WO2019051813A1 - 一种目标识别方法、装置和智能终端 - Google Patents

一种目标识别方法、装置和智能终端 Download PDF

Info

Publication number
WO2019051813A1
WO2019051813A1 PCT/CN2017/101966 CN2017101966W WO2019051813A1 WO 2019051813 A1 WO2019051813 A1 WO 2019051813A1 CN 2017101966 W CN2017101966 W CN 2017101966W WO 2019051813 A1 WO2019051813 A1 WO 2019051813A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
recognition result
tested
recognition
target
Prior art date
Application number
PCT/CN2017/101966
Other languages
English (en)
French (fr)
Inventor
廉士国
刘兆祥
王宁
Original Assignee
达闼科技（北京）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 达闼科技（北京）有限公司 filed Critical 达闼科技（北京）有限公司
Priority to JP2020514952A priority Critical patent/JP7104779B2/ja
Priority to CN201780002585.2A priority patent/CN108064389B/zh
Priority to PCT/CN2017/101966 priority patent/WO2019051813A1/zh
Publication of WO2019051813A1 publication Critical patent/WO2019051813A1/zh
Priority to US16/818,837 priority patent/US11036990B2/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/96—Management of image or video recognition tasks
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J9/00—Programme-controlled manipulators
    • B25J9/16—Programme controls
    • B25J9/1694—Programme controls characterised by use of sensors other than normal servo-feedback from position, speed or acceleration sensors, perception control, multi-sensor controlled systems, sensor fusion
    • B25J9/1697—Vision controlled systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24317—Piecewise classification, i.e. whereby each classification requires several discriminant rules
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/809—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of classification results, e.g. where the classifiers operate on the same input data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/61—Control of cameras or camera modules based on recognised objects
    • H04N23/611—Control of cameras or camera modules based on recognised objects where the recognised objects include parts of the human body

