• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A first deep learning model is trained based on general facial images. A second deep learning model is trained based on facial images extracted from general facial images. Automatic face detection is performed based on the first deep learning model trained to obtain a first prediction score and the second deep learning model trained to obtain a second prediction score. A prediction score result is generated based on the first prediction score and the second prediction score, and the prediction score result is compared with a threshold to determine an automatic face detection result for the extracted facial images.
    公开号:BR112019009219A2
    申请号:R112019009219
    申请日:2018-06-07
    公开日:2019-08-13
    发明作者:Ma Chenguang
    申请人:Alibaba Group Holding Ltd；
    IPC主号:
    专利说明:

    “METHOD FOR FACIAL RECOGNITION, APPLIANCE AND ELECTRONIC DEVICE” [001] This application claims priority for Chinese Patent Application No. 201710421333.5, filed on June 7, 2017, which is hereby incorporated by reference in its entirety.
    Field of Invention [002] The present application relates to the field of computer software technologies and, in particular, to a method, apparatus and electronic device for automatic face detection.
    Background of the Invention [003] An automatic face detection technology is used to determine whether the current user is the authentic user using facial recognition techniques to intercept counterfeit attacks, such as screen repeat attacks, printed photo attacks and security attacks. three-dimensional modeling.
    [004] Currently, automatic face detection technology can be classified into automatic intrusive face detection technology and non-intrusive automatic face detection technology. In intrusive automatic face detection technology, a user needs to cooperatively complete some specific live actions such as blinking, turning his head, or opening his mouth. When performing facial recognition based on the instructions given, the automatic detection module can determine whether an operator accurately completes the live operation and whether the operator is the authentic user. In non-intrusive automatic face detection technology, a user does not need to cooperatively complete a live action, so the user experience is better, but the technical complexity is higher. In addition, automatic detection is performed mainly depending on information about a single input frame image
    Petition 870190062179, of 07/04/2019, p. 8/52
    2/34 or information about other device sensors.
    [005] In the automatic non-intrusive face detection technology described in existing technology, supervised training is generally performed on a single model of deep learning using live and non-live facial images and then automatic prediction is performed face in the single entry frame image using the trained model.
    [006] However, this technical solution depends to a large extent on a type of training data for facial spoofing attacks and is limited by an objective condition of insufficient training data. It is difficult to fully extract a live face image feature. As a result, this model cannot fully express a live face feature, and the accuracy of an automatic face detection result is reduced.
    Brief Description of the Invention [007] The embodiments of the present application provide a method, apparatus and electronic device for automatic face detection to solve the following technical problems in the existing technology. In a technical solution based on a single deep learning model, it is difficult to fully extract a live face image feature. As a result, this model cannot fully express a live face feature, and the accuracy of an automatic face detection result is reduced.
    [008] To solve the technical problems described, the embodiments of this application are implemented as follows:
    [009] An embodiment of the present application provides an automatic face detection method, including: training a first deep learning model based on general facial images; train one
    Petition 870190062179, of 07/04/2019, p. 9/52
    3/34 according to a deep learning model based on extracted facial images cut from general facial images; and perform automatic face detection based on the first trained deep learning model and the second trained deep learning model.
    [0010] An embodiment of the present application provides an automatic face detection device, including: a training module, configured to: train a first model of deep learning based on general facial images; and train a second model of deep learning based on the extracted facial images cut from the general facial images; and a detection module, configured to perform automatic face detection based on the first trained deep learning model and the second trained deep learning model.
    [0011] At least one technical solution used in the embodiments of the present patent application can achieve the following beneficial effects. One such benefit is that more features of live face image are extracted. Compared to an existing technology model, the first trained deep learning model and the second trained deep learning model better express the live face feature together, thereby improving the accuracy of the automatic face detection result. Therefore, some or all of the problems in existing technology can be resolved.
