Image processing method, storage medium and computer terminal

Abstract

This application discloses an image processing method, storage medium, and computer terminal. The method comprises: acquiring a target image, wherein the target image contains a target object; performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; determining at least one target position set corresponding to the target object based on the feature vectors of the at least one feature space, wherein each target position set contains multiple target positions that represent the position of the target object in each feature space; and determining the target position of the target object in the target image based on the at least one target position set. This application solves the technical problem of low accuracy of target image detection results in related technologies.

Description

Technical Field

[0001] The present application relates to the field of image processing, and in particular to an image processing method, a storage medium, and a computer terminal. Background Art

[0002] Currently, ReDet (rotation equivariant detector) uses a horizontal candidate box network to solve the task of rotating target detection. First, the horizontal candidate box network is used to obtain the horizontal detection box of the target object, and then the horizontal detection box of the target object is adjusted at an angle to obtain the target detection box of the target object. The whole process is divided into two steps, which is not only inefficient but also has low accuracy of the final target detection box.

[0003] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0004] The embodiments of the present application provide an image processing method, a storage medium, and a computer terminal to at least solve the technical problem of low accuracy of target image detection results in the related art.

[0005] According to one aspect of an embodiment of the present application, there is provided an image processing method, comprising: acquiring a target image, wherein the target image contains a target object; performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; based on the feature vector of the at least one feature space, determining at least one target position set corresponding to the target object, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; based on the at least one target position set, determining the target position of the target object in the target image.

[0006] According to another aspect of an embodiment of the present application, an image processing method is also provided, including: acquiring a vehicle image, wherein the vehicle is contained in the vehicle image; performing feature extraction on the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; based on the feature vector of at least one feature space, determining at least one target position set corresponding to the vehicle, wherein the multiple target positions contained in each target position set are used to characterize the position of the vehicle in each feature space; based on the at least one target position set, determining the target position of the vehicle in the vehicle image.

[0007] According to another aspect of an embodiment of the present application, an image processing method is also provided, including: a cloud server receives a target image uploaded by a client, wherein the target image contains a target object; the cloud server performs feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the cloud server determines at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; the cloud server determines the target position of the target object in the target image based on at least one target position set; and the cloud server sends the target position to the client.

[0008] According to one aspect of an embodiment of the present application, an image processing device is provided, including: an acquisition module for acquiring a target image, wherein the target image includes a target object; an extraction module for performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; a prediction module for determining at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions included in each target position set are used to characterize the position of the target object in each feature space; and a determination module for determining the target position of the target object in the target image based on the at least one target position set.

[0009] According to another aspect of an embodiment of the present application, an image processing device is also provided, including: an acquisition module for acquiring a vehicle image, wherein the vehicle image contains a vehicle; an extraction module for performing feature extraction on the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; a detection module for determining at least one target position set corresponding to the vehicle based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the vehicle in each feature space; and a determination module for determining the target position of the vehicle in the vehicle image based on at least one target position set.

[0010] According to another aspect of an embodiment of the present application, an image processing device is also provided, including: a receiving module for receiving a target image uploaded by a client through a cloud server, wherein the target image includes a target object; an extraction module for performing feature extraction on the target image through the cloud server to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; a first determination module for determining at least one target position set corresponding to the target object based on the feature vector of at least one feature space through the cloud server, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; a second determination module for determining the target position of the target object in the target image based on at least one target position set through the cloud server; and a sending module for sending the target position to the client through the cloud server.

[0011] According to another aspect of an embodiment of the present application, a computer-readable storage medium is further provided. The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned image processing method.

[0012] According to another aspect of an embodiment of the present application, a computer terminal is further provided, including a memory and a processor, wherein the processor is configured to run a program stored in the memory, wherein the above-mentioned image processing method is executed when the program is run.

[0013] First, a target image can be obtained, wherein the target image contains a target object; feature extraction is performed on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; based on the feature vector of the at least one feature space, at least one target position set corresponding to the target object is determined, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; based on the at least one target position set, the target position of the target object in the target image is determined, thereby achieving the effect of improving the accuracy of the detection result. It is easy to notice that since the target position is determined by the at least one target position set of the target object, the low detection accuracy caused by the discontinuity of the angle of the detection frame corresponding to the target position can be avoided. By obtaining the target position set, the target position can be directly determined without performing other operations, which can improve the accuracy of the target position, thereby solving the low accuracy of the detection result caused by the discontinuity of the angle regression of the target detection frame corresponding to the target position in the related art, and further solving the technical problem of low accuracy of the detection result of the target image in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0015] Figure 1 It is a hardware structure block diagram of a computer terminal (or mobile device) for implementing an image processing method according to the prior art;

[0016] Figure 2 is a flowchart of an image processing method according to Example 1 of the present application;

[0017] Figure 3 is a flowchart of another image processing method according to Example 1 of the present application;

[0018] Figure 4 is a schematic diagram of an image after vehicles in a remote sensing image are annotated according to Example 1 of the present application;

[0019] Figure 5 is a schematic structural diagram of a model according to Example 1 of the present application;

[0020] Figure 6 is a flowchart of an image processing method according to Example 2 of the present application;

[0021] Figure 7 is a flowchart of an image processing method according to Example 3 of the present application;

[0022] Figure 8 is a schematic diagram of an image processing device according to Embodiment 4 of the present application;

[0023] Figure 9 is a schematic diagram of an image processing device according to Embodiment 5 of the present application;

[0024] Figure 10 is a schematic diagram of an image processing device according to Example 6 of the present application;

[0025] Figure 11 This is a structural block diagram of a computer terminal according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following interpretations:

[0029] Rotating target detection: It can be used to extract the minimum bounding rectangle of the target to be detected;

[0030] PointRPN (Point Region Proposal Network): can extract candidate boxes for points.