Definitions

  • the embodiments of the present invention relate to the field of intelligent identification technologies, and in particular, to a target identification method, apparatus, and intelligent terminal.
  • the smart terminal may take a long time to collect more information before it can be analyzed in more detail. Identify the results. In this case, if the intelligent terminal is required to output a higher-level recognition result, it may take a long time to meet the timeliness of the target recognition; if the intelligent terminal is specified in order to meet the timeliness of the target recognition If the recognition result must be output in one time, it is possible to obtain only a low-level recognition result, which is also not conducive to the user's friendly experience.
  • the embodiment of the present application provides a target recognition method, device, and intelligent terminal, which can solve the problem of how to achieve a compromise between the timeliness and the level of detail of the target recognition.
  • the information about the object to be tested collected in the preset time period is used as the judgment information, and the object to be tested includes at least two types of attributes, and the priority relationship is set between the at least two types of attributes;
  • the judgment information is used as the judgment information, and the data obtained based on the judgment information and before the judgment information is collected is returned, and the Describe the result of the recognition of the current time of the target, and output the result of the recognition.
  • the embodiment of the present application provides a target identification device, which is applied to an intelligent terminal, and includes:
  • the information collecting unit is configured to use the information about the target to be measured collected in the preset time period as the judgment information, where the object to be tested includes at least two attribute types, and the at least two attribute types are set with a priority relationship;
  • the identification unit includes: an identification module and an output module, the identification module is configured to acquire a recognition result of the current time of the object to be tested based on the determination information and data obtained before collecting the determination information, the output module And configured to output the recognition result, where the recognition result corresponds to one of the attribute types;
  • a determining unit configured to determine whether an attribute type corresponding to the identification result is an attribute type having the highest priority among the at least two attribute types
  • an intelligent terminal including:
  • At least one processor and,
  • the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the target recognition method as described above.
  • an embodiment of the present application provides a non-transitory computer readable storage medium, where the non-transitory computer readable storage medium stores computer executable instructions for causing a smart terminal to execute the above The target recognition method.
  • the target identification method, apparatus, and intelligent terminal provided by the embodiments of the present application divide multiple attributes with priority order for the attributes of the target to be tested according to the degree of detail of the description of the object to be tested.
  • Type and, in the process of identification, the information collected for the target to be measured in the preset time period is used as the judgment information, and based on the judgment information and the data obtained before the judgment information is collected, the information is acquired and outputted.
  • the recognition result of the target at the current time is measured, and if the priority of the attribute type corresponding to the recognition result at the current time is not the highest level, that is, if the recognition result obtained at the current time is not the most detailed recognition result, then at the next preset time
  • the segment continues to collect information for the object to be tested, and the information is used as the judgment information, and the above-mentioned identification steps and determination steps are repeated, and the recognition result of the object to be measured can be output in time under different recognition scenarios, and at the same time, The most detailed recognition results are obtained, and the information collection time is tired. Gradually based on a richer more detailed information output recognition results, thus, be able to reach a compromise between the level of detail and timeliness of target recognition, enhance the user experience.
  • FIG. 1 is a schematic flowchart diagram of a target recognition method according to an embodiment of the present application
  • FIG. 2 is a schematic flow chart of another object recognition method provided by an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of a target recognition apparatus according to an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of hardware of an intelligent terminal according to an embodiment of the present application.
  • the embodiment of the present application provides a target recognition method, device, and intelligent terminal, which can be applied to any application field related to target recognition, such as: intelligent guide blind, welcome robot, service robot, intrusion object detection, semantic recognition, etc. It is especially suitable for application areas such as intelligent guide blind, welcome robot, service robot and other human-computer interaction experience.
  • the object recognition method provided by the embodiment of the present application is capable of outputting the recognition result based on the information collected within the preset time period in time, and determining the “priority” of the attribute type corresponding to the recognition result obtained at the current time. Whether to continue the information collection to further optimize the detailed level of the target identification intelligent optimization identification method, by assigning a plurality of priority order attribute types to the attributes of the target to be tested according to the degree of detail of the description of the object to be tested (where The higher the priority attribute type, the higher the level of detail of the recognition result, and in the process of identification, the information about the target to be measured collected in the preset time period is used as the judgment information, and based on the judgment information.
  • the smart terminal is within the first preset time period (ratio For example, within the first 5s, that is, within 5 seconds after triggering the smart terminal to start collecting information, a clear face image can be collected, and the "person name" of the tested person can be identified based on the clear face image.
  • the smart terminal can output the “person name” of the tested person to interact with the user; meanwhile, since the most detailed recognition result has been obtained at this time, the information collected for the measured person can be stopped.
  • the detected person side faces the camera of the intelligent terminal, and the intelligent terminal can only collect.
  • the smart terminal outputs the first recognition result: the "gender” of the tested person, so as to facilitate In time, the "sex" is not the most detailed recognition result. Therefore, in the embodiment of the present application, the smart terminal continues to be targeted for the next preset time period (for example, within the second 5s).
  • the person collecting information if the smart terminal can collect the positive face image of the measured person in the second preset time period, the side face image collected in the first preset time period and the second pre-preparation can be combined Set the positive face image collected during the time period, identify the “person name” of the tested person, and then output the second recognition result: the “person name” of the tested person, thereby obtaining more detailed recognition results, which is convenient for the intelligent terminal to further Tune User interaction with the content, and enhance the user experience.
  • the object recognition method and apparatus provided by the embodiments of the present application can be applied to any type of smart terminal, such as a robot, a guide glasses, a smart helmet, a smart phone, a tablet computer, a server, and the like.
  • the smart terminal can include any suitable type of storage medium for storing data, such as a magnetic disk, a compact disc (CD-ROM), a read-only memory or a random access memory.
  • the smart terminal may also include one or more logical computing modules that perform any suitable type of function or operation in parallel, such as viewing a database, image processing, etc., in a single thread or multiple threads.
  • the logic operation module may be any suitable type of electronic circuit or chip-type electronic device capable of performing logical operation operations, such as a single core processor, a multi-core processor, a graphics processing unit (GPU), or the like.
  • FIG. 1 is a schematic flowchart of a target identification method according to an embodiment of the present disclosure, which may be performed by any type of smart terminal. Specifically, refer to FIG. 1 , where the method includes but is not limited to:
  • Step 110 The information about the object to be tested collected in the preset time period is used as the judgment information.
  • the “target to be tested” may include, but is not limited to, a person, an animal, an object, and the like. According to the degree of detail of the description of the object to be measured, at least two different layers can be divided for the object to be tested.
  • the attribute type of the level, and the priority level of these attribute types is set according to the level of detail of the description of the object to be tested.
  • the attribute type with difficulty in identifying is relatively high in detail, and the difficulty level of recognition can be sorted according to the recognition rate of different attribute types under the same conditions (for example, inputting the same picture) (for example)
  • person name recognition is difficult to identify by gender, gender recognition is difficult for face/human recognition); or, it can be sorted according to the mutual inclusion relationship between attribute types (for example, to identify the gender, the presence of the face must be recognized first).
  • the attribute type of the person may be set according to the degree of detail of the description of the object to be tested, including: “person name”, “gender”, and “whether it is a person”, and according to the difficulty level of identification, Set the priority order of these attribute types to: L1 (person name) > L2 (gender) > L3 (whether it is human).
  • the smart terminal continuously collects information for the target to be tested, and outputs the recognition result of the target to be tested at a plurality of preset time nodes, and then the “preset time period” is one of the target recognition targets.
  • the “information” collected by the smart terminal is a judgment basis that can reflect the attribute of the target to be tested, and based on the information, the attribute of the target to be tested can be identified.
  • the type of the information may include, but is not limited to, image information, sound information, thermal infrared images, near-infrared images, ultrasonic signals, electromagnetic reflection signals, etc., which may be acquired by one or more sensors, for example, by camera acquisition.
  • image information of the target is measured, the sound information for the target to be tested is collected by the microphone, the thermal infrared image for the target to be tested is collected by the thermal infrared sensor, and the like.
  • the smart terminal collects the back image a1 of the measured person during the time period of (0, t1), and collects the side face image a2 of the measured person during the time period of (t1, t2), at (t2,
  • the positive face image a3 of the person to be tested is acquired during the time period of T), where 0 ⁇ t1 ⁇ t2 ⁇ T.
  • the “information for the target to be measured collected in the preset time period” described in this step 110 may be the back image a1 collected by the smart terminal during the time period of (0, t1), or may be
  • the side face image a2 acquired during the period of (t1, t2) may also be the face image a3 acquired during the period of (t2, T).
  • the smart terminal when the object to be measured is identified, the smart terminal may be configured to continuously collect information for the object to be tested according to actual conditions, and perform target recognition at a specific time node. Specifically, when the target identification is performed at a certain time node, the information collected by the smart terminal for the object to be tested in the time period between the current time node and the previous time node may be used as the judgment information, and the smart terminal is based on the determination information. The following step 120 is performed to obtain the recognition result of the target to be tested at the time node.
  • the smart terminal may be configured to receive the information acquisition command as the time start node (ie, time 0), and perform target recognition at time t1, t2, and T.
  • the information collected for the object to be tested in the time period is used as the judgment information, and the following step 120 is performed based on the judgment information to obtain the recognition result of the target to be tested at time t1; at time t2, the time period is (t1, t2)
  • the information collected for the target to be tested is used as the judgment information, and the following step 120 is performed based on the judgment information to obtain the recognition result of the target to be tested at time t2; at time T, the target is to be measured in the (t2, T) time period.
  • the collected information is used as the judgment information, and the following step 120 is performed based on the judgment information to obtain the recognition result of the target to be tested at time T. It can be understood that, in practical applications, the duration between any two time nodes may be equal (that is, the target identification is performed periodically), or may be unequal, and the embodiment of the present application does not specifically limited.
  • Step 120 Acquire a recognition result of the current time of the object to be tested based on the judgment information and data obtained before the judgment information is collected, and output the recognition result.
  • the target information is determined based on the determination information acquired within the preset time period before the time node and the data obtained before the determination information is collected.
  • the “decision information” may be information collected by the smart terminal in any preset time period for acquiring the recognition result of the object to be tested at the current time.
  • the “current time” refers to a time at which the collection of the determination information is completed and the target recognition is performed based on the determination information.
  • the “data obtained before collecting the judgment information” may be information collected for the object to be tested before collecting the judgment information; or may be a recognition result acquired before the judgment information is collected (ie, The recognition result that has been obtained before the current time).
  • the collected judgment information is the side face image a2 collected by the smart terminal during the time period of (t1, t2)
  • the time t2 at which the acquisition of the side face image a2 is completed is the “current time”
  • the data obtained before the acquisition of the determination information that is, the data obtained before the time t1 (including the time t1)
  • the data may be the back image a1 collected during the (0, t1) time period, or may be the smart terminal.
  • the "recognition result” corresponds to one of the attribute types of the object to be tested.
  • the attribute types include: “person name”, “gender”, and “whether it is a person”; if the obtained recognition result is “Li Si”, the attribute type corresponding to the recognition result “Li Si” is “person name”; if the obtained recognition result is "male”, the attribute type corresponding to the recognition result "male” is “gender”; if the obtained recognition result is "person”, the recognition result "person” corresponds The attribute type is "whether it is a person”. And, in order to meet the requirements of the timeliness of target recognition, The most detailed recognition result is possible to be output.
  • the time node for performing target recognition is based on the collected judgment information and the data obtained before the judgment information is collected.
  • Obtaining the most detailed recognition result that is, the judgment result corresponding to the attribute type having the highest priority after performing the operation on the judgment information and the data
  • the judgment results “person", "male” and "Li Si” can be obtained.
  • the attribute type "person name” corresponding to "Li Si” has the highest priority, so it can be acquired and output at this time. Identify the result "Li Si”.
  • the specific implementation manner of obtaining the identification result of the current time of the target to be tested may include, but is not limited to, the following three implementation manners: based on the determination information and the data obtained before the determination information is collected.
  • the data obtained before the collecting the judgment information includes the information about the object to be tested collected before the judgment information is collected.
  • the feature fusion may be adopted. The recognition result of the current time of the target is measured.
  • the features for identifying the attributes of the object to be tested may be extracted from the judgment information collected in different time periods, and the features are merged, and then based on a suitable recognition algorithm, such as a neural network algorithm, based on The merged feature obtains the recognition result of the target to be measured at the current moment.
  • the data obtained before the obtaining the judgment information includes the recognition result acquired before the judgment information is collected.
  • the “result fusion” may be used to obtain the current time of the target to be tested. result.
  • the recognition result corresponding to the determination information is acquired; and then, the identification with the highest priority of the attribute type is selected from the recognition result corresponding to the determination information and the recognition result acquired before the collection of the determination information.
  • the result is the recognition result of the current time of the target to be tested.
  • the obtaining a recognition result corresponding to the determination information that is, obtaining a recognition result of the object to be tested based on the determination information.
  • the recognition result obtained before the collection of the judgment information includes: “person” and “male”, and the recognition result obtained based on the judgment information acquired at the current time is “male”, and may be from “person” or “male”.
  • “Male” is selected from the three recognition results of "male” (corresponding attribute class) The type has the highest priority) as the recognition result of the current time of the target to be tested.
  • the collected judgment information is information collected for the target to be tested in the (0, t1) time period. If there is no data before the judgment information is collected, at this time, the recognition result of the current time of the object to be tested can be obtained only based on the collected judgment information, that is, in the embodiment, the collected judgment information The corresponding recognition result is the recognition result of the current time of the target to be tested.
  • the judgment result corresponding to each attribute type of the target to be tested is set to one for A confidence level that characterizes the reliability (or credibility) of the result of the determination.
  • the recognition result obtained in the step 120 is a judgment result corresponding to one of the attribute types, the confidence of the determination result satisfies a preset condition, and the attribute type corresponding to the recognition result satisfies the confidence
  • the priority of the attribute type corresponding to the judgment result of the preset condition is the highest.
  • the confidence level of the judgment result can be determined by the similarity degree of the feature comparison, and the higher the degree of similarity, the higher the confidence degree.
  • the “preset condition” may be set according to an actual application scenario, and used to identify the reliability of a certain judgment result. Specifically, the preset condition may be: a confidence level of the determination result is greater than or equal to a confidence threshold corresponding to the corresponding attribute type.