    Brief Description of the Drawings [0012] In order to describe the technical solutions in embodiments of the present application or in the existing technology more clearly, the following briefly presents the accompanying drawings necessary to describe the embodiments or the existing technology. Apparently, the drawings attached in the following description only show some shapes
    Petition 870190062179, of 07/04/2019, p. 10/52
    4/34 of carrying out the present application, and a person with current knowledge in the art can still derive other drawings from these attached drawings without creative efforts.
    [0013] Figure 1 is a schematic flowchart illustrating an example of a model training stage in a solution of the present application;
    [0014] Figure 2 is a schematic flowchart illustrating an example of an automatic detection stage in a solution of the present application;
    [0015] Figure 3 is a schematic flowchart illustrating an automatic face detection method according to an embodiment of the present application;
    [0016] Figure 4 is a schematic diagram illustrating the comparison between a general facial image and a facial image extracted according to an embodiment of the present application;
    [0017] Figure 5 is a schematic structural diagram illustrating an automatic face detection apparatus corresponding to Figure 3 according to an embodiment of the present application; and [0018] Figure 6 is a flow chart illustrating an example of a method implemented by computer to determine the authenticity of the user with automatic face detection, in accordance with an embodiment of the present invention.
    Detailed Description of the Invention [0019] The embodiments of the present application provide a method, apparatus and electronic device for automatic face detection.
    [0020] To enable a technician in the subject to better understand the technical solutions in this application, the following describes clearly and completely the technical solutions in the ways of carrying out the
    Petition 870190062179, of 07/04/2019, p. 11/52
    5/34 this application with reference to the accompanying drawings in the embodiments of this application. Apparently, the described embodiments are merely a part and not all embodiments of the present application. All other embodiments obtained by a person skilled in the art based on the embodiments of this application without creative efforts must be encompassed within the protection of this application.
    [0021] All deep learning models in a solution of this application are based on a neural network. To facilitate the description, a central idea of the solution of the present application is described for the first time based on an example and with reference to Figure 1 and Figure
    2.
    [0022] In this example, the solution of the present application can be classified into a model training stage and an automatic detection stage.
    [0023] Figure 1 is a schematic flowchart illustrating an example of a model training stage in a solution of the present application. In a model training stage, two independent models of deep learning are trained using live and non-live samples (belonging to a training data set) on a facial image: a first model of deep learning and a second model of deep learning. An input image of the first deep learning model is a general facial image collected, and an input image of the second deep learning model can be a facial image extracted from the overall facial image. The first deep learning model and the second deep learning model can use different deep learning network structures (that is, a structure of a neural network in which a
    Petition 870190062179, of 07/04/2019, p. 12/52
    6/34 model is based). Different network structures are differently sensitive to different image characteristics. Live and non-live training data sets are used to complete the training of the first deep learning model and the second deep learning model based on a deep learning method.
    [0024] Figure 2 is a schematic flowchart illustrating an example of an automatic detection stage in a solution of the present application. In an automatic detection stage, a user's facial scan image is collected as a user's overall facial image, and a first deep learning model is inserted to obtain a PA prediction score. In addition, face detection is performed on the user's facial scan image, an extracted facial image is cut out of the user's facial scan image based on a detection result, and a second deep learning model is inserted into the image extracted face to obtain a PB prediction score. Subsequently, for example, a prediction score result of (PA + PB) can be compared with a given threshold (for example, the threshold can be 1), to make a joint decision to determine an automatic face detection result for the user's scan image.
    [0025] Based on the main idea described, the following describes the solution of the present application in detail.
    [0026] Figure 3 is a schematic flowchart illustrating an automatic face detection method according to an embodiment of the present application. From a program perspective, the procedure can be performed by a program on a server or a terminal, for example, an identity authentication program or an e-commerce application. From the point of view of a device, the procedure is performed by at least one of the following devices that
    Petition 870190062179, of 07/04/2019, p. 13/52
    7/34 can be used as a server or a terminal: an access control device, a personal computer, an average computer, a computer cluster, a cell phone, a tablet computer, a smart dressing device, an automobile machine or a point of sale (POS).
