Image recognition method, device and equipment

Abstract

The embodiments of the present application relate to the field of computers, and disclose an image recognition method, device, and equipment. Obtain image information of the recognition target, each image information includes: color information of each recognition target; obtain first coordinate information of the recognition target based on the color information of the recognition target; obtain depth map information of the recognition target; obtain spatial transformation features corresponding to the recognition target; obtain second coordinate information corresponding to the recognition target based on the first coordinate information and spatial transformation features corresponding to the recognition target; obtain dimension information based on the second coordinate information corresponding to the recognition target. It can be seen that the technical solution converts the original 3D recognition object recognition process into 2D image information recognition by utilizing the image information of the target object and the corresponding spatial feature transformation. In this way, the recognition process of the target object can be made simpler and the application difficulty can be reduced; on the other hand, the training of 3D recognition objects using 2D image information is realized, thereby improving training efficiency.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computers and to an image recognition method, device and equipment. Background Art

[0002] Image recognition technology refers to the technology of using computers to process, analyze and understand images to identify objects of various patterns. The objects of existing image recognition technology are usually divided into two-dimensional (2D) objects and three-dimensional (3D) objects according to the dimension. In some detection and recognition scenarios, computers are usually required to identify and distinguish 2D objects and 3D objects based on image information.

[0003] For the recognition of 2D objects, 2D image information is usually used in combination with corresponding learning methods to train the computer. The method is relatively mature and the recognition process is relatively simple. For the recognition of 3D objects, multiple image acquisition devices set at different angles are usually used to obtain depth information of the 3D object, and the corresponding spatial features of the 3D object are obtained in combination with point cloud technology, so that the computer can complete the recognition of the 3D object through the corresponding spatial features.

[0004] Currently, for the recognition of 3D objects, on the one hand, the process of obtaining depth information and applying point cloud technology is relatively complicated, which makes practical application difficult; on the other hand, the training objects are real objects, so the available training set resources are less than those for the recognition of 2D objects, resulting in low training efficiency. Summary of the invention

[0005] The embodiments of the present application provide an image recognition method, apparatus and device to solve the problems of difficulty in distinguishing and recognizing 3D images and low training efficiency in existing detection and recognition scenarios.

[0006] In a first aspect, an embodiment of the present application provides an image recognition method, the method comprising:

[0007] Acquire image information of at least one recognition target, wherein each recognition target corresponds to one image information, and each of the image information includes: color information corresponding to each recognition target;

[0008] Obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target;

[0009] Acquire depth map information of the at least one recognition target, wherein each recognition target corresponds to one depth map information;

[0010] Obtaining a spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target;

[0011] According to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, second coordinate information corresponding to each identified target is obtained, and the second coordinate information acts on the depth map information;

[0012] The dimension information of the identified target is obtained according to the second coordinate information corresponding to the identified target.

[0013] In this way, a spatial transformation feature relationship is established by identifying the image information and depth map information of the target, and the pixel coordinates obtained from the image information can be used to calculate the dimension information of the identified target, thereby reducing the difficulty of identifying 3D objects.

[0014] In some possible implementations, the image information is acquired through an image acquisition device, and the depth map information is acquired through a grayscale map acquisition device.

[0015] In this way, the collected image information and depth map information can be made independent of each other, which facilitates the subsequent establishment of spatial transformation features.

[0016] In some possible implementations, the first coordinate information of each recognized target in the corresponding image information includes: four pixel coordinate information of the outer contour of each recognized target in the corresponding image information.

[0017] In this way, the rectangular area where the outer contour of the recognition target is located can be obtained through the four pixel coordinate information, which is convenient for obtaining the second coordinate information through the spatial transformation feature later.

[0018] In some possible implementations, the spatial transformation feature is obtained by a distortion correction matrix.