  • the confidence threshold corresponding to each attribute type may be the same. For example, the confidence thresholds corresponding to the attribute types “person name”, “gender”, and “whether or not” are both 70%, and the judgment result of the target to be tested is obtained.
  • the recognition result of the object to be tested is the judgment result "Zhang San” corresponding to the attribute type "person name” having the highest priority among the three.
  • the confidence threshold corresponding to each attribute type may also be different. For example, the confidence threshold corresponding to the attribute type “person name” may be preset to be 75%, corresponding to the attribute type “gender”. The confidence threshold is 85%, and the confidence threshold corresponding to the attribute type “is it human” is 95%.
  • the recognition result of the target to be tested is The highest priority attribute type "gender” corresponds to the judgment result "male”.
  • the current time of the target to be measured is obtained based on the collected judgment information (or based on the collected judgment information and the information collected for the target to be measured before collecting the judgment information)
  • Specific implementation manners of the identification result may include, but are not limited to, the following two implementation manners:
  • each attribute type of the target to be tested may be acquired first based on the collected judgment information (or based on the collected judgment information and the information collected for the target to be measured before collecting the judgment information). Corresponding judgment result and confidence of each judgment result; and then outputting a judgment result corresponding to the attribute type having the highest priority among the judgment results satisfying the preset condition as the recognition result of the object to be tested.
  • each attribute type of the object to be tested is acquired.
  • the result of the judgment can be achieved by using a suitable algorithm (for example, a neural network).
  • the judgment information collected by the smart terminal is the image information of the person
  • the intelligence The terminal can iteratively calculate the judgment result corresponding to the attribute type "whether it is human", "gender” and "person name” from the image, for example, firstly, the feature for discriminating "whether is a person” is calculated by the bottom layer of the neural network.
  • the feature 1 the "whether or not” corresponding judgment result and the confidence of the judgment result are obtained; then, the feature 2 for discriminating "gender” is calculated based on the feature 1 in the middle layer of the neural network, and the “sex” is obtained according to the feature 2 The corresponding judgment result and the confidence of the judgment result; finally, the feature 3 for discriminating the "person name” is calculated based on the feature 2 at the uppermost layer of the neural network, and the judgment result corresponding to the "person name” is obtained according to the feature 3 and The confidence level of the judgment result.
  • the judgment result that the confidence degree satisfies the preset requirement is first selected, and then the judgment result with the highest level of detail (that is, the highest priority of the corresponding attribute type) is selected as the current target to be tested. The result of the identification of the moment.
  • the second implementation manner based on the collected judgment information (or based on the collected judgment information and the information collected for the object to be tested collected before the judgment information is collected), according to the priority from high to low
  • the judgment result corresponding to each attribute type of the object to be tested and the confidence of each judgment result are sequentially obtained step by step until the first confidence level satisfies the preset condition, and the first confidence level is satisfied to meet the preset.
  • the judgment result of the condition is used as the recognition result of the current time of the target to be tested.
  • the judgment information for the object to be tested is collected (or when the collected judgment information and the information for the object to be tested collected before the judgment information is acquired), first based on the collected information ( Or, based on the collected judgment information and the information collected for the object to be tested collected before collecting the judgment information, obtaining the first-level judgment result corresponding to the attribute type with the highest priority and the first-level judgment.
  • the first-level confidence of the result if the first-level confidence meets the preset condition (for example, the first-level confidence is greater than or equal to the first-level confidence threshold), the first-level judgment result is directly output as the recognition result of the current time of the target to be tested.
  • the second-level judgment result corresponding to the attribute type of the next level and the second-level confidence level of the second-level judgment result are obtained; if the second-level confidence degree satisfies a preset condition (for example, the second-level confidence level) If the value is greater than or equal to the second-level confidence threshold, the second-level judgment result is output as the recognition result of the current time of the target to be tested, otherwise, the judgment result corresponding to the attribute type of the next level is obtained based on the collected judgment information and Confidence, and so on, until the judgment result that the confidence level satisfies the preset condition is obtained.
  • a preset condition for example, the second-level confidence level
  • different characteristics may be extracted from the collected judgment information (or the collected judgment information and the information collected for the target to be measured before collecting the judgment information) for different levels.
  • the target to be measured is a vehicle and the collected information is image information for the vehicle
  • the feature a can be extracted from the image information to identify whether there is a car in the image
  • the feature b is extracted.
  • the feature c is extracted for identifying the type of car (car, truck, bus, etc.) and the like.
  • the judgment result corresponding to each attribute type of the object to be tested and the confidence thereof are obtained step by step according to the order of priority from high to low, when the first confidence level satisfies the judgment result of the preset condition.
  • the judgment result that the first confidence degree satisfies the preset condition is directly used as the recognition result of the current time of the target to be tested, and does not need to identify and judge each attribute type, thereby reducing the amount of data processing without affecting the recognition.
  • the collected judgment information may also include at least two types of information sources.
  • the “information source” refers to a source of information capable of reflecting an attribute of a target to be tested.
  • the “at least two information sources” may be at least two different types of judgment information, such as any two or more of image information, sound information, thermal infrared images, near infrared images, ultrasonic signals, or electromagnetic reflection signals.
  • the "at least two information sources” may also be some type of information collected from at least two angles, for example, image information (or sound information) of the target to be measured is collected from multiple angles, each Image information (or sound information) acquired from the angle of view can be used as a source of information.
  • the “at least two information sources” may also be a combination of the above two forms.
  • the collected judgment information includes image information collected from multiple angles and is collected from an angle. Sound information.
  • the collected judgment information includes at least two kinds of information sources, the same can be referred to the above description.
  • the method of "integration” or “result fusion” is based on these information sources to obtain the recognition result of the current time of the object to be tested.
  • the smart terminal after outputting the recognition result of the current time of the target to be tested, the smart terminal also sends a corresponding to the recognition result.
  • Interactive signal For some application scenarios that can perform human-computer interaction, such as smart guide blind, welcome robot, service robot, etc., after outputting the recognition result of the current time of the target to be tested, the smart terminal also sends a corresponding to the recognition result. Interactive signal.
  • the user may be given a voice prompt “the person in front of the first preset time node”. "If the recognition result outputted by the node at the second preset time is "male”, you can continue to give the user a voice prompt "The person in front is male” at the second preset time node, if in the third preset If the recognition result of the time node output is "Zhang San”, the voice prompt "This male is Zhang San” can be sent to the user at the third preset time node.
  • Step 130 Determine whether the attribute type corresponding to the recognition result is the attribute type with the highest priority among the at least two attribute types.
  • the recognition result when a recognition result is obtained at a predetermined time node, the recognition result is output, and it is determined whether the attribute type corresponding to the recognition result is the highest priority among the at least two attribute types.
  • the attribute type if yes, performs the following step 140; if not, the following step 150 is performed.
  • the recognition result obtained at a predetermined time node is “male”
  • the attribute type corresponding to “male” is “gender”
  • the attribute type having the highest priority It is a "person name”
  • Step 140 Stop collecting information for the object to be tested.
  • Step 140 can also be performed in other manners.
  • the recognition result may also be first determined. Whether it is the most detailed recognition result obtained for the first time, if yes, perform the following step 150; if not, the most detailed recognition result currently obtained and the most detailed recognition result obtained before If the verification is successful, the information collection for the target to be tested is stopped. If the verification is unsuccessful, the following step 150 is continued.
  • Step 150 The information about the object to be tested collected in the next preset time period is used as the judgment information.
  • the intelligent terminal continues to collect information for the target to be tested when the next preset is reached.
  • the time node is, for example, the second preset time node
  • the next preset time period ie, the time period between the second preset time node and the first preset time node
  • the information about the object to be tested is used as the judgment information, and the process returns to the above step 130, so that the smart terminal acquires the recognition result when the target to be tested is at the “next preset time node” (the second preset time node).
  • the object recognition method provided by the embodiment of the present application is that the target identification method provided by the embodiment of the present application divides multiple attributes with priority order according to different levels of detail of the description of the object to be tested. Attribute type, and in the process of identification, the information about the object to be tested collected in the preset time period is used as the judgment information, and based on the judgment information and the data obtained before the judgment information is acquired, the The recognition result of the target to be tested at the current time, and if the priority of the attribute type corresponding to the recognition result at the current time is not the highest level, that is, if the recognition result obtained at the current time is not the most detailed recognition result, then the next preset The information for the object to be tested is continuously collected during the time period, and the information is used as the judgment information, and the above identification step is repeated.
  • the step of judging and judging can output the recognition result of the object to be tested in time in different recognition scenarios. At the same time, if the most detailed recognition result is not obtained, the information acquisition time is gradually accumulated based on the accumulation of information. The information output is more detailed in identifying the results, thereby enabling a compromise between the timeliness and level of detail of the target recognition and improving the user experience.
  • the second embodiment of the present application further provides another A target recognition method is different from the target recognition method provided in the first embodiment in that, before outputting the recognition result, it is also required to determine whether the priority of the attribute type corresponding to the recognition result of the current time of the target to be measured is high.
  • a schematic flowchart of another object identification method provided by an embodiment of the present application may include, but is not limited to, the following steps:
  • Step 210 The information about the target to be detected collected in the preset time period is used as the judgment information.
  • Step 220 Acquire a recognition result of the current time of the object to be tested based on the judgment information and the data obtained before the judgment information is collected.
  • Step 230 Determine whether the priority of the attribute type corresponding to the recognition result of the current time of the object to be tested is higher than the attribute type corresponding to the recognition result of the previous time of the object to be tested.
  • the smart terminal when the recognition result of the target to be tested is obtained at a preset time node (ie, the current time), first, by determining, the smart terminal acquires the preset time node (ie, the current time). Whether the priority of the attribute type corresponding to the recognition result is higher than the attribute type corresponding to the recognition result obtained by the previous preset time node (ie, the previous time), and if so, the recognition result obtained at the current time is higher than the previous time The obtained recognition result is more detailed, so that step 231 is executed to output the recognition result acquired at the current time; if not, step 232 is performed, and the recognition result acquired at the current time is not output.
  • Step 231 Output the recognition result.
  • Step 232 The recognition result is not output.
  • Step 240 Determine whether the attribute type corresponding to the identification result is the attribute type with the highest priority among the at least two attribute types.
  • the step 240 may be performed in synchronization with the step 230. If the attribute type corresponding to the recognition result is the attribute type with the highest priority among the at least two attribute types, step 241 is performed; If the attribute type corresponding to the result is not the attribute type with the highest priority among the at least two attribute types, step 242 is performed.
  • Step 241 Stop collecting information for the object to be tested.
  • Step 242 The information about the object to be tested collected in the next preset time period is used as the judgment information, and the process returns to step 220.
  • steps 210, 220, 240, 241, and 242 have the same or similar technical features as the steps 110, 120, 130, 140, and 150 in the first embodiment, respectively.
  • the specific embodiment in the first embodiment is also applicable to the embodiment, and the embodiment will not be described in detail.
  • the object of the present application is that the target identification method provided by the embodiment of the present application determines whether the priority of the attribute type corresponding to the recognition result of the current time of the target is determined before outputting the obtained recognition result.
  • FIG. 3 is a schematic structural diagram of a target recognition apparatus according to an embodiment of the present application.
  • the target recognition apparatus 3 includes an information collection unit 31, an identification unit 32, and a determination unit 33.
  • the information collecting unit 31 is configured to use the information about the object to be tested collected in the preset time period as the judgment information, where the object to be tested includes at least two types of attributes, and the at least two attribute types are set with priority Level relationship.
  • the identification unit 32 includes an identification module 321 and an output module 322.
  • the identification module 321 is configured to acquire the recognition result of the current time of the object to be tested based on the determination information and the data obtained before the determination information is collected; the output module 322 is configured to output the recognition result, and the identification The result corresponds to one of the attribute types.
  • the recognition result is a determination result corresponding to one of the attribute types, and the confidence of the determination result satisfies a preset condition.
  • the attribute type corresponding to the recognition result has the highest priority among the attribute types corresponding to the determination result that the confidence degree satisfies the preset condition.
  • the determining unit 33 is configured to determine whether the attribute type corresponding to the recognition result is the attribute type with the highest priority among the at least two attribute types; if not, the control information collecting unit 31 collects the next preset time period. The obtained information for the object to be tested is sent to the identification unit 32 as the judgment information.
  • the information is collected by the information collecting unit 31 for the object to be tested, and the information about the object to be tested collected in the preset time period is input as the judgment information into the recognition unit 32.
  • the recognition result of the current time of the object to be tested is acquired by the identification module 321 based on the determination information and the data obtained before the determination information is collected, and the recognition result is output through the output module 322.
  • the collected information for the object to be tested is sent back to the identification unit 32 as the judgment information.
  • the data obtained before the collecting the determination information includes: information collected by the object to be tested collected before the determining the information is collected, and the identification module 321 is specifically configured to: Combining the judgment information with the feature in the information about the target to be measured collected before the collecting the judgment information; acquiring the recognition result of the current time of the target to be tested based on the merged feature.
  • the data obtained before the collecting the determination information includes: information collected by the object to be tested collected before the determining the information is collected, and the identification module 321 is specifically configured to: Obtaining a recognition result corresponding to the judgment information; selecting, from the recognition result corresponding to the judgment information and the recognition result acquired before collecting the judgment information, a recognition result having the highest priority of the attribute type as the target to be tested The recognition result of the current moment.
  • the target recognition device 3 further includes:
  • the object recognition apparatus divides the attributes of the target to be tested into multiple priority orders by different degrees of detail according to the description of the object to be tested.
  • the type of the attribute, and in the process of the identification, the information about the object to be tested collected in the preset time period is used as the judgment information by the information collecting unit 31, and based on the judgment information and the judgment in the identification unit 32.
  • the data obtained before the information obtains and outputs the recognition result of the object to be tested at the current time.
  • the determining unit 33 determines whether the priority of the attribute type corresponding to the recognition result at the current time is the highest level, and if so, the control information is collected.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Signal Processing (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Robotics (AREA)
  • Mechanical Engineering (AREA)
  • Image Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Image Processing (AREA)

