Method and device for identifying object attributes, storage medium and electronic device

Abstract

The present disclosure provides a method and device for identifying object attributes, a storage medium and an electronic device, including obtaining an image to be processed, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and the actual distance between the target object and multiple objects around it, then inputting the information to be identified into a first object attribute recognition model, and finally obtaining a first object attribute label. The technical solution of the embodiment of the invention ensures that the first object attribute recognition model can combine and consider the influence between the target object and multiple objects around the target object, so that the obtained first object attribute label is more accurate and stable.

Description

Technical Field

[0001] The present disclosure relates to the field of computer vision technology, and in particular to a method and device for identifying object attributes, a storage medium, and an electronic device. Background Art

[0002] Object recognition is a basic research in the field of computer vision. The recognition of object attributes is the basis and prerequisite of object recognition, and it is also an important direction in the field of object recognition.

[0003] In related technologies, object attribute recognition methods mostly use human action sequences and labels of target objects to recognize object attributes. However, target objects do not exist in isolation. Objects are closely related to each other and have significant mutual influence and effects. Ignoring the interaction and influence between objects will lead to inaccurate and unstable object attribute recognition.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The purpose of the present disclosure is to provide a method and device for identifying object attributes, a storage medium and an electronic device, which can at least to some extent overcome the problem that the object attribute identification is not accurate and stable due to the limitation of related technologies that ignores the mutual influence between the target object and the objects around the target object.

[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0007] According to one aspect of the present disclosure, a method for identifying an attribute of an object is provided, comprising:

[0008] Get the image to be processed;

[0009] Preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around the target object;

[0010] The information to be identified is input into a first object attribute identification model to obtain a first object attribute label.

[0011] In one embodiment of the present disclosure, the identification method further includes:

[0012] Preprocessing the image to be processed to form a sample to be trained, wherein the sample to be trained includes a target object image, multiple object images around the target object, and an actual distance between the target object and multiple surrounding objects;

[0013] Inputting the sample to be trained into a first preset model to obtain a first initial object attribute label;

[0014] According to the comparison between the target object standard attribute label and the first initial object attribute label, the first preset model parameters are modified until the training condition of the first preset model meets the preset conditions, thereby obtaining the trained first object attribute recognition model.

[0015] In one embodiment of the present disclosure, the first object attribute recognition model includes:

[0016] A relationship graph, wherein the relationship graph is generated by the actual distance between the target object and multiple surrounding objects, the target object image and multiple object images surrounding the target object, wherein each object is regarded as a node on the relationship graph, some nodes are marked with object attribute labels, and the link labels between the target object node and multiple object nodes surrounding the target object have weights;

[0017] A graph label propagation calculation module, used to receive the weights and the object attribute labels of the partial node labels as input, and output a preliminary attribute feature vector through a graph label propagation algorithm;

[0018] The graph attention mechanism processing module is used to receive the preliminary attribute feature vector input, output the final attribute feature vector through the graph attention mechanism algorithm, and generate a first object attribute label.

[0019] In one embodiment of the present disclosure, the first object attribute label includes: a hardness attribute label, a volume attribute label, and a shape attribute label of the object.

[0020] In one embodiment of the present disclosure, the identification method further includes:

[0021] Preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object and images of multiple objects around the target object;

[0022] The information to be identified is input into a second object attribute identification model to obtain a second object attribute label.

[0023] In one embodiment of the present disclosure, the second object attribute recognition model includes:

[0024] A gray level co-occurrence matrix calculation module is used to receive the target object image and multiple object image inputs around the target object, and output ten features of the gray level co-occurrence matrix of each object through a gray level co-occurrence matrix algorithm;

[0025] The support vector machine processing module is used to receive the ten feature inputs, output a final object feature vector through the support vector machine algorithm, and generate a second object attribute label.

[0026] In one embodiment of the present disclosure, the second object attribute label includes: a smooth attribute label and a rough attribute label of the object.

