WO2022063270A1 - 人脸图像的属性特征的识别方法、装置及电子设备 - Google Patents

人脸图像的属性特征的识别方法、装置及电子设备 Download PDF

Info

Publication number
WO2022063270A1
WO2022063270A1 PCT/CN2021/120650 CN2021120650W WO2022063270A1 WO 2022063270 A1 WO2022063270 A1 WO 2022063270A1 CN 2021120650 W CN2021120650 W CN 2021120650W WO 2022063270 A1 WO2022063270 A1 WO 2022063270A1
Authority
WO
WIPO (PCT)
Prior art keywords
face image
quality
attribute feature
attribute
recognition result
Prior art date
Application number
PCT/CN2021/120650
Other languages
English (en)
French (fr)
Inventor
陶帅
Original Assignee
维沃移动通信有限公司
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 维沃移动通信有限公司 filed Critical 维沃移动通信有限公司
Publication of WO2022063270A1 publication Critical patent/WO2022063270A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • 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
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30168—Image quality inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G06T2207/30201—Face

Definitions

  • the present application belongs to the field of communication technologies, and in particular relates to a method, a device and an electronic device for identifying attribute features of a face image.
  • the electronic device can recognize the attribute characteristics of the shooting object when previewing the shooting object, and then perform specific optimization on the shooting object based on the attribute characteristics of the shooting object.
  • the electronic device updates the shooting preview image in real time, identifies the attribute feature based on the updated image, and then updates the identification result of the shooting object attribute in real time.
  • the inventor found that the prior art has at least the following problems: when the state of the photographed object changes, such as when the person bows his head or faces sideways during the photographing process, the recognition result may change, thereby As a result, the reliability of the recognition result of the attribute feature of the photographed object is low.
  • Embodiments of the present invention provide a method, device, and electronic device for identifying attribute features of a face image, to solve the problem of low reliability of identification results of attribute features of the shooting object when the state of the shooting object changes.
  • an embodiment of the present invention provides a method for identifying an attribute feature of a face image, the method comprising:
  • the quality of the face image is evaluated, and the quality evaluation result of the face image is obtained;
  • an embodiment of the present invention further provides a device for identifying attribute features of a face image, the device comprising:
  • an evaluation module configured to evaluate the quality of the face image when the face image is collected, and obtain a quality evaluation result of the face image
  • An obtaining module configured to obtain the target recognition result of the attribute feature of the face image based on the quality evaluation result.
  • embodiments of the present application provide an electronic device, the electronic device includes a processor, a memory, and a program or instruction stored on the memory and executable on the processor, the program or instruction being The processor implements the steps of the method according to the first aspect when executed.
  • an embodiment of the present application provides a readable storage medium, where a program or an instruction is stored on the readable storage medium, and when the program or instruction is executed by a processor, the steps of the method according to the first aspect are implemented .
  • an embodiment of the present application provides a chip, the chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is configured to run a program or an instruction, and implement the first aspect the method described.
  • the quality of the face image is evaluated, and a quality evaluation result of the face image is obtained; based on the quality evaluation result, the Object recognition results of attribute features of face images. Since the recognition result of the attribute feature of the face image is obtained based on the quality evaluation result of the face image, the recognition result of the attribute feature can be obtained based on the face image with higher quality, so that the recognition result of the attribute feature of the face image can be improved. accuracy.
  • FIG. 1 is a flowchart of a method for identifying attributes of a face image provided by an embodiment of the present invention
  • FIG. 2 is a structural diagram of a device for identifying attributes of a face image provided by an embodiment of the present invention
  • FIG. 3 is a structural diagram of an acquisition module in an apparatus for identifying attributes and features of a face image provided by an embodiment of the present invention
  • FIG. 4 is a structural diagram of an evaluation module in an apparatus for identifying attribute features of a face image provided by an embodiment of the present invention
  • FIG. 5 is a structural diagram of an electronic device provided by an embodiment of the present invention.
  • FIG. 1 is a flowchart of a method for identifying attribute features of a face image provided by an embodiment of the present invention. As shown in FIG. 1, the method includes the following steps:
  • Step 101 under the condition that a face image is collected, evaluate the quality of the face image, and obtain a quality evaluation result of the face image.
  • the electronic device may identify based on the obtained preview image, for example, identify each frame of the obtained image in real time, or obtain one frame of image for identification every preset number of frames.
  • the quality of the face image can be evaluated.
  • the characteristics of the face image can reflect the quality of the face image. For example, the quality of the face image with a large degree of head up, head down, side face, occlusion, blur, backlight and dark light is lower; , A face with basically no occlusion and suitable lighting is a high-quality face image.
  • the quality of face images can be evaluated based on the above principles.
  • Step 102 Based on the quality evaluation result, obtain a target recognition result of the attribute feature of the face image.
  • the attribute features of the face image can reflect the inherent characteristics of the face image, such as gender, age, race and so on.
  • the quality of the face image can be obtained based on the features.
  • the accuracy of the attribute features identified based on the face image is higher. Therefore, based on the characteristics of the face image and the quality evaluation results of the face image, the accuracy of the face image can be obtained.
  • the object recognition result of the attribute feature of the image is based on the characteristics of the face image and the quality evaluation results of the face image.
  • the attribute features are re-identified based on the face image as the target recognition result, or after the quality evaluation result of the face image is obtained, based on the quality evaluation result, the previous The identified attribute features are obtained, and at least part of the attribute features with high reliability are obtained as the target identification result.
  • a face image with higher quality indicated by the quality assessment result can be obtained, and attribute features can be identified based on the face image; or only when the quality assessment result indicates a person
  • the quality of the face image is higher than the preset quality
  • the attribute features of the face image are identified; according to the correlation between the quality evaluation result of the face image and the accuracy of attribute feature identification, the calculation accuracy of each attribute feature identification result can be calculated. and output the accuracy, which is convenient for users to control the accuracy by themselves.
  • the quality assessment result includes a quality score
  • the quality assessment result of the face image includes N quality scores for the face images in N frames of images
  • the quality assessment result is based on the quality assessment result.
  • the N quality scores include at least K quality scores whose quality scores are greater than or equal to the first threshold, obtain the face image based on at least K frames of images corresponding to the at least K quality scores The target recognition result of the attribute feature;
  • N is an integer greater than 1
  • K is an integer greater than 0
  • N ⁇ K is an integer greater than 1
  • the quality of the face image can be scored, and the quality evaluation result of the face image is determined according to the scoring result, and the quality score can be used to reflect the quality of the face image.
  • the electronic device can use preset rules to obtain target recognition results of attribute features.
  • the above N frames of images may be continuous or discontinuous multiple frames of images.
  • the electronic device may recognize the face image of each frame of the N frames of images to obtain N quality scores. Based on N quality scores, different rules are adopted:
  • the N quality scores include a score with a higher quality score, that is, if there are at least K quality scores with a quality score greater than or equal to the first threshold, it can be based on at least K frames corresponding to the at least K quality scores.
  • the image is used to identify the attribute feature of the face image, and the image whose quality score is lower than the first threshold can be directly discarded without identifying the attribute feature of the face image, so that the reliability of the identification result can be improved.