Abstract

一种目标识别方法、装置和智能终端。其中，所述方法包括：将预设时间段内采集到的针对待测目标的信息作为判断信息；基于该判断信息以及在采集该判断信息之前获得的数据，获取并输出待测目标当前时刻的识别结果；判断该识别结果对应的属性类型的优先级是否为最高级；若否，则将下一预设时间段内采集到的针对该待测目标的信息作为判断信息，并且返回基于该判断信息以及在采集该判断信息之前获得的数据，获取并输出待测目标当前时刻的识别结果的步骤。通过上述技术方案，能够在目标识别的及时性和详细程度之间达到折中，提升用户体验。

Description

一种目标识别方法、装置和智能终端 技术领域
本申请实施例涉及智能识别技术领域，尤其涉及一种目标识别方法、装置和智能终端。
背景技术
随着机器智能化进程的推进，人与智能终端之间的交互越来越频繁，人机交互的自然体验问题也随之变得越来越重要。其中，影响人机交互的自然体验的两个重要因素就是智能终端对待测目标的识别的及时性以及详细程度。
当前，大多数智能终端都被希望能够输出人名、车的型号(或者系列)、车牌号码、猫的品种等详细程度较高的目标识别结果，以提升人机交互体验。
然而，在实际场景中，环境是多变的，而智能终端的识别能力是有限的，在某些场景下智能终端有可能需要花费较长的时间采集更多的信息才可以分析得到较详细的识别结果。在这种情况下，如果强制要求智能终端输出详细程度较高的识别结果，有可能需要很长的时间，无法满足目标识别的及时性；如果为了满足目标识别的及时性而规定智能终端在某一时间内必须输出识别结果，则有可能只能得到一个详细程度较低的识别结果，这同样不利于用户的友好体验。
由此，如何在目标识别的及时性与详细程度之间达到折中是现有的智能识别技术亟待解决的问题。
发明内容
本申请实施例提供一种目标识别方法、装置和智能终端，能够解决如何在目标识别的及时性与详细程度之间达到折中的问题。
第一方面，本申请实施例提供了一种目标识别方法，应用于智能终端，包括：
将预设时间段内采集到的针对待测目标的信息作为判断信息，所述待测目标包括至少两种属性类型，所述至少两种属性类型之间设置有优先级关系；
基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，输出所述识别结果，所述识别结果与其中一种所述属性类型对应；
判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；
若否，则将下一预设时间段内采集到的针对所述待测目标的信息作为判断信息，并且返回所述基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，输出所述识别结果的步骤。
第二方面，本申请实施例提供一种目标识别装置，应用于智能终端，包括：
信息采集单元，用于将预设时间段内采集到的针对待测目标的信息作为判断信息，所述待测目标包括至少两种属性类型，所述至少两种属性类型之间设置有优先级关系；
识别单元，包括：识别模块和输出模块，所述识别模块用于基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，所述输出模块用于输出所述识别结果，所述识别结果与其中一种所述属性类型对应；
判断单元，用于判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；
若否，则控制所述信息采集单元将下一预设时间段内采集到的针对所述待测目标的信息作为所述判断信息发送至所述识别单元。
第三方面，本申请实施例提供一种智能终端，包括：
至少一个处理器；以及，
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如上所述的目标识别方法。
第四方面，本申请实施例提供一种非暂态计算机可读存储介质，所述非暂态计算机可读存储介质存储有计算机可执行指令，所述计算机可执行指令用于使智能终端执行如上所述的目标识别方法。
第五方面，本申请实施例还提供了一种计算机程序产品，所述计算机程序产品包括存储在非暂态计算机可读存储介质上的计算机程序，所述计算机程序包括程序指令，当所述程序指令被智能终端执行时，使所述智能终端执行如上所述的目标识别方法。
本申请实施例的有益效果在于：本申请实施例提供的目标识别方法、装置和智能终端通过根据对待测目标的描述的详细程度的不同为待测目标的属性划分多个具有优先级顺序的属性类型，并且，在识别的过程中，将预设时间段内采集到的针对待测目标的信息作为判断信息，并基于该判断信息以及在采集该判断信息之前获得的数据，获取以及输出该待测目标在当前时刻的识别结果，同时，若当前时刻的识别结果对应的属性类型的优先级不是最高级，即，若当前时刻获得的识别结果不是最详细的识别结果，则在下一预设时间段内继续采集针对该待测目标的信息，并将该信息作为判断信息，重复上述的识别步骤和判断步骤，能够在不同的识别场景下都及时地输出对待测目标的识别结果，同时，若获取到的不是最详细的识别结果，则随着信息采集时间的累积，逐渐基于更加丰富的信息输出更加详细的识别结果，从而，能够在目标识别的及时性和详细程度之间达到折中，提升用户体验。
附图说明
一个或多个实施例通过与之对应的附图中的图片进行示例性说明，这些示例性说明并不构成对实施例的限定，附图中具有相同参考数字标号的元件表示为类似的元件，除非有特别申明，附图中的图不构成比例限制。
图1是本申请实施例提供的一种目标识别方法的流程示意图；
图2是本申请实施例提供的另一种目标识别方法的流程示意图；
图3是本申请实施例提供的一种目标识别装置的结构示意图；
图4是本申请实施例提供的一种智能终端的硬件结构示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
需要说明的是，如果不冲突，本申请实施例中的各个特征可以相互结合，均在本申请的保护范围之内。另外，虽然在装置示意图中进行了功能模块划分，在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于装置中的模块划分，或流程图中的顺序执行所示出或描述的步骤。
本申请实施例提供了一种目标识别方法、装置和智能终端，能够适用于任意与目标识别相关的应用领域，比如：智能导盲、迎宾机器人、服务机器人、入侵对象探测、语义识别等，尤其适用于智能导盲、迎宾机器人、服务机器人等注重人机交互体验的应用领域。
其中，本申请实施例提供的目标识别方法是一种能够及时地基于在预设时间段内采集到的信息输出识别结果，并且根据当前时刻获得的识别结果对应的属性类型的“优先级”决定是否继续进行信息采集以进一步优化目标识别的详细程度的智能优化识别方法，通过根据对待测目标的描述的详细程度的不同为待测目标的属性划分多个具有优先级顺序的属性类型(其中，优先级越高的属性类型对应的识别结果的详细程度越高)，并且，在识别的过程中，将预设时间段内采集到的针对待测目标的信息作为判断信息，并基于该判断信息以及在采集该判断信息之前获得的数据，获取以及输出该待测目标在当前时刻的识别结果，若当前时刻的识别结果对应的属性类型的优先级不是最高级，即，若当前时刻获得的识别结果不是最详细的识别结果，则在下一预设时间段内继续采集针对该待测目标的信息，并将该信息作为判断信息，重复上述的识别步骤和判断步骤，能够在不同的识别场景下都及时地输出对待测目标的识别结果，同时，若获取到的不是最详细的识别结果，则随着信息采集时间的累积，逐渐基于更加丰富的信息输出更加详细的识别结果，从而，能够在目标识别的及时性和详细程度之间达到折中，提升用户体验。
由此，采用本申请实施例提供的目标识别方法、装置和智能终端在识别相同的人/物(待测目标)时，在不同识别环境下都能够及时地输出待测目标的识别结果，并且，随着时间的推移，能够输出越来越详细的识别结果。但需要说明的是：在不同的识别环境下，智能终端在同一预设时间段内输出的识别结果的详细程度有可能不一样。
例如，以通过采集到的图像识别人为例，在光照好，距离近，并且被测人正对着智能终端的摄像头的识别环境中，智能终端在第一个预设时间段内(比 如，第一个5s内，即触发智能终端开始采集信息后5s内)即可采集到一张清晰的人脸图像，进而基于该清晰的人脸图像即可识别出被测人的“人名”，此时，智能终端可以输出该被测人的“人名”以与用户进行交互；同时，由于此时已经获得了最详细的识别结果，从而可以停止针对该被测人采集信息。而在另一些识别环境中，比如，在智能终端进行信息采集的第一个预设时间段内(比如，第一个5s内)被测人侧对着智能终端的摄像头，智能终端只能采集到被测人的侧脸图像，进而基于该侧脸图像只能识别出该被测人的“性别”，此时，智能终端输出第一个识别结果：被测人的“性别”，以便于及时地向用户反馈，同时，由于“性别”不是最详细的识别结果，因此，在本申请实施例中，智能终端还继续在下一预设时间段内(比如，第二个5s内)针对被测人采集信息；若智能终端在第二个预设时间段内能够采集到被测人的正脸图像，则可以结合第一个预设时间段内采集到的侧脸图像和第二个预设时间段内采集到的正脸图像，识别出被测人的“人名”，继而输出第二个识别结果：该被测人的“人名”，从而得到更加详细的识别结果，便于智能终端进一步调整与用户交互的内容，提升用户体验。
本申请实施例提供的目标识别方法和装置能够应用于任意类型的智能终端，比如：机器人、导盲眼镜、智能头盔、智能手机、平板电脑、服务器等。该智能终端可以包括任何合适类型的，用以存储数据的存储介质，例如磁碟、光盘(CD-ROM)、只读存储记忆体或随机存储记忆体等。该智能终端还可以包括一个或者多个逻辑运算模块，单线程或者多线程并行执行任何合适类型的功能或者操作，例如查看数据库、图像处理等。所述逻辑运算模块可以是任何合适类型的，能够执行逻辑运算操作的电子电路或者贴片式电子器件，例如：单核心处理器、多核心处理器、图形处理器(GPU)等。
具体地，下面结合附图，对本申请实施例作进一步阐述。
实施例一
图1是本申请实施例提供的一种目标识别方法的流程示意图，可以由任意类型的智能终端执行，具体地，请参阅图1，该方法包括但不限于：
步骤110：将预设时间段内采集到的针对待测目标的信息作为判断信息。
在本实施例中，所述“待测目标”可以包括但不限于：人、动物、物体等。根据对待测目标的描述的详细程度的不同可以为待测目标划分至少两个不同层 级的属性类型，并且，按照其对待测目标的描述的详细程度的高低，为这些属性类型设置优先级关系。其中，可以认为识别难度较大的属性类型对应的详细程度较高，而识别的难易程度可以依据不同属性类型的识别算法在相同条件下(例如输入相同的图片)的识别率来排序(例如，通常人名识别难于性别识别，性别识别难于人脸/人体识别)；或者，也可以依据属性类型间的相互包含关系来排序(例如，要识别性别需先识别到人脸的存在)。例如：假设待测目标为人，可以根据对待测目标的描述的详细程度的不同，设置人的属性类型包括：“人名”、“性别”以及“是否为人”，而根据识别的难易程度，可以设置这些属性类型的优先级顺序为：L1(人名)＞L2(性别)＞L3(是否为人)。
在本实施例中，智能终端持续针对待测目标采集信息，并在多个预设时间节点输出待测目标的识别结果，那么，所述“预设时间段”即进行目标识别的其中一个预设时间节点与该预设时间节点的上一个预设时间节点之间的时间阶段，该时间阶段可以是智能终端进行信息采集的进程中的任意一个时间阶段。此外，在本实施例中，智能终端采集到的“信息”是能够反映待测目标的属性的判断依据，基于该信息可以识别出待测目标的属性。该信息的类型可以包括但不限于：图像信息、声音信息、热红外画面、近红外画面、超声信号、电磁反射信号等，可以通过一种或者多种传感器采集得到，比如，通过摄像头采集针对待测目标的图像信息、通过麦克风采集针对待测目标的声音信息、通过热红外传感器采集针对待测目标的热红外画面等。
例如：假设智能终端在(0，t1)的时间段内采集到被测人的背影图像a1，在(t1，t2)的时间段内采集到被测人的侧脸图像a2，在(t2，T)的时间段内采集到被测人的正脸图像a3，其中，0＜t1＜t2＜T。则，在本步骤110中所述的“预设时间段内采集到的针对待测目标的信息”可以是智能终端在(0，t1)的时间段内采集到的背影图像a1，也可以是在(t1，t2)的时间段内采集到的侧脸图像a2，还可以是在(t2，T)的时间段内采集到的正脸图像a3。
在本实施例中，对待测目标进行识别时，可以规定智能终端根据实际情况持续针对待测目标采集信息，并在特定的时间节点进行目标识别。具体地，在某一时间节点进行目标识别时，可以将智能终端在当前时间节点和上一时间节点之间的时间段内针对待测目标采集到的信息作为判断信息，智能终端基于该判断信息执行下述步骤120可以获取到该待测目标在该时间节点的识别结果。 比如，可以规定智能终端以接收到信息采集指令的那一刻为时间起始节点(即，0时刻)，在t1、t2以及T时刻进行目标识别，则，可以在t1时刻，将(0，t1)时间段内针对待测目标采集的信息作为判断信息，基于该判断信息执行下述步骤120后可以得到待测目标在t1时刻的识别结果；在t2时刻，将(t1，t2)时间段内针对待测目标采集的信息作为判断信息，基于该判断信息执行下述步骤120后可以得到待测目标在t2时刻的识别结果；在T时刻，将(t2，T)时间段内针对待测目标采集的信息作为判断信息，基于该判断信息执行下述步骤120后可以得到待测目标在T时刻的识别结果。其中，可以理解的是，在实际应用中，任意两个时间节点之间的时长可以是相等的(即，周期性进行目标识别)，也可以是不相等的，本申请实施例对此不作具体限定。
