CN111125545A

CN111125545A - Target object determination method and device and electronic equipment

Info

Publication number: CN111125545A
Application number: CN201911362297.5A
Authority: CN
Inventors: 程皓
Original assignee: Qingdao Guangshi Technology Co Ltd; Beijing Kuangshi Technology Co Ltd
Current assignee: Beijing Kuangshi Technology Co Ltd
Priority date: 2019-12-24
Filing date: 2019-12-24
Publication date: 2020-05-08
Also published as: CN113343013A

Abstract

The invention provides a method, a device and electronic equipment for determining a target object, wherein the method is applied to a retrieval server, the retrieval server stores at least one archive of an existing object, the archive comprises object information of the existing object to which the archive belongs, each archive corresponds to at least one object tag, and the object tags are obtained according to object information statistics, and the method comprises the following steps: acquiring target object information of a target object; calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object; and determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object. The method and the device can reduce the limitation in the process of determining the target object and improve the retrieval efficiency of the target object.

Description

Target object determination method and device and electronic equipment

Technical Field

The present invention relates to the field of internet technologies, and in particular, to a method and an apparatus for determining a target object, and an electronic device.

Background

Object files are now built. The existing method for searching object archives includes an accurate searching method such as image searching or attribute searching, and a user inputs a target image, that is, returns an object with a high similarity to the target image, or inputs an accurate target attribute, that is, returns an object satisfying the attribute. However, when the user cannot acquire the target image, the image retrieval cannot be performed; and if the attribute acquired by the user has deviation, the target object cannot be retrieved through the attribute. For example, a user wants to search for a target object wearing a red jacket and black trousers, but the user is in a weak color, the jacket color of the target object is actually brown, and the user cannot obtain an accurate search result by using the attribute because the object of the brown jacket is filtered by the condition of the red jacket.

Disclosure of Invention

In view of this, the present invention provides a method, an apparatus and an electronic device for determining a target object, so as to reduce limitations in the process of determining the target object and improve retrieval efficiency of the target object.

In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:

in a first aspect, an embodiment of the present invention provides a method for determining a target object, which is applied to a search server, where the search server stores a profile of at least one existing object, where the profile includes object information of the existing object to which the profile belongs, each profile corresponds to at least one object tag, and the object tag is obtained by statistics according to the object information, where the method includes: acquiring target object information of a target object; calculating the fitting degree of the existing object to the target object based on the target object information and the stored object label of the existing object; and determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object.

In one embodiment, the target object information and at least one of the object tags each have at least one type, each type includes at least one subclass, the target object information of a certain subclass is target object sub-information corresponding to the subclass, the object tag of a certain subclass is an object tag corresponding to the subclass, and the target object sub-information and the object tag corresponding to the same subclass have a corresponding relationship; calculating a degree of fit of the existing object to the target object based on the target object information and an object tag of the existing object, comprising: calculating the matching degree of the first target object sub-information and the first object label corresponding to the same subclass to obtain the matching degree corresponding to the subclass, thereby obtaining the matching degree corresponding to each subclass; and calculating the fitting degree of the existing object to the target object according to the matching degree corresponding to each subclass.

In one embodiment, at least one of the object tags is a qualitative tag or a quantitative tag, wherein the qualitative tag is a tag that only contains the attribute and has a unique attribute, and the quantitative tag is a tag whose tag value contains the attribute and a ratio of the attribute.

In one embodiment, the step of calculating the matching degree between the first target object sub-information and the first object tag corresponding to the same sub-class includes: when the first object tag is a qualitative tag, if the attribute contained in the first target object sub information is matched with the attribute contained in the first object tag, determining that the matching degree of the first target object sub information and the first object tag is a preset first matching value; otherwise, determining the matching degree of the first target object sub information and the first object label as a preset second matching value; and/or when the first object tag is a quantitative tag, if the proportion of the attributes contained in the first object tag is 100% and the attributes contained in the first target object sub-information are matched with the attributes contained in the first object tag, determining that the matching degree of the first target object sub-information and the first object tag is a preset first matching value; if the attribute contained in the first object tag does not contain the attribute contained in the first target object sub information, determining the matching degree of the first target object sub information and the first object tag as the preset second matching value; and if the attribute contained in the first object tag contains the attribute contained in the first target object sub information and the proportion of the attribute is less than 100%, determining that the matching degree of the first target object sub information and the first object tag is a third matching value, wherein the third matching value is calculated according to the attribute contained in the first target object sub information and the attribute contained in the first object tag and the proportion of the attribute, and the third matching value is greater than the second matching value and less than the first matching value.

In one embodiment, the step of calculating the matching degree between the first target object sub-information and the first object tag corresponding to the same sub-class includes: and calculating the matching degree of the first target object sub-information and the first object label according to the proximity degree between the first attribute contained in the first target object sub-information and the second attribute contained in the first object label, wherein the proximity degree between every two attributes is preset.

In one embodiment, the target object information includes target object essential information, and the calculating the degree of fitting of the existing object to the target object based on the target object information and the object tag of the existing object includes: screening a first existing object from the existing objects, wherein the first existing object corresponds to an object label matched with the necessary information of the target object; calculating a fitness of the first existing object to the target object based on the target object information and an object label of the first existing object; the step of determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object comprises the following steps: and determining a first existing object corresponding to the target object based on the fitting degree of the first existing object to the target object.

In one embodiment, the step of calculating the degree of fitting of the existing object to the target object according to the matching degree corresponding to each of the subclasses includes: and carrying out weighted summation on the matching degrees corresponding to the subclasses based on the preset weight corresponding to each subclass to obtain the fitting degree of the existing object to the target object.

In one embodiment, the step of determining the existing object corresponding to the target object based on the degree of fitting of the existing object to the target object includes: and determining the existing objects with the degree of fitting to the target object larger than a set threshold value and/or determining the first N existing objects with the maximum degree of fitting to the target object as the existing objects corresponding to the target object.

In one embodiment, the object information includes at least one of an object image, photographing information of the object image, and spatiotemporal information of an existing object; the method further comprises the following steps: detecting an existing object and/or an accessory of the existing object in an object image contained in the archive through target detection, and performing attribute identification on the existing object and/or the accessory of the existing object in the object image to obtain the attribute of the existing object and/or the accessory of the existing object corresponding to the object image; counting attributes of existing objects and/or accessories of the existing objects corresponding to the object images to obtain object attribute counting results; and obtaining at least one type of object tag in a basic attribute type, a wearing attribute type and an accessory attribute type according to the object attribute statistical result corresponding to the object image.

In one embodiment, the method further comprises: counting the shooting information and/or the time-space information contained in the archives to obtain a behavior attribute counting result; and obtaining an object label of the behavior attribute type according to the behavior attribute statistical result.

In one embodiment, the method further comprises: acquiring the information of the same row of the file according to the shooting information contained in the file; the peer information comprises images and shooting information of the images, wherein the images are within a preset time threshold range of shooting time interval of the object images and within a preset proximity threshold range of shooting place proximity; carrying out statistical analysis on the peer information to obtain a social attribute statistical result; and obtaining an object tag of the social attribute type according to the social attribute statistical result.

In one embodiment, the type of the object tag of the existing object includes at least one of a basic attribute type, a wearing attribute type, a behavior attribute type, a social attribute type, and an accessory type of the object to which the profile belongs.

In a second aspect, an embodiment of the present invention further provides an apparatus for determining a target object, which is applied to a search server, where the search server stores a profile corresponding to at least one existing object, where the profile includes object information of the existing object to which the profile belongs, each profile corresponds to at least one object tag, and the object tag is obtained by statistics according to the object information, and the apparatus includes: the information acquisition module is used for acquiring target object information of a target object; the fitting degree calculation module is used for calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object; and the object determining module is used for determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object.