[0031] Currently, ReDet still uses a horizontal proposal box network for rotated object detection. ReDet splits the prediction of rotated proposal boxes into two parts, which is not only inefficient but also produces low-quality proposal boxes, limiting the accuracy of the final rotated box detection. ReDet also directly regresses the angle of the rotated box when predicting rotated proposal boxes, resulting in discontinuities in the angle prediction.

[0032] To address the above issues, this application provides an image processing method. Unlike ReDet's two-step prediction of rotated candidate boxes, this application directly predicts rotated candidate boxes in PointRPN, thereby providing more high-quality rotated candidate boxes. Unlike ReDet, which directly regresses the angle of the box when predicting rotated candidate boxes, this application obtains the rotated box by predicting a set of points, eliminating the problem of angle discontinuity.

[0033] Example 1

[0034] According to an embodiment of the present application, an embodiment of an image processing method is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0035] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing an image processing method. Figure 1 As shown, the computer terminal 10 (or mobile device 10) may include one or more (illustrated as 102a, 102b, ..., 102n) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0036] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be fully or partially integrated into any of the other components of the computer terminal 10 (or mobile device). The data processing circuitry may function as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0037] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the image processing method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implementing the above-mentioned image processing method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0038] The transmission device 106 is used to receive or send data via a network. A specific example of the network may include a wireless network provided by the communications provider of the computer terminal 10. In one embodiment, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0039] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0040] It should be noted that, in some optional embodiments, the above Figure 1 The computer device (or mobile device) shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the aforementioned computer device (or mobile device).

[0041] Under the above operating environment, this application provides Figure 2 The image processing method shown. Figure 2 FIG. 1 is a flow chart of an image processing method according to an embodiment of the present application. Figure 2 As shown, the method may include the following steps:

[0042] Step S202: Acquire a target image.

[0043] The target image contains the target object.

[0044] The target image in the above steps can be a remote sensing image, an aerial image taken by a drone, or a radar image, but is not limited to these. In different application scenarios, the type of target object contained in the target image varies. For example, in a transportation scenario, the target object contained in the target image can be a vehicle; in an urban planning application scenario, the target object contained in the target image can be a road network, land parcels, buildings, etc.; in an agricultural and forestry application scenario, the target object contained in the target image can be land parcels, greenhouses, etc.; in a meteorological application scenario, the target object contained in the target image can be clouds; in a water conservancy application scenario, the target object contained in the target image can be dams, rivers, lakes, etc.; in a disaster application scenario, the target object contained in the target image can be mountains, dams, etc.

[0045] In an optional embodiment, the target image can be captured by a satellite or drone and transmitted to a server via a network for processing by the server. The target image can also be displayed to the user and displayed within the image acquisition area. In another optional embodiment, the target image can be captured by a satellite or drone and uploaded to the server by the user and processed by the server. The user can upload the target image to the server by clicking the "Upload Image" button in the interactive interface or by directly dragging the target image into the dotted box. Moreover, the image uploaded by the user can be displayed within the image acquisition area. The server here can be a server deployed locally or in the cloud.

[0046] Step S204: extract features from the target image to obtain a feature vector of at least one feature space.

[0047] Among them, the resolutions of multiple feature spaces are different.

[0048] In an optional embodiment, a feature extraction network may be used to extract features from the target image to obtain a feature vector of at least one feature space.

[0049] The feature extraction network mentioned above can be ResNet50+FPN (residual network and feature pyramid network).

[0050] In another optional embodiment, to improve the accuracy of the target object detection results in the target image, the target image can be downsampled using ResNet50+FPN to obtain multiple feature spaces with different resolutions. Feature extraction is then performed on the multiple feature spaces to obtain a feature vector corresponding to each feature space. Since the resolutions of the multiple feature spaces are different, the accuracy of the prediction results can be improved by predicting the target object in the target image based on the feature vectors of the multiple feature spaces.

[0051] Step S206: determining at least one target position set corresponding to the target object based on the feature vector of at least one feature space.

[0052] The multiple target positions contained in each target position set are used to represent the position of the target object in each feature space.

[0053] The aforementioned target position set may be a position set of features related to the target object in the feature space.

[0054] In an optional embodiment, when feature vectors of multiple feature spaces are obtained, the feature vectors of each feature space can be predicted to obtain a target position set of the target object in each feature space. Specifically, PointRPN can be used to predict a preset number of pixel points of the target object in each feature space, and the above-mentioned target position set can be generated based on the preset number of pixel points, wherein the preset number of pixel points can be used to represent the position of the target object in each feature space.

[0055] Step S208: determining a target position of the target object in the target image based on at least one target position set.

[0056] Optionally, a target detection frame can be used to represent the target position. The target detection frame can be used to mark the position of the target object in the target image. The target detection frame can frame the target object in the target image. This allows prediction of the target object in the target detection frame to obtain a detection result for the target object.

[0057] In an optional embodiment, after determining at least one target position set corresponding to the target object based on the feature vectors of each feature space, the target position of the target object in the target image can be directly determined based on the at least one target position set. Since the target position set is used to represent the position of the target object in the feature space, the target position determined using the target position set does not need to be adjusted again based on the orientation of the target object, i.e., the determined target position has a high degree of accuracy. In an optional embodiment, when feature vectors of multiple feature spaces are obtained, the feature vectors of each feature space can be predicted to obtain a target position set for the target object in each feature space, and a target detection frame for each feature space can be generated based on the target position set for the target object in each feature space. By predicting the target object in the target detection frame in each feature space, a detection result for the target object contained in each feature space is obtained. The multiple detection results can then be fused to determine a final target object detection result. Since the resolution of each feature space is different, by fusing the detection results of each feature space, a better prediction effect can be achieved, thereby making the final target object detection result more accurate.