步骤120：基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，输出所述识别结果。
在本实施例中，在特定的时间节点，基于该时间节点之前预设时间段内获取到的判断信息以及在采集该判断信息之前获得的数据进行目标识别。所述“判断信息”可以是智能终端在任意一个预设时间段内采集到的用于获取当前时刻该待测目标的识别结果的信息。所述“当前时刻”是指完成该判断信息的采集并基于该判断信息进行目标识别的时刻。所述“在采集所述判断信息之前获得的数据”可以是在采集该判断信息之前采集到的针对该待测目标的信息；也可以是在采集该判断信息之前获取到的识别结果(即，当前时刻之前已经获取到的识别结果)。比如，假设采集到的判断信息为智能终端在(t1，t2)的时间段内采集到的侧脸图像a2，则，完成该侧脸图像a2的采集的时刻t2即“当前时刻”，而“在采集该判断信息之前获得的数据”即t1时刻之前(包括t1时刻)获得的数据，该数据可以是(0，t1)时间段内采集到的背影图像a1，或者，也可以是智能终端在t1时刻获取到的识别结果。
在本实施例中，所述“识别结果”与待测目标的其中一种属性类型相对应。比如，以识别人为例，其属性类型包括：“人名”、“性别”以及“是否为人”；如果获取到的识别结果为“李四”，则该识别结果“李四”对应的属性类型即“人名”；如果获取到的识别结果为“男性”，则该识别结果“男性”对应的属性类型即“性别”；如果获取到的识别结果为“人”，则该识别结果“人”对应的属性类型即“是否为人”。并且，为了在满足目标识别的及时性的要求的同时，尽 可能地输出最详细的识别结果，在本实施例中，在每一个进行目标识别(即执行本步骤120)的时间节点，均基于采集到的判断信息以及在采集所述判断信息之前获得的数据获取最详细的识别结果(即，对这些判断信息和数据进行运算之后得到优先级最高的属性类型对应的判断结果)，比如，在某一时间节点，基于采集到的判断信息以及在采集所述判断信息之前获得的数据，可以得到判断结果“人”、“男性”和“李四”，其中，“李四”对应的属性类型“人名”的优先级最高，因此，此时可以获取并输出识别结果“李四”。
具体地，在本实施例中，基于判断信息以及在采集该判断信息之前获得的数据，获取该待测目标当前时刻的识别结果的具体实施方式可以包括但不限于以下三种实施方式：
在其中一种实施方式中，在采集所述判断信息之前获得的数据包括在采集所述判断信息之前采集到的针对该待测目标的信息，此时，可以采用“特征融合”的方式获取待测目标当前时刻的识别结果。
具体为：首先，融合所述判断信息和所述在采集所述判断信息之前采集到的针对所述待测目标的信息中的特征，然后，基于融合后的特征获取所述待测目标当前时刻的识别结果。具体地，可以分别从不同时间段内采集到的判断信息中提取出用于识别待测目标的属性的特征，并对这些特征进行融合，然后通过合适的识别算法，比如，神经网络算法，基于融合后的特征获得待测目标在当前时刻的识别结果。
在另一种实施方式中，在采集所述判断信息之前获得的数据包括在采集该判断信息之前获取到的识别结果，此时，可以采用“结果融合”的方式获取待测目标当前时刻的识别结果。
具体为：首先，获取与所述判断信息对应的识别结果；然后，从与所述判断信息对应的识别结果和在采集所述判断信息之前获取到的识别结果中选择属性类型优先级最高的识别结果作为所述待测目标当前时刻的识别结果。其中，所述获取与所述判断信息对应的识别结果，即：基于该判断信息得到该待测目标的识别结果。比如，在采集所述判断信息之前获取到的识别结果包括：“人”和“男性”，基于当前时刻获取到的判断信息得到的识别结果为“男性”，则可以从“人”、“男性”、“男性”这三个识别结果中选择出“男性”(对应的属性类 型优先级最高)作为该待测目标当前时刻的识别结果。
在又一种实施方式中，若当前的时刻是第一个进行目标识别的时间节点，比如，采集到的判断信息为(0，t1)时间段内采集到的针对待测目标的信息，则，在采集该判断信息之前没有任何的数据，此时，可以仅基于采集到的判断信息获取该待测目标当前时刻的识别结果，也就是说，在该实施方式中，与采集到的判断信息对应的识别结果即该待测目标当前时刻的识别结果。
特别地，在一些实施例中，为了在保证目标识别的及时性和详细程度的情况下，同时保证识别结果的可靠性，待测目标的每一属性类型对应的判断结果都设置有一个用于表征该判断结果的可靠性(或，可信性)的置信度。而本步骤120所获得的识别结果为其中一种所述属性类型对应的判断结果，所述判断结果的置信度满足预设条件，并且，所述识别结果对应的属性类型在置信度满足所述预设条件的判断结果对应的属性类型中优先级最高。
其中，判断结果的置信度可以通过特征比对的相似程度来确定，相似程度越高，置信度越高。所述“预设条件”可以根据实际应用场景而设置，用于鉴定某一判断结果的可靠程度。具体地，该预设条件可以是：判断结果的置信度大于或者等于与其对应的属性类型所对应的置信阈值。其中，每一属性类型对应的置信阈值可以是相同的，比如，与属性类型“人名”、“性别”和“是否为人”对应的置信阈值均为70％，若获取到待测目标的判断结果包括：“张三”(置信度为70％)，“男性”(置信度为89％)，“人”(置信度为100％)，则，判断结果“张三”、“男性”和“人”的置信度均满足预设条件，此时，该待测目标的识别结果为这三者中优先级最高的属性类型“人名”对应的判断结果“张三”。或者，在另一些实施例中，每一属性类型对应的置信阈值也可以是不相同的，比如，可以预设与属性类型“人名”对应的置信阈值为75％，与属性类型“性别”对应的置信阈值为85％，与属性类型“是否为人”对应的置信阈值为95％，若获取到待测目标的判断结果同样是：“张三”(置信度为70％)，“男性”(置信度为89％)，“人”(置信度为100％)，则，置信度满足预设条件的判断结果仅包括“男性”和“人”，此时，该待测目标的识别结果为这两者中优先级最高的属性类型“性别”对应的判断结果“男性”。
在该实施例中，基于采集到的判断信息(或者，基于采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)获取待测目标当前时刻 的识别结果的具体实施方式可以包括但不限于以下两种实施方式：
在第一种实施方式中，可以首先基于采集到的判断信息(或者，基于采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)获取待测目标每一属性类型对应的判断结果以及每一判断结果的置信度；然后输出置信度满足预设条件的判断结果中优先级最高的属性类型对应的判断结果作为所述待测目标的识别结果。
其中，在该实施例方式中，基于采集到的判断信息(或者，基于采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)获取待测目标每一属性类型对应的判断结果可以通过使用合适的算法(比如，神经网络)来实现。比如，假设待测目标为人，智能终端采集到的判断信息(或者，采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)为该人的图像信息，则，智能终端可以从该图像中迭代式地计算出属性类型“是否为人”、“性别”和“人名”对应的判断结果，比如，首先通过神经网络的底层计算出用于判别“是否为人”的特征1，并根据特征1得到“是否为人”对应判断结果及该判断结果的置信度；然后，在神经网络的中间层基于特征1计算用于判别“性别”的特征2，并根据特征2得到“性别”对应的判断结果及该判断结果的置信度；最后，在神经网络的最上层基于特征2计算出用于判别“人名”的特征3，并根据特征3得到“人名”对应的判断结果及该判断结果的置信度。当获取到所有判断结果及其置信度之后，首先筛选出置信度满足预设要求的判断结果，然后选择详细程度最高(即，所对应的属性类型优先级最高)的判断结果作为待测目标当前时刻的识别结果。
在第二种实施方式中，可以基于采集到的判断信息(或者，基于采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)，根据优先级从高到低的顺序逐级获取待测目标每一属性类型对应的判断结果以及每一判断结果的置信度，直至第一个置信度满足预设条件的判断结果出现时，输出该第一个置信度满足预设条件的判断结果作为所述待测目标当前时刻的识别结果。即：当采集到针对待测目标的判断信息时(或者，当获取到采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息时)，首先基于采集到的信息(或者，基于采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)获取优先级最高的属性类型对应的一级判断结果以及一级判断 结果的一级置信度，如果该一级置信度满足预设条件(比如，一级置信度大于或等于一级置信阈值)，则直接输出该一级判断结果作为待测目标当前时刻的识别结果，否则，基于采集到的判断信息获取下一级别的属性类型对应的二级判断结果以及二级判断结果的二级置信度；如果该二级置信度满足预设条件(比如，二级置信度大于或者等于二级置信阈值)，则，输出该二级判断结果作为待测目标当前时刻的识别结果，否则，继续基于采集到的判断信息获取再下一级别的属性类型对应的判断结果及其置信度，如此循环，直至获取到置信度满足预设条件的判断结果。
其中，在该实施方式中，可以从采集到的判断信息(或者，采集到的判断信息和在采集该判断信息之前采集到的针对待测目标的信息)中提取出不同的特征用于不同级别的判断，例如，假设待测目标为车，采集到的信息为针对该车的图像信息，则，可以从该图像信息中提取出特征a用于识别图像中是否有车，提取出特征b用于识别图像中车的颜色，提取特征c用于识别车的类型(轿车、货车、公交车等)等。
在该实施方式中，通过根据优先级从高到低的顺序逐级获取待测目标每一属性类型对应的判断结果及其置信度，当出现第一个置信度满足预设条件的判断结果时，就直接将该第一个置信度满足预设条件的判断结果作为待测目标当前时刻的识别结果，而不需要对每一个属性类型进行识别判断，能够减少数据处理量，在不影响识别的详细程度和可靠性的前提下，提升识别效率。
再者，在实际应用中，为了提升识别的准确度和识别效率，采集到的判断信息也可以包括至少两种信息源。其中，所述“信息源”是指能够反映待测目标的属性的信息来源。所述“至少两种信息源”可以是至少两种不同类型的判断信息，比如，图像信息、声音信息、热红外画面、近红外画面、超声信号或电磁反射信号中的任意两种或者多种；或者，所述“至少两种信息源”也可以是从至少两个角度采集到的某一类型的信息，比如，从多个角度采集待测目标的图像信息(或者声音信息)，每一视角采集到的图像信息(或者声音信息)均可作为一种信息源。当然，可以理解的是，所述“至少两种信息源”也可以是上述两种形式的组合，比如，采集到的判断信息中包括从多个角度采集到的图像信息和从一个角度采集到的声音信息。
当采集到的判断信息包括至少两种信息源时，同样可以参照以上描述的“特 征融合”或者“结果融合”的方式基于这些信息源获取待测目标当前时刻的识别结果。
此外，针对一些可以进行人机交互的应用场景，如，智能导盲、迎宾机器人、服务机器人等，智能终端在输出待测目标当前时刻的识别结果之后，还发送与所述识别结果对应的交互信号。
例如：对于用于导盲的智能眼镜或者智能头盔，如果在第一个预设时间节点输出的识别结果为“人”，可以在第一个预设时间节点向用户发出语音提示“前面有个人”，如果在第二个预设时间节点输出的识别结果为“男性”，可以继续在第二个预设时间节点向用户发出语音提示“前面这个人是男性”，如果在第三个预设时间节点输出的识别结果为“张三”，则可以继续在第三个预设时间节点向用户发出语音提示“这个男性是张三”。
又如，对于用于迎宾或者提供服务的机器人，如果在第一预设时间节点输出的识别结果为“人”，可以对待测目标说“您好！请问我有什么可以帮到您？”并提供普适***，随着采集到的信息的增多，如果在第二预设时间节点输出识别结果“男性”，则调整与该待测目标交谈的内容为针对男性的内容，比如说“请问您是否要查找最新的电子产品”；如果在第三预设时间节点输出识别结果“张三”，则继续调整与待测目标交谈的内容为针对张三的内容，比如说“您最近关注的产品已经到货，需要试用吗？”。
步骤130：判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型。
在本实施例中，当在某一预定的时间节点获取到一个识别结果时，输出该识别结果，并且，判断该识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型，如果是，则执行下述步骤140；如果不是，则执行下述步骤150。比如，以识别人为例，假设在某一预定的时间节点获取到的识别结果为“男性”，“男性”对应的属性类型为“性别”，而对于人物识别来说，优先级最高的属性类型为“人名”，从而，此时需要继续针对待测目标采集信息，并且执行下述步骤150。
步骤140：停止针对所述待测目标采集信息。
当在某一预设时间节点获取到待测目标优先级最高的属性类型对应的识别结果时，说明已经得到了最详细的识别结果，因此，在本实施例中，为了减少 不必要的计算量以及能耗，可以在获取到最详细的识别结果时，停止针对该待测目标采集信息。