  • the value of K can be preset by the electronic device or set based on user operations. When the number of images with high quality scores reaches K or more than K, the attribute characteristics of the face image can be performed based on these images. Identify and reduce errors.
  • two thresholds are set for the face quality score, namely TH1 and TH2.
  • Faces with a quality score less than TH1 are of poor quality, and the face attribute recognition results at this time may change frequently; faces greater than TH1 and less than TH2 are of better quality, and the reliability and stability of the face attribute recognition results at this time are relatively high.
  • the quality of face images is low, and the recognition results of attribute features may change frequently.
  • the attribute features recognized for the first frame of face image can be obtained as the target recognition result, and the updating of the recognition result can be suspended.
  • the quality of the face images varies.
  • face images with a quality score lower than TH1 can be discarded, and a face image with a quality score greater than TH1 can be obtained, and attribute feature recognition is performed based on this part of the face images.
  • the reliability of the target recognition result of the attribute feature of the face image obtained in this way is improved.
  • the top K recognition results with higher scores may be obtained, and the average value fusion of the K recognition results may be performed.
  • obtaining the target recognition result of the attribute feature of the face image based on the at least K frames of images corresponding to the at least K quality scores including:
  • P is an integer greater than 1.
  • At least K frames of images with higher quality scores may be further processed, so as to obtain a recognition result of the attribute feature of the face image based on at least part of the K frame images, that is, the P frame images.
  • At least K quality scores corresponding to at least K frames of images may be sorted, and P quality scores ranked in the top P positions are obtained. Since the image quality of the face images corresponding to the P quality scores is relatively high, the attribute feature recognition of the face images may be performed based on the P frame images corresponding to the P quality scores. For example, 5 quality scores are sorted in descending order of scores, and the 5 quality scores ranked in the top 5 are obtained. Based on the 5 frames of images corresponding to the 5 quality scores, attribute feature recognition is performed on the face image.
  • processing can be performed based on the P recognition results to obtain the final recognition result of the attribute features of the final face image, that is, the target recognition result.
  • the average value of the P recognition results is fused to obtain the recognition result of the attribute feature; or if the P recognition results are the same, any recognition result is used as the attribute feature of the face image, and the P recognition results are used as the attribute feature of the face image.
  • the recognition results after the sorting are removed, and most of the remaining recognition results are used as the attribute features of the face image; or the attribute features are obtained in other ways for the P recognition results.
  • the attribute feature of the face image is determined based on the multiple recognition results obtained from the face image with the highest quality score ranking, which can further improve the reliability of the recognition result of the attribute feature of the face image.
  • the method further includes:
  • the second threshold is greater than the first threshold.
  • the reliability of the current recognition result is relatively high, and the recognition of the attribute features of the face image can be stopped. , and stop updating the target recognition result of the attribute feature of the face image, which can save energy consumption and improve the reliability of the recognition result of the face attribute feature.
  • evaluating the quality of the face image to obtain a quality evaluation result of the face image including:
  • the quality of the face image is evaluated, and the quality evaluation result of the face image is obtained;
  • obtaining the target recognition result of the attribute feature of the face image including:
  • the attribute feature of the face image is identified, and the target recognition result of the attribute feature of the face image is obtained.
  • the features of the face image can be identified, and the quality of the face image can be scored according to the identified features.
  • the integrity of the facial features in the face can be recognized.
  • the quality score of the face image is higher, indicating that the quality of the face image is better; otherwise, the quality score of the face image is higher. If it is lower, it indicates that the quality of the face image is poor.
  • the target recognition result of the attribute feature of the face image can be obtained based on the feature of the recognized face image and combined with the quality evaluation result. For example, based on the identified features, including hairstyle, face outline, facial features and other features, and combined with the integrity of the facial features, it is determined that the attribute feature of the face image is male.
  • the feature recognition of face image is realized based on deep neural network.
  • the recognition process of face image is mainly based on two modules: feature extraction module and face attribute category regression module, in which the main computational cost is in the feature extraction module.
  • the embodiment of the present application adds recognition of face quality assessment on the basis of face attribute feature recognition. Face quality assessment and face attribute recognition are implemented based on multi-task neural networks. Since face attribute recognition and face quality assessment share the feature extraction network, the face quality assessment module has almost no additional computational overhead.
  • the identification of attribute features based on the quality-assessed face image can improve the accuracy of attribute feature identification.
  • the method further includes:
  • At least one of the quality evaluation result of the face image and the target recognition result of the attribute feature of the face image is output.
  • the quality evaluation result of the face image is output, so that the user can obtain the reliability of the identification result of the attribute feature based on the quality evaluation result.
  • the quality evaluation result indicates that the quality of the face image is low, the user can adjust the pose to obtain a high-quality face image.
  • the target recognition result of the attribute feature of the face image is output, which is convenient for the user to obtain the recognition result. And because the recognition result is determined based on the quality evaluation result of the face image, the accuracy of the recognition result can be improved.
  • the face attribute recognition result when the quality of the face in the preview interface is continuously low, the face attribute recognition result does not change frequently; when the quality of the face fluctuates, the attribute feature is determined based on the face image with a higher quality score. , the recognition results of attribute features are more accurate and stable; when there are high-quality face images, the recognition of attribute features of face images is stopped, which not only ensures the high reliability of the recognition results, but also reduces unnecessary computational overhead.
  • the execution subject may be the device for recognizing the attribute feature of the face image, or the method for executing the recognizing device for the attribute feature of the face image.
  • Load the control module of the recognition method of the attribute feature of the face image Load the control module of the recognition method of the attribute feature of the face image.
  • the method for recognizing the attribute feature of the face image provided by the embodiment of the present application is described by taking the method for recognizing the attribute feature of the face image that is performed by the device for recognizing the attribute feature of the face image as an example.
  • FIG. 2 is a structural diagram of an apparatus for identifying attributes of a face image provided by an embodiment of the present invention.
  • the apparatus 200 for identifying attributes of a face image includes:
  • An evaluation module 201 configured to evaluate the quality of the human face image under the condition of collecting the human face image, and obtain the quality evaluation result of the human face image;
  • the obtaining module 202 is configured to obtain, based on the quality evaluation result, a target recognition result for the attribute feature of the face image.
  • the quality evaluation result includes a quality score
  • the quality evaluation result of the face image includes N quality scores for the face images in N frames of images, as shown in FIG. 3 .
  • the The acquisition module 202 includes:
  • the first obtaining sub-module 2021 is configured to obtain the target recognition result of the attribute feature of the face image based on the first frame image in the N frame images under the condition that the N quality scores are all less than the first threshold ;
  • the second obtaining sub-module 2022 is configured to, when the N quality scores include at least K quality scores whose quality scores are greater than or equal to the first threshold, based on at least K corresponding to the at least K quality scores frame image, and obtain the target recognition result of the attribute feature of the face image;