    [0027] The procedure in Figure 3 can include the following steps.
    [0028] (S301) Train a first model of deep learning based on general facial images.
    [0029] In this embodiment of the present application, the general facial images used to train the first deep learning model may include a plurality of samples. In the plurality of samples, some are live facial images that are collected by photographing a live face and that can be used as positive samples, and some are non-live facial images that are collected when photographing a non-live face, such as a photo. face or a face model, which can be used as negative samples.
    [0030] In this embodiment of the present application, the first deep learning model is a classification model, and general facial images are used as inputs to the classification model. After processing the model, general facial images can be classified into at least the live facial image category or the non-live facial image category. One goal of training the first deep learning model is to improve the accuracy of the classification of the first deep learning model.
    [0031] (S302) Train a second model of deep learning based on the extracted facial images cut out from the general facial images.
    [0032] In this embodiment of this application, in addition to
    Petition 870190062179, of 07/04/2019, p. 14/52
    8/34 a complete facial region, the general facial image usually includes some unrelated regions, such as a background region and a human body, except for one face. The extracted facial image can exclude unrelated regions, and can include at least one extracted facial region, for example, an entire facial region, an eye region or a nasal region. There may be one or more second models of deep learning, and each second model of deep learning can correspond to a type of facial regions.
    [0033] Figure 4 is a schematic diagram illustrating the comparison between a general facial image and a facial image extracted according to an embodiment of the present application.
    [0034] In Figure 4, (a) is a general facial image. For ease of understanding, an extracted facial image is marked in (a), using dashed lines, and (a) can be correspondingly cropped to obtain an extracted facial image shown in (b).
    [0035] Furthermore, when the extracted facial image is an image including only a partial facial region, the general facial image can also be an image including an entire facial region and, basically, excluding an unrelated region.
    [0036] In this embodiment of the present application, the extracted facial image used to train the second deep learning model can also include a variety of samples. In the sample range, some are live facial images that can be used as positive samples, and some are non-live facial images that can be used as negative samples.
    [0037] In this embodiment of the present application, the second model of deep learning is also a model of classification and the extracted facial images are used as input of the model of
    Petition 870190062179, of 07/04/2019, p. 15/52
    9/34 rating. After model processing, extracted facial images can be classified into at least the live facial image category or the non-live facial image category. One objective of training the second deep learning model is to improve the accuracy of the classification of the second deep learning model.
    [0038] In addition to being cut out from the general facial image, the extracted facial image can be obtained through special collection, without depending on the general facial image.
    [0039] In this embodiment of the present application, the first deep learning model and the second deep learning model can be different models or the same model before training.
    [0040] A sequence of execution of the step (S301) and the step (S302) is not limited in the present application and the step (S301) and the step (S302) can be carried out simultaneously or successively.
    [0041] (S303) Performs automatic face detection based on the first trained deep learning model and the second trained deep learning model.
    [0042] Each step of Figure 3 can be performed by the same device or by the same program, or it can be performed by different devices or different programs. For example, step (S301) to step (S303) are performed by a device (1). For another example, step (S301) and step (S302) are performed by a device (1) and step (S303) is performed by a device (2); etc.
    [0043] According to the method of Figure 3, more characteristics of live face image are extracted. Compared with an existing technology model, the first trained deep learning model and the second jointly trained deep learning model better express a live face feature, thereby improving the accuracy of
    Petition 870190062179, of 07/04/2019, p. 16/52
    10/34 an automatic face detection result. Therefore, some or all of the problems in existing technology can be resolved.
    [0044] Based on the method of Figure 3, this embodiment of the present application further provides some method specific embodiment solutions and an extension solution, which are described below.
    [0045] In this embodiment of the present application, to implement a difference between sensitivity of the first deep learning model to an image characteristic and sensitivity of the second deep learning model to an image characteristic, the first deep learning model and the according to the deep learning model they can preferably use different deep learning network structures.
