image to be identified and the corresponding reference image belong to the same viewing angle image; When the image to be identified and the corresponding reference image belong to the same viewing angle, extracting the same viewing angle features of the image to be identified from the features of the image to be identified; When the image to be identified and the corresponding reference image belong to images with different viewing angles, different viewing angle features of the image to be identified are extracted from the features of the image to be identified.

3. The method according to claim 1 or 2, It is characterized in that When the viewing angle feature distance includes the same viewing angle feature distance and the different viewing angle feature distance, determining whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle feature distance between the image to be identified and the corresponding reference image includes: Fusing the same-viewing angle feature distance and the different-viewing angle feature distance to obtain a fused viewing angle feature distance between the image to be identified and the corresponding reference image; According to the fusion viewing angle feature distance, it is determined whether the image to be identified and the corresponding reference image contain the same target object.

4. The method according to claim 1, It is characterized in that The determining, based on the viewing angle information of the image to be recognized, the viewing angle information of the reference image, the feature of the image to be recognized, and the feature of the reference image, whether the image to be recognized and the corresponding reference image contain the same target object comprises: Inputting the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image into a preset image recognition network to identify the target object, and determining whether the image to be identified and the corresponding reference image contain the same target object; The image recognition network includes a same-viewing angle recognition network and a different-viewing angle recognition network, the same-viewing angle recognition network and the different-viewing angle recognition network are connected in parallel, and the parameters of the same-viewing angle recognition network and the different-viewing angle recognition network are not shared.

5. The method according to claim 4, It is characterized in that The same-view recognition network is used to: Extracting the same-viewing angle features of the image to be identified from the features of the image to be identified; Calculating a same-viewing angle feature distance between the image to be identified and the corresponding reference image according to the same-viewing angle feature of the image to be identified and the feature of the corresponding reference image; When the same-viewing angle feature distance is less than a first preset distance, it is determined that the image to be identified and the corresponding reference image contain the same target object; when the same-viewing angle feature distance is greater than or equal to the first preset distance, it is determined that the image to be identified and the corresponding reference image do not contain the same target object.

6. The method according to claim 4, It is characterized in that The different-viewing angle recognition network is used for: Extracting different viewing angle features of the image to be identified from the features of the image to be identified; Calculating a different-viewing angle feature distance between the image to be identified and the corresponding reference image according to different-viewing angle features of the image to be identified and features of the corresponding reference image; When the different-viewing angle feature distance of the image to be identified is less than a second preset distance, it is determined that the image to be identified and the corresponding reference image contain the same target object; when the different-viewing angle feature distance is greater than or equal to the second preset distance, it is determined that the image to be identified and the corresponding reference image do not contain the same target object.

7. The method according to claim 1, It is characterized in that The obtaining of viewing angle information of at least one image to be identified and viewing angle information of a reference image includes: Inputting the image to be identified into a preset visual discrimination network to obtain the viewing angle information of the image to be identified; the visual discrimination network is used to discriminate the shooting angle of the input image; The reference image is input into the visual discrimination network to obtain the viewing angle information of the reference image.

8. The method according to claim 1, It is characterized in that The acquiring the features of the image to be recognized and the features of the reference image includes: Using a preset image feature extraction network to perform shallow feature extraction on the image to be identified, to obtain features of the image to be identified; The image feature extraction network is used to perform shallow feature extraction on the reference image to obtain features of the reference image.

9. The method according to claim 4, It is characterized in that The method of training the image recognition network includes: Obtaining viewing angle information of the sample image, viewing angle information of the sample reference image, features of the sample image, and features of the sample reference image; Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image, and the features of the sample reference image into an initial same-viewing angle recognition network and an initial different-viewing angle recognition network, respectively, to obtain a target loss value; According to the target loss value, the parameters of the initial same-view recognition network and the parameters of the different-view recognition network are adjusted simultaneously until the target loss value meets the first preset training condition, thereby obtaining the trained same-view recognition network and the different-view recognition network.