  • N is an integer greater than 1
  • K is an integer greater than 0
  • N ⁇ K is an integer greater than 1
  • the second acquisition submodule includes:
  • a sorting unit for sorting the at least K quality scores
  • a first obtaining unit configured to obtain, among the at least K quality scores, P quality scores that are ranked within a preset range
  • a second obtaining unit configured to obtain P identification results of the attribute features of the face image based on the P frame images corresponding to the P quality scores
  • a third obtaining unit configured to obtain the target recognition result of the attribute feature of the face image based on the P recognition results
  • P is an integer greater than 1.
  • the device further includes:
  • a stopping module configured to stop updating the target recognition result of the attribute feature of the face image when the P quality scores are all greater than or equal to the second threshold
  • the second threshold is greater than the first threshold.
  • the evaluation module 201 includes:
  • An evaluation submodule 2012 configured to evaluate the quality of the human face image according to the feature of the human face image, to obtain a quality evaluation result of the human face image
  • the acquisition module is specifically used for:
  • the attribute feature of the face image is identified, and the target recognition result of the attribute feature of the face image is obtained.
  • the apparatus 200 for identifying the attribute feature of a face image can implement the various processes implemented by the electronic device in the above method embodiments and achieve the same beneficial effects. To avoid repetition, details are not described here.
  • the device for identifying the attribute feature of the face image in the embodiment of the present application may be a device, or may be a component, an integrated circuit, or a chip in a terminal.
  • the apparatus may be a mobile electronic device or a non-mobile electronic device.
  • the mobile electronic device may be a mobile phone, a tablet computer, a notebook computer, a palmtop computer, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (personal digital assistant).
  • UMPC ultra-mobile personal computer
  • netbook or a personal digital assistant
  • non-mobile electronic devices can be servers, network attached storage (Network Attached Storage, NAS), personal computer (personal computer, PC), television (television, TV), teller machine or self-service machine, etc., this application Examples are not specifically limited.
  • Network Attached Storage NAS
  • personal computer personal computer, PC
  • television television
  • teller machine or self-service machine etc.
  • the device for identifying the attribute feature of the face image in the embodiment of the present application may be a device having an operating system.
  • the operating system may be an Android (Android) operating system, an ios operating system, or other possible operating systems, which are not specifically limited in the embodiments of the present application.
  • the device for recognizing the attribute feature of a face image provided by the embodiment of the present application can realize the various processes implemented by the device for recognizing the attribute feature of a face image in the method embodiment corresponding to FIG. 1 and achieve the same beneficial effect. To avoid repetition, here No longer.
  • an embodiment of the present application further provides an electronic device, including a processor, a memory, a program or an instruction stored in the memory and executable on the processor, and the program or instruction is executed by the processor to implement the above.
  • an electronic device including a processor, a memory, a program or an instruction stored in the memory and executable on the processor, and the program or instruction is executed by the processor to implement the above.
  • the electronic devices in the embodiments of the present application include the above-mentioned mobile electronic devices and non-mobile electronic devices.
  • FIG. 5 is a schematic diagram of a hardware structure of an electronic device implementing an embodiment of the present application.
  • the electronic device 500 includes but is not limited to: a radio frequency unit 501, a network module 502, an audio output unit 503, an input unit 504, a sensor 505, a display unit 506, a user input unit 507, an interface unit 508, a memory 509, and a processor 510, etc. part.
  • the electronic device 500 may also include a power supply (such as a battery) for supplying power to various components, and the power supply may be logically connected to the processor 510 through a power management system, so as to manage charging, discharging, and power management through the power management system. consumption management and other functions.
  • a power supply such as a battery
  • the structure of the electronic device shown in FIG. 5 does not constitute a limitation on the electronic device.
  • the electronic device may include more or less components than the one shown, or combine some components, or arrange different components, which will not be repeated here. .
  • processor 510 is used for:
  • the quality of the face image is evaluated to obtain the quality evaluation result of the face image
  • the recognition result of the attribute feature of the face image is obtained based on the quality evaluation result of the face image, the recognition result of the attribute feature can be obtained based on the face image with higher quality, so that the recognition result of the attribute feature of the face image can be improved. accuracy.
  • the quality evaluation result includes a quality score
  • the processor 510 executes the The quality assessment result is obtained, and the target recognition result of the attribute feature of the face image is obtained, including:
  • the N quality scores include at least K quality scores whose quality scores are greater than or equal to the first threshold, obtain the face image based on at least K frames of images corresponding to the at least K quality scores The target recognition result of the attribute feature;
  • N is an integer greater than 1
  • K is an integer greater than 0
  • N ⁇ K is an integer greater than 1
  • the processor 510 executes the at least K frames of images corresponding to the at least K quality scores to obtain the target recognition result of the attribute feature of the face image, including:
  • P is an integer greater than 1.
  • processor 510 is further configured to:
  • the second threshold is greater than the first threshold.
  • the processor 510 performs the evaluation on the quality of the face image to obtain a quality evaluation result of the face image, including:
  • the quality of the face image is evaluated, and the quality evaluation result of the face image is obtained;
  • the processor 510 executes the target recognition result based on the quality assessment result to obtain the attribute feature of the face image, including:
  • the attribute feature of the face image is identified, and the target recognition result of the attribute feature of the face image is obtained.
  • the input unit 504 may include a graphics processor (Graphics Processing Unit, GPU) 5041 and a microphone 5042. Such as camera) to obtain still pictures or video image data for processing.
  • the display unit 506 may include a display panel 5061, which may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like.
  • the user input unit 507 includes a touch panel 5071 and other input devices 5072 .
  • the touch panel 5071 is also called a touch screen.
  • the touch panel 5071 may include two parts, a touch detection device and a touch controller.
  • Other input devices 5072 may include, but are not limited to, physical keyboards, function keys (such as volume control keys, switch keys, etc.), trackballs, mice, and joysticks, which will not be repeated here.
  • Memory 509 may be used to store software programs as well as various data, including but not limited to application programs and operating systems.
  • the processor 510 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the like, and the modem processor mainly processes wireless communication. It can be understood that, the above-mentioned modulation and demodulation processor may not be integrated into the processor 510.
  • An embodiment of the present application further provides a readable storage medium, where a program or an instruction is stored on the readable storage medium, and when the program or instruction is executed by a processor, each of the above-mentioned embodiments of the method for recognizing an attribute feature of a face image is implemented process, and can achieve the same technical effect, in order to avoid repetition, it will not be repeated here.
  • the processor is the processor in the electronic device described in the foregoing embodiments.
  • the readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, and the like.
  • An embodiment of the present application further provides a chip, where the chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is used for running a program or an instruction to realize the attribute feature of the above-mentioned face image
  • the chip includes a processor and a communication interface
  • the communication interface is coupled to the processor
  • the processor is used for running a program or an instruction to realize the attribute feature of the above-mentioned face image
  • the chip mentioned in the embodiments of the present application may also be referred to as a system-on-chip, a system-on-chip, a system-on-a-chip, or a system-on-a-chip, or the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