In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory; the memory has stored thereon a computer program which, when executed by the processor, performs the method of any one of the aspects as provided in the first aspect.

In a fourth aspect, an embodiment of the present invention provides a computer storage medium for storing computer software instructions for use in any one of the methods provided in the first aspect.

The embodiment of the invention provides a method, a device and electronic equipment for determining a target object, which are applied to a retrieval server, wherein the retrieval server stores archives corresponding to at least one existing object, each archive comprises object information of the existing object to which the archive belongs, each archive corresponds to at least one object label, and the object labels are obtained through statistics according to the object information. Compared with the prior art in which the target object is accurately searched according to the accurate information, the method disclosed by the embodiment of the invention better realizes fuzzy attribute search on the target object, and effectively avoids missing detection and false detection caused by inaccurate attribute information.

Additional features and advantages of embodiments of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of embodiments of the invention as set forth above.

In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

Fig. 1 is a schematic structural diagram of an electronic device according to an embodiment of the present invention;

fig. 2 is a flowchart illustrating a method for determining a target object according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of an object tag provided by an embodiment of the present invention;

fig. 4 is a flowchart illustrating another method for determining a target object according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram illustrating a target object determining apparatus according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of another target object determination apparatus according to an embodiment of the present invention.

Detailed Description

To make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some, not all, embodiments of the present invention.

In view of the fact that the method for determining the existing object corresponding to the target object by accurately searching and determining the existing object in the prior art has great limitations and the problem that the existing object is difficult to determine, embodiments of the present invention provide a method, an apparatus, and an electronic device for determining the target object.

Example one

Referring first to fig. 1, a schematic structural diagram of an electronic device 100 for implementing the method and apparatus for determining a target object according to the embodiment of the present invention is shown, where the electronic device 100 includes one or more processors 102, one or more storage devices 104, an input device 106, an output device 108, and an image capturing device 110, and these components are interconnected through a bus system 112 and/or other types of connection mechanisms (not shown). It should be noted that the components and structure of the electronic device 100 shown in fig. 1 are only exemplary and not limiting, and the electronic device may have some of the components shown in fig. 1 and may also have other components and structures not shown in fig. 1, as desired.

The processor 102 may be implemented in at least one hardware form of a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), the processor 102 may be one or a combination of several of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), or other forms of processing units having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 100 to perform desired functions.

The storage 104 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. On which one or more computer program instructions may be stored that may be executed by processor 102 to implement client-side functionality (implemented by the processor) and/or other desired functionality in embodiments of the invention described below. Various applications and various data, such as various data used and/or generated by the applications, may also be stored in the computer-readable storage medium.

The input device 106 may be a device used by a user to input instructions and may include one or more of a keyboard, a mouse, a microphone, a touch screen, and the like.

The output device 108 may output various information (e.g., images or sounds) to the outside (e.g., a user), and may include one or more of a display, a speaker, and the like.

The image capture device 110 may take images (e.g., photographs, videos, etc.) desired by the user and store the taken images in the storage device 104 for use by other components.

Exemplary electronic devices for implementing the method and apparatus for determining a target object according to the embodiments of the present invention may be implemented as smart terminals such as smart phones, tablet computers, and the like.

Example two

Referring to fig. 2, a schematic flow chart of a method for determining a target object is shown, where the method is applied to a search server, the search server stores archives of existing objects, the archives include object information of the existing objects to which the archives belong, each archive corresponds to at least one object tag, and the object tags are obtained according to the object information. The archive of each object can be obtained by conventional archiving algorithms, for example, clustering archive by distance between features in the image of the object, classifying objects whose movement trajectories match into the same archive, and classifying information associated to the same object by prior information into the same archive. Theoretically, one existing object corresponds to one archive, and object information contained in one archive belongs to the same existing object. However, due to the accuracy of the archive algorithm, there may be cases where object information of a plurality of existing objects is contained in one archive or where object information of one existing object is distributed in a plurality of archives. The existing object may include, for example, a person, an animal, or a vehicle, the object tag may describe a plurality of attributes of the existing object, taking the existing object is a person as an example, the object tag may be used to describe one or more of a basic attribute, a wearing attribute, a behavior attribute, a social attribute, and an accessory attribute of the person object, and by adding a plurality of object tags to the existing object, the attributes of the existing object may be more detailed and comprehensive, so that the existing object more similar to the target object may be determined, and the method for determining the target object provided in the embodiment of the present invention mainly includes the following steps S202 to S206:

in step S202, target object information of the target object is acquired.

The target object information may be an attribute of a currently grasped target object, for example, the obtained target object information includes a general attribute such as a time and place where the target object frequently appears or a partial social relationship of the existing target object, or the obtained target object information includes an accurate attribute such as the number of times the target object appears in the place a. Specifically, an upload channel of the target object information can be provided for the user. In one embodiment, descriptive characters uploaded by a user and aiming at a target object can be received, and semantic analysis is carried out on the uploaded characters to obtain target object information of the target object. In another embodiment, one or more selectable attribute options may be provided for the user, and the attributes selected by the user based on the attribute options are obtained, so as to obtain the target object information of the target object. For example, if the target object is a person, attribute options of the types of clothing color option, age zone, and the like may be provided; if the target object is an animal, attribute options such as species options, hair color options and the like can be provided; if the target object is a vehicle, attribute options for the type of vehicle, vehicle color, etc. may be provided. In another embodiment, an image or video uploaded by a user may be acquired, and target object information of a target object may be obtained by identifying an attribute of the target object in the image or video.

And step S204, calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object.

In one embodiment, the degree of fitting of each existing object stored in the retrieval server to the target object may be calculated, for example, for each existing object, a to-be-compared tag corresponding to the target object information may be determined from object tags of the existing object, and the degree of similarity between the to-be-compared tag and the target object information is determined, so that the degree of fitting between the existing object and the target object is determined based on the degree of similarity. In another embodiment, the degrees of fitting of the existing objects in the search server to the target object may be sequentially calculated, and if the calculated degrees of fitting are greater than or equal to a preset threshold, the calculation of the degrees of fitting of the remaining existing objects in the search server to the target object is stopped, for example, if it is determined that the degree of fitting of a certain existing object to the target object is 100%, the degrees of fitting of the remaining existing objects in the search server to the target object are not calculated.

Step S206, the existing object corresponding to the target object is determined based on the fitting degree of the existing object to the target object.

In practical application, in order to quickly determine the existing object corresponding to the target object, the existing object with the highest degree of fitting may be determined as the existing object corresponding to the target object, a fitting threshold may be preset, the existing object with the degree of fitting higher than the fitting threshold may be determined as the existing object corresponding to the target object, in addition, the existing objects may be sorted according to the degree of fitting, and the top N existing objects in the obtained sorting result may be determined as the existing objects corresponding to the target object.

The method for determining the target object, provided by the embodiment of the invention, includes the steps of firstly obtaining target object information of the target object, then calculating the fitting degree of the existing object and the target object based on the target object information and the stored object label of the existing object, and determining the existing object corresponding to the target object based on the fitting degree. Compared with the prior art in which the target object is accurately searched according to the accurate information, the method disclosed by the embodiment of the invention better realizes fuzzy attribute search on the target object, and effectively avoids missing detection and false detection caused by inaccurate attribute information.