[0019] In some possible implementations, the second coordinate information corresponding to each identified target includes: four depth value information of the outer contour of each identified target in the corresponding grayscale image, and the pixel coordinate information of each outer contour of each identified target corresponds one-to-one to the depth value information of the outer contour of each identified target.

[0020] In this way, the first coordinate information can be converted into the second coordinate information through the spatial transformation feature, and the features of the identified target in the picture information can be reflected in the grayscale image.

[0021] In some possible implementations, the obtaining of the identification target dimension information according to the second coordinate information corresponding to the identification target may be implemented by:

[0022] If the second coordinate information is greater than or equal to a preset threshold, it is determined that the dimension information corresponding to the identified target is a three-dimensional target;

[0023] If the second coordinate information is less than a preset threshold, it is determined that the dimension information corresponding to the identified target is a two-dimensional target.

[0024] In some possible implementations, obtaining the first coordinate information of each identified target in the corresponding image information based on the color information corresponding to each identified target is implemented through a target detection neural network, R-CNN.

[0025] In a second aspect, an embodiment of the present application further provides an image recognition device, the device comprising:

[0026] A first acquisition module is used to acquire image information of at least one recognition target, wherein each recognition target corresponds to a piece of image information, and each piece of image information includes: color information corresponding to each recognition target;

[0027] A first processing module, used for obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target;

[0028] A second acquisition module is used to acquire the depth map information of the at least one recognition target, wherein each recognition target corresponds to a depth map information;

[0029] A feature construction module, used to obtain the spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target;

[0030] A second processing module, configured to obtain second coordinate information corresponding to each identified target according to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, wherein the second coordinate information acts on the depth map information;

[0031] A determination module is used to obtain the dimension information of the identified target based on the second coordinate information corresponding to the identified target.

[0032] In a third aspect, an embodiment of the present application further provides an electronic device, the electronic device comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus;

[0033] The memory is used to store executable instructions, and when the executable instructions are executed, the processor executes the test method in any possible implementation of the first aspect or the second aspect.

[0034] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein executable instructions are stored in the storage medium, and when the executable instructions are executed, a computing device executes the test method in any possible implementation of the first aspect or the second aspect.

[0035] An embodiment of the present application provides a technical solution for an image recognition method. In this solution, first, picture information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of picture information, and each piece of picture information includes: color information corresponding to each recognition target; first coordinate information of each recognition target in the corresponding picture information is obtained based on the color information corresponding to each recognition target; depth map information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of depth map information; spatial transformation features corresponding to each recognition target are obtained based on the picture information corresponding to each recognition target and the depth map information corresponding to each recognition target; second coordinate information corresponding to each recognition target is obtained based on the first coordinate information corresponding to each recognition target and the spatial transformation features corresponding to each recognition target, and the second coordinate information acts on the depth map information; and dimension information of the recognition target is obtained based on the second coordinate information corresponding to the recognition target. Since the color information of the identified target is used in the process to obtain the coordinates of the area where the identified target is located in the image information, and according to the spatial feature transformation between the image information and the depth map information, the second spatial coordinate corresponding to the identified target in the grayscale image can be calculated, and the size of the second spatial coordinate value is combined with the threshold preset in the corresponding grayscale image for comparison, and the final judgment result is used as the confirmation of the dimensional information of the identified target. It can be seen that the technical solution of the present application converts the original 3D recognition object recognition process into a 2D image information recognition process by utilizing the image information of the target object and the corresponding spatial feature transformation. In this way, on the one hand, the target object recognition process can be made simpler and the application difficulty can be reduced; on the other hand, the training of 3D recognition object recognition using 2D image information is realized, thereby improving the training efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flowchart of the image processing method provided by the embodiment of the present application;

[0037] Figure 2 is a schematic diagram of an exemplary composition of an image processing device provided in an embodiment of the present application;

[0038] Figure 3 It is a schematic diagram of an exemplary structure of an image processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] The terms used in the following embodiments of the present application are for the purpose of describing optional implementation modes, and are not intended to be used as limitations to the present application. As used in the specification and the appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions. It should also be understood that although the terms first, second, etc. may be used to describe a certain class of objects in the following embodiments, the objects are not limited to these terms. These terms are used to distinguish the specific objects of this class of objects. For example, the same is true for other classes of objects that may be described using the terms first, second, etc. in the following embodiments, which will not be repeated here.