[0058] In another optional embodiment, PointRPN can be used to predict a preset number of pixel points of the target object in each feature space, and the above-mentioned target position set can be generated based on the preset number of pixel points. The minimum enclosing rectangle algorithm can be used to obtain the rotated candidate box corresponding to the target position set. Furthermore, the target object in the rotated candidate box can be identified first to obtain the category of the target object. After obtaining the category of the target object, the rotated candidate box can be fine-tuned to obtain a target detection box, and the target object in the target detection box can be predicted to obtain the detection result of the target object. Fine-tuning the rotated candidate box according to the category of the target object can improve the accuracy of the obtained target detection box, thereby improving the accuracy of the detection result of the target object.

[0059] In another optional embodiment, after performing target detection on the target image, the server can directly display the detection results to the user for viewing, and the category of the target object and the target detection box corresponding to the target object displayed in the detection results can be displayed in the target image in the result feedback area; in another optional embodiment, after performing target detection on the target image, the server can feed back the detection results to the user's client through the network, and the client can display them to the user for viewing. Specifically, the client can display the category of the target object and the target detection box corresponding to the target object on the target image. If the user believes that the category of the target object and the position of the target detection box in the detection result are incorrect, the user can adjust the position of the target detection box, re-enter the category of the target object, and upload it to the server, so that the server can re-train the model used for target detection based on the user's feedback, thereby improving the server performance.

[0060] In a transportation scenario, the target image may be a traffic image, wherein the target objects contained in the traffic image may be vehicles such as airplanes and vehicles. A traffic image may be first acquired, and then features may be extracted from the traffic image using ResNet50+FPN to obtain feature spaces of varying resolutions. Vehicles in the traffic image may be detected based on the feature vectors of each feature space. The positions of multiple pixels of the vehicle in the target image may be determined based on the characteristics of the vehicle, and a corresponding target position set may be generated. The positions of the vehicles in the traffic image may be annotated using a target detection frame, and the category of the vehicle in the target detection frame may be predicted, wherein the category of the vehicle may be a vehicle, airplane, train, etc.

[0061] In another optional embodiment, the application scenario of the above scheme can be to use remote sensing image target detection as the main application scenario, wherein the remote sensing image can be an image of a city street scene, traffic, water body, etc., and the target object in the remote sensing image can be detected to determine the target position of the target in the remote sensing image. Specifically, a target detection frame can be used to represent the target position, that is, a target detection frame is used to mark the target object in the remote sensing image.

[0062] Through the above steps, first, a target image can be obtained, wherein the target image contains a target object; feature extraction is performed on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; based on the feature vector of at least one feature space, at least one target position set corresponding to the target object is determined, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; based on the at least one target position set, the target position of the target object in the target image is determined, thereby achieving the effect of improving the accuracy of the detection result. It is easy to notice that since the target position is determined by at least one target position set of the target object, the low detection accuracy caused by the discontinuity of the angle of the detection frame corresponding to the target position can be avoided. By obtaining the target position set, the target position can be directly determined without performing other operations, which can improve the accuracy of the target position, thereby solving the low accuracy of the detection result caused by the discontinuity of the angle regression of the target detection frame corresponding to the target position in the related art, and further solving the technical problem of low accuracy of the detection result of the target image in the related art.

[0063] In the above embodiment of the present application, determining the target position of the target object in the target image based on at least one target position set includes: determining an initial detection frame of the target object in the target image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; determining a target feature vector corresponding to the initial detection frame from a feature vector of at least one feature space; and adjusting the initial detection frame based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to represent the target position.

[0064] The target positions may be positions of pixels of the target object in the target image, wherein the target positions may be nine target positions corresponding to nine pixels of the target object, and the number of target positions is not limited.

[0065] In an optional embodiment, a result generation module can be used to determine an initial detection box of the target object in the target image based on at least one target position set, wherein the result generation module can be PointRPN, and PointRPN can be composed of multiple consecutive convolutional layers. Specifically, PointRPN can be composed of three consecutive convolutional layers.

[0066] In another optional embodiment, the result generation module can be used to predict the feature vector of at least one feature space to obtain multiple pixel points corresponding to the target object in the feature space, and generate a target position set of multiple pixel points. The minimum bounding rectangle algorithm can be used to calculate the minimum bounding rectangle of the target position set, that is, the initial detection frame mentioned above. The target feature vector marked by the initial detection frame in the feature vector of at least one feature space can be obtained, and the target feature vector can be used for object recognition to obtain the category of the target object. At the same time, the initial detection frame can be adjusted according to the target feature vector to obtain a target detection frame with higher accuracy, that is, the target detection frame mentioned above for characterizing the target position.

[0067] Furthermore, the initial detection frame can be further adjusted according to the category of the target object, so that the obtained target detection frame can accurately mark the target object belonging to the category, thereby improving the accuracy of the target detection frame.

[0068] Furthermore, after obtaining the initial detection frame, object recognition can be directly performed on the target feature vector corresponding to the target object in the feature space. It is not limited to the recognition of the feature vectors contained in the initial detection frame. This can avoid missing feature vectors related to the target object that are not marked in the initial detection frame, thereby improving the recognition accuracy of the target feature vector and thus improving the accuracy of classifying the target object.