本申请公开了一种人脸图像的属性特征的识别方法、装置及电子设备，属于通信技术领域，以解决当拍摄对象的状态发生变化时，导致拍摄对象的属性特征的识别结果的可靠度较低的问题。其中，该方法包括：在采集到人脸图像的情况下，对人脸图像的质量进行评估，得到人脸图像的质量评估结果；基于质量评估结果，获取对人脸图像的属性特征的目标识别结果。

Description

人脸图像的属性特征的识别方法、装置及电子设备
相关申请的交叉引用
本申请主张在2020年9月27日在中国提交的中国专利申请号No.202011030431.4的优先权，其全部内容通过引用包含于此。
技术领域
本申请属于通信技术领域，具体涉及一种人脸图像的属性特征的识别方法、装置及电子设备。
背景技术
随着拍摄技术的发展，用户对图像的拍摄效果提出了更高的要求。为了提高图像的拍摄效果，电子设备在对拍摄对象进行拍摄预览时，能够识别拍摄对象的属性特征，然后基于拍摄对象的属性特征对拍摄对象进行特定的优化。在拍摄预览过程中，电子设备实时更新拍摄预览图像，并基于更新的图像进行属性特征的识别，然后实时更新对拍摄对象属性的识别结果。
然而，在上述过程中，发明人发现现有技术至少存在以下问题：当拍摄对象的状态发生变化时，如人物在拍摄过程中低头、侧脸等时，则可能会导致识别结果发生改变，从而导致拍摄对象的属性特征的识别结果的可靠度较低。
发明内容
本发明实施例提供一种人脸图像的属性特征的识别方法、装置及电子设备，以解决当拍摄对象的状态发生变化时，导致拍摄对象的属性特征的识别结果的可靠度较低的问题。
为了解决上述技术问题，本发明是这样实现的：
第一方面，本发明实施例提供了一种人脸图像的属性特征的识别方法，该方法包括：
在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所 述人脸图像的质量评估结果；
基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
第二方面，本发明实施例还提供一种人脸图像的属性特征的识别装置，该装置包括：
评估模块，用于在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
获取模块，用于基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
第三方面，本申请实施例提供了一种电子设备，该电子设备包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如第一方面所述的方法的步骤。
第四方面，本申请实施例提供了一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如第一方面所述的方法的步骤。
第五方面，本申请实施例提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如第一方面所述的方法。
本申请实施例中，在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。由于人脸图像的属性特征的识别结果是基于对人脸图像的质量评估结果得到的，可以基于质量较高的人脸图像获得属性特征的识别结果，从而能够提高对人脸图像属性特征识别的准确性。
附图说明
图1是本发明实施例提供的人脸图像的属性特征的识别方法的流程图；
图2是本发明实施例提供的人脸图像的属性特征的识别装置的结构图；
图3是本发明实施例提供的人脸图像的属性特征的识别装置中的获取模块的结构图；
图4是本发明实施例提供的人脸图像的属性特征的识别装置中的评估模块的结构图；
图5是本发明实施例提供的电子设备的结构图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请的说明书和权利要求书中的术语“第一”、“第二”等是用于区别类似的对象，而不用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外，说明书以及权利要求中“和/或”表示所连接对象的至少其中之一，字符“/”，一般表示前后关联对象是一种“或”的关系。
下面结合附图，通过具体的实施例及其应用场景对本申请实施例提供的人脸图像的属性特征的识别方法进行详细地说明。
参见图1，图1是本发明实施例提供的人脸图像的属性特征的识别方法的流程图，如图1所示，包括以下步骤：
步骤101、在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果。
在对人脸图像进行拍摄预览时，电子设备可以基于获取的预览图像进行识别，例如，实时对获取的每一帧图像进行识别，或者每隔预设帧数获取一帧图像进行识别。基于人脸图像的特征，可以对人脸图像的质量进行评估。人脸图像的特征可以反映人脸图像的质量，例如，较大程度的仰头、低头、侧脸、遮挡、模糊、逆光和暗光的人脸图像的质量较低；而姿态正、成像清晰、基本无遮挡、光照合适的人脸为高质量的人脸图像。可以基于上述原则 对人脸图像的质量进行评估。
步骤102、基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
人脸图像的属性特征可以反映人脸图像的固有特征，如性别、年龄、人种等。在对人脸图像的特征进行识别时，可以基于特征获取人脸图像的质量。当人脸图像的质量越高时，基于该人脸图像识别到的属性特征的准确性更高，因此，可以基于人脸图像的特征并结合对人脸图像的质量评估结果，获取对人脸图像的属性特征的目标识别结果。具体可以是在得出人脸图像的质量评估结果后，基于人脸图像再次识别获取属性特征作为目标识别结果，还可以是在得出人脸图像的质量评估结果后，基于质量评估结果得出前已识别的属性特征，获取至少部分可靠性较高的属性特征作为目标识别结果。
例如，在获取到多个质量评估结果不同的人脸图像时，可以获取质量评估结果指示的质量较高的人脸图像，并基于该人脸图像识别属性特征；或者仅在质量评估结果指示人脸图像的质量高于预设质量时，对人脸图像的属性特征进行识别；还可以根据人脸图像的质量评估结果与属性特征识别准确性的关联性，针对每个属性特征识别结果计算准确度，并输出该准确度，便于用户自行对准确度进行把控。