当然，可以理解的是，当在某一预设时间节点获取到最详细的识别结果时，停止针对所述待测目标采集信息这一实施方式仅为其中一种实施方式，在实际应用中，也可以采用其他的方式执行步骤140。比如，在一些实施例中，为了保证输出的识别结果的准确性，当在某一预设时间节点获取到待测目标优先级最高的属性类型对应的识别结果时，也可以首先判断该识别结果是否为第一次获取到的最详细的识别结果，如果是，则执行下述步骤150；如果不是，则对当前获取到的最详细的识别结果与之前获取到的最详细的识别结果进行校验，如果校验成功，则停止针对该待测目标采集信息，如果校验不成功，则继续执行下述步骤150。
步骤150：将下一预设时间段内采集到的针对所述待测目标的信息作为判断信息。
在本实施例中，若在某一预设时间节点，比如，第一个预设时间节点，没有获取到待测目标优先级最高的属性类型对应的识别结果，说明此时还没有得到最详细的识别结果，还需对待测目标进行进一步的识别，以获取更加详细的识别结果，因此，当步骤130的判断结果为否时，智能终端继续针对待测目标采集信息，当到达下一个预设时间节点时，比如，第二个预设时间节点，将该下一预设时间段(即，第二个预设时间节点和第一个预设时间节点之间的时间段)内采集到的针对该待测目标的信息作为判断信息，返回上述步骤130，以使智能终端获取到该待测目标在“下一个预设时间节点”(第二个预设时间节点)时的识别结果。
通过上述技术方案可知，本申请实施例的有益效果在于：本申请实施例提供的目标识别方法通过根据对待测目标的描述的详细程度的不同为待测目标的属性划分多个具有优先级顺序的属性类型，并且，在识别的过程中，将预设时间段内采集到的针对待测目标的信息作为判断信息，并基于该判断信息以及在采集该判断信息之前获得的数据，获取以及输出该待测目标在当前时刻的识别结果，同时，若当前时刻的识别结果对应的属性类型的优先级不是最高级，即，若当前时刻获得的识别结果不是最详细的识别结果，则在下一预设时间段内继续采集针对该待测目标的信息，并将该信息作为判断信息，重复上述的识别步 骤和判断步骤，能够在不同的识别场景下都及时地输出对待测目标的识别结果，同时，若获取到的不是最详细的识别结果，则随着信息采集时间的累积，逐渐基于更加丰富的信息输出更加详细的识别结果，从而，能够在目标识别的及时性和详细程度之间达到折中，提升用户体验。
实施例二
考虑到在一些实际应用场景中，有可能存在连续两个预定时间节点都获得相同的识别结果的情况，为了避免重复输出相同的识别结果，提升用户体验，本申请第二实施例还提供了另一种目标识别方法，该方法与实施例一所提供的目标识别方法的不同之处在于：在输出识别结果之前，还需判断待测目标当前时刻的识别结果对应的属性类型的优先级是否高于待测目标上一时刻的识别结果对应的属性类型；如果是，才输出该识别结果；如果不是，就不输出该识别结果。
具体地，如图2所示，为本申请实施例提供的另一种目标识别方法的流程示意图，该方法可以包括但不限于以下步骤：
步骤210：将预设时间段内采集到的针对待测目标的信息作为判断信息。
步骤220：基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果。
在本实施例中，获取到待测目标在某一预设时间点的识别结果之后，同时执行下述步骤230和步骤240。
步骤230：判断所述待测目标当前时刻的识别结果对应的属性类型的优先级是否高于所述待测目标上一时刻的识别结果对应的属性类型。
在本实施例中，当获取到待测目标在某一预设时间节点(即，当前时刻)的识别结果时，首先通过判断智能终端在该预设时间节点(即，当前时刻)获取到的识别结果对应的属性类型的优先级是否高于上一个预设时间节点(即，上一时刻)获取到的识别结果对应的属性类型，若是，则说明当前时刻获取到的识别结果比上一时刻获取到的识别结果更详细，从而执行步骤231，输出当前时刻获取到的识别结果；若否，则执行步骤232，不输出当前时刻获取到的识别结果。
步骤231：输出所述识别结果。
步骤232：不输出所述识别结果。
步骤240：判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型。
在本实施例中，本步骤240可以与步骤230同步进行，若所述识别结果对应的属性类型是所述至少两种属性类型中优先级最高的属性类型，则执行步骤241；若所述识别结果对应的属性类型不是所述至少两种属性类型中优先级最高的属性类型，则执行步骤242。
步骤241：停止针对所述待测目标采集信息。
步骤242：将下一预设时间段内采集到的针对所述待测目标的信息作为判断信息，并返回步骤220。
需要说明的是，在本实施例中，上述步骤210、220、240、241和242分别与实施例一中的步骤110、120、130、140和150具有相同或相似的技术特征，因此，实施例一中的具体实施方式同样适用于本实施例，本实施例便不再详述。
通过上述技术方案可知，本申请实施例的有益效果在于：本申请实施例提供的目标识别方法通过在输出获得的识别结果之前先判断待测目标当前时刻的识别结果对应的属性类型的优先级是否高于待测目标上一时刻的识别结果对应的属性类型，如果是，才输出该识别结果，能够避免重复输出相同的识别结果给用户造成困扰的情况，提升用户体验。
实施例三
图3是本申请实施例提供的一种目标识别装置的结构示意图，请参阅图3，该目标识别装置3包括：信息采集单元31、识别单元32和判断单元33。
信息采集单元31，用于将预设时间段内采集到的针对待测目标的信息作为判断信息，所述待测目标包括至少两种属性类型，所述至少两种属性类型之间设置有优先级关系。
识别单元32，包括：识别模块321和输出模块322。其中，识别模块321用于基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果；输出模块322用于输出所述识别结果，所述识别结果与其中一种所述属性类型对应。其中，在一些实施例中，所述识别结果为其中一种所述属性类型对应的判断结果，所述判断结果的置信度满足预设条件， 并且，所述识别结果对应的属性类型在置信度满足所述预设条件的判断结果对应的属性类型中优先级最高。
判断单元33，用于判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；若否，则控制信息采集单元31将下一预设时间段内采集到的针对所述待测目标的信息作为所述判断信息发送至识别单元32。
在本实施例中，当需要进行目标识别时，首先通过信息采集单元31针对待测目标采集信息，并将预设时间段内采集到的针对待测目标的信息作为判断信息输入识别单元32，在识别单元32中，通过识别模块321基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，进而通过输出模块322输出所述识别结果，以及，在判断单元33中判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；若否，则控制信息采集单元31将下一预设时间段内采集到的针对所述待测目标的信息作为所述判断信息发送回识别单元32。
其中，在一些实施例中，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前采集到的针对所述待测目标的信息，则，识别模块321具体用于：融合所述判断信息和所述在采集所述判断信息之前采集到的针对所述待测目标的信息中的特征；基于融合后的特征获取所述待测目标当前时刻的识别结果。
其中，在一些实施例中，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前采集到的针对所述待测目标的信息，则，识别模块321具体用于：获取与所述判断信息对应的识别结果；从与所述判断信息对应的识别结果和在采集所述判断信息之前获取到的识别结果中选择属性类型优先级最高的识别结果作为所述待测目标当前时刻的识别结果。
其中，在一些实施例中，识别单元32还包括：判断模块323。
该判断模块323用于判断所述待测目标当前时刻的识别结果对应的属性类型的优先级是否高于所述待测目标上一时刻的识别结果对应的属性类型；若是，则通过输出模块322输出所述识别结果；若否，则控制输出模块322不输出所述识别结果。
此外，在一些实施例中，该目标识别装置3还包括：
交互单元34，用于发送与所述识别结果对应的交互信号。
需要说明的是，由于所述目标识别装置与上述方法实施例中的目标识别方法基于相同的发明构思，因此，上述方法实施例的相应内容以及有益效果同样适用于本装置实施例，此处不再详述。
通过上述技术方案可知，本申请实施例的有益效果在于：本申请实施例提供的目标识别装置通过根据对待测目标的描述的详细程度的不同为待测目标的属性划分多个具有优先级顺序的属性类型，并且，在识别的过程中，通过信息采集单元31将预设时间段内采集到的针对待测目标的信息作为判断信息，并在识别单元32中基于该判断信息以及在采集该判断信息之前获得的数据，获取以及输出该待测目标在当前时刻的识别结果，同时，通过判断单元33判断当前时刻的识别结果对应的属性类型的优先级是否为最高级，若是，则控制信息采集单元31在下一预设时间段内继续采集针对该待测目标的信息，并将该信息作为判断信息发送至识别单元32，能够在不同的识别场景下都及时地输出对待测目标的识别结果，同时，若获取到的不是最详细的识别结果，则随着信息采集时间的累积，逐渐基于更加丰富的信息输出更加详细的识别结果，从而，能够在目标识别的及时性和详细程度之间达到折中，提升用户体验。
实施例四
图4是本申请实施例提供的一种智能终端的硬件结构示意图，该智能终端400可以是任意类型的智能终端，如：机器人、导盲眼镜、智能头盔、智能手机、平板电脑、服务器等，能够执行上述方法实施例一和实施例二所提供的目标识别方法。
具体地，请参阅图4，该智能终端400包括：
一个或多个处理器401以及存储器402，图4中以一个处理器401为例。
处理器401和存储器402可以通过总线或者其他方式连接，图4中以通过总线连接为例。
存储器402作为一种非暂态计算机可读存储介质，可用于存储非暂态软件程序、非暂态性计算机可执行程序以及模块，如本申请实施例中的目标识别方法对应的程序指令/模块(例如，附图3所示的信息采集单元31、识别单元32判断单元33和交互单元34)。处理器401通过运行存储在存储器402中的非暂 态软件程序、指令以及模块，从而执行目标识别装置的各种功能应用以及数据处理，即实现上述任一方法实施例的目标识别方法。
存储器402可以包括存储程序区和存储数据区，其中，存储程序区可存储操作***、至少一个功能所需要的应用程序；存储数据区可存储根据目标识别装置的使用所创建的数据等。此外，存储器402可以包括高速随机存取存储器，还可以包括非暂态存储器，例如至少一个磁盘存储器件、闪存器件、或其他非暂态固态存储器件。在一些实施例中，存储器402可选包括相对于处理器401远程设置的存储器，这些远程存储器可以通过网络连接至智能终端400。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
所述一个或者多个模块存储在所述存储器402中，当被所述一个或者多个处理器401执行时，执行上述任意方法实施例中的目标识别方法，例如，执行以上描述的图1中的方法步骤110至步骤150，图2中的方法步骤210至步骤242，实现图3中的单元31-34的功能。
实施例五
本申请实施例还提供了一种非暂态计算机可读存储介质，所述非暂态计算机可读存储介质存储有计算机可执行指令，该计算机可执行指令被一个或多个处理器执行，例如被图4中的一个处理器401执行，可使得上述一个或多个处理器执行上述任意方法实施例中的目标识别方法，例如，执行以上描述的图1中的方法步骤110至步骤150，图2中的方法步骤210至步骤242，实现图3中的单元31-34的功能。
以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。
通过以上的实施方式的描述，本领域普通技术人员可以清楚地了解到各实施方式可借助软件加通用硬件平台的方式来实现，当然也可以通过硬件。本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一非暂态计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其 中，所述的存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory,ROM)或随机存储记忆体(Random Access Memory,RAM)等。
上述产品可执行本申请实施例所提供的目标识别方法，具备执行目标识别方法相应的功能模块和有益效果。未在本实施例中详尽描述的技术细节，可参见本申请实施例所提供的目标识别方法。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；在本申请的思路下，以上实施例或者不同实施例中的技术特征之间也可以进行组合，步骤可以以任意顺序实现，并存在如上所述的本申请的不同方面的许多其它变化，为了简明，它们没有在细节中提供；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的范围。