[0069] In another optional embodiment, the result generation module can be used to predict the feature vectors of each feature space to obtain multiple positions of multiple pixel points corresponding to the target object in the target image, and based on the target position set corresponding to each feature space, the minimum enclosing rectangle algorithm is used to calculate the initial detection frames corresponding to multiple positions in the target position set. The angle of the initial detection frame can be directly determined without the need to specifically adjust the angle of the initial detection frame, thereby avoiding the discontinuity problem caused by angle regression, thereby improving the accuracy of the obtained initial detection frame.

[0070] In a transportation scenario, the feature vector corresponding to the vehicle marked in the feature vector of at least one feature space of the initial detection frame can be obtained, and the feature vector can be identified to obtain the category of the vehicle, so that the initial detection frame can be further adjusted according to the category of the vehicle, so that the target detection frame can accurately mark the vehicles belonging to the category, thereby improving the accuracy of the target detection frame. Specifically, when the detected category of the vehicle is an aircraft, the initial detection frame can be adjusted so that all parts of the aircraft can be framed in the adjusted target detection frame. In this way, in the subsequent recognition process of the target detection frame, the category identified is more accurate.

[0071] In the above embodiment of the present application, determining the target feature vector corresponding to the initial detection frame from the feature vectors of at least one feature space includes: determining the feature vector located in the initial detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0072] In an optional embodiment, since the initial detection frame is generated based on the target position corresponding to the target object, the features contained in the initial detection frame have a strong correlation with the features of the target object, that is, the category of the target object can be more accurately predicted through the target feature vector contained in the initial detection frame. Therefore, the target feature vector corresponding to the target object can be obtained from the initial detection frame, thereby improving the accuracy of the detection result of the target object.

[0073] In the above embodiment of the present application, before determining the feature vector located in the initial detection frame from the feature vectors of at least one feature space to obtain the target feature vector, the method also includes: filtering the initial detection frame based on the confidence of at least one target position set to obtain a filtered detection frame, wherein the confidence is used to characterize the degree of matching between the at least one target position set and the target object; determining the feature vector located in the filtered detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0074] In an optional embodiment, at least one target position set in each feature space can be obtained first. It should be noted that there may be multiple target objects in a target image. Therefore, the feature vector corresponding to each target object in the feature space can be obtained, and the result generation module can be used to predict the feature vector corresponding to each object to obtain the confidence of the target position set corresponding to the feature vector of each object. The initial detection frame corresponding to the target position set can be screened according to the confidence, and the initial detection frame corresponding to the target position set with a higher confidence is determined as the screened detection frame, and the feature vector located in the screened detection frame is determined from the screened detection frame to obtain the target feature vector, thereby improving the accuracy of the target feature vector.

[0075] In the above embodiment of the present application, the method also includes: obtaining a training sample, wherein the training sample includes: a training image, and a preset detection frame corresponding to the training object contained in the training image; using a target detection model to predict the training image to obtain a generated detection frame corresponding to the training object, wherein the target detection model includes: a feature extraction module and a result generation module, the feature extraction module is used to extract a training feature vector of at least one feature space of the input training image, and the result generation module is used to generate a generated detection frame based on the training feature vector of at least one feature space; based on the preset detection frame and the generated detection frame, a total loss function of the result generation module is generated; based on the total loss function, the model parameters of the result generation module are adjusted.

[0076] In an optional embodiment, the result generation module can be trained by obtaining corresponding training samples for target images of different scenes to be recognized. For example, if the result generation module is to be used to predict detection boxes corresponding to vehicles in a transportation scenario, the result generation module can be trained using training samples related to transportation scenarios.

[0077] In another optional embodiment, a training sample can be obtained first, and the feature extraction module can be used to extract features from the training image to obtain training feature vectors of feature spaces corresponding to different resolutions. The training feature vectors of each feature space can then be predicted using the result generation module to obtain a generated detection frame corresponding to the training object. The total loss function of the result generation module can be generated based on the preset detection frame contained in the training sample and the predicted generated detection frame, so that the model parameters of the result generation module can be adjusted according to the total loss function, thereby improving the model accuracy of the result generation module.

[0078] In the above embodiment of the present application, based on the preset detection frame and the generated detection frame, the total loss function of the result generation module includes: obtaining the intersection-over-union ratio of the preset detection frame and the generated detection frame to generate a first loss function; obtaining the distance between the preset detection frame and the generated detection frame to generate a second loss function; based on the first loss function and the second loss function, generating a total loss function.

[0079] In an optional embodiment, the intersection over union (IOU) between the preset detection frame and the generated detection frame can be obtained to measure the degree of overlap between the preset detection frame and the generated detection frame, so as to generate a first loss function based on the degree of overlap; the distance deviation and angle deviation between the preset detection frame and the generated detection frame can also be obtained to generate a second loss function based on the distance deviation and the angle deviation. The total loss function obtained based on the first loss function and the second loss function can better reflect the current prediction accuracy of the result generation module. Therefore, by adjusting the model parameters through the total loss function, the model accuracy of the result generation module can be further improved.

[0080] In the above embodiment of the present application, after determining at least one target position set corresponding to the target object based on the feature vector of at least one feature space, the method also includes: outputting at least one target position set; receiving a feedback position set corresponding to the at least one target position set, wherein the feedback position set is obtained by modifying the at least one target position set; and determining the target position of the target object in the target image based on the feedback position set.

[0081] In an optional embodiment, after determining at least one target position set corresponding to the target object, the at least one target position set can be output to the user's client so that the user can modify the at least one target position set to improve the accuracy of the at least one target position set. After the user's modification, a feedback position set can be generated and received through the server so that the target position of the target object in the target image can be determined based on the feedback position set, thereby improving the accuracy of the target position.