可选的，所述质量评估结果包括质量评分，在所述人脸图像的质量评估结果包括对N帧图像中的人脸图像的N个质量评分的情况下，所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
在所述N个质量评分均小于第一阈值的情况下，基于所述N帧图像中的第一帧图像获取所述人脸图像的属性特征的目标识别结果；
在所述N个质量评分中包括质量评分大于或等于所述第一阈值的至少K个质量评分的情况下，基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果；
其中，N为大于1的整数，K为大于0的整数，且N≥K。
在该实施方式中，可以对人脸图像的质量进行评分，根据评分结果确定对人脸图像的质量评估结果，质量评分能够用于反映人脸图像的质量。针对质量评分较高和较低的人脸图像，电子设备可以采用预设的规则获取属性特 征的目标识别结果。
上述N帧图像可以是连续或者是不连续的多帧图像。电子设备可以针对N帧图像中的每帧图像的人脸图像进行识别，得到N个质量评分。基于N个质量评分，采取不同的规则：
规则一、当N个质量评分均小于第一阈值时，说明质量评分较低，电子设备基于第一帧图像获取人脸图像的属性特征的目标识别结果，从而可以减少识别结果发生频繁跳变导致的识别结果的可靠性差。
规则二、当N个质量评分中包括质量评分较高的评分，即存在质量评分大于或等于第一阈值的至少K个质量评分的情况下，可以基于该至少K个质量评分对应的至少K帧图像进行人脸图像的属性特征的识别，而对于质量评分低于第一阈值的图像可以直接丢弃而不进行人脸图像属性特征的识别，从而能够提高识别结果的可靠性。其中，K的值可以是电子设备预先设置的或者是基于用户操作设定的，当质量评分较高的图像的数量达到K个或大于K个时，可以基于这些图像进行人脸图像的属性特征识别，减少误差。
例如，将人脸质量评分设置两个阈值，分别为TH1和TH2。质量评分小于TH1的人脸质量差，此时的人脸属性识别结果可能发生频繁跳变；大于TH1且小于TH2的人脸质量较好，此时的人脸属性识别结果可靠性和稳定性较好；大于TH2的人脸质量极好，此时人脸属性识别结果可靠性极高。
当针对多帧人脸图像获取的多个质量评分均小于TH1，人脸图像的质量较低，属性特征的识别结果可能发生频繁跳变。可以获取针对第一帧人脸图像识别的属性特征作为目标识别结果，并暂停更新识别结果。
当针对多帧人脸图像获取的多个质量评分包括小于TH1的质量评分和大于TH1的质量评分时，人脸图像的质量存在高低不等的情况下。此时，可以摒弃质量评分低于TH1的人脸图像，而获取质量评分大于TH1的人脸图像，并基于这部分人脸图像进行属性特征识别。这样获得到的人脸图像属性特征的目标识别结果的可靠性提高。进一步地，可以在这部分人脸图像中，获取分数较高的前K个识别结果，并对该K个识别结果进行平均值融合。
本实施方式，根据人脸图像的质量评分的大小，采用不同的规则获取人脸图像的属性特征识别结果，能够提高处理灵活性，从而提高识别结果的可 靠性。
可选的，所述基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果，包括：
对所述至少K个质量评分进行排序；
在所述至少K个质量评分中，获取排序位于预设范围内的P个质量评分；
基于所述P个质量评分对应的P帧图像，获取对人脸图像的属性特征的P个识别结果；
基于所述P个识别结果，获取所述人脸图像的属性特征的目标识别结果；
其中，P为大于1的整数。
在该实施方式中，可以进一步对质量评分较高的至少K帧图像进行处理，从而基于该K帧图像中的至少部分图像，即P帧图像得到对人脸图像的属性特征的识别结果。
具体地，可以将至少K帧图像对应的至少K个质量评分进行排序，并获取排序位于前P位的P个质量评分。由于该P个质量评分对应的人脸图像的图像质量较高，可以基于该P个质量评分对应的P帧图像进行人脸图像的属性特征识别。例如，将5个质量评分按照分数从高到低排序，获取排序位于前5位的5个质量评分。基于该5个质量评分对应的5帧图像，对人脸图像进行属性特征识别。
在对P帧图像进行识别，得到P个识别结果后，可以基于P个识别结果进行处理，得到最终的人脸图像的属性特征的最终识别结果，即目标识别结果。
例如，对P个识别结果的平均值进行融合，得到属性特征的识别结果；或者在该P个识别结果相同的情况下，以任一个识别结果作为人脸图像的属性特征，而在P个识别结果不同的情况下，去掉排序较后的识别结果，按照剩余的相同的大多数识别结果作为人脸图像的属性特征；或者对P个识别结果按照其他方式获取属性特征。
本实施方式，基于质量评分排序靠前的人脸图像获取的多个识别结果，确定人脸图像的属性特征，能够进一步提高人脸图像的属性特征的识别结果的可靠性。
可选的，所述方法还包括：
在所述P个质量评分均大于或等于第二阈值的情况下，停止更新所述人脸图像的属性特征的目标识别结果；
其中，所述第二阈值大于所述第一阈值。
在该实施方式中，在P帧图像的P个质量评分较高，即均大于或等于第二阈值的情况下，当前的识别结果的可靠性较高，可以停止对人脸图像属性特征的识别，并停止更新人脸图像属性特征的目标识别结果，能够节约能耗，且提高人脸属性特征识别结果的可靠性。
可选的，所述对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果，包括：
识别所述人脸图像的特征；
根据所述人脸图像的特征，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
基于所述质量评估结果和所述人脸图像的特征，对所述人脸图像的属性特征进行识别，得到所述人脸图像的属性特征的目标识别结果。
在该实施方式中，在对人脸图像的质量进行评估时，可以识别人脸图像的特征，并根据识别的特征对人脸图像的质量评分。例如，可以识别人脸中的五官的完整度，当五官的完整度较高时，则人脸图像的质量评分较高，表示人脸图像的质量较好；反之，则人脸图像的质量评分较低，表示人脸图像的质量较差。