[0040] An embodiment of the present application provides a technical solution for an image recognition method. In this solution, first, picture information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of picture information, and each piece of picture information includes: color information corresponding to each recognition target; first coordinate information of each recognition target in the corresponding picture information is obtained based on the color information corresponding to each recognition target; depth map information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of depth map information; spatial transformation features corresponding to each recognition target are obtained based on the picture information corresponding to each recognition target and the depth map information corresponding to each recognition target; second coordinate information corresponding to each recognition target is obtained based on the first coordinate information corresponding to each recognition target and the spatial transformation features corresponding to each recognition target, and the second coordinate information acts on the depth map information; and dimension information of the recognition target is obtained based on the second coordinate information corresponding to the recognition target. Since the color information of the identified target is used in the process to obtain the coordinates of the area where the identified target is located in the image information, and according to the spatial feature transformation between the image information and the depth map information, the second spatial coordinate corresponding to the identified target in the grayscale image can be calculated, and the size of the second spatial coordinate value is combined with the threshold preset in the corresponding grayscale image for comparison, and the final judgment result is used as the confirmation of the dimensional information of the identified target. It can be seen that the technical solution of the present application converts the original 3D recognition object recognition process into a 2D image information recognition process by utilizing the image information of the target object and the corresponding spatial feature transformation. In this way, on the one hand, the target object recognition process can be made simpler and the application difficulty can be reduced; on the other hand, the training of 3D recognition object recognition using 2D image information is realized, thereby improving the training efficiency.

[0041] Any electronic device involved in the embodiments of the present application may be an electronic device such as a mobile phone, a tablet computer, a wearable device (e.g., a smart watch, a smart bracelet, etc.), a laptop computer, a desktop computer, and a vehicle-mounted device. The electronic device is pre-installed with a software deployment application. It is understood that the embodiments of the present application do not impose any restrictions on the specific type of electronic device.

[0042] Image recognition technology refers to the technology of using computers to process, analyze and understand images to identify objects of various patterns. The objects of existing image recognition technology are usually divided into two-dimensional (2D) objects and three-dimensional (3D) objects according to the dimension. In some detection and recognition scenarios, computers are usually required to identify and distinguish 2D objects and 3D objects based on image information.

[0043] For the recognition of 2D objects, 2D image information is usually used in combination with corresponding learning methods to train the computer. The method is relatively mature and the recognition process is relatively simple. For the recognition of 3D objects, multiple image acquisition devices set at different angles are usually used to obtain depth information of the 3D object, and the corresponding spatial features of the 3D object are obtained in combination with point cloud technology, so that the computer can complete the recognition of the 3D object through the corresponding spatial features.

[0044] Currently, for the recognition of 3D objects, on the one hand, the process of obtaining depth information and applying point cloud technology is relatively complicated, which makes practical application difficult; on the other hand, the training objects are real objects, so the available training set resources are less than those for the recognition of 2D objects, resulting in low training efficiency.

[0045] The following is a description of several exemplary implementations, illustrating the technical solutions of the embodiments of the present application and the technical effects produced by the technical solutions of the present application.

[0046] In the first aspect of the present application, an image recognition method is proposed, see Figure 1 , Figure 1 The image processing method provided in the embodiment of the present application includes the following steps:

[0047] Acquire image information of at least one recognition target, wherein each recognition target corresponds to one image information, and each of the image information includes: color information corresponding to each recognition target;

[0048] Obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target;

[0049] Acquire depth map information of the at least one recognition target, wherein each recognition target corresponds to one depth map information;

[0050] Obtaining a spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target;

[0051] According to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, second coordinate information corresponding to each identified target is obtained, and the second coordinate information acts on the depth map information;

[0052] The dimension information of the identified target is obtained according to the second coordinate information corresponding to the identified target.