[0082] In the above embodiment of the present application, outputting at least one target location set includes: determining a preset display mode corresponding to each target location set based on a comparison result of the confidence of at least one target location set and a preset confidence; and outputting at least one target location set according to the preset display mode.

[0083] The above preset display modes can be set by yourself. Among them, the target position sets of different types of target objects can be displayed in different colors and different textures, so that the type of target object can be determined according to the output target position set.

[0084] The aforementioned preset confidence level may be the confidence level of the target location set corresponding to each target object. The type of the target object represented by the target location set may be determined based on the aforementioned preset confidence level.

[0085] In an optional embodiment, a comparison result between the confidence of at least one target position set and a preset confidence can be obtained to determine the category of the target object corresponding to each target position set, so as to determine the preset display mode of the target position set according to the category of the target object, wherein the preset display modes corresponding to target objects of different categories can be pre-set, and the target position sets corresponding to target objects of different categories can be output according to the preset display modes.

[0086] The following combination Figures 3 to 5 A preferred embodiment of the present application is described in detail. The method can be executed by a computer terminal or a server. Figure 3 As shown, the method includes the following steps:

[0087] Step S301, acquiring a remote sensing image;

[0088] Step S302, segmenting the remote sensing image to obtain a target image;

[0089] Optionally, the size of the target image may be 1024×1024.

[0090] Step S303: extract features from the target image using a feature extraction module to obtain feature vectors with different spatial resolutions.

[0091] The above-mentioned feature extraction module can be ResNet50+FPN.

[0092] Step S304: input the above-mentioned feature vectors into the result generation module to predict the target position set corresponding to each feature space;

[0093] The above-mentioned result generation module may be PointRPN.

[0094] Step S305, calculating an initial detection frame corresponding to the target position set according to a minimum bounding rectangle algorithm;

[0095] Step S306, obtaining a target feature vector corresponding to the initial detection frame from the feature space;

[0096] Step S307, using the result generation module to predict the target feature vector to obtain the category of the target object in the initial detection frame;

[0097] Step S308, adjusting the initial detection frame according to the category of the target object to obtain a target detection frame;

[0098] Step S309 : annotating the target object in the remote sensing image based on the target detection frame to generate an annotated remote sensing image.

[0099] like Figure 4Shown is an image after the vehicles in the remote sensing image are annotated.

[0100] like Figure 5 The figure shows the structure of the PointRPN model. First, a feature vector F∈R of at least one feature space is obtained from the feature extraction network. H×W×C , and then input the feature vector into the PointRPN model, the position coordinates of 9 points P∈R can be predicted in each feature space H×W×18 And the corresponding confidence score S∈R H×W×1 , where the 9 points predicted in the feature space can be the target positions of the more critical feature vectors. The minimum circumscribed rectangular box corresponding to the 9 points in the feature space can be calculated, and the predicted 9 points can be changed into the rotation candidate box, that is, the initial detection box B∈R H×W×5 , where the initial detection frame includes the coordinates (x, y) of the center point and the size of the initial detection frame. The initial detection frames can be filtered based on the confidence level of the category of each initial detection frame to obtain filtered detection frames. Feature vectors located within the filtered detection frames are then determined from feature vectors in at least one feature space to obtain a target feature vector, and the category of the target feature vector is predicted.

[0101] The proposed PointRPN model can directly provide a high-quality initial detection frame. By predicting a set of points through the PointRPN model to predict the initial detection frame, the discontinuity problem of angle regression can be avoided, and the accuracy of target detection predicted in remote sensing images can be improved.

[0102] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0103] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0104] Example 2

[0105] According to an embodiment of the present application, an image processing method embodiment is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0106] Figure 6 FIG. 1 is a flow chart of an image processing method according to an embodiment of the present invention. Figure 6 As shown, the method may include the following steps:

[0107] Step S602: Acquire a vehicle image.

[0108] The vehicle image includes a vehicle.

[0109] The above-mentioned means of transportation include but are not limited to airplanes, cars, subways, trains and other means of transportation.

[0110] Step S604 : extracting features from the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different.

[0111] Step S606: Determine at least one target position set corresponding to the vehicle based on the feature vector of at least one feature space.

[0112] The multiple target positions contained in each target position set are used to represent the position of the vehicle in each feature space.

[0113] Step S608 : determining the target position of the vehicle in the vehicle image based on at least one target position set.

[0114] In the above embodiment of the present application, determining the target position of a vehicle in a vehicle image based on at least one target position set includes: determining an initial detection frame of the vehicle in the vehicle image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to characterize the initial position of the vehicle in the vehicle image; determining a target feature vector corresponding to the initial detection frame from a feature vector of at least one feature space; and adjusting the initial detection frame based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to characterize the target position.

[0115] In the above embodiment of the present application, determining the target feature vector corresponding to the initial detection frame from the feature vectors of at least one feature space includes: determining the feature vector located in the initial detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0116] In the above embodiment of the present application, before determining the feature vector located in the initial detection frame from the feature vectors of at least one feature space to obtain the target feature vector, the method also includes: filtering the initial detection frame based on the confidence of at least one target position set to obtain a filtered detection frame, wherein the confidence is used to characterize the degree of matching between the at least one target position set and the vehicle; determining the feature vector located in the filtered detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0117] In the above embodiment of the present application, after determining at least one target position set corresponding to the vehicle based on the feature vector of at least one feature space, the method also includes: outputting at least one target position set; receiving a feedback position set corresponding to the at least one target position set, wherein the feedback position set is obtained by modifying the at least one target position set; and determining the target position of the vehicle in the vehicle image based on the feedback position set.