In order to determine an existing object corresponding to a target object through fuzzy search, an embodiment of the present invention may generate an object tag of the existing object in advance according to a profile, where the object tag is obtained according to object information, where the object information includes at least one of an object image, shooting information of the object image, and spatiotemporal information of the existing object, and the type of the object tag includes at least one of an existing basic attribute type, a wearing attribute type, a behavior attribute type, a social attribute type, and an accessory type. The narrow object image may be an image including an existing object, and the wide object image may be an image including only an associated object of the existing object (for example, an accessory of the existing object). For example, a vehicle may be in the possession of a human object, and an image of the vehicle (which may contain only the vehicle and not the owner of the vehicle) may be placed in the vehicle owner's profile. The shooting information of the object image refers to shooting information such as shooting time and shooting location (commonly-used shooting equipment representation) of the object image, and the space-time information of the existing object refers to space-time information of the object which is obtained by signal acquisition means except for the image such as International Mobile Subscriber Identity (IMSI), Radio Frequency (RF) and the like. By setting multiple types of object labels, more dimensions can be depicted on the existing object, the retrievable attributes of the existing object are more diverse, and the fitting degree of the existing object to the target object can be evaluated in a finer granularity. The embodiment of the invention provides a specific implementation mode for generating an object label of an object according to object information, which comprises the following steps:

(1) obtaining at least one type of object label in the basic attribute type, the wearing attribute type and the accessory type of the existing object according to the following steps 1.1 to 1.3.

Step 1.1, detecting an existing object in an object image contained in the archive through object detection, and performing attribute identification on the existing object and/or an accessory of the existing object in the object image to obtain an attribute of the existing object and/or the accessory of the existing object corresponding to the object image.

For example, the existing object is a human object, and the accessory is a car/non-motor vehicle. And performing face/pedestrian/motor vehicle/non-motor vehicle/license plate detection on each object image contained in the file, and performing attribute identification on the detected face/pedestrian/motor vehicle/non-motor vehicle/license plate to obtain basic attributes, wearing attributes and accessory attributes which are in one-to-one correspondence with the object images. The basic attributes can comprise the subclasses of gender, age, ethnicity and the like of the existing object, and the wearing attributes can comprise the subclasses of clothing style and clothing color of the existing object, whether the style color of a backpack and a backpack is adopted, whether a hat and a hat are adopted, whether beards and beard type colors are left and the like; the accessory attributes may include a number of sub-categories of the vehicle, the license plate number of the non-vehicle, the brand, the model, the color, the pendant, the driving information of the vehicle, the non-vehicle (e.g., whether the existing object is the driver or the passenger, if the existing object is the driver/the passenger, who/what attributes/which image the corresponding passenger/driver is/has), etc. related to the existing object. It is understood that a carry-on or a backpack carried by an existing object may be used as an accessory of the existing object to generate an object tag of an accessory attribute type, and may also be used as a wearing item of the existing object to generate an object tag of a wearing attribute type.

And 1.2, counting the attributes of the existing object and/or the accessory of the existing object corresponding to the object image to obtain an attribute counting result.

In a specific implementation manner, for the attribute of each seed class, the attributes corresponding to a plurality of object images are counted, and the attribute with the largest occurrence frequency is used as the attribute statistical result of the attribute of the seed class, or each attribute and the proportion occupied by the attribute are used as the attribute statistical result of the attribute of the seed class; or, the attribute with the largest occurrence frequency in the attributes meeting the condition is used as the attribute statistical result of the attribute of the seed class, or the attributes meeting the condition and the proportion thereof are used as the attribute statistical result of the attribute of the seed class. Thereby obtaining the attribute statistics of each subclass attribute. Wherein, the proportion of an attribute is the proportion of the number of times of the attribute appearing in all the attributes of the seed. An object image is associated with the attribute, and the attribute is considered to appear 1 time.

In one example, the file has 10 target images in total, from which gender attributes can be detected. Regarding the attributes of the subclass under the basic attribute type of gender attribute, the gender attributes corresponding to 10 target images obtained in step 1.1 are male, female, and female, respectively, and the "female" with the largest number of occurrences can be taken as the statistical result of the gender attributes, or 90% of the males and females can be taken as the statistical result of the gender attributes (the attributes corresponding to the subclass are male and female, and the proportion of the attributes is 10% and 90%, respectively). In another example, the file contains 10 target images in total, from which the color attribute of the jacket can be detected. For the attributes of the subclasses under the wearing attribute type of the jacket color attribute, the jacket color attributes corresponding to 10 target images obtained in step 1.1 are red, yellow, green, and green, respectively, the jacket color that appears more than 3 times can be determined as the jacket color that satisfies the condition (regular wearing), and the yellow and green are the jacket colors that satisfy the condition, and then the attribute with the largest number of occurrences among the attributes (yellow and green) that satisfy the condition is used as the attribute statistical result of the subclass attribute of the regular wearing jacket color, or the attributes (yellow and green) that satisfy the condition and the ratio thereof (yellow 60% green 30%, or yellow 66.7% green 33.3%) are used as the attribute statistical result of the subclass attribute of the regular wearing jacket color. In one example, the file has 10 object images (for example, vehicle mount images) that can detect an existing object and an accessory such as a vehicle at the same time. In step 1.1, whether the existing object is a driver or a passenger can be determined based on the relative position between the existing object and the vehicle, the vehicle information can be determined based on the license plate, so that the riding attributes corresponding to the 10 object images obtained in step 1.1 are respectively riding a vehicle, riding B vehicle, riding C vehicle, riding D vehicle and riding D vehicle, and "riding a vehicle 20%, riding a vehicle 10%, riding C vehicle 40% and riding D vehicle 30%" are taken as the attribute statistics of the subclass attributes of the riding vehicles.

The vehicle attribute and the riding information of the human figure object can be obtained by counting a plurality of images, and for example, the human figure object takes the vehicle a 9 times and takes the vehicle B1 time. Vehicle C was driven 9 times and vehicle D was driven 1 time.

In another specific implementation manner, for the attributes of some subclasses, statistics is performed on the attributes corresponding to the object image in combination with the time of attribute occurrence to obtain an attribute statistical result. The time when the attribute appears refers to the shooting time of the image corresponding to the attribute.

In one example, the file has 10 target images in total, in which the color attribute of the jacket can be detected, and the target image that was captured most recently is an a image. If the upper garment color attribute corresponding to the a image is green, which is the attribute of the sub-category of the wearing attribute type, which is the upper garment color attribute, obtained in step 1.1, the green is used as the attribute statistical result of the sub-category attribute of the color of the upper garment worn recently.

In another specific implementation manner, for the attributes of some subclasses, the attributes corresponding to the object image are counted in combination with the prior information to obtain an attribute counting result. The time when the attribute appears refers to the shooting time of the object image corresponding to the attribute.

In one example, the existing object corresponding to the file is M, and a vehicle with a license plate number of jing M00000 under its name can be obtained. The archive has 10 object images capable of detecting the attribute of the vehicle pendant corresponding to the license plate, the 10 object images have pendants of a puppy, a doll, a robot cat and a puppy, and the statistical result of the attribute of the vehicle pendant corresponding to the license plate can be ' 30% of the puppy, 60% of the robot cat and ' 10% of the puppy, the robot cat and the puppy '.