[0027] According to another aspect of the present disclosure, there is provided a device for identifying attributes of an object, comprising:

[0028] An image acquisition module, used for acquiring an image to be processed;

[0029] An image processing module, used for preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around it;

[0030] The attribute recognition module is used to input the information to be recognized into the first object attribute recognition model to obtain the first object attribute label.

[0031] According to another aspect of the present disclosure, there is provided an electronic device, characterized in that it includes:

[0032] Processor; and

[0033] A memory, configured to store executable instructions of the processor;

[0034] Wherein, the processor is configured to perform the above-mentioned object attribute recognition method by executing the executable instructions.

[0035] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned object attribute recognition method is implemented.

[0036] The embodiments of the present disclosure provide a method and device for identifying object attributes, a storage medium, and an electronic device, including obtaining an image to be processed, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and the actual distance between the target object and multiple objects around it, and then inputting the information to be identified into a first object attribute recognition model, and finally obtaining a first object attribute label. The technical solution of the embodiment of the invention ensures that the first object attribute recognition model can combine and consider the influence between the target object and multiple objects around the target object, so that the obtained first object attribute label is more accurate and stable.

[0037] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0039] Figure 1 A flow chart of a method for identifying object attributes in an embodiment of the present disclosure is shown;

[0040] Figure 2 yes Figure 1 A flow chart of a method for obtaining a trained first object attribute recognition model;

[0041] Figure 3 yes Figure 1 A flowchart of a first object attribute recognition model recognizing object attributes;

[0042] Figure 4 yes Figure 1 The structural block diagram of the first object attribute recognition model;

[0043] Figure 5 A flow chart showing another method for identifying object attributes in an embodiment of the present disclosure is shown;

[0044] Figure 6 yes Figure 5 A flow chart of a method for obtaining a trained second object attribute recognition model;

[0045] Figure 7 yes Figure 5 A flowchart of a second object attribute recognition model recognizing object attributes;

[0046] Figure 8 yes Figure 5 The structural block diagram of the second object attribute recognition model;

[0047] Fig. 9 A flow chart showing another method for identifying object attributes in an embodiment of the present disclosure is shown;

[0048] Fig.10 Schematically illustrating an electronic device for implementing the above-mentioned object attribute recognition method according to an exemplary embodiment of the present invention DETAILED DESCRIPTION

[0049] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0050] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0051] The exemplary embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0052] Figure 1 A flowchart of a method for identifying object attributes disclosed in the present invention is schematically shown.

[0053] refer to Figure 1 , the recognition method applied to an object attribute may include:

[0054] Step S101, obtaining an image to be processed;

[0055] Step S102, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around the target object;

[0056] Step S103: input the information to be identified into the first object attribute recognition model 400 to obtain a first object attribute label.

[0057] The embodiments of the present disclosure provide a method and device for identifying object attributes, a storage medium, and an electronic device, including obtaining an image to be processed, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and the actual distance between the target object and multiple objects around it, and then inputting the information to be identified into a first object attribute recognition model 400, and finally obtaining a first object attribute label. The technical solution of the embodiment of the invention ensures that the first object attribute recognition model 400 can combine and consider the influence between the target object and multiple objects around the target object, so that the obtained first object attribute label is more accurate and stable.

[0058] First, the purpose of the invention of the exemplary embodiment of the present invention is explained and described.

[0059] The current object attribute recognition methods mainly focus on the target object. The basic idea is to first obtain the image of the target object, process the image of the target object, and then perform feature extraction. Then, based on the extracted texture features, shape features, etc., traditional machine learning methods or emerging deep learning methods are used to train a judgment model. Finally, the trained judgment model is used to test and judge unlabeled, unknown data to identify the category of the object attribute.

[0060] However, the target object does not exist in isolation. Further refinement and research on the various properties of the target object only ignores the close relationship between the target object and the surrounding objects, and the significant mutual influence and interaction between them, which will lead to inaccurate and unstable object attribute recognition.