Claims (15)

  1. 一种目标识别方法，应用于智能终端，其特征在于，包括：
    将预设时间段内采集到的针对待测目标的信息作为判断信息，所述待测目标包括至少两种属性类型，所述至少两种属性类型之间设置有优先级关系；
    基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，输出所述识别结果，所述识别结果与其中一种所述属性类型对应；
    判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；
    若否，则将下一预设时间段内采集到的针对所述待测目标的信息作为判断信息，并且返回所述基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，输出所述识别结果的步骤。
  2. 根据权利要求1所述的目标识别方法，其特征在于，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前采集到的针对所述待测目标的信息，
    则，所述基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，包括：
    融合所述判断信息和所述在采集所述判断信息之前采集到的针对所述待测目标的信息中的特征；
    基于融合后的特征获取所述待测目标当前时刻的识别结果。
  3. 根据权利要求1所述的目标识别方法，其特征在于，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前获取到的识别结果，
    则，所述基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，包括：
    获取与所述判断信息对应的识别结果；
    从与所述判断信息对应的识别结果和在采集所述判断信息之前获取到的识别结果中选择属性类型优先级最高的识别结果作为所述待测目标当前时刻的识别结果。
  4. 根据权利要求1-3任一项所述的目标识别方法，其特征在于，所述识别结果为其中一种所述属性类型对应的判断结果，所述判断结果的置信度满足预设条件，并且，所述识别结果对应的属性类型在置信度满足所述预设条件的判断结果对应的属性类型中优先级最高。
  5. 根据权利要求1-3任一项所述的目标识别方法，其特征在于，所述输出所述识别结果的步骤之前，还包括：
    判断所述待测目标当前时刻的识别结果对应的属性类型的优先级是否高于所述待测目标上一时刻的识别结果对应的属性类型；
    若是，则输出所述识别结果。
  6. 根据权利要求1-3任一项所述的目标识别方法，其特征在于，所述输出所述识别结果的步骤之后，还包括：
    发送与所述识别结果对应的交互信号。
  7. 一种目标识别装置，应用于智能终端，其特征在于，包括：
    信息采集单元，用于将预设时间段内采集到的针对待测目标的信息作为判断信息，所述待测目标包括至少两种属性类型，所述至少两种属性类型之间设置有优先级关系；
    识别单元，包括：识别模块和输出模块，所述识别模块用于基于所述判断信息以及在采集所述判断信息之前获得的数据，获取所述待测目标当前时刻的识别结果，所述输出模块用于输出所述识别结果，所述识别结果与其中一种所述属性类型对应；
    判断单元，用于判断所述识别结果对应的属性类型是否为所述至少两种属性类型中优先级最高的属性类型；
    若否，则控制所述信息采集单元将下一预设时间段内采集到的针对所述待测目标的信息作为所述判断信息发送至所述识别单元。
  8. 根据权利要求7所述的目标识别装置，其特征在于，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前采集到的针对所述待测目标的信息，
    则，所述识别模块具体用于：
    融合所述判断信息和所述在采集所述判断信息之前采集到的针对所述待测目标的信息中的特征；
    基于融合后的特征获取所述待测目标当前时刻的识别结果。
  9. 根据权利要求7所述的目标识别装置，其特征在于，所述在采集所述判断信息之前获得的数据包括：在采集所述判断信息之前获取到的识别结果，
    则，所述识别模块具体用于：
    获取与所述判断信息对应的识别结果；
    从与所述判断信息对应的识别结果和在采集所述判断信息之前获取到的识别结果中选择属性类型优先级最高的识别结果作为所述待测目标当前时刻的识别结果。
  10. 根据权利要求7-9任一项所述的目标识别装置，其特征在于，所述识别结果为其中一种所述属性类型对应的判断结果，所述判断结果的置信度满足预设条件，并且，所述识别结果对应的属性类型在置信度满足所述预设条件的判断结果对应的属性类型中优先级最高。
  11. 根据权利要求7-9任一项所述的目标识别装置，其特征在于，所述识别单元，还包括：
    判断模块，用于判断所述待测目标当前时刻的识别结果对应的属性类型的优先级是否高于所述待测目标上一时刻的识别结果对应的属性类型；
    若是，则通过所述输出模块输出所述识别结果。
  12. 根据权利要求7-9任一项所述的目标识别装置，其特征在于，所述目标识别装置还包括：
    交互单元，用于发送与所述识别结果对应的交互信号。
  13. 一种智能终端，其特征在于，包括：
    至少一个处理器；以及，
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如权利要求1-6任一项 所述的目标识别方法。
  14. 一种非暂态计算机可读存储介质，其特征在于，所述非暂态计算机可读存储介质存储有计算机可执行指令，所述计算机可执行指令用于使智能终端执行如权利要求1-6任一项所述的目标识别方法。
  15. 一种计算机程序产品，其特征在于，所述计算机程序产品包括存储在非暂态计算机可读存储介质上的计算机程序，所述计算机程序包括程序指令，当所述程序指令被智能终端执行时，使所述智能终端执行如权利要求1-6任一项所述的目标识别方法。
PCT/CN2017/101966 2017-09-15 2017-09-15 一种目标识别方法、装置和智能终端 WO2019051813A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
JP2020514952A JP7104779B2 (ja) 2017-09-15 2017-09-15 対象認識方法、装置及びインテリジェント端末
CN201780002585.2A CN108064389B (zh) 2017-09-15 2017-09-15 一种目标识别方法、装置和智能终端
PCT/CN2017/101966 WO2019051813A1 (zh) 2017-09-15 2017-09-15 一种目标识别方法、装置和智能终端
US16/818,837 US11036990B2 (en) 2017-09-15 2020-03-13 Target identification method and apparatus, and intelligent terminal