And 1.3, obtaining at least one type of object tag in the basic attribute type, the wearing attribute type and the accessory type according to the attribute statistical result corresponding to the object image. In one embodiment, the attribute statistics of each subclass can be used as the object label of the subclass. The object tags of the basic attribute types can comprise object tags of subclasses such as gender, age, ethnicity and the like of the existing object, and the wearing attributes can comprise subclasses such as recent clothing style and clothing color, frequent clothing style and clothing color, whether backpack and backpack style colors, whether hat and hat styles are worn, whether beard and beard types are left, and the like of the existing object; the accessory attributes may include a number of sub-categories of the existing object recently/often/ever driven or under the name of the existing object, the license plate number of the non-motor vehicle, the brand, the model, the color, the pendant, driving information of the motor vehicle, the non-motor vehicle (e.g., whether the existing object is the driver or the passenger, if the existing object is the driver/the passenger, who/what attributes the corresponding passenger/driver is/what images correspond), etc. It is understood that a carry-on or a backpack carried by an existing object may be used as an accessory of the existing object to generate an object tag of an accessory attribute type, and may also be used as a wearing item of the existing object to generate an object tag of a wearing attribute type.

(2) The step of generating an object tag of the behavior attribute type of the object according to the object information is performed according to the following steps 2.1 to 2.2:

and 2.1, counting the shooting information and/or the space-time information contained in the file to obtain a behavior attribute counting result.

The photographing information of the object image includes time/place information of photographing of the object image, which is spatio-temporal information acquired from the object image. The space-time information of the existing object is acquired by signal acquisition means except for images, such as IMSI, RF and the like, and can be effectively supplemented, and the space-time information acquired by the object can be used together to more accurately depict the space-time information of the existing object. The behavior attribute may include two subclasses, a spatio-temporal attribute and a behavior analysis attribute.

The spatio-temporal attribute subclasses may include one or more subclasses of period activity, site activity, and number of days active. The time period activity is the statistics of the activity times of the existing object in each hour in the activity times of the whole day, the place activity is the statistics of the times of the existing object appearing in each place, and the activity days are the days in which the existing object is captured or the signals of the existing object are acquired. The number of times of activities of the existing object and the number of times of occurrence in a certain place are obtained by counting shooting information and/or time-space information contained in the archive. For example, an existing object is photographed once, or its signal is collected once, it is considered to be active once, an existing object is photographed once at a certain location, or its signal is collected once at a certain location, it is considered to be present once at a certain location. The time when the existing object is shot or the signal is acquired is the activity time, and the place where the signal is shot or acquired is the activity place.

The behavior analysis attribute can be obtained by analyzing the temporal and spatial attributes, and comprises one or more of diurnal night appearance, frequent occurrence and disappearance of a certain area, personnel gathering of a certain area and abnormal residence of a certain area. If the ratio of the night activity times to the day activity times of the existing object is higher than a first preset threshold value, the existing object has a day-night emergence attribute; if the frequency of the activity of the existing object in a certain area is higher than a second preset threshold, the existing object has the attribute of frequent emergence in the area; if the existing object is captured in a certain area or signals of the existing object are acquired in the certain area, and the number of the objects capturing/acquiring the signals in the area is statistically displayed to be higher than a third preset threshold, the existing object has the personnel gathering attribute in the area; if the length of the existing object in the certain area is longer than the fourth preset threshold, the existing object has the attribute of abnormal residence in the certain area.

And 2.2, obtaining an object label of the behavior attribute type according to the behavior attribute statistical result. In one embodiment, the behavior attribute statistics may be used as object tags for the behavior attribute types. For example, the object tags of the behavior attribute type include a sub-tag of "8 o 'clock-9 o' clock 50%, 18 o 'clock-19 o' clock 50%" for activity in the time period, and "A area appears and disappears frequently" for sub-tag.

(3) The step of generating an object tag of the social attribute type of the existing object from the object information may be performed according to the following steps 3.1 to 3.3:

and 3.1, acquiring the information of the same row of the existing object according to the shooting information contained in the file. The information of the same line comprises images and shooting information of the images, wherein the images and the shooting information of the images are within a preset time threshold range of the shooting time interval of the images of the object and within a preset proximity threshold range of the proximity of the shooting place.

And 3.2, carrying out statistical analysis on the information of the same row to obtain a social attribute statistical result.

The object tags of the social attribute type may include at least one sub-type object tag of friend names, friend numbers, and friend scores corresponding to the friend names of existing objects, where the friend scores may represent the closeness degree of the existing objects and the friends, and a higher friend score indicates a higher closeness degree of the existing objects and the friends.

In one embodiment, the peer information includes a peer object and a peer score with the object. Acquiring the peer information of the existing object according to the shooting information contained in the archive comprises acquiring the peer object of the existing object and the peer integral of the object according to the shooting information contained in the archive. For example, if the object image contains the existing object and the object a at the same time, the existing object and the object a are recorded in the same line once; if the image within the preset time threshold range of the shooting time interval with the object image and within the preset proximity threshold range of the shooting place proximity is obtained according to the shooting time and the shooting place of the object image, and the image comprises the object A, the existing object and the object A are recorded to be in the same row. Once per row, the row-wise integral is incremented by 1. Carrying out statistical analysis on the peer information to obtain a social attribute statistical result, wherein the statistical result comprises the following steps: and when the collinear integral between the existing object and the object A is larger than a preset integral threshold value, determining that the existing object and the object A are friends, and taking the collinear integral of the existing object and the object A as friend integral.

And 3.3, obtaining the object label of the social attribute type according to the social attribute statistical result. In one embodiment, the social attribute statistics may be treated as object tags of the social attribute type.

In order to facilitate understanding of the object labels generated in the above process, an embodiment of the present invention provides a schematic diagram of object labels, and as shown in fig. 3, the object labels of multiple subclasses obtained through the above steps can depict the existing object in more dimensions, so that the retrievable attributes of the existing object are more diverse, and the fitting degree of the existing object to the target object can be evaluated in a finer granularity.

On the basis of the object tag, the embodiment of the invention can better retrieve the existing object corresponding to the target object in a fuzzy search mode according to the acquired target object information, wherein the type of the target object information comprises at least one of a basic attribute type, a wearing attribute type, a behavior attribute type, a social attribute type and an accessory type of the target object. In a specific implementation, the object tag contains at least one type, such as the basic attribute type, the wearing attribute type, the accessory attribute type, the behavior attribute type, and the social attribute type described above, each type containing at least one subclass, for example, the basic attribute type may include subclasses of age, gender, and ethnicity, the wearing attribute type may include subclasses of clothing color and clothing type, the accessory attribute type may include subclasses of vehicle type, vehicle color, and the like, and the behavior attribute type may include subclasses such as time activity, place activity, active days, night and day, night, frequent presence in a certain area, people in a certain area, and abnormal residency in a certain area. It will be appreciated that if an object tag of a type has only one subclass, its type may be identical to the subclass. The type/subclass of the target object information and the type/subclass of the object tag are at least partially consistent, so that the matching degree between the attributes corresponding to the same type/subclass of target object information and the object tag can be used for measuring the fitting degree of the existing object to the target object.