[0061] In addition, the exemplary embodiment of the present invention uses information about the target object and multiple objects around the target object to train a preset model and perform a parameter adjustment process from the perspective of an object attribute recognition model. This allows the subsequently generated object attribute recognition model to obtain object attribute labels that are more accurate and stable when performing object recognition.

[0062] Next, the object attribute recognition method involved in the exemplary embodiment of the present invention is explained and illustrated.

[0063] Figure 2Schematically shown Figure 1 FIG. 4 is a flow chart of a method for obtaining a trained first object attribute recognition model 400 .

[0064] Specifically, refer to Figure 2 The object attribute recognition method also includes a method for obtaining a trained first object attribute recognition model 400, the steps of which are as follows:

[0065] Step S201, obtaining an image to be processed;

[0066] Specifically, a camera or other device may be used to collect image information of the target object and multiple objects around the target object.

[0067] Step S202, preprocessing the image to be processed to form a sample to be trained, wherein the sample to be trained includes a target object image, multiple object images around the target object, and an actual distance between the target object and multiple objects around the target object;

[0068] Specifically, the image to be processed can be preprocessed, and each object in the image information of the target object and multiple objects around the target object can be segmented using image segmentation technology, and the actual distance between each object can be calculated. Among them, image segmentation technology is an important research direction in the field of computer vision and an important part of image semantic understanding. From a mathematical point of view, image segmentation is the process of dividing an image into non-overlapping areas. The technology-related scene object segmentation, human foreground and background segmentation, face and body parsing, 3D reconstruction and other technologies have been widely used in industries such as unmanned driving, augmented reality, and security monitoring.

[0069] Step S203, inputting the sample to be trained into a first preset model to obtain a first initial object attribute label;

[0070] Specifically, first, the preprocessed target object image, the images of multiple objects around the target object, and the actual distance between the target object and multiple objects around it are mapped to a static graph, and each object is regarded as a node on the graph to generate a relationship graph 401. Whether there is a link between two nodes is determined by the distance between the two corresponding objects. If the actual distance is greater than a certain set value, we believe that there is no link between the nodes corresponding to the two objects, otherwise there is a link. The smaller the distance, the greater the weight of the link, and the weight of the link corresponding to two objects that fit completely is considered to be the largest. Calculate the link weight between the target object node and the multiple object nodes around the target object. The larger the link weight, the greater the similarity between the two nodes, and annotate some labels of the object attributes.

[0071] Secondly, the weights and object attribute labels of some nodes are input into the graph label propagation calculation module 402, and semi-supervised learning is performed through the graph label propagation algorithm in the module to calculate the preliminary attribute characteristic vector. Among them, the label propagation algorithm (LPA) is a graph-based semi-supervised learning method, and its basic idea is to use the label information of labeled nodes to predict the label information of unlabeled nodes. The relationship between samples is used to establish a relational complete graph model. In the complete graph, the nodes include labeled and unlabeled data, that is, the above-mentioned link weight is the similarity between two nodes, and the node label is passed to other nodes according to the similarity. Label data is like a source, which can be used to label unlabeled data. The greater the similarity of the nodes, the easier it is to propagate the label.

[0072] Finally, the preliminary attribute feature vector is input into the graph attention mechanism processing module 403, and after being trained by the graph attention mechanism algorithm, the final attribute feature vector is output to generate the first initial object attribute label. Among them, the attention mechanism is a data processing method in machine learning, which is widely used in various types of machine learning tasks such as natural language processing (NLP), image processing (CV) and speech recognition. According to the different domains in which the attention mechanism is applied, that is, the different ways and positions of applying the attention weight, the attention mechanism is divided into three types: spatial domain, channel domain and mixed domain. Weighted aggregation of neighbors can be achieved, which is not only more robust to noisy neighbors, but also gives the model a certain degree of interpretability. The multi-headed graph attention mechanism algorithm can be used to mine the low-order relationships and high-order connection patterns in the learning relationship graph 401, and different weights are also given to neighbor nodes of the same order to achieve differentiated treatment, so as to learn more fine-grained relationships.