在对人脸图像的属性特征进行识别时，可以基于已识别的人脸图像的特征，并结合质量评估结果，获取人脸图像的属性特征的目标识别结果。例如，基于已经识别的特征，包括发型、脸轮廓，五官轮廓等特征，并结合五官的完整度，确定人脸图像的属性特征为男性。
人脸图像的特征识别基于深度神经网络实现，人脸图像的识别过程主要基于两个模块：特征提取模块和人脸属性类别回归模块，其中主要的计算开销在特征提取模块。本申请实施例在人脸属性特征识别的基础上，增加了一 个人脸质量评估的识别。人脸质量评估和人脸属性识别基于多任务神经网络实现，由于人脸属性识别和人脸质量评估共用特征提取网络，所以人脸质量评估模块几乎没有增加额外的计算开销。而基于质量评估后的人脸图像进行属性特征的识别，能够提高属性特征识别的准确度。
可选的，在所述获取对所述人脸图像的属性特征的目标识别结果之后，所述方法还包括：
输出对所述人脸图像的质量评估结果和对所述人脸图像的属性特征的目标识别结果中的至少一项。
本实施方式，输出对人脸图像的质量评估结果，便于用户基于质量评估结果获取属性特征的识别结果的可靠性。当质量评估结果指示人脸图像的质量较低时，用户可以调整姿态，以获得高质量的人脸图像。
输出人脸图像的属性特征的目标识别结果，便于用户获取该识别结果。且由于该识别结果基于人脸图像的质量评估结果确定，能够提高识别结果的准确性。
本申请实施例，在预览界面的人脸质量持续低时，人脸属性识别结果不会发生频繁跳变；人脸质量忽高忽低时，由于基于质量评分较高的人脸图像确定属性特征，属性特征的识别结果更加准确、稳定；当存在高质量人脸图像时，停止识别人脸图像的属性特征，在保证了识别结果的高可靠性的同时，又减少了不必要的计算开销。
需要说明的是，本申请实施例提供的人脸图像的属性特征的识别方法，执行主体可以为人脸图像的属性特征的识别装置，或者该人脸图像的属性特征的识别装置中的用于执行加载人脸图像的属性特征的识别方法的控制模块。本申请实施例中以人脸图像的属性特征的识别装置执行加载人脸图像的属性特征的识别方法为例，说明本申请实施例提供的人脸图像的属性特征的识别方法。
参见图2，图2是本发明实施例提供的人脸图像的属性特征的识别装置的结构图，如图2所示，人脸图像的属性特征的识别装置200包括：
评估模块201，用于在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
获取模块202，用于基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
可选的，所述质量评估结果包括质量评分，在所述人脸图像的质量评估结果包括对N帧图像中的人脸图像的N个质量评分的情况下，如图3所示，所述获取模块202包括：
第一获取子模块2021，用于在所述N个质量评分均小于第一阈值的情况下，基于所述N帧图像中的第一帧图像获取所述人脸图像的属性特征的目标识别结果；
第二获取子模块2022，用于在所述N个质量评分中包括质量评分大于或等于所述第一阈值的至少K个质量评分的情况下，基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果；
其中，N为大于1的整数，K为大于0的整数，且N≥K。
可选的，所述第二获取子模块包括：
排序单元，用于对所述至少K个质量评分进行排序；
第一获取单元，用于在所述至少K个质量评分中，获取排序位于预设范围内的P个质量评分；
第二获取单元，用于基于所述P个质量评分对应的P帧图像，获取对人脸图像的属性特征的P个识别结果；
第三获取单元，用于基于所述P个识别结果，获取所述人脸图像的属性特征的目标识别结果；
其中，P为大于1的整数。
可选的，所述装置还包括：
停止模块，用于在所述P个质量评分均大于或等于第二阈值的情况下，停止更新所述人脸图像的属性特征的目标识别结果；
其中，所述第二阈值大于所述第一阈值。
可选的，如图4所示，所述评估模块201包括：
识别子模块2011，用于识别所述人脸图像的特征；
评估子模块2012，用于根据所述人脸图像的特征，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
所述获取模块具体用于：
基于所述质量评估结果和所述人脸图像的特征，对所述人脸图像的属性特征进行识别，得到所述人脸图像的属性特征的目标识别结果。
人脸图像的属性特征的识别装置200能够实现上述方法实施例中电子设备实现的各个过程以及达到相同的有益效果，为避免重复，这里不再赘述。
本申请实施例中的人脸图像的属性特征的识别装置可以是装置，也可以是终端中的部件、集成电路、或芯片。该装置可以是移动电子设备，也可以为非移动电子设备。示例性的，移动电子设备可以为手机、平板电脑、笔记本电脑、掌上电脑、车载电子设备、可穿戴设备、超级移动个人计算机(ultra-mobile personal computer，UMPC)、上网本或者个人数字助理(personal digital assistant，PDA)等，非移动电子设备可以为服务器、网络附属存储器(Network Attached Storage，NAS)、个人计算机(personal computer，PC)、电视机(television，TV)、柜员机或者自助机等，本申请实施例不作具体限定。
本申请实施例中的人脸图像的属性特征的识别装置可以为具有操作系统的装置。该操作系统可以为安卓(Android)操作系统，可以为ios操作系统，还可以为其他可能的操作系统，本申请实施例不作具体限定。
本申请实施例提供的人脸图像的属性特征的识别装置能够实现图1对应的方法实施例中人脸图像的属性特征的识别装置实现的各个过程以及达到相同的有益效果，为避免重复，这里不再赘述。
可选的，本申请实施例还提供一种电子设备，包括处理器，存储器，存储在存储器上并可在所述处理器上运行的程序或指令，该程序或指令被处理器执行时实现上述人脸图像的属性特征的识别方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
需要注意的是，本申请实施例中的电子设备包括上述所述的移动电子设 备和非移动电子设备。
图5为实现本申请实施例的一种电子设备的硬件结构示意图。