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/101966 WO2019051813A1 (zh) 2017-09-15 2017-09-15 一种目标识别方法、装置和智能终端

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/818,837 Continuation US11036990B2 (en) 2017-09-15 2020-03-13 Target identification method and apparatus, and intelligent terminal

Publications (1)

Publication Number Publication Date
WO2019051813A1 true WO2019051813A1 (zh) 2019-03-21

Family

ID=62142062

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/101966 WO2019051813A1 (zh) 2017-09-15 2017-09-15 一种目标识别方法、装置和智能终端

Country Status (4)

Country Link
US (1) US11036990B2 (zh)
JP (1) JP7104779B2 (zh)
CN (1) CN108064389B (zh)
WO (1) WO2019051813A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111581436A (zh) * 2020-03-30 2020-08-25 西安天和防务技术股份有限公司 目标识别方法、装置、计算机设备和存储介质

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109073537A (zh) * 2018-07-16 2018-12-21 深圳达闼科技控股有限公司 一种物质检测的方法、装置、终端和可读存储介质
CN109543569A (zh) * 2018-11-06 2019-03-29 深圳绿米联创科技有限公司 目标识别方法、装置、视觉传感器及智能家居***
CN111200534B (zh) * 2018-11-16 2022-01-18 中国电信股份有限公司 终端信息的确定方法和装置
CN114666169B (zh) * 2022-05-24 2022-08-12 杭州安恒信息技术股份有限公司 一种扫描探测类型的识别方法、装置、设备及介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110116685A1 (en) * 2009-11-16 2011-05-19 Sony Corporation Information processing apparatus, setting changing method, and setting changing program
US20120139950A1 (en) * 2010-12-01 2012-06-07 Sony Ericsson Mobile Communications Japan, Inc. Display processing apparatus
US20130262565A1 (en) * 2012-03-27 2013-10-03 Sony Corporation Server, client terminal, system, and storage medium
CN105844283A (zh) * 2015-01-16 2016-08-10 阿里巴巴集团控股有限公司 用于识别图像类目归属的方法、图像搜索方法及装置
CN106934817A (zh) * 2017-02-23 2017-07-07 中国科学院自动化研究所 基于多属性的多目标跟踪方法及装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7711145B2 (en) * 2006-01-27 2010-05-04 Eastman Kodak Company Finding images with multiple people or objects
US8315463B2 (en) * 2006-11-14 2012-11-20 Eastman Kodak Company User interface for face recognition
JP5697139B2 (ja) 2009-11-25 2015-04-08 Kddi株式会社 ２次コンテンツ提供システムおよび方法
WO2012176317A1 (ja) 2011-06-23 2012-12-27 サイバーアイ・エンタテインメント株式会社 画像認識システムを組込んだ関連性検索によるインタレスト・グラフ収集システム
US9055276B2 (en) * 2011-07-29 2015-06-09 Apple Inc. Camera having processing customized for identified persons
US20130027569A1 (en) * 2011-07-29 2013-01-31 Kenneth Alan Parulski Camera having processing customized for recognized persons
CN102521558B (zh) * 2011-11-18 2014-04-02 深圳市宝捷信科技有限公司 一种条码识别方法及装置
WO2018032970A1 (zh) * 2016-08-19 2018-02-22 腾讯科技（深圳）有限公司 基于虚拟现实场景的认证方法、虚拟现实设备及存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110116685A1 (en) * 2009-11-16 2011-05-19 Sony Corporation Information processing apparatus, setting changing method, and setting changing program
US20120139950A1 (en) * 2010-12-01 2012-06-07 Sony Ericsson Mobile Communications Japan, Inc. Display processing apparatus
US20130262565A1 (en) * 2012-03-27 2013-10-03 Sony Corporation Server, client terminal, system, and storage medium
CN105844283A (zh) * 2015-01-16 2016-08-10 阿里巴巴集团控股有限公司 用于识别图像类目归属的方法、图像搜索方法及装置
CN106934817A (zh) * 2017-02-23 2017-07-07 中国科学院自动化研究所 基于多属性的多目标跟踪方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111581436A (zh) * 2020-03-30 2020-08-25 西安天和防务技术股份有限公司 目标识别方法、装置、计算机设备和存储介质
CN111581436B (zh) * 2020-03-30 2024-03-22 西安天和防务技术股份有限公司 目标识别方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
CN108064389A (zh) 2018-05-22
US20200218897A1 (en) 2020-07-09
CN108064389B (zh) 2019-02-01
JP2020535501A (ja) 2020-12-03
JP7104779B2 (ja) 2022-07-21
US11036990B2 (en) 2021-06-15

Similar Documents

Publication Publication Date Title
WO2019051813A1 (zh) 一种目标识别方法、装置和智能终端
US11501781B2 (en) Methods and systems for passive wakeup of a user interaction device
US20220020127A1 (en) Dynamic image recognition model updates
WO2019051814A1 (zh) 一种目标识别方法、装置和智能终端
WO2020042993A1 (zh) 语音控制方法、装置及***
WO2022105243A1 (zh) 事件检测方法、装置、电子设备及存储介质
WO2019056267A1 (zh) 一种分级交互决策方法、交互终端以及云端服务器
US10991372B2 (en) Method and apparatus for activating device in response to detecting change in user head feature, and computer readable storage medium
TW201935309A (zh) 入離場狀態檢測方法和裝置
US11120707B2 (en) Cognitive snapshots for visually-impaired users
US11023717B2 (en) Method, apparatus, device and system for processing commodity identification and storage medium
US11620995B2 (en) Voice interaction processing method and apparatus
US20190290493A1 (en) Intelligent blind guide method and apparatus
WO2023201947A1 (en) Methods, systems, and storage media for task dispatch
US20190087381A1 (en) Method and apparatus for controlling serial peripheral interface of fingerprint sensor, and mobile terminal
CN109815405A (zh) 灰度分流方法与***
US20190346929A1 (en) Attention Levels in a Gesture Control System
CN109922311A (zh) 基于音视频联动的监控方法、装置、终端及存储介质
CN116824455A (zh) 事件检测方法、装置、设备和存储介质
CN116246402A (zh) 一种监控方法及装置
CN113721232B (zh) 目标对象检测方法、装置、电子设备及介质
CN116311546A (zh) 活体检测方法和***
US20190392821A1 (en) Artificial intelligence (ai) based voice response system etiquette
CN109711260A (zh) 疲劳状态的检测方法、终端设备及介质
CN113095110B (zh) 人脸数据动态入库的方法、装置、介质及电子设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17924921

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2020514952

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17924921

Country of ref document: EP

Kind code of ref document: A1