    [0046] Different network structures of two deep learning models may indicate that the two deep learning models include one or more different network structure parameters. The network structure parameter can include, for example, a number of hidden variable layers, a type of hidden variable layer, a number of neuronal nodes, a number of input layer nodes, or a number of output layer nodes.
    [0047] Certainly, some specific models of deep learning can also include specific corresponding parameters. For example, for a deep learning model based on a convolutional neural network widely used in the field of imaging today, a size of a convolution nucleus of a convolution unit is also a network structure parameter specific to that deep learning model .
    [0048] For the solution of the present application, generally, the different deep learning network structures include at least
    Petition 870190062179, of 07/04/2019, p. 17/52
    11/34 one of the following parameters: a number of hidden variable layers, a type of hidden variable layer, a number of neuronal nodes or a size of a convolution nucleus of a convolution unit.
    [0049] In this embodiment of the present application, to improve the efficiency of model training and the reliability of model training, model training can be performed in a supervised training manner.
    [0050] For example, in a supervised training manner, for step (S301), the general facial image includes a first label, and the first label indicates whether a general facial image corresponding to the first label is a live facial image.
    [0051] Training a first deep learning model based on a general facial image may include: introducing the first deep learning model for the general facial image, where the first deep learning model extracts a feature from the general facial image, and predicts, based on the extracted feature, whether the overall facial image is a live facial image; and adjust the first deep learning model based on a prediction result and the first label of the overall facial image. Generally, when the prediction result is inconsistent with the first label, the first deep learning model is adjusted, so that the first adjusted deep learning model can obtain, through re-prediction, a prediction result consistent with the first label.
    [0052] The feature extracted by the first deep learning model in a training process may preferably include an image structure feature of the overall facial image, for example, a screen photo border or face distortion in the facial image general.
    Petition 870190062179, of 07/04/2019, p. 18/52
    12/34 [0053] For another example, similarly, in a supervised training manner, for step (S302), the extracted facial image includes a second label, and the second label indicates whether an extracted facial image corresponding to the second label is a live facial image.
    [0054] The training of a second model of deep learning based on the extracted facial images cut out from the general facial images may include: obtaining the extracted facial images cut out from the general facial images; apply the second deep learning model to the extracted facial image obtained, in which the second deep learning model extracts a characteristic from the extracted facial image, and predicts, based on the extracted characteristic, whether the extracted facial image is a live facial image ; and adjust the second model of deep learning based on a prediction result and the second label of the extracted facial image. Generally, when the prediction result is inconsistent with the second label, the second deep learning model is adjusted, so that the second adjusted deep learning model can obtain a prediction result consistent with the second label through new prediction.
    [0055] The characteristic extracted by the second model of deep learning in a training process may preferably include a characteristic of the image material of the extracted facial image, for example, blur, texture or color distortion in the extracted facial image.
    [0056] In the two examples described above, the first deep learning model and the second deep learning model are differently sensitive to different image characteristics. The first model of deep learning is more sensitive to the image structure characteristic, and the second model of deep learning is more sensitive to
    Petition 870190062179, of 07/04/2019, p. 19/52
    13/34 characteristic of image material. For a face image, the image structure characteristic is relatively global and generalized, and the image material characteristic is relatively local and refined.
    [0057] Therefore, the first trained deep learning model and the second trained deep learning model can jointly extract a facial image feature in a more hierarchical and abundant way, in order to make a joint decision to obtain an automatic detection result more accurate facial.
    [0058] In this embodiment of the present application, the corresponding training data sets and / or corresponding deep learning network structures are different, so that the first deep learning model and the second deep learning model can be differently sensitive to the different characteristics of the image.