该电子设备500包括但不限于：射频单元501、网络模块502、音频输出单元503、输入单元504、传感器505、显示单元506、用户输入单元507、接口单元508、存储器509、以及处理器510等部件。
本领域技术人员可以理解，电子设备500还可以包括给各个部件供电的电源(比如电池)，电源可以通过电源管理系统与处理器510逻辑相连，从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。图5中示出的电子设备结构并不构成对电子设备的限定，电子设备可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置，在此不再赘述。
其中，处理器510用于：
在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
由于人脸图像的属性特征的识别结果是基于对人脸图像的质量评估结果得到的，可以基于质量较高的人脸图像获得属性特征的识别结果，从而能够提高对人脸图像属性特征识别的准确性。
可选的，所述质量评估结果包括质量评分，在所述人脸图像的质量评估结果包括对N帧图像中的人脸图像的N个质量评分的情况下，处理器510执行所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
在所述N个质量评分均小于第一阈值的情况下，基于所述N帧图像中的第一帧图像获取所述人脸图像的属性特征的目标识别结果；
在所述N个质量评分中包括质量评分大于或等于所述第一阈值的至少K个质量评分的情况下，基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果；
其中，N为大于1的整数，K为大于0的整数，且N≥K。
可选的，处理器510执行所述基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果，包括：
对所述至少K个质量评分进行排序；
在所述至少K个质量评分中，获取排序位于预设范围内的P个质量评分；
基于所述P个质量评分对应的P帧图像，获取对人脸图像的属性特征的P个识别结果；
基于所述P个识别结果，获取所述人脸图像的属性特征的目标识别结果；
其中，P为大于1的整数。
可选的，处理器510还用于：
在所述P个质量评分均大于或等于第二阈值的情况下，停止更新所述人脸图像的属性特征的目标识别结果；
其中，所述第二阈值大于所述第一阈值。
可选的，处理器510执行所述对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果，包括：
识别所述人脸图像的特征；
根据所述人脸图像的特征，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
处理器510执行所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
基于所述质量评估结果和所述人脸图像的特征，对所述人脸图像的属性特征进行识别，得到所述人脸图像的属性特征的目标识别结果。
应理解的是，本申请实施例中，输入单元504可以包括图形处理器(Graphics Processing Unit，GPU)5041和麦克风5042，图形处理器5041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元506可包括显示面板5061，可以采用液晶显示器、有机发光二极管等形式来配置显示面板5061。用户输入单元507包括触控面板5071以及其他输入设备5072。触控面板5071，也称为触摸屏。触控面板5071可包括触摸检测装置和触摸控制器两个部分。其他输 入设备5072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆，在此不再赘述。存储器509可用于存储软件程序以及各种数据，包括但不限于应用程序和操作系统。处理器510可集成应用处理器和调制解调处理器，其中，应用处理器主要处理操作系统、用户界面和应用程序等，调制解调处理器主要处理无线通信。可以理解的是，上述调制解调处理器也可以不集成到处理器510中。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有程序或指令，该程序或指令被处理器执行时实现上述人脸图像的属性特征的识别方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
其中，所述处理器为上述实施例中所述的电子设备中的处理器。所述可读存储介质，包括计算机可读存储介质，如计算机只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等。
本申请实施例另提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现上述人脸图像的属性特征的识别方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
应理解，本申请实施例提到的芯片还可以称为系统级芯片、系统芯片、芯片系统或片上系统芯片等。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外，需要指出的是，本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能，还可 包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能，例如，可以按不同于所描述的次序来执行所描述的方法，并且还可以添加、省去、或组合各种步骤。另外，参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述，但是本申请并不局限于上述的具体实施方式，上述的具体实施方式仅仅是示意性的，而不是限制性的，本领域的普通技术人员在本申请的启示下，在不脱离本申请宗旨和权利要求所保护的范围情况下，还可做出很多形式，均属于本申请的保护之内。