[0118] In the above embodiment of the present application, outputting at least one target location set includes: determining a preset display mode corresponding to each target location set based on a comparison result of the confidence of at least one target location set and a preset confidence; and outputting at least one target location set according to the preset display mode.

[0119] It should be noted that the preferred implementation scheme involved in the above embodiments of this application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0120] Example 3

[0121] According to an embodiment of the present application, an image processing method embodiment is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0122] Figure 7 This is a flow chart of an image processing method according to embodiment 3 of the present application. Figure 7 As shown, the method may include the following steps:

[0123] Step S702: The cloud server receives the target image uploaded by the client.

[0124] The target image contains the target object.

[0125] In step S704 , the cloud server performs feature extraction on the target image to obtain a feature vector of at least one feature space.

[0126] Among them, the resolutions of multiple feature spaces are different.

[0127] Step S706: The cloud server determines at least one target location set corresponding to the target object based on the feature vector of at least one feature space.

[0128] Among them, each target position set contains multiple target positions used to represent the position of the target object in each feature space

[0129] Step S708: The cloud server determines the target position of the target object in the target image based on at least one target position set.

[0130] Step S710: The cloud server sends the target location to the client.

[0131] In the above embodiment of the present application, the cloud server determines the target position of the target object in the target image based on at least one target position set, including: the cloud server determines the initial detection frame of the target object in the target image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; the cloud server determines the target feature vector corresponding to the initial detection frame from the feature vector of at least one feature space; the cloud server adjusts the initial detection frame based on the target feature vector to obtain the target detection frame, wherein the target detection frame is used to represent the target position.

[0132] In the above embodiment of the present application, the cloud server determines the target feature vector corresponding to the initial detection frame from the feature vector of at least one feature space, including: the cloud server determines the feature vector located in the initial detection frame from the feature vector of at least one feature space to obtain the target feature vector.

[0133] In the above embodiment of the present application, before the cloud server determines the feature vector located within the initial detection frame from the feature vectors of at least one feature space and obtains the target feature vector, the method also includes: the cloud server filters the initial detection frame based on the confidence of at least one target position set to obtain a filtered detection frame, wherein the confidence is used to characterize the degree of matching between the at least one target position set and the target object; the cloud server determines the feature vector located within the filtered detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0134] In the above embodiment of the present application, after the cloud server determines at least one target position set corresponding to the target object based on the feature vector of at least one feature space, the method also includes: the cloud server outputs at least one target position set; the cloud server receives a feedback position set corresponding to the at least one target position set, wherein the feedback position set is obtained by modifying the at least one target position set; the cloud server determines the target position of the target object in the target image based on the feedback position set.

[0135] In the above embodiment of the present application, the cloud server outputs at least one target location set, including: the cloud server determines the preset display mode corresponding to each target location set based on the comparison result of the confidence of at least one target location set and the preset confidence; the cloud server outputs at least one target location set according to the preset display mode.

[0136] It should be noted that the preferred implementation scheme involved in the above embodiments of this application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0137] Example 4

[0138] According to an embodiment of the present application, an image processing device for implementing the above image processing method is also provided. Figure 8 As shown, the device 800 includes: an acquisition module 802 , an extraction module 804 , a prediction module 806 , and a determination module 808 .

[0139] Among them, the acquisition module 802 is used to acquire the target image, wherein the target image contains the target object; the extraction module 804 is used to extract features from the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the prediction module 806 is used to determine at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; the determination module 808 is used to determine the target position of the target object in the target image based on at least one target position set.

[0140] It should be noted that the acquisition module 802, extraction module 804, prediction module 806, and determination module 808 correspond to steps S202 to S208 in Example 1. The examples and application scenarios implemented by the four modules and the corresponding steps are the same, but are not limited to the contents disclosed in Example 1. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in Example 1.

[0141] In the above embodiment of the present application, the determination module includes: a determination unit and an adjustment unit.

[0142] The determination unit is used to determine an initial detection frame of the target object in the target image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; the determination unit is used to determine a target feature vector corresponding to the initial detection frame from a feature vector of at least one feature space; and the adjustment unit is used to adjust the initial detection frame based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to represent the target position.

[0143] In the above embodiment of the present application, the determining unit is further configured to determine a feature vector located within the initial detection frame from feature vectors in at least one feature space to obtain a target feature vector.

[0144] In the above embodiments of the present application, the device further includes: a screening module.

[0145] Among them, the screening module is used to screen the initial detection frame based on the confidence of at least one target position set to obtain a screened detection frame, wherein the confidence is used to characterize the degree of matching between the at least one target position set and the target object; the determination module is used to determine the feature vector located in the screened detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0146] In the above embodiments of the present application, the device further includes: an output module and a receiving module.

[0147] Among them, the output module is used to output at least one target position set; the receiving module is used to receive a feedback position set corresponding to the at least one target position set, wherein the feedback position set is obtained by modifying the at least one target position set; the determination module is used to determine the target position of the target object in the target image based on the feedback position set.

[0148] In the above embodiments of the present application, the output module includes: a determination unit and an output unit.

[0149] The determining unit is used to determine a preset display mode corresponding to each target position set based on a comparison result between the confidence level of at least one target position set and a preset confidence level; and the output unit is used to output at least one target position set according to the preset display mode.

[0150] It should be noted that the preferred implementation scheme involved in the above embodiments of this application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0151] Example 5

[0152] According to an embodiment of the present application, an image processing device for implementing the above image processing method is also provided. Figure 9 As shown, the device 900 includes: an acquisition module 902 , an extraction module 904 , a detection module 906 , and a determination module 908 .