[0053] Optionally, the image information is acquired through an image acquisition device (such as a high-definition camera), and the depth map information is acquired through a depth map (such as a depth camera) acquisition device.

[0054] Optionally, the depth map information obtained can be presented in the form of a grayscale image. Taking the K device of a certain platform as an example, the amount of depth map data information obtained is the same as the number of pixels, and the value range can be 0 to 65535 frames. According to the different device precisions of the actual application, the value range is mapped to 0 to 1 frame to obtain the corresponding grayscale image.

[0055] Optionally, the first coordinate information of each recognized target in the corresponding picture information includes: four pixel coordinate information of the outer contour of each recognized target in the corresponding picture information.

[0056] Optionally, the four pixel coordinate information can be the horizontal and vertical coordinate information corresponding to two pixel points respectively, for example: the coordinates (x1, y1) corresponding to pixel point A, and the coordinates (x2, y2) corresponding to pixel point B. The four pixel coordinate information are: x1, y1, x2 and y2.

[0057] Optionally, the four pixel coordinate information includes: x (the horizontal coordinate of the rectangular coordinate system of the plane where the target is located), y (the vertical coordinate of the rectangular coordinate system of the plane where the target is located), width (the horizontal width corresponding to the target) and height (the vertical height corresponding to the target).

[0058] Optionally, the spatial transformation feature is obtained by a distortion correction matrix.

[0059] Optionally, the second coordinate information corresponding to each identified target includes: four depth value information of the outer contour of each identified target in the corresponding grayscale image, and the pixel coordinate information of each outer contour of each identified target corresponds one-to-one to the depth value information of the outer contour of each identified target.

[0060] Optionally, the obtaining of the identification target dimension information according to the second coordinate information corresponding to the identification target is implemented by:

[0061] If the second coordinate information is greater than or equal to a preset threshold, it is determined that the dimension information corresponding to the identified target is a three-dimensional target;

[0062] If the second coordinate information is less than a preset threshold, it is determined that the dimension information corresponding to the identified target is a two-dimensional target.

[0063] Optionally, obtaining the first coordinate information of each identified target in the corresponding image information based on the color information corresponding to each identified target is implemented through a target detection neural network, R-CNN.

[0064] For example, a company needs to identify the specific number of beverages on a shelf (i.e., a three-dimensional recognition object), but needs to distinguish pictures of beverage bottles on a billboard placed near the shelf (i.e., a two-dimensional recognition object) to avoid interference when comparing the quantity, that is, to achieve the distinction between the two-dimensional information and the three-dimensional information of the recognition target.

[0065] First, you can use the image acquisition device and the grayscale image acquisition device to obtain the color information and depth map information of the beverage bottle image on the billboard. In this way, you can get the color information and depth map information corresponding to the two-dimensional recognition target;

[0066] Through the color information and in combination with the training of the corresponding neural network (for example, the neural network can identify the location of the beverage bottle in the picture according to the color information), the pixel coordinates corresponding to the outline of the beverage bottle in the billboard (i.e., the first coordinate information) or the four pixel vertex coordinates of the rectangular frame of the area where the beverage bottle is located in the billboard (which can also be used as the first coordinate information) are obtained;

[0067] According to the acquired image information and depth map information, a distortion correction matrix (i.e., spatial transformation feature) corresponding to the beverage bottle in the billboard is obtained, and the relationship between the pixel-by-pixel corresponding grayscale value of the identified object (i.e., the beverage bottle in the billboard) in the image information can be obtained through the matrix;