Claims (12)

  1. 一种人脸图像的属性特征的识别方法，包括：
    在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
    基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
  2. 根据权利要求1所述的方法，其中，所述质量评估结果包括质量评分，在所述人脸图像的质量评估结果包括对N帧图像中的人脸图像的N个质量评分的情况下，所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
    在所述N个质量评分均小于第一阈值的情况下，基于所述N帧图像中的第一帧图像获取所述人脸图像的属性特征的目标识别结果；
    在所述N个质量评分中包括质量评分大于或等于所述第一阈值的至少K个质量评分的情况下，基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果；
    其中，N为大于1的整数，K为大于0的整数，且N≥K。
  3. 根据权利要求2所述的方法，其中，所述基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果，包括：
    对所述至少K个质量评分进行排序；
    在所述至少K个质量评分中，获取排序位于预设范围内的P个质量评分；
    基于所述P个质量评分对应的P帧图像，获取对人脸图像的属性特征的P个识别结果；
    基于所述P个识别结果，获取所述人脸图像的属性特征的目标识别结果；
    其中，P为大于1的整数。
  4. 根据权利要求3所述的方法，还包括：
    在所述P个质量评分均大于或等于第二阈值的情况下，停止更新所述人脸图像的属性特征的目标识别结果；
    其中，所述第二阈值大于所述第一阈值。
  5. 根据权利要求1所述的方法，其中，所述对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果，包括：
    识别所述人脸图像的特征；
    根据所述人脸图像的特征，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
    所述基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果，包括：
    基于所述质量评估结果和所述人脸图像的特征，对所述人脸图像的属性特征进行识别，得到所述人脸图像的属性特征的目标识别结果。
  6. 一种人脸图像的属性特征的识别装置，包括：
    评估模块，用于在采集到人脸图像的情况下，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
    获取模块，用于基于所述质量评估结果，获取对所述人脸图像的属性特征的目标识别结果。
  7. 根据权利要求6所述的装置，其中，所述质量评估结果包括质量评分，在所述人脸图像的质量评估结果包括对N帧图像中的人脸图像的N个质量评分的情况下，所述获取模块包括：
    第一获取子模块，用于在所述N个质量评分均小于第一阈值的情况下，基于所述N帧图像中的第一帧图像获取所述人脸图像的属性特征的目标识别结果；
    第二获取子模块，用于在所述N个质量评分中包括质量评分大于或等于所述第一阈值的至少K个质量评分的情况下，基于所述至少K个质量评分对应的至少K帧图像，获取所述人脸图像的属性特征的目标识别结果；
    其中，N为大于1的整数，K为大于0的整数，且N≥K。
  8. 根据权利要求7所述的装置，其中，所述第二获取子模块包括：
    排序单元，用于对所述至少K个质量评分进行排序；
    第一获取单元，用于在所述至少K个质量评分中，获取排序位于预设范围内的P个质量评分；
    第二获取单元，用于基于所述P个质量评分对应的P帧图像，获取对人脸图像的属性特征的P个识别结果；
    第三获取单元，用于基于所述P个识别结果，获取所述人脸图像的属性特征的目标识别结果；
    其中，P为大于1的整数。
  9. 根据权利要求8所述的装置，还包括：
    停止模块，用于在所述P个质量评分均大于或等于第二阈值的情况下，停止更新所述人脸图像的属性特征的目标识别结果；
    其中，所述第二阈值大于所述第一阈值。
  10. 根据权利要求6所述的装置，其中，所述评估模块包括：
    识别子模块，用于识别所述人脸图像的特征；
    评估子模块，用于根据所述人脸图像的特征，对所述人脸图像的质量进行评估，得到所述人脸图像的质量评估结果；
    所述获取模块具体用于：
    基于所述质量评估结果和所述人脸图像的特征，对所述人脸图像的属性特征进行识别，得到所述人脸图像的属性特征的目标识别结果。
  11. 一种电子设备，包括处理器，存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求1至5任一项所述的人脸图像的属性特征的识别方法的步骤。
  12. 一种可读存储介质，其中，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如权利要求1至5任一项所述的人脸图像的属性特征的识别方法的步骤。
PCT/CN2021/120650 2020-09-27 2021-09-26 人脸图像的属性特征的识别方法、装置及电子设备 WO2022063270A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011030431.4 2020-09-27
CN202011030431.4A CN112150444A (zh) 2020-09-27 2020-09-27 人脸图像的属性特征的识别方法、装置及电子设备

Publications (1)

Publication Number Publication Date
WO2022063270A1 true WO2022063270A1 (zh) 2022-03-31

Family

ID=73894907

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/120650 WO2022063270A1 (zh) 2020-09-27 2021-09-26 人脸图像的属性特征的识别方法、装置及电子设备

Country Status (2)

Country Link
CN (1) CN112150444A (zh)
WO (1) WO2022063270A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112150444A (zh) * 2020-09-27 2020-12-29 维沃移动通信（杭州）有限公司 人脸图像的属性特征的识别方法、装置及电子设备
CN113591767A (zh) * 2021-08-09 2021-11-02 浙江大华技术股份有限公司 图像识别评估值的确定方法及装置、存储介质、电子装置
CN113869357A (zh) * 2021-08-17 2021-12-31 浙江大华技术股份有限公司 属性类别识别方法、设备及计算机存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108875542A (zh) * 2018-04-04 2018-11-23 北京旷视科技有限公司 一种人脸识别方法、装置、系统及计算机存储介质
CN110163171A (zh) * 2019-05-27 2019-08-23 北京字节跳动网络技术有限公司 用于识别人脸属性的方法和装置
US20200029011A1 (en) * 2018-07-23 2020-01-23 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and image capture apparatus
CN112150444A (zh) * 2020-09-27 2020-12-29 维沃移动通信（杭州）有限公司 人脸图像的属性特征的识别方法、装置及电子设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108875542A (zh) * 2018-04-04 2018-11-23 北京旷视科技有限公司 一种人脸识别方法、装置、系统及计算机存储介质
US20200029011A1 (en) * 2018-07-23 2020-01-23 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and image capture apparatus
CN110163171A (zh) * 2019-05-27 2019-08-23 北京字节跳动网络技术有限公司 用于识别人脸属性的方法和装置
CN112150444A (zh) * 2020-09-27 2020-12-29 维沃移动通信（杭州）有限公司 人脸图像的属性特征的识别方法、装置及电子设备

Also Published As

Publication number Publication date
CN112150444A (zh) 2020-12-29

Similar Documents

Publication Publication Date Title
WO2022063270A1 (zh) 人脸图像的属性特征的识别方法、装置及电子设备
WO2021135601A1 (zh) 辅助拍照方法、装置、终端设备及存储介质
US9779527B2 (en) Method, terminal device and storage medium for processing image
WO2019120029A1 (zh) 智能调节屏幕亮度的方法、装置、存储介质及移动终端
CN109213732B (zh) 一种改善相册分类的方法、移动终端及计算机可读存储介质
CN111242273B (zh) 一种神经网络模型训练方法及电子设备
CN112004117B (zh) 视频播放方法及装置
CN105303149B (zh) 人物图像的展示方法和装置
WO2021092808A1 (zh) 网络模型的训练方法、图像的处理方法、装置及电子设备
CN112364799A (zh) 一种手势识别方法及装置
WO2021159935A1 (zh) 图像显示方法及相关产品
CN112214112A (zh) 参数调节方法及装置
WO2022111458A1 (zh) 图像拍摄方法和装置、电子设备及存储介质
CN110147742B (zh) 一种关键点定位方法、装置及终端
WO2022078241A1 (zh) 拍摄方法、装置及电子设备
WO2021254251A1 (zh) 输入显示方法、装置及电子设备
CN111405180A (zh) 拍照方法、装置、存储介质及移动终端
CN107168536A (zh) 试题搜索方法、试题搜索装置及电子终端
CN112532885A (zh) 防抖方法、装置及电子设备
US20170300514A1 (en) Method and terminal for implementing image sequencing
CN111134686A (zh) 一种人体病症确定方法、装置、存储介质及终端
CN112949467B (zh) 人脸检测方法、装置、电子设备和存储介质
CN112511890A (zh) 视频图像处理方法、装置及电子设备
CN106528149A (zh) 一种锁屏方法及装置
WO2022042502A1 (zh) 美颜功能开启方法、装置及电子设备

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21871639

Country of ref document: EP

Kind code of ref document: A1