[0153] Among them, the acquisition module 902 is used to acquire a vehicle image, wherein the vehicle image contains a vehicle; the extraction module 904 is used to extract features from the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the detection module 906 is used to determine at least one target position set corresponding to the vehicle based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the vehicle in each feature space; the determination module 908 is used to determine the target position of the vehicle in the vehicle image based on at least one target position set.

[0154] It should be noted that the acquisition module 902, extraction module 904, detection module 906, and determination module 908 correspond to steps S602 to S608 in Example 2. The examples and application scenarios implemented by the four modules and the corresponding steps are the same, but are not limited to the contents disclosed in Example 2. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in Example 1.

[0155] It should be noted that the preferred implementation scheme involved in the above embodiments of this application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0156] Example 6

[0157] According to an embodiment of the present application, an image processing device for implementing the above image processing method is also provided. Figure 10 As shown, the apparatus 1000 includes: a receiving module 1002 , an extracting module 1004 , a first determining module 1006 , a second determining module 1008 , and a sending module 1010 .

[0158] Among them, the receiving module 1002 is used to receive the target image uploaded by the client through the cloud server, wherein the target image includes the target object; the extraction module 1004 is used to perform feature extraction on the target image through the cloud server to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the first determination module 1006 is used to determine at least one target position set corresponding to the target object based on the feature vector of at least one feature space through the cloud server, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; the second determination module 1008 is used to determine the target position of the target object in the target image based on at least one target position set through the cloud server; the sending module 1010 is used to send the target position to the client through the cloud server.

[0159] It should be noted that the receiving module 1002, extraction module 1004, first determination module 1006, second determination module 1008, and sending module 1010 correspond to steps S702 to S710 in Example 3. The examples and application scenarios implemented by the five modules and the corresponding steps are the same, but are not limited to the contents disclosed in Example 3. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in Example 1.

[0160] It should be noted that the preferred implementation scheme involved in the above embodiments of this application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0161] Example 7

[0162] The embodiment of the present application can provide a computer terminal, which can be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal can also be replaced by a terminal device such as a mobile terminal.

[0163] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of a computer network.

[0164] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the image processing method: obtaining a target image, wherein the target image contains a target object; performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; based on the feature vector of at least one feature space, determining at least one target position set corresponding to the target object, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; based on at least one target position set, determining the target position of the target object in the target image.

[0165] Optionally, Figure 11 This is a structural block diagram of a computer terminal according to an embodiment of the present application. Figure 11 As shown, the computer terminal 1100 may include: one or more (only one is shown in the figure) processors and a memory.

[0166] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the image processing method and device in the embodiments of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned image processing method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal 1100 via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof.

[0167] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain a target image, wherein the target image contains a target object; perform feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; based on the feature vector of at least one feature space, determine at least one target position set corresponding to the target object, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; based on at least one target position set, determine the target position of the target object in the target image.

[0168] Optionally, the processor may also execute the program code of the following steps: determining an initial detection frame of the target object in the target image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; determining a target feature vector corresponding to the initial detection frame from a feature vector of at least one feature space; adjusting the initial detection frame based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to represent the target position.

[0169] Optionally, the processor may further execute program code of the following steps: determining a feature vector located within the initial detection frame from feature vectors in at least one feature space to obtain a target feature vector.

[0170] Optionally, the processor may also execute the program code of the following steps: based on the confidence of at least one target position set, filtering the initial detection frame to obtain a filtered detection frame, wherein the confidence is used to characterize the degree of matching between at least one target position set and the target object; determining the feature vector located within the filtered detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0171] Optionally, the above-mentioned processor can also execute the program code of the following steps: outputting at least one target position set; receiving a feedback position set corresponding to at least one target position set, wherein the feedback position set is obtained by modifying at least one target position set; and determining the target position of the target object in the target image based on the feedback position set.

[0172] Optionally, the processor may further execute program code of the following steps: determining a preset display mode corresponding to each target position set based on a comparison result between the confidence level of at least one target position set and a preset confidence level; and outputting at least one target position set according to the preset display mode.

[0173] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtaining a vehicle image, wherein the vehicle is contained in the vehicle image; performing feature extraction on the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; based on the feature vector of at least one feature space, determining at least one target position set corresponding to the vehicle, wherein the multiple target positions contained in each target position set are used to characterize the position of the vehicle in each feature space; based on at least one target position set, determining the target position of the vehicle in the vehicle image.

[0174] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: the cloud server receives the target image uploaded by the client, wherein the target image contains the target object; the cloud server extracts features from the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the cloud server determines at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; the cloud server determines the target position of the target object in the target image based on at least one target position set; the cloud server sends the target position to the client.

[0175] It can be understood by those skilled in the art that Figure 11 The structure shown is for illustration only, and the computer terminal may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 11 It does not limit the structure of the above electronic device. For example, the computer terminal 1100 may also include Figure 11 More or fewer components (such as network interfaces, display devices, etc.) shown in, or with Figure 11 Different configurations shown.

[0176] A person skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0177] Example 8

[0178] The embodiment of the present application further provides a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the image processing method provided in the above embodiment.

[0179] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0180] Optionally, in this embodiment, the storage medium is configured to store program code for executing the following steps: acquiring a target image, wherein the target image contains a target object; performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; determining at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; and determining the target position of the target object in the target image based on the at least one target position set.

[0181] Optionally, the storage medium is further configured to store program code for executing the following steps: determining an initial detection frame of the target object in the target image based on at least one target position set, wherein all target positions in the same target position set are located in the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; determining a target feature vector corresponding to the initial detection frame from a feature vector of at least one feature space; and adjusting the initial detection frame based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to represent the target position.