[0073] Step S204, based on the comparison between the target object standard attribute label and the first initial object attribute label, the first preset model parameters are modified until the training condition of the first preset model meets the preset conditions, thereby obtaining the trained first object attribute recognition model 400.

[0074] In the embodiment of the present disclosure, when training the first preset model, information about the target object and multiple objects around the target object is added to the training sample, so as to avoid studying only the various attributes of the target object while ignoring the factors that affect each other between objects. The first initial object attribute label output after the first preset model is trained is compared with the standard attribute label of the actual target object, and the first preset model parameters are corrected according to the comparison result analysis until the training of the first preset model meets the preset conditions, and the trained first object attribute recognition model 400 is obtained. After training, the parameters of the first preset model are corrected, so that the trained first object attribute recognition model 400 can recognize the object attributes more accurately and stably.

[0075] Figure 3 Schematically shown Figure 1 Flowchart of the first object attribute recognition model 400 recognizing object attributes.

[0076] Specifically, refer to Figure 3 The first object attribute recognition model 400 recognizes the object attribute method S103, and the steps are as follows:

[0077] Step S301, inputting the target object image, the images of multiple objects around the target object, and the actual distance between the target object and the multiple objects around it into the relationship graph 401, and obtaining the object attribute labels of the link labels with weights and partial node labels between the target object node and the multiple object nodes around the target object;

[0078] Step S302, inputting the weights and object attribute labels of some nodes into the graph label propagation calculation module 402, and outputting a preliminary attribute feature vector through the graph label propagation algorithm;

[0079] Step S303, input the preliminary attribute feature vector into the graph attention mechanism processing module 403, output the final attribute feature vector through the graph attention mechanism algorithm, and generate the first object attribute label.

[0080] Specifically, the final attribute feature vector includes multiple dimensions, including: hardness attribute feature vector, volume attribute feature vector and shape attribute feature vector, which respectively correspond to the hardness attribute label, volume attribute label and shape attribute label of the target object in the generated first object attribute label. For example, the shape attribute label can be a rectangle, a circle, a polygon, etc.

[0081] In the embodiments of the present disclosure, through the relationship graph 401 in the first object recognition model, the semi-supervised learning of the graph label propagation algorithm of the graph label propagation calculation module 402 and the graph attention mechanism algorithm of the graph attention mechanism processing module 403, after the preliminary learning of the graph label propagation algorithm, the graph attention mechanism is added for further processing, so that the hardness, shape and volume of the attributes of the object can be identified more accurately.

[0082] Figure 4 Schematically shown Figure 1 4. A structural block diagram of the first object attribute recognition model 400.

[0083] Specifically, refer to Figure 4 The first object attribute recognition model 400 includes a relationship graph 401, a graph label propagation calculation module 402 and a graph attention mechanism processing module 403.

[0084] A relationship graph 401, where the relationship graph 401 is generated by the actual distance between the target object and multiple surrounding objects, and the target object image and multiple object images surrounding the target object, wherein each object is regarded as a node on the relationship graph 401, some nodes are marked with object attribute labels, and the link labels between the target object node and multiple object nodes surrounding the target object have weights;

[0085] A graph label propagation calculation module 402 is used to receive the object attribute label input of the weight and some node labels, and output a preliminary attribute feature vector through a graph label propagation algorithm;

[0086] The graph attention mechanism processing module 403 is used to receive the preliminary attribute feature vector input, output the final attribute feature vector through the graph attention mechanism algorithm, and generate the first object attribute label.

[0087] Figure 5 A flowchart of another method for identifying object attributes in an embodiment of the present disclosure is schematically shown.