    [0059] For example, if the first deep learning model and the second deep learning model are based on a convolutional neural network, a convolution nucleus of a convolutional unit in a convolutional neural network in which the first deep learning model is based can be relatively large, so that the first deep learning model extracts an image structure characteristic from the overall facial image. Correspondingly, a convolution core of a convolution unit in a convolutional neural network on which the second deep learning model is based can be relatively small, so that the second deep learning model extracts a material feature from the facial image. extracted. So, in this example, the convolution nucleus of the convolution unit in the convolutional neural network in which the first model
    Petition 870190062179, of 07/04/2019, p. 20/52
    14/34 deep learning is based is greater than the convolution nucleus of the convolution unit in the convolutional neural network that the second deep learning model is based on.
    [0060] It should be noted that the size of the convolution core is merely an example of a parameter that can affect sensitivity, and another parameter of the network structure can also affect sensitivity.
    [0061] In this embodiment of the present application, for step (S303), the first trained deep learning model and the second trained deep learning model jointly make the decision to carry out automatic face detection. There are a variety of ways to make specific decisions. For example, a separate decision is made separately using the first deep learning model and the second deep learning model, and then a final decision result is determined by the synthesis of all separate decision results. For another example, a separate decision can be made first using the first deep learning model and the second deep learning model. When a separate decision result satisfies a specific condition, the separate decision result can be used directly as a final decision result; otherwise, a decision is made comprehensively in combination with another remaining model, to obtain a final decision result; etc.
    [0062] If a first way described in the previous paragraph is used, an example is the following:
    [0063] For example, for step (S303), automatic face detection based on the first trained deep learning model and the second deep learning model can include: obtaining the overall facial image (which is usually a raster image of a user) collected for automatic face detection; introduce the first model of
    Petition 870190062179, of 07/04/2019, p. 21/52
    15/34 deep learning trained in the general facial image collected for processing, to obtain the first corresponding prediction data; obtain the extracted facial image cut from the collected general facial image, and introduce the second model of deep learning trained for processing, to obtain second corresponding prediction data; and making a joint decision based on the first prediction data and the second prediction data, to obtain an automatic face detection result for the user's facial scan image.
    [0064] The first prediction data can be, for example, the prediction score described PA, and the second prediction data can be, for example, the prediction score described PB. Certainly, the prediction score is merely an example of a form of expression of the first prediction data and the second prediction data, or there may be another form of expression, for example, a probability value or a Boolean value.
    [0065] The above is the automatic face detection method provided in this embodiment of the present application. As shown in Figure 5, based on the same idea of the invention, an embodiment of the present application further provides a corresponding apparatus.
    [0066] Figure 5 is a schematic structural diagram illustrating an automatic face detection apparatus corresponding to Figure 3 according to an embodiment of the present application. The device can be located in a body carrying out the procedure in Figure 3, including: a training module (501), configured to: train a first model of deep learning based on general facial images; and train a second model of deep learning based on the extracted facial images cut from the general facial images; and a detection module (502), configured to perform automatic detection of
    Petition 870190062179, of 07/04/2019, p. 22/52
    16/34 face based on the first trained deep learning model and the second trained deep learning model.
    [0067] Optionally, the first deep learning model and the second deep learning model use different deep learning network structures.
    [0068] Optionally, the different deep learning network structures include at least one of the following parameters: a number of hidden variable layers, a type of a hidden variable layer, a number of neuronal nodes or a size of a convolution core. a convolution unit.
    [0069] Optionally, the general facial image includes a first label, and the first label indicates whether a general facial image corresponding to the first label is a live facial image.
    [0070] Training, by the training module (501), a first deep learning model based on general facial images includes: introducing, by the training module (501), the first deep learning model for the general facial image, where the first deep learning model predicts, based on an image structure characteristic of the overall facial image, whether the overall facial image is a live facial image; and adjust the first deep learning model based on a prediction result and the first label of the overall facial image.
    [0071] Optionally, the extracted facial image includes a second label, and the second label indicates whether an extracted facial image corresponding to the second label is a live facial image.