[0068] Furthermore, by acquiring the depth map information of the two-dimensional recognition target (including the second coordinate information and the corresponding depth value information), a threshold for distinguishing the dimensional information of the recognition target can be obtained and set. For example, when the depth value information of the recognition target is less than this threshold, it will be judged as a two-dimensional target (that is, the dimensional information of the recognition target is a two-dimensional target), and when the depth value information of the recognition target is greater than or equal to this threshold, it will be judged as a three-dimensional target (that is, the dimensional information of the recognition target is a three-dimensional target). In the process, multiple depth image acquisition devices are not used, and the point cloud method is not used to distinguish and identify the recognition target in two dimensions and three dimensions, which simplifies the process of distinguishing between two-dimensional and three-dimensional recognition targets. In addition, this process can obtain a training set of two-dimensional targets, which is used as the system's training and recognition of two-dimensional recognition targets.

[0069] Similarly, the three-dimensional recognition target (i.e., the beverage bottle on the shelf) is recognized, as described above, and will not be described here one by one. Finally, the training set of the three-dimensional target is obtained. Since the three-dimensional target also uses image information and grayscale image information, the training set of the three-dimensional target is also a picture set. In this way, the training of the three-dimensional target using the picture set (two-dimensional training set) can be realized, which improves the training efficiency.

[0070] An embodiment of the present application provides a technical solution for an image recognition method. In this solution, first, picture information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of picture information, and each piece of picture information includes: color information corresponding to each recognition target; first coordinate information of each recognition target in the corresponding picture information is obtained based on the color information corresponding to each recognition target; depth map information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of depth map information; spatial transformation features corresponding to each recognition target are obtained based on the picture information corresponding to each recognition target and the depth map information corresponding to each recognition target; second coordinate information corresponding to each recognition target is obtained based on the first coordinate information corresponding to each recognition target and the spatial transformation features corresponding to each recognition target, and the second coordinate information acts on the depth map information; and dimension information of the recognition target is obtained based on the second coordinate information corresponding to the recognition target. Since the color information of the target to be identified is used in the process to obtain the coordinates of the area where the target to be identified is located in the image information, and according to the spatial feature transformation between the image information and the depth map information, the second spatial coordinate corresponding to the target to be identified in the grayscale image can be calculated, and the size of the second spatial coordinate value is combined with the threshold preset in the corresponding grayscale image for comparison, and the final judgment result is used as the confirmation of the dimensional information of the target to be identified. It can be seen that by utilizing the image information of the target object and the corresponding spatial feature transformation, the original 3D recognition object recognition process is converted into a 2D image information recognition process. In this way, on the one hand, the target object recognition process can be made simpler and the application difficulty can be reduced; on the other hand, the training of 3D recognition object recognition using 2D image information is realized, which improves the training efficiency.

[0071] The above embodiments introduce various implementation methods of the image processing method provided by the embodiments of the present application from the aspects of obtaining picture information of at least one identified target, obtaining first coordinate information, obtaining depth map information of at least one identified target, obtaining spatial transformation features, obtaining second coordinate information, and obtaining target dimension information. It should be understood that the processing steps of obtaining picture information of at least one identified target, obtaining first coordinate information, obtaining depth map information of at least one identified target, obtaining spatial transformation features, obtaining second coordinate information, and obtaining target dimension information, etc., the embodiments of the present application can implement the above functions in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0072] For example, if the above implementation steps implement the corresponding functions through software modules. Figure 2 As shown, the image recognition device may include a first acquisition module, a first processing module, a second acquisition module, a feature construction module, a second processing module and a determination module. The image processing device may be used to perform part or all of the operations of the above-mentioned image processing method.

[0073] For example:

[0074] A first acquisition module is used to acquire image information of at least one recognition target, wherein each recognition target corresponds to a piece of image information, and each piece of image information includes: color information corresponding to each recognition target;

[0075] A first processing module, used for obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target;

[0076] A second acquisition module is used to acquire the depth map information of the at least one recognition target, wherein each recognition target corresponds to a depth map information;

[0077] A feature construction module, used to obtain the spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target;

[0078] A second processing module, configured to obtain second coordinate information corresponding to each identified target according to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, wherein the second coordinate information acts on the depth map information;

[0079] A determination module is used to obtain the dimension information of the identified target based on the second coordinate information corresponding to the identified target.