[0088] Specifically, refer to Figure 5 , the recognition method applied to an object attribute may also include:

[0089] Step S501, obtaining an image to be processed;

[0090] Step S502, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object and images of multiple objects around the target object;

[0091] Step S503: input the information to be identified into the second object attribute identification model 800 to obtain a second object attribute label.

[0092] In the embodiment of the present disclosure, the image to be processed is obtained, the image to be processed is pre-processed to form information to be identified, wherein the information to be identified includes the target object image and multiple object images around the target object, and then the information to be identified is input into the second object attribute recognition model 800, and finally the second object attribute label is obtained. The technical solution of the embodiment of the invention ensures that the second object attribute recognition model 800 can combine and consider the influence between the target object and multiple objects around the target object, so that the first object attribute label obtained is more accurate and stable.

[0093] Figure 6 yes Figure 5 FIG. 8 is a flow chart of a method for obtaining a trained second object attribute recognition model 800 .

[0094] Specifically, refer to Figure 6 The object attribute recognition method also includes a method for obtaining a trained second object attribute recognition model 800, the steps of which are as follows:

[0095] Step S601, obtaining an image to be processed;

[0096] Step S602, preprocessing the image to be processed to form a sample to be trained, wherein the sample to be trained includes a target object image and multiple object images around the target object;

[0097] Specifically, the preprocessing of the image to be processed has been described in detail above, please refer to the introduction of step S202, and will not be repeated here.

[0098] Step S603, inputting the sample to be trained into a second preset model to obtain a second initial object attribute label;

[0099] Specifically, the target object image and multiple object images around the target object are first input into the gray level co-occurrence matrix calculation module 802, and the gray level co-occurrence matrix algorithm is used to output ten features of the gray level co-occurrence matrix of each object. The ten features include: mean and standard deviation of energy, mean and standard deviation of entropy, mean and standard deviation of contrast, mean and standard deviation of inverse moment, and mean and standard deviation of correlation. Among them, the gray level co-occurrence matrix (GLCM, Gray-level co-occurrence matrix) refers to a common method for describing texture by studying the spatial correlation characteristics of gray. Since texture is formed by the repeated occurrence of gray distribution in spatial position, there will be a certain gray relationship between two pixels separated by a certain distance in the image space, that is, the spatial correlation characteristics of gray in the image. The gray level co-occurrence matrix method can obtain its co-occurrence matrix by calculating the gray image, and then obtain some eigenvalues ​​of the matrix by calculating the co-occurrence matrix to represent certain texture features of the image respectively.

[0100] Step S604, based on the comparison between the target object standard attribute label and the second initial object attribute label, the second preset model parameters are modified until the training condition of the second preset model meets the preset conditions, thereby obtaining the trained second object attribute recognition model 800.

[0101] In the embodiment of the present disclosure, the training result of the second preset model is compared with the standard attribute label of the target object, and then the second preset model parameters are corrected until the preset conditions are met to obtain the second object attribute recognition model 800, which can increase the stability of object attribute recognition.

[0102] Figure 7 Schematically shown Figure 5 Flowchart of the second object attribute recognition model 800 recognizing object attributes.

[0103] Specifically, refer to Figure 7The second object attribute recognition model 800 recognizes the object attribute method S503, and the steps are as follows:

[0104] Step S701, input the target object image and multiple images around the target object into the gray level co-occurrence matrix calculation module 802, and output ten features of the gray level co-occurrence matrix of each object through the gray level co-occurrence matrix algorithm.

[0105] Step S702: input the ten features into the support vector machine processing module 803, output the final object attribute feature vector through the support vector machine algorithm, and generate a second object attribute label.

[0106] Specifically, the final object attribute feature vector includes multiple dimensions, including: a smooth attribute feature vector and a rough attribute feature vector, which respectively correspond to: a smooth attribute label and a rough attribute label of the target object in the generated second object attribute label.