    [0072] The training, by the training module (501), a second model of deep learning based on the extracted facial images cut out from the general facial images includes: obtaining, through the module
    Petition 870190062179, of 07/04/2019, p. 23/52
    17/34 of training (501), the extracted facial image cut out from the general facial image; and introducing the second model of deep learning in the extracted facial image, where the second model of deep learning predicts, based on a material characteristic of the extracted facial image, if the extracted facial image is a live facial image; and adjust the second model of deep learning based on a prediction result and the second label of the extracted facial image.
    [0073] Optionally, the first deep learning model and the second deep learning model are based on a convolutional neural network.
    [0074] A convolution nucleus of a convolution unit in a convolutional neural network in which the first deep learning model is based on and is larger than a convolution nucleus of a convolution unit in a convolutional neural network in which the second the deep learning model is based, so that the deep learning model extracts an image structure characteristic from the overall facial image, and the second deep learning model extracts a material image characteristic from the extracted facial image.
    [0075] Optionally, the realization, by the detection module (502), of automatic face detection based on the first trained deep learning model and the second trained deep learning model includes: obtaining, by the detection module (502), the general facial image collected to detect the automatic face; introduce the first trained deep learning model in the general facial image collected for processing, to obtain the first corresponding prediction data; obtain extracted facial image cut out from the collected general facial image, and insert the second model of deep learning trained for processing, to obtain the second prediction data
    Petition 870190062179, of 07/04/2019, p. 24/52
    Corresponding 18/34; and make a joint decision based on the first prediction data and the second prediction data, to obtain an automatic face detection result for a user's facial scan image.
    [0076] Based on the same idea of the invention, an embodiment of the present application further provides a corresponding electronic device, including: at least one processor; and a memory communicatively connected to at least one processor.
    [0077] Memory stores an instruction that can be executed by at least one processor, and the instruction is executed by at least one processor, to allow at least one processor: to train a first deep learning model based on general facial images ; train a second model of deep learning based on the extracted facial images cut out from the general facial images; and perform automatic face detection based on the first trained deep learning model and the second trained deep learning model.
    [0078] Based on the same idea of the disclosure, an embodiment of the present application further provides a corresponding non-volatile computer storage medium, wherein the non-volatile computer storage medium stores a computer executable instruction and the instruction computer executable is defined to: train a first deep learning model based on general facial images; train a second model of deep learning based on the extracted facial images cut out from the general facial images; and perform automatic face detection based on the first trained deep learning model and the second trained deep learning model.
    [0079] The embodiments in this specification are all described in a progressive manner, for equal parts or
    Petition 870190062179, of 07/04/2019, p. 25/52
    19/34 similar in the embodiments, reference can be made to these embodiments, and each embodiment is focused on a difference from other embodiments. In particular, an embodiment of the apparatus, an embodiment of the electronic device, an embodiment of the non-volatile computer storage medium are basically similar to an embodiment of the method and, therefore, briefly described; for related parties, reference is made to partial descriptions in the method's realization.
    [0080] The apparatus, the electronic device and the non-volatile computer storage medium provided in the embodiments of the present application correspond to the method. Therefore, the apparatus, the electronic device and the non-volatile computer storage medium also have beneficial technical effects similar to a beneficial technical effect of the corresponding method. The beneficial technical effect of the method is described in detail above, so that the beneficial technical effects of the corresponding apparatus, electronic device and non-volatile computer storage medium are not described here again.