On the basis of obtaining the object labels of the respective objects through the above-mentioned means, for each existing object, the embodiment of the present invention provides a specific implementation manner for calculating the degree of fitting between the existing object and the target object based on the target object information and the object label of the existing object, see the following steps a to b:

step a, calculating the matching degree of the first target object sub information and the first object label corresponding to the same subclass, and obtaining the matching degree corresponding to the subclass, thereby obtaining the matching degree corresponding to each subclass. The matching degree is used for representing the similarity degree between the first target object sub information and the first object label corresponding to the same subclass, and the higher the matching degree between the first target object sub information and the first object label is, the higher the similarity degree between the first target object sub information and the first object label is. In practical applications, the target object information of a certain subclass may be used as the target object sub information corresponding to the subclass, the object label of the certain subclass may be used as the first object label corresponding to the subclass, and the target object sub information corresponding to the same subclass and the first object label may have a corresponding relationship. In one embodiment, the user may grasp partial information of the target object, and the partial information grasped by the user is taken as target object information, and the target object information includes at least one target object sub-information. For example, if the user grasps the target object sub-information of some subclasses of the target object, and if there are also object tags of existing objects, the subclasses shared by the two are set as shared subclasses, the target object sub-information and the object tag corresponding to a subclass in the shared subclasses are sequentially set as the first target object sub-information and the first object tag having a correspondence relationship, and the matching degree between the two is calculated as the matching degree of the subclasses corresponding to the two, thereby obtaining the matching degree corresponding to each subclass in the shared subclasses. In one example, the user only grasps the target object sub-information corresponding to the sub-category of friend names in the social property type and the target object sub-information corresponding to the sub-category of location liveness in the behavioral property type of the target object, if the object tag of an existing object also has a subclass of friend names in a social property type and location liveness in a behavioral property type, and calculating the matching degree of the first target object sub-information and the first object label so as to obtain the matching degree corresponding to each subclass in the shared subclass.

In addition, the object label provided by the embodiment of the invention can be a qualitative label or a quantitative label. The qualitative label is a label that only contains attributes and that the contained attributes are unique, for example, an object label of a gender subclass in a certain basic attribute type is "woman", the object label only includes a unique attribute that the gender of a known object is known to be woman, that is, the object label is a qualitative label; the quantitative label is a label whose label value includes attributes and the proportion of the attributes, for example, the object label of gender subclass in a certain basic attribute type is "woman (attribute): 90% (attribute ratio); male (attribute): 10% (ratio of attribute) ", the object label is a quantitative label. In practical applications, the object label of the known object can be set to be a qualitative label or a quantitative label based on actual conditions. Usually, the quantitative label is obtained by statistics according to the attributes of a plurality of object images in the archive, and the specific statistical method is as described above and will not be described herein again.

In a specific embodiment, the matching degree between the first target object sub-information and the first object tag corresponding to the same sub-class may be calculated according to the following steps a1 to a 2:

a1, when the first object label is a qualitative label, if the attribute contained in the first object sub information matches with the attribute contained in the first object label, determining the matching degree of the first object sub information and the first object label as a preset first matching value; otherwise, determining the matching degree of the first target object sub information and the first object label as a preset second matching value. The attribute contained in the first target object sub information is matched with the attribute contained in the first object tag, which can also be understood as that the attribute contained in the first target object sub information is completely the same as the attribute contained in the first object tag. In practical applications, a first matching value and a second matching value may be preset, for example, the first matching value is set to 1, the second matching value is set to 0, if it is known that the attribute contained in the first target object sub-information of the target object is the normal wearing color "yellow" and the attribute contained in the first object label is also the normal wearing color "yellow", it is determined that the first target object sub-information corresponding to the subclass of the normal wearing color completely matches the first object label, and a large matching degree corresponding to the subclass of the normal wearing color is returned to be the first matching value 1; and if the attribute contained in the first object label is the constant-wearing color 'black', determining that the first target object sub-information corresponding to the subclass of the constant-wearing color is not matched with the first object label, and returning that the large matching degree corresponding to the subclass of the constant-wearing color is a second matching value of 0.

In step a2, when the first object label is a quantitative label, the matching degree between the first object sub-information and the first object label corresponding to the same subclass can be calculated according to the following three cases. It should be noted that although the first object label is a quantitative label, the corresponding target object sub-information is usually qualitative, for example, the target object sub-information corresponding to the sub-category of the color of the wearing apparel is usually "yellow", rather than "yellow 40% green 60%".

The first condition is as follows: and if the proportion of the attributes contained in the first object tag is 100% and the attributes contained in the first target object sub information are matched with the attributes contained in the first object tag, determining the matching degree of the first target object sub information and the first object tag as a preset first matching value. For example, if the first object tag of the gender subclass includes that the proportion of the known object is 100% and the attribute included in the first object sub-information indicates that the object is a woman, it is determined that the first object sub-information corresponding to the gender subclass matches the first object tag, and the matching degree corresponding to the gender subclass is determined to be a first matching value.

Case two: and if the attributes contained in the first object tag do not contain the attributes contained in the first target object sub information, determining the matching degree of the first target object sub information and the first object tag as a preset second matching value. And if the attribute contained in the first object label is that the known object is a male object, and the attribute contained in the second first target object sub-information is that the target object is a female object, namely the attribute contained in the first object label corresponding to the gender subclass does not contain the attribute contained in the first target object sub-information, determining the matching degree corresponding to the gender subclass as a second matching value.

Case three: and if the attribute contained in the first object tag contains the attribute contained in the first target object sub information and the proportion of the attribute is less than 100%, determining that the matching degree of the first target object sub information and the first object tag is a third matching value. The third matching value is calculated according to the attribute contained in the first target object sub information and the attribute and the proportion of the attribute contained in the first object tag, the third matching value is greater than the second matching value and smaller than the first matching value, and if the second matching value is 0 and the first matching value is 1, the third matching value may be a value between 0 and 1 calculated according to the attribute contained in the first target object sub information and the attribute and the proportion of the attribute contained in the first object tag. For example, if the attribute included in the first target object sub-information is that the target object always carries a red handbag, and if the probability that the attribute included in the first object tag of the handbag color subclass includes that the carried handbag of the existing object is red (attribute) is 70% (the ratio of the attributes), and the probability that the carried handbag color is the other is 30%, it can be determined that the third matching value is 1 × 70% +0 × 30% — 0.7, that is, the matching degree of the first target object sub-information corresponding to the handbag color subclass and the first object tag is 0.7. Of course, the way of calculating the third matching value according to the attribute contained in the first target object sub information and the attribute and the ratio of the attribute contained in the first object tag is not limited to this, and is not limited herein.

It is understood that the steps a1 and a2 may be executed separately or in any order, and this embodiment is not limited.

Considering that the attributes contained in the target object information may not be conveniently and accurately described, for example, the color of clothing or vehicle is very specific and is easily affected by light, and the attributes contained in the object tags of the color subclasses do not completely include the attributes contained in the target object sub-information, such as the clothing color in the first target sub-information is brick red or purple, and the clothing color in the first object tag includes only red, black, white and blue, in this case, if the matching degree of the first target object sub-information and the first object tag corresponding to the same subclass is calculated according to the above-mentioned steps a1 to a2, the obtained matching value is 0. However, actually, brick red or purple in the first target sub information and red in the first object tag are relatively close to each other and have a certain matching degree, so that another implementation for calculating the matching degree between the first target sub information and the first object tag corresponding to the same subclass is provided in the embodiment of the present invention, and the matching degree between the first target sub information and the first object tag may be calculated according to the proximity degree between the first attribute included in the first target sub information and the second attribute included in the first object tag. Wherein the proximity between two attributes is preset. Taking color as an example, the proximity of complementary colors on a newton color ring may be set to a small value, the proximity of neighboring colors to a large value, and the proximity of the same color to a maximum value. For example, the proximity between red and green is set to a small value (e.g., 0), the proximity between red and purple is set to a large value (e.g., 0.8), and the proximity between red and red is set to a maximum value (e.g., 1). If the attributes of the first object label corresponding to the color subclass of the frequent-wear jacket are 50% purple and 50% green, and the attribute of the first object sub-information is red, the matching degree of the first object sub-information and the first object label is 0.8 × 50% +0 × 50% — 0.4.