[0107] In the embodiment of the present disclosure, based on the target object image and multiple object images around the target object, the gray level co-occurrence matrix calculation module 802 is used to obtain ten features of the gray level co-occurrence matrix of each object, and multiple objects around the target object are also taken into account, so that the recognition result is more reliable and the recognition instability will not be caused by ignoring the influence between objects.

[0108] Figure 8 The schematic diagram shows Figure 5 800 is a structural block diagram of the second object attribute recognition model 800.

[0109] Specifically, refer to Figure 8 The second object attribute recognition model 800 includes a gray level co-occurrence matrix calculation module 802 and a support vector machine processing module 803 .

[0110] A gray level co-occurrence matrix calculation module 802 is used to receive an input of a target object image and multiple object images around the target object, and output ten features of a gray level co-occurrence matrix of each object through a gray level co-occurrence matrix algorithm;

[0111] The support vector machine processing module 803 is used to receive ten feature inputs, output a final object feature vector through a support vector machine algorithm, and generate a second object attribute label.

[0112] Fig. 9 A flow chart of another method for identifying object attributes in an embodiment of the present disclosure is shown.

[0113] Specifically, refer to Fig. 9 , a method for identifying object attributes also includes:

[0114] Step S901, obtaining an image to be processed;

[0115] Step S902, preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around the target object;

[0116] Step S903, input the target object image, the images of multiple objects around the target object, and the actual distance between the target object and the multiple objects around it into the first object attribute recognition model 400 to obtain a first object attribute label; input the target object image and the images of multiple objects around the target object into the second object attribute recognition model 800 to obtain a second object attribute label;

[0117] Step S904: Integrate the first object attribute label and the second object attribute label to output an object attribute label.

[0118] Specifically, the object attribute label includes a hardness attribute label, a volume attribute label, a shape attribute label, a smoothness attribute label, and a roughness attribute label of the target object.

[0119] In the embodiment of the present disclosure, the object attribute recognition model includes a first object attribute recognition model 400 and a second object attribute recognition model 800. By inputting the information to be recognized into the object attribute recognition model, and then integrating the first object attribute label and the second object attribute label to output the object attribute label, more object attribute recognition results can be recognized.

[0120] A system for identifying object attributes includes an image acquisition module, an image processing module and an attribute identification module.

[0121] An image acquisition module, used for acquiring an image to be processed;

[0122] An image processing module, used for preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around it;

[0123] The attribute recognition module is used to input the information to be recognized into the first object attribute recognition model 400 to obtain the first object attribute label.

[0124] Since the functions of the recognition system applied to the object attributes have been described in detail in the corresponding method embodiments, the present disclosure will not elaborate on them here.

[0125] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0126] Refer to the following Fig.10 The electronic device 1000 according to this embodiment of the present invention will be described. Fig.10 The electronic device 1000 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0127] like Fig.10 As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device 1000 may include but are not limited to: the at least one processing unit 1010, the at least one storage unit 1020, and a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010).

[0128] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 performs the steps according to various exemplary embodiments of the present invention described in the above “Exemplary Method” section of this specification. For example, the processing unit 1010 can perform the following steps: Figure 1 In step S101, an image to be processed is obtained; in step S102, the image to be processed is preprocessed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and an actual distance between the target object and multiple objects around the target object; in step S103, the information to be identified is input into a first object attribute recognition model to obtain a first object attribute label.

[0129] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 1021 and / or a cache memory unit 1022 , and may further include a read-only memory unit (ROM) 1023 .

[0130] The storage unit 1020 may also include a program / utility 1024 having a set (at least one) of program modules 1025, such program modules 1025 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0131] Bus 1030 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0132] The electronic device 1000 may also communicate with one or more external devices 1100 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1000, and / or may communicate with any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 1060. As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0133] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0134] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible implementations, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary embodiments of the present invention described in the above "Exemplary Method" section of the present specification.