[0080] It can be seen from this that

[0081] An embodiment of the present application provides a technical solution for an image recognition method. In this solution, first, picture information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of picture information, and each piece of picture information includes: color information corresponding to each recognition target; first coordinate information of each recognition target in the corresponding picture information is obtained based on the color information corresponding to each recognition target; depth map information of at least one recognition target is obtained, wherein each recognition target corresponds to a piece of depth map information; spatial transformation features corresponding to each recognition target are obtained based on the picture information corresponding to each recognition target and the depth map information corresponding to each recognition target; second coordinate information corresponding to each recognition target is obtained based on the first coordinate information corresponding to each recognition target and the spatial transformation features corresponding to each recognition target, and the second coordinate information acts on the depth map information; and dimension information of the recognition target is obtained based on the second coordinate information corresponding to the recognition target. Since the color information of the identified target is used in the process to obtain the coordinates of the area where the identified target is located in the image information, and according to the spatial feature transformation between the image information and the depth map information, the second spatial coordinate corresponding to the identified target in the grayscale image can be calculated, and the size of the second spatial coordinate value is combined with the threshold preset in the corresponding grayscale image for comparison, and the final judgment result is used as the confirmation of the dimensional information of the identified target. It can be seen that the technical solution of the present application converts the original 3D recognition object recognition process into a 2D image information recognition process by utilizing the image information of the target object and the corresponding spatial feature transformation. In this way, on the one hand, the target object recognition process can be made simpler and the application difficulty can be reduced; on the other hand, the training of 3D recognition object recognition using 2D image information is realized, thereby improving the training efficiency.

[0082] It is understandable that the functions of the above modules can be integrated into hardware entities for implementation. For example, the acquisition module can be integrated into a transceiver for implementation, the generation module, the selection module and the construction module can be integrated into a processor for implementation, and the programs and instructions for implementing the functions of the above modules can be maintained in a memory. Figure 3 As shown, an electronic device is provided, the electronic device includes a processor, a transceiver and a memory, wherein the transceiver is used to execute the image generation method for acquiring image information and object pictures, and the memory is used to store the program / code pre-installed in the above deployment device, and can also store the code for the processor to execute. When the processor runs the code stored in the memory, the electronic device executes part or all of the operations of the software deployment method in the above method.

[0083] The specific process is described in the embodiment of the above method and will not be described in detail here.

[0084] In a specific implementation, corresponding to the aforementioned electronic device, an embodiment of the present application further provides a computer storage medium, wherein the computer storage medium provided in the electronic device may store a program, and when the program is executed, some or all of the steps in each embodiment of the aforementioned software deployment method may be implemented. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0085] One or more of the above modules or units can be implemented by software, hardware or a combination of the two. When any of the above modules or units is implemented by software, the software exists in the form of computer program instructions and is stored in a memory, and the processor can be used to execute the program instructions and implement the above method flow. The processor may include but is not limited to at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor and other types of computing devices that run software, each of which may include one or more cores for executing software instructions to perform operations or processing. The processor may be built into an SoC (system on chip) or an application specific integrated circuit (ASIC), or it may be an independent semiconductor chip. In addition to the core used to execute software instructions for operations or processing in the processor, it may further include necessary hardware accelerators, such as field programmable gate arrays (FPGA), PLDs (programmable logic devices), or logic circuits that implement dedicated logic operations.

[0086] When the above modules or units are implemented in hardware, the hardware can be any one or any combination of a CPU, a microprocessor, a DSP, an MCU, an artificial intelligence processor, an ASIC, a SoC, an FPGA, a PLD, a dedicated digital circuit, a hardware accelerator or a non-integrated discrete device, which can run the necessary software or not rely on the software to execute the above method flow.