And b, calculating the fitting degree of the existing object to the target object according to the matching degree corresponding to each subclass. If only one subclass is included in the target object information, the matching degree between the first object label corresponding to the subclass and the first target object sub information can be directly determined as the fitting degree of the existing object to the target object. If the target object information comprises a plurality of subclasses, weighting and summing the matching degrees corresponding to the subclasses based on the preset weights corresponding to the subclasses to obtain the fitting degree of the existing object to the target object, for example, if the preset weight of the friend number subclass in the social property is 0.4 and the preset weight of the place liveness subclass in the behavior property is 0.2, calculating a first product of the preset weight of the friend number subclass and the matching degree and a second product of the preset weight of the place liveness subclass and the matching degree respectively, and taking the sum of the first product and the second product as the fitting degree of the existing object to the target object.

In another embodiment, the following steps (1) to (3) may be performed to obtain the fitting degree of the existing object to the target object. (1) Calculating the matching degree of the first object label corresponding to each existing object in the same subclass and the sub information of the first target object; (2) sorting each existing object according to the matching degree corresponding to each subclass respectively to obtain a sorting value of each existing object based on the subclass; (3) and calculating the fitting degree of the existing object to the target object according to the corresponding ranking values of the subclasses. For example, the target object information includes 2 subclasses, the friends whose target object sub-information is the target object include friend a, friend B, friend C and friend D, and the target object sub-information of the second subclass is that the target object often appears in area S in month 10. Firstly, calculating the matching degree of a first object label corresponding to each existing object in a first subclass (friend name subclass in a social attribute type) and a second subclass (a region frequently appearing and disappearing subclass in a behavior attribute type) and first target object sub information; then, according to the matching degree corresponding to the first subclass of each existing object, sequencing each existing object to obtain a first sequencing value P1 of each existing object, wherein the first sequencing value of the existing object which is sequenced at the front is higher than the first sequencing value of the existing object which is sequenced at the back; sorting the existing objects according to the matching degree corresponding to the second subclass of the existing objects to obtain a second sorting value P2 of the existing objects, wherein the second sorting value of the existing object which is sorted in the front is higher than the second sorting value of the existing object which is sorted in the back; and finally, weighting and summing the P1 and the P2 of each existing object to obtain the fitting degree of each existing object to the target object. Wherein, the same or different weights may be configured for each sub-class when performing the weighted summation.

In another embodiment, the following steps (a) to (b) may be performed to obtain the fitting degree of the existing object to the target object.

And (I) screening a first existing object from the existing objects, wherein the first existing object corresponds to an object label matched with necessary information of the target object. The first existing object corresponds to an object tag matched with the necessary information of the target object, the target object information comprises the necessary information of the target object, and the necessary information of the target object can be understood as the attribute which a user determines the target object to have, namely the attribute which the existing object must accord with. For example, if the target object is determined to be a male, the "male" may be determined as the necessary information of the target object in the target object information, and the first existing object whose object tag includes "male 100%" (when the object tag of the subclass of the existing object is a quantitative tag) or whose object tag includes "male" (when the object tag of the subclass of the existing object is a qualitative tag) is screened out from the existing object, and the fitting degree of the existing object to the target object is not calculated for other existing objects (such as "female 100%" or "male 80%", and female 20% "), so that the user may more flexibly and accurately configure the screening condition of the target object, and at the same time, reduce the amount of calculation required for calculating the fitting degree to some extent, thereby improving the efficiency of determining the existing object corresponding to the target object.

And (II) calculating the fitting degree of the first existing object to the target object based on the target object information and the object label of the first existing object. After the first existing object is determined, the method and the device can only calculate the fitting degree of the first existing object to the target object so as to reduce the calculation amount required for calculating the fitting degree, in practical applications, the fitting degree of the first existing object to the target object can be calculated by referring to the steps of calculating the fitting degree of the existing object to the target object provided in the foregoing embodiments, for example, when the object label of the first existing object is a qualitative label, the degree of fitting of the first existing object to the target object is calculated with reference to the aforementioned step a1, when the object label of the first existing object is a quantitative label, the degree of fitting of the first existing object to the target object is calculated by referring to the step a2, or the degree of closeness between the attribute of the object label of each subclass of the first existing object and the attribute of the target object information may be calculated to calculate the degree of fitting of the first existing object to the target object, which is not described herein again in detail.

After the degree of fitting of the object to the target is determined according to the above steps, the existing object corresponding to the target object may be determined based on the degree of fitting of the existing object to the target object, and the embodiment of the present invention provides the following manner for determining the existing object corresponding to the target object:

the proximity degree determines the existing objects with the degree of fitting to the target object larger than a set threshold value and/or the first N existing objects with the maximum degree of fitting to the target object as the existing objects corresponding to the target object. In one embodiment, it may be determined that the target object corresponds to an existing object when the degree of fit of the existing object to the target object is greater than a set threshold. For example, when the degree of fit of the existing object to the target object is greater than 90%, the existing object is determined as the existing object corresponding to the target object. In another embodiment, the top N existing objects with the highest fitting degree to the target object may also be determined as the existing objects corresponding to the determination target object. For example, a plurality of existing objects with high degree of fitting are all determined as existing objects corresponding to the target object, and in order to facilitate a user to obtain existing objects of the target object from the determined existing objects, embodiments of the present invention provide a specific implementation manner for determining existing objects corresponding to the target object based on the degree of fitting between the target object and each existing object, which may be sorted according to the degree of fitting between the target object and each existing object, and then select a preset number of existing objects as existing objects corresponding to the target object starting from the existing object with the highest degree of fitting. For example, each existing object is sorted in an order of a degree of fitting from large to small, and the existing object of top100 in the sorting is determined as the existing object corresponding to the target object.

After the existing object corresponding to the target object is determined in the above manner, the existing object corresponding to the target object may also be displayed, for example, the determined existing object is displayed for the user through a preset page, and/or the existing object corresponding to the target object is transmitted to the designated device, where the designated device may be office equipment of the user, such as a computer or a mobile phone, so that the user can browse the existing object corresponding to the target object and further delete the wrong existing object therefrom, thereby determining the existing object corresponding to the manually screened target object.

In summary, the method for determining a target object provided in the embodiments of the present invention fully utilizes the object picture stored in the archive, the shooting information of the object image, and the spatio-temporal data of the existing object to generate the multidimensional object tag of each existing object, and can implement fuzzy search on the existing object corresponding to the target object by setting the target object information of the target object. In addition, the embodiment of the invention changes from the accurate retrieval depending on a single picture or limited attribute information into the fuzzy retrieval using mastered fuzzy and inaccurate generalized information, effectively utilizes the archive, and the finally determined existing object corresponding to the target object can be applied to various application scenes such as multi-dimensional study and judgment, data fusion and the like, thereby being convenient for quickly and conveniently retrieving the existing object corresponding to the target object.

EXAMPLE III

On the basis of the foregoing embodiments, the present invention provides a specific example of a method for determining a target object, where the target object is a human object, and the method may include the following steps S402 to S412, with reference to a flowchart of another method for determining a target object shown in fig. 4:

step S402, acquiring a portrait file. The portrait file includes a target image of a person target, a collection time (i.e., the shooting time) and a collection location (i.e., the shooting location) of the target image, and spatiotemporal information of the target.