[0182] Optionally, the storage medium is further configured to store program codes for executing the following steps: determining a feature vector located within the initial detection frame from feature vectors in at least one feature space to obtain a target feature vector.

[0183] Optionally, the storage medium is also configured to store program code for executing the following steps: filtering the initial detection frame based on the confidence of at least one target position set to obtain a filtered detection frame, wherein the confidence is used to characterize the degree of matching between the at least one target position set and the target object; determining the feature vector located within the filtered detection frame from the feature vectors of at least one feature space to obtain the target feature vector.

[0184] Optionally, the above-mentioned storage medium is also configured to store program code for performing the following steps: outputting at least one target position set; receiving a feedback position set corresponding to at least one target position set, wherein the feedback position set is obtained by modifying at least one target position set; and determining the target position of the target object in the target image based on the feedback position set.

[0185] Optionally, the above-mentioned storage medium is also configured to store program code for executing the following steps: determining a preset display mode corresponding to each target position set based on a comparison result of the confidence of at least one target position set and a preset confidence; and outputting at least one target position set according to the preset display mode.

[0186] Optionally, in this embodiment, the storage medium is configured to store program code for executing the following steps: acquiring a vehicle image, wherein the vehicle image contains a vehicle; performing feature extraction on the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; determining at least one target position set corresponding to the vehicle based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the vehicle in each feature space; and determining the target position of the vehicle in the vehicle image based on the at least one target position set.

[0187] Optionally, in this embodiment, the storage medium is configured to store program code for executing the following steps: the cloud server receives a target image uploaded by the client, wherein the target image contains a target object; the cloud server performs feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of multiple feature spaces are different; the cloud server determines at least one target position set corresponding to the target object based on the feature vector of at least one feature space, wherein the multiple target positions contained in each target position set are used to characterize the position of the target object in each feature space; the cloud server determines the target position of the target object in the target image based on at least one target position set; and the cloud server sends the target position to the client.

[0188] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0189] In the above embodiments of the present application, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0190] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0191] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0192] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0193] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0194] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. An image processing method, characterized in that: include: Acquire a target image, wherein the target image includes a target object; Performing feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; Predicting a preset number of pixel points of the target object in the at least one feature space using a point extraction candidate box, and generating at least one target position set based on the preset number of pixel points, wherein each target position set includes multiple target positions used to represent the position of the target object in each feature space; Based on the at least one target position set, a target position of the target object in the target image is determined, wherein the target position is a position of a pixel point of the target object in the target image.

2. The method according to claim 1, characterized in that Determining the target position of the target object in the target image based on the at least one target position set includes: determining, based on the at least one target position set, an initial detection frame of the target object in the target image, wherein all target positions in the same target position set are located within the same initial detection frame, and the initial detection frame is used to represent the initial position of the target object in the target image; Determine a target feature vector corresponding to the initial detection frame from the feature vectors of the at least one feature space; The initial detection frame is adjusted based on the target feature vector to obtain a target detection frame, wherein the target detection frame is used to represent the target position.

3. The method according to claim 2, characterized in that Determining the target feature vector corresponding to the initial detection frame from the feature vectors of the at least one feature space includes: A feature vector located within the initial detection frame is determined from the feature vectors of the at least one feature space to obtain the target feature vector.

4. The method according to claim 3, characterized in that Before determining the feature vector located within the initial detection frame from the feature vectors of the at least one feature space to obtain the target feature vector, the method further includes: Based on the confidence level of the at least one target position set, the initial detection frame is filtered to obtain a filtered detection frame, wherein the confidence level is used to represent a degree of matching between the at least one target position set and the target object; A feature vector located within the filtered detection frame is determined from the feature vectors of the at least one feature space to obtain the target feature vector.

5. The method according to claim 1, wherein The method further comprises: outputting the at least one target position set; receiving a feedback position set corresponding to the at least one target position set, wherein the feedback position set is obtained by modifying the at least one target position set; The target position of the target object in the target image is determined based on the feedback position set.

6. The method according to claim 5, characterized in that Outputting the at least one target location set includes: Determining a preset display mode corresponding to each target location set based on a comparison result of the confidence level of the at least one target location set and a preset confidence level; The at least one target location set is output according to the preset display manner.

7. An image processing method, characterized in that: include: Acquire a vehicle image, wherein the vehicle image includes a vehicle; Performing feature extraction on the vehicle image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; Predicting a preset number of pixel points of the vehicle in the at least one feature space using a point extraction candidate box, and generating at least one target position set based on the preset number of pixel points, wherein each target position set includes multiple target positions used to represent the position of the vehicle in each feature space; Based on the at least one target position set, a target position of the vehicle in the vehicle image is determined, wherein the target position is a position of a pixel point of the target object in the vehicle image.

8. An image processing method, characterized in that: include: The cloud server receives a target image uploaded by a client, wherein the target image includes a target object; The cloud server performs feature extraction on the target image to obtain a feature vector of at least one feature space, wherein the resolutions of the multiple feature spaces are different; The cloud server predicts a preset number of pixel points of the target object in the at least one feature space using a point extraction candidate box, and generates at least one target position set based on the preset number of pixel points, wherein each target position set includes multiple target positions used to represent the position of the target object in each feature space; The cloud server determines a target position of the target object in the target image based on the at least one target position set, wherein the target position is a position of a pixel point of the target object in the target image; The cloud server sends the target location to the client.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the image processing method according to any one of claims 1 to 8.

10. A computer terminal, characterized in that: include: A memory and a processor, wherein the processor is configured to run a program stored in the memory, wherein the image processing method according to any one of claims 1 to 8 is executed when the program is run.

Images