[0087] Furthermore, Figure 3The bus interface may also include a bus interface, which may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by the processor and memory represented by the memory. The bus interface may also link various other circuits such as peripherals, voltage regulators, and power management circuits together, which are all well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver provides a unit for communicating with various other devices over a transmission medium. The processor is responsible for managing the bus architecture and general processing, and the memory may store data used by the processor when performing operations.

[0088] When the above modules or units are implemented using software, they can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integration. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0089] It should be understood that in various embodiments of the present application, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment.

[0090] Each part of this specification is described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment part.

[0091] Although the optional embodiments of the present application have been described, those skilled in the art, once informed of the basic creative concept, may make additional changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiment and all changes and modifications falling within the scope of the present application.

[0092] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the present application should be included in the scope of protection of the present invention.

Claims

1. An image recognition method, characterized in that: The method comprises: Acquire image information of at least one recognition target, wherein each recognition target corresponds to one image information, and each of the image information includes: color information corresponding to each recognition target; Obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target; Acquire depth map information of the at least one recognition target, wherein each recognition target corresponds to one depth map information; Obtaining a spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target; According to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, second coordinate information corresponding to each identified target is obtained, and the second coordinate information acts on the depth map information; Obtaining dimension information of the identified target according to the second coordinate information corresponding to the identified target; The second coordinate information corresponding to each identified target includes: four depth value information of the outer contour of each identified target in the corresponding grayscale image, and the pixel coordinate information of the outer contour of each identified target corresponds one-to-one to the depth value information of the outer contour of each identified target.

2. The image recognition method according to claim 1, characterized in that: The image information is acquired through an image acquisition device, and the depth map information is acquired through a depth map acquisition device.

3. The image recognition method according to claim 1, characterized in that: The first coordinate information of each recognition target in the corresponding picture information includes: four pixel coordinate information of the outer contour of each recognition target in the corresponding picture information.

4. The image recognition method according to claim 1, characterized in that: The spatial transformation features are obtained from a distortion correction matrix.

5. The image recognition method according to claim 1, characterized in that: The obtaining of the dimension information of the identified target according to the second coordinate information corresponding to the identified target is implemented by: If the depth value information is greater than or equal to a preset threshold, it is determined that the dimension information corresponding to the identified target is a three-dimensional target; If the depth value information is less than a preset threshold, the dimension information corresponding to the identified target is determined to be a two-dimensional target.

6. The image recognition method according to claim 1, characterized in that: The first coordinate information of each identified target in the corresponding image information is obtained based on the color information corresponding to each identified target through the target detection neural network, R-CNN.

7. An image recognition device, characterized in that: The device comprises: A first acquisition module is used to acquire image information of at least one recognition target, wherein each recognition target corresponds to a piece of image information, and each piece of image information includes: color information corresponding to each recognition target; A first processing module, used for obtaining first coordinate information of each recognition target in the corresponding image information according to the color information corresponding to each recognition target; A second acquisition module is used to acquire the depth map information of the at least one recognition target, wherein each recognition target corresponds to a depth map information; A feature construction module, used to obtain the spatial transformation feature corresponding to each recognition target according to the image information corresponding to each recognition target and the depth map information corresponding to each recognition target; A second processing module, configured to obtain second coordinate information corresponding to each identified target according to the first coordinate information corresponding to each identified target and the spatial transformation feature corresponding to each identified target, wherein the second coordinate information acts on the depth map information; A determination module is used to obtain the dimension information of the identified target based on the second coordinate information corresponding to the identified target; the second coordinate information corresponding to each identified target includes: four depth value information of the outer contour of each identified target in the corresponding grayscale image, and the pixel coordinate information of the outer contour of each identified target corresponds one-to-one to the depth value information of the outer contour of each identified target.

8. An electronic device, characterized in that: The electronic device comprises: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 6.

Images