In step S404, portrait label types (i.e., the aforementioned object labels) of the character object in multiple dimensions are determined based on preset character portrait rules. The portrait tag type may include one or more of a basic attribute type, a wearing attribute type (which may also be referred to as a clothing type), a behavior attribute type (which may include a behavior rule and a system analysis tag), a social attribute type (which may also be referred to as a friend relationship), and an accessory attribute type (which includes a driving vehicle), where the behavior rule is the aforementioned temporal-spatial attribute, and the system analysis tag is the aforementioned behavior analysis attribute. In practical application, a character portrait rule can be established in advance, and portrait label types of character objects are divided into portrait labels with six dimensions, namely a basic attribute type, a clothing type, a behavior rule type, a system analysis label type, a friend relationship type and a driving vehicle type according to the character portrait rule.

In step S406, object detection and attribute recognition are performed on each object image, and attributes such as sex, age, nationality, clothing style, clothing color, vehicle type, whether the person object is a driver or a passenger, and the like of the person object in each object image are obtained. The attributes in each object image are counted according to a predetermined portrait label type, and character portrait information (that is, the object label) of each character object is generated. Specifically, reference may be made to the process of generating the object tag according to the object information provided in the foregoing embodiment, which is not described herein again in the embodiments of the present invention.

In step S408, image information of a target object (i.e., the target object information) is acquired. For example, the friends of the target object include friend a, friend B, friend C, and friend D, and the target object often appears in region S1 for month 10.

In step S410, a fitting degree calculation is performed based on the image information of the target object and the object label of each character object. For example, the character objects are sorted according to the friend relationship subclass and the behavior rule subclass to obtain a sorting value P1 corresponding to the friend relationship of each character object and a sorting value P2 corresponding to the behavior rule, and the sorting value P1 and the sorting value P2 are weighted to obtain the fitting degree P of each character object to the target object.

In step S412, a portrait archive corresponding to the target object is determined based on the degree of fitting. In practical application, the portrait files can be sorted in the descending order of the fitting degree, and the portrait file of top100 in the sorting result is returned.

The method for determining the target object, provided by the embodiment of the invention, comprises the steps of obtaining a portrait file, performing attribute identification on an object image in the portrait file, then determining image tag types of multiple dimensions of the character object based on a preset character image rule, performing statistics on the portrait file according to the image tag types to generate an object tag of each character object, performing fitting degree calculation on the portrait file according to the image information of the target object and the object tag of each character object after obtaining the image information of the target object, and further determining the character object corresponding to the target object based on the fitting degree. By the aid of the method, fuzzy search of the character objects is well achieved, limitation in the process of accurately searching the character objects can be effectively reduced, difficulty in determining the character objects corresponding to the target objects can be effectively reduced, and accordingly retrieval efficiency of the character objects is improved.

Example four

As to the method for determining a target object provided in the second embodiment, an embodiment of the present invention provides a device for determining a target object, where the device is applied to a search server, the search server stores at least one archive of an existing object, the archive includes object information of the existing object to which the archive belongs, each archive corresponds to at least one object tag, and the object tags are obtained by statistics according to the object information, referring to a schematic structural diagram of a device for determining a person object shown in fig. 5, the device includes the following modules:

an information obtaining module 502 is configured to obtain target object information of a target.

And a fitting degree calculation module 504, configured to calculate a fitting degree of the existing object to the target object based on the target object information and the object tag of the existing object.

And an object determining module 506, configured to determine an existing human object corresponding to the target object based on a degree of fitting of the existing human object to the target object and each human object.

The device for determining the target object provided by the embodiment of the invention utilizes the pre-stored object tag corresponding to the existing object and the target object information of the target object to calculate the fitting degree between the existing object and the target object, and determines the existing object of the target object based on the fitting degree.

In one embodiment, the target object information and the at least one object tag each have at least one type, each type includes at least one subclass, the target object information of a certain subclass is target object sub-information corresponding to the subclass, the object tag of a certain subclass is an object tag corresponding to the subclass, and the target object sub-information and the object tag corresponding to the same subclass have a corresponding relationship; the fitness calculating module 504 is further configured to: calculating the matching degree of the first target object sub-information and the first object label corresponding to the same subclass to obtain the matching degree corresponding to the subclass, thereby obtaining the matching degree corresponding to each subclass; and calculating the fitting degree of the existing object to the target object according to the matching degree corresponding to each subclass.

In one embodiment, at least one object tag is a qualitative tag or a quantitative tag, wherein the qualitative tag is a tag that only contains the attribute and has a unique attribute, and the quantitative tag is a tag whose tag value contains the attribute and a ratio of the attribute.

In one embodiment, the fitting degree calculation module 504 is further configured to: when the first object label is a qualitative label, if the attribute contained in the first object sub information is matched with the attribute contained in the first object label, determining the matching degree of the first object sub information and the first object label to be a preset first matching value; otherwise, determining the matching degree of the first target object sub information and the first object label as a preset second matching value; and/or when the first object label is a quantitative label, if the proportion of the attributes contained in the first object label is 100% and the attributes contained in the first target object sub-information are matched with the attributes contained in the first object label, determining the matching degree of the first target object sub-information and the first object label to be a preset first matching value; if the attributes contained in the first object tag do not contain the attributes contained in the first target object sub information, determining the matching degree of the first target object sub information and the first object tag as a preset second matching value; and if the attributes contained in the first object tag contain the attributes contained in the first target object sub information and the proportion of the attributes is less than 100%, determining that the matching degree of the first target object sub information and the first object tag is a third matching value, wherein the third matching value is calculated according to the attributes contained in the first target object sub information and the attributes and proportion of the attributes contained in the first object tag, and the third matching value is greater than the second matching value and less than the first matching value.

In one embodiment, the fitting degree calculation module 504 is further configured to: and calculating the matching degree of the first target object sub-information and the first object label according to the proximity degree between the first attribute contained in the first target object sub-information and the second attribute contained in the first object label, wherein the proximity degree between every two attributes is preset.

In one embodiment, the target object information includes target object essential information; the fitness calculating module 504 is further configured to: screening a first existing object from the existing objects, wherein the first existing object corresponds to an object label matched with necessary information of a target object; calculating the fitting degree of the first existing object to the target object based on the target object information and the object label of the first existing object; the object determination module 506 is further configured to: and determining a first existing object corresponding to the target object based on the fitting degree of the first existing object to the target object.

In one embodiment, the fitting degree calculation module 504 is further configured to: and carrying out weighted summation on the matching degrees corresponding to the subclasses based on the preset weight corresponding to each subclass to obtain the fitting degree of the existing object and the target object.

In one embodiment, the object determining module 506 is further configured to determine existing objects having a degree of fitting to the target object greater than a set threshold and/or the top N existing objects having the greatest degree of fitting to the target object as existing objects corresponding to the determined target object.

In one embodiment, the object information includes at least one of an object image, shooting information of the object image, and spatiotemporal information of an existing object; on the basis of fig. 5, an embodiment of the present invention provides another apparatus for determining a target object, referring to a schematic structural diagram of another apparatus for determining a target object shown in fig. 6, the apparatus may further include a tag generation module 602, configured to detect an existing object in an object image included in the archive through object detection, and perform attribute identification on the existing object in the object image and/or an accessory of the existing object, so as to obtain an attribute of the existing object and/or the accessory of the existing object corresponding to the object image; counting attributes of an existing object and/or an accessory of the existing object corresponding to the object image to obtain an object attribute counting result; and obtaining at least one type of object tag in the basic attribute type, the wearing attribute type and the accessory attribute type according to the object attribute statistical result corresponding to the object image.

In an embodiment, the tag generating module 602 is further configured to perform statistics on shooting information and/or temporal-spatial information included in the archive to obtain a behavior attribute statistical result; and obtaining the object label of the behavior attribute type according to the behavior attribute statistical result.