10. The method according to claim 9, It is characterized in that The viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image, and the features of the sample reference image are respectively input into an initial same-viewing angle recognition network and an initial different-viewing angle recognition network to obtain a target loss value, including: Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-viewing angle recognition network, and calculating the same-viewing angle feature distance between the sample image and the corresponding reference image; Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial different viewing angle recognition network, and calculating the different viewing angle feature distance between the sample image and the corresponding reference image; Fusing the same-viewing angle feature distance and the different-viewing angle feature distance to obtain a fused viewing angle feature distance; The same-viewing angle feature distance, the different-viewing angle feature distance and the fused-viewing angle feature distance are weightedly added to calculate the target loss value.

11. The method according to claim 9, It is characterized in that The viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image, and the features of the sample reference image are respectively input into an initial same-viewing angle recognition network and an initial different-viewing angle recognition network to obtain a target loss value, including: Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-viewing angle recognition network, and calculating the same-viewing angle feature distance between the sample image and the corresponding reference image; Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial different viewing angle recognition network, and calculating the different viewing angle feature distance between the sample image and the corresponding reference image; The same-viewing angle feature distance and the different-viewing angle feature distance are merged to obtain the target loss value.

12. The method according to claim 9, It is characterized in that Obtaining viewing angle information of the sample image, viewing angle information of the sample reference image, features of the sample image, and features of the sample reference image, including: Acquire the sample image and the sample reference image; Inputting the sample image and the sample reference image into a trained visual discrimination network to obtain viewing angle information of the sample image and the viewing angle information of the sample reference image; The sample image and the sample reference image are input into a trained feature extraction network to obtain features of the sample image and features of the sample reference image.

13. The method according to claim 9, It is characterized in that The method of training the image recognition network includes: Obtaining viewing angle information of the sample image, viewing angle information of the sample reference image, features of the sample image, and features of the sample reference image; Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial same-viewing angle recognition network to obtain a same-viewing angle loss value; and adjusting the parameters of the initial same-viewing angle recognition network according to the same-viewing angle loss value until the same-viewing angle loss value meets a second preset training condition, thereby obtaining a trained same-viewing angle recognition network; Inputting the viewing angle information of the sample image, the viewing angle information of the sample reference image, the features of the sample image and the features of the sample reference image into the initial different viewing angle recognition network to obtain a different viewing angle loss value; and adjusting the parameters of the initial different viewing angle recognition network according to the different viewing angle loss value until the different viewing angle loss value meets a third preset training condition, thereby obtaining a trained different viewing angle recognition network; The trained same-view recognition network and the trained different-view recognition network are connected in parallel to obtain the image recognition network.

14. A target object recognition device, It is characterized in that The device comprises: A first acquisition module, used to acquire viewing angle information of at least one image to be recognized and viewing angle information of a reference image; A second acquisition module, used to acquire the features of the image to be identified and the features of the reference image; an identification module, configured to determine whether the image to be identified and the corresponding reference image contain the same target object according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified and the features of the reference image; The identification module is specifically used for: Determine a viewing angle feature distance between the image to be identified and the corresponding reference image according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the feature of the image to be identified and the feature of the reference image; the viewing angle feature distance represents the distance between the feature vector of the image to be identified and the feature vector of the corresponding reference image; Determining whether the image to be identified and the corresponding reference image contain the same target object according to the visual feature distance between the image to be identified and the corresponding reference image; The identification module is further specifically used for: Obtaining same-viewing angle features and / or different-viewing angle features of the image to be identified according to the viewing angle information of the image to be identified, the viewing angle information of the reference image, the features of the image to be identified, and the features of the reference image; Calculating the distance between the same-viewing angle feature of the image to be identified and the corresponding feature of the reference image to obtain the same-viewing angle feature distance between the image to be identified and the corresponding reference image; And / or, calculating the distance between the different-viewing angle features of the image to be identified and the corresponding features of the reference image to obtain the different-viewing angle feature distance between the image to be identified and the corresponding reference image.

15. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 13 are implemented.

16. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 13 are implemented.