[0135] A program product for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0136] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0137] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0138] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0139] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0140] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0141] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0142] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0143] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.

Claims

1. A method for identifying object attributes, characterized in that: include: Get the image to be processed; Preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around the target object; Inputting the information to be identified into a first object attribute recognition model to obtain a first object attribute label; The first object attribute recognition model includes: A relationship graph, wherein the relationship graph is generated by the actual distance between the target object and multiple surrounding objects, the target object image and multiple object images surrounding the target object, wherein each object is regarded as a node on the relationship graph, some nodes are marked with object attribute labels, and the link labels between the target object node and multiple object nodes surrounding the target object have weights; A graph label propagation calculation module, used to receive the weights and the object attribute labels of the partial node labels as input, and output a preliminary attribute feature vector through a graph label propagation algorithm; A graph attention mechanism processing module, configured to receive the preliminary attribute feature vector input, output a final attribute feature vector through a graph attention mechanism algorithm, and generate a first object attribute label; The first object attribute label includes: a hardness attribute label, a volume attribute label and a shape attribute label of the object.

2. The object attribute recognition method according to claim 1, characterized in that: The identification method further comprises: Preprocessing the image to be processed to form a sample to be trained, wherein the sample to be trained includes a target object image, multiple object images around the target object, and an actual distance between the target object and multiple surrounding objects; Inputting the sample to be trained into a first preset model to obtain a first initial object attribute label; According to the comparison between the target object standard attribute label and the first initial object attribute label, the first preset model parameters are modified until the training condition of the first preset model meets the preset conditions, thereby obtaining the trained first object attribute recognition model.

3. The object attribute recognition method according to claim 1, characterized in that: The identification method further comprises: Preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object and images of multiple objects around the target object; The information to be identified is input into a second object attribute identification model to obtain a second object attribute label.

4. The object attribute recognition method according to claim 3, characterized in that: The second object attribute recognition model includes: A gray level co-occurrence matrix calculation module is used to receive the target object image and multiple object image inputs around the target object, and output ten features of the gray level co-occurrence matrix of each object through a gray level co-occurrence matrix algorithm, the ten features including: mean and standard deviation of energy, mean and standard deviation of entropy, mean and standard deviation of contrast, mean and standard deviation of inverse moment, and mean and standard deviation of correlation; The support vector machine processing module is used to receive the ten feature inputs, output a final object feature vector through the support vector machine algorithm, and generate a second object attribute label.

5. The object attribute recognition method according to claim 4, characterized in that: The second object attribute label includes: a smooth attribute label and a rough attribute label of the object.

6. A device for identifying object attributes, characterized in that: include: An image acquisition module, used for acquiring an image to be processed; An image processing module, used for preprocessing the image to be processed to form information to be identified, wherein the information to be identified includes an image of a target object, images of multiple objects around the target object, and actual distances between the target object and multiple objects around it; An attribute recognition module, used for inputting the information to be recognized into a first object attribute recognition model to obtain a first object attribute label; The first object attribute recognition model includes: A relationship graph, wherein the relationship graph is generated by the actual distance between the target object and multiple surrounding objects, the target object image and multiple object images surrounding the target object, wherein each object is regarded as a node on the relationship graph, some nodes are marked with object attribute labels, and the link labels between the target object node and multiple object nodes surrounding the target object have weights; A graph label propagation calculation module, used to receive the weights and the object attribute labels of the partial node labels as input, and output a preliminary attribute feature vector through a graph label propagation algorithm; A graph attention mechanism processing module, configured to receive the preliminary attribute feature vector input, output a final attribute feature vector through a graph attention mechanism algorithm, and generate a first object attribute label; The first object attribute label includes: a hardness attribute label, a volume attribute label and a shape attribute label of the object.

7. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the object attribute recognition method according to any one of claims 1 to 5 by executing the executable instructions.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the object attribute recognition method described in any one of claims 1 to 5 is implemented.

Images