In an embodiment, the tag generating module 602 is further configured to: acquiring the information of the same row of the file according to the shooting information contained in the file; the peer information comprises images and shooting information of the images, wherein the images and the shooting information of the images are within a preset time threshold range of shooting time interval with the object image and within a preset proximity threshold range of shooting place proximity; carrying out statistical analysis on the information of the same row to obtain a social attribute statistical result; and obtaining the object label of the social attribute type according to the social attribute statistical result.

In one embodiment, the type of the object tag includes at least one of a basic attribute type, a wearing attribute type, a behavior attribute type, a social attribute type, and an accessory type of the object to which the profile belongs.

The device provided by the embodiment has the same implementation principle and technical effect as the foregoing embodiment, and for the sake of brief description, reference may be made to the corresponding contents in the foregoing method embodiment for the portion of the embodiment of the device that is not mentioned.

EXAMPLE five

The method and apparatus for determining a target object and the computer program product of the electronic device provided in the embodiments of the present invention include a computer-readable storage medium storing a program code, where instructions included in the program code may be used to execute the method described in the foregoing method embodiments, and specific implementation may refer to the method embodiments, and will not be described herein again.

In addition, in the description of the embodiments of the present invention, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present invention, which are used for illustrating the technical solutions of the present invention and not for limiting the same, and the protection scope of the present invention is not limited thereto, although the present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.

Claims

1. A method for determining a target object is applied to a retrieval server, the retrieval server stores at least one archive of existing objects, the archive includes object information of the existing objects to which the archive belongs, each archive corresponds to at least one object tag, and the object tags are obtained through statistics according to the object information, the method includes:

acquiring target object information of a target object;

calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object;

and determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object.

2. The method of claim 1, wherein the object information and at least one of the object labels each have at least one type, each type includes at least one subclass, the object information of a certain subclass is object sub-information corresponding to the subclass, the object label of a certain subclass is an object label corresponding to the subclass, and the object sub-information and the object label corresponding to the same subclass have a corresponding relationship;

calculating a degree of fit of the existing object to the target object based on the target object information and an object tag of the existing object, comprising:

calculating the matching degree of the first target object sub-information and the first object label corresponding to the same subclass to obtain the matching degree corresponding to the subclass, thereby obtaining the matching degree corresponding to each subclass;

and calculating the fitting degree of the existing object to the target object according to the matching degree corresponding to each subclass.

3. The method of claim 2, wherein at least one of the object tags is a qualitative tag or a quantitative tag, wherein the qualitative tag is a tag that only contains the attribute and the contained attribute is unique, and the quantitative tag is a tag whose tag value contains the attribute and a ratio of the attribute.

4. The method of claim 3, wherein the step of calculating the matching degree between the sub-information of the first target object and the label of the first object corresponding to the same sub-class comprises:

when the first object tag is a qualitative tag,

if the attribute contained in the first target object sub information is matched with the attribute contained in the first object tag, determining the matching degree of the first target object sub information and the first object tag as a preset first matching value;

otherwise, determining the matching degree of the first target object sub information and the first object label as a preset second matching value;

and/or the presence of a gas in the gas,

when the first object label is a quantitative label,

if the proportion of the attributes contained in the first object tag is 100% and the attributes contained in the first target object sub information are matched with the attributes contained in the first object tag, determining the matching degree of the first target object sub information and the first object tag as a preset first matching value;

if the attribute contained in the first object tag does not contain the attribute contained in the first target object sub information, determining the matching degree of the first target object sub information and the first object tag as the preset second matching value;

and if the attribute contained in the first object tag contains the attribute contained in the first target object sub information and the proportion of the attribute is less than 100%, determining that the matching degree of the first target object sub information and the first object tag is a third matching value, wherein the third matching value is calculated according to the attribute contained in the first target object sub information and the attribute contained in the first object tag and the proportion of the attribute, and the third matching value is greater than the second matching value and less than the first matching value.

5. The method according to claim 2 or 3, wherein the step of calculating the matching degree between the sub-information of the first target object and the label of the first object corresponding to the same sub-class comprises:

and calculating the matching degree of the first target object sub-information and the first object label according to the proximity degree between the first attribute contained in the first target object sub-information and the second attribute contained in the first object label, wherein the proximity degree between every two attributes is preset.

6. The method according to any one of claims 1 to 5, wherein the target object information includes target object necessary information,

the calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object comprises:

screening a first existing object from the existing objects, wherein the first existing object corresponds to an object label matched with the necessary information of the target object;

calculating a fitness of the first existing object to the target object based on the target object information and an object label of the first existing object;

the step of determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object comprises the following steps:

and determining a first existing object corresponding to the target object based on the fitting degree of the first existing object to the target object.

7. The method according to any one of claims 1 to 6, wherein the step of calculating the degree of fitting of the existing object to the target object according to the matching degree corresponding to each subclass comprises:

and carrying out weighted summation on the matching degrees corresponding to the subclasses based on the preset weight corresponding to each subclass to obtain the fitting degree of the existing object to the target object.

8. The method according to any one of claims 1-7, wherein the step of determining the existing object corresponding to the target object based on the fitness of the existing object to the target object comprises:

and determining the existing objects with the degree of fitting to the target object larger than a set threshold value and/or determining the first N existing objects with the maximum degree of fitting to the target object as the existing objects corresponding to the target object.

9. The method according to claim 1, wherein the object information includes at least one of an object image, photographing information of the object image, spatiotemporal information of an existing object;

the method further comprises the following steps:

detecting an existing object and/or an accessory of the existing object in an object image contained in the archive through target detection, and performing attribute identification on the existing object and/or the accessory of the existing object in the object image to obtain the attribute of the existing object and/or the accessory of the existing object corresponding to the object image;

counting attributes of existing objects and/or accessories of the existing objects corresponding to the object images to obtain attribute counting results;

and obtaining at least one type of object tag in the basic attribute type, the wearing attribute type and the accessory attribute type according to the attribute statistical result corresponding to the object image.

10. The method of claim 9, further comprising:

counting the shooting information and/or the time-space information contained in the archives to obtain a behavior attribute counting result;

and obtaining an object label of the behavior attribute type according to the behavior attribute statistical result.

11. The method according to claim 9 or 10, characterized in that the method further comprises:

acquiring the information of the same row of the file according to the shooting information contained in the file; the peer information comprises images and shooting information of the images, wherein the images and the shooting information of the images are within a preset time threshold range of a shooting time interval of the object image and within a preset proximity threshold range of a shooting place proximity;

carrying out statistical analysis on the peer information to obtain a social attribute statistical result;

and obtaining an object tag of the social attribute type according to the social attribute statistical result.

12. The method of claim 2, wherein the type of the object tag comprises at least one of a base attribute type, a wear attribute type, a behavior attribute type, a social attribute type, and an accessory type.

13. An apparatus for determining a target object, applied to a search server, wherein the search server stores a profile corresponding to at least one existing object, the profile includes object information of the existing object to which the profile belongs, each profile corresponds to at least one object tag, and the object tag is obtained statistically according to the object information, the apparatus includes:

the information acquisition module is used for acquiring target object information of a target object;

the fitting degree calculation module is used for calculating the fitting degree of the existing object to the target object based on the target object information and the object label of the existing object;

and the object determining module is used for determining the existing object corresponding to the target object based on the fitting degree of the existing object to the target object.

14. An electronic device comprising a processor and a memory;

the memory has stored thereon a computer program which, when executed by the processor, performs the method of any of claims 1 to 12.

15. A computer storage medium storing computer software instructions for use in the method of any one of claims 1 to 12.