    [0081] In the 90s, if technology improvement is hardware improvement (for example, improvement of a circuit structure, such as a diode, transistor or switch) or software improvement (improvement of a method procedure) can obviously be distinguished. However, as technologies develop, the improvement of many current method procedures can be seen as a direct improvement of a hardware circuit structure. A designer usually schedules an improved method procedure for a hardware circuit to obtain a corresponding hardware circuit structure. Therefore, a method procedure can be improved by hardware entity modules. For example, a programmable logic device (PLD) (for example,
    Petition 870190062179, of 07/04/2019, p. 26/52
    20/34 an FPGA (field programmable port arrangement) is an integrated circuit, and a logic function of the programmable logic device is determined by a user through device programming. The designer performs the programming to “integrate” a digital system into a PLD without asking a chip manufacturer to design and produce an application-specific integrated circuit chip. In addition, programming is mainly implemented by modifying the “logical compiler” software instead of manually making an integrated circuit chip. This is similar to a software compiler used to develop and compose a program. However, the original code obtained before compilation is also written in a specific programming language, and this is referred to as the Hardware Description Language, HDL. However, there are several HDLs, such as an ABEL (Advanced Boolean Expression Language), an AHDL (Altera Hardware Description Language), Confluence, a CUPL (Cornell University Programming Language), HDCal, a JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM and a RHDL (Ruby Hardware Description Language). Currently, a VHDL (High Speed Integrated Circuit Hardware Description Language) and Verilog are the most popular. A person skilled in the art should also understand that only logical programming needs to be performed in the method procedure using the various hardware description languages described, and the various hardware description languages are programmed into an integrated circuit, so that a circuit hardware that implements the logical method procedure can be easily obtained.
    [0082] A controller can be implemented in any appropriate way. For example, the controller can use a microprocessor or a processor and can store shapes of a computer-readable medium, a logic port, a switch, an integrated circuit
    Petition 870190062179, of 07/04/2019, p. 27/52
    21/34 application-specific (ASIC), a programmable logic controller and an embedded microcontroller that are computer-readable program code (for example, software or hardware) that can be executed by the (micro) processor. Examples of controllers include, but are not limited to, the following micro controllers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 or Silicone Labs C8051F320. A memory controller can also be implemented as part of the logic control of memory. A person skilled in the art also knows that, in addition to implementing the controller in a pure computer-readable program code manner, logical programming can be completely performed using the method step, so that the controller implements the same function in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, a built-in microcontroller, etc. Therefore, the controller can be considered as a hardware component, and an apparatus for implementing various functions in the controller can also be considered as a structure in a hardware component. Alternatively, a device configured to implement several functions can be considered as a software module or a structure in a hardware component that can implement the method.
    [0083] The system, apparatus, module or unit described in the described embodiments can be implemented by a computer chip or an entity, or implemented by a product with a function. A typical deployment device is a computer. Specifically, the computer can be, for example, a personal computer, a portable computer, a mobile phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an e- game console, a tablet computer, or a wearable device, or a
    Petition 870190062179, of 07/04/2019, p. 28/52
    22/34 combination of any of these devices.
    [0084] For ease of description, the device described is described by dividing functions into several units. Certainly, when the present application is implemented, the functions of each unit can be implemented in one or more pieces of software and / or hardware.
    [0085] A person skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system or a computer program product. Accordingly, the present invention can use one form of hardware-only embodiments, only software-based embodiments, or embodiments with a combination of software and hardware. In addition, the present invention can use a form of computer program product that is implemented on one or more storage media usable per computer (including but not limited to disk memory, CDROM, optical memory, etc.) that include program code usable by computer.
    [0086] The present description is described with reference to the flowcharts and / or block diagrams of the method, the device (system) and the computer program product according to the embodiments of the present invention. It should be understood that computer program instructions can be used to implement each process and / or each block in flowcharts and / or block diagrams and a combination of a process and / or a block in flowcharts and / or block diagrams . These computer program instructions can be provided for a general purpose computer, a dedicated computer, an embedded processor or a processor of any other programmable data processing device to generate a machine, so that instructions executed by a computer or a any processor
    Petition 870190062179, of 07/04/2019, p. 29/52
    23/34 another programmable data processing device generates a device to implement a specific function in one or more processes in flowcharts or in one or more blocks in block diagrams.
    [0087] These computer program instructions can be stored in a computer-readable memory that can instruct the computer or any other programmable data processing device to function in a specific way, so that the instructions stored in the computer-readable memory generate an artifact that includes an instructional apparatus. The instruction apparatus implements a specific function in one or more processes in the flowcharts and / or in one or more blocks in the block diagrams.
    [0088] These computer program instructions can be loaded onto a computer or other programmable data processing device, so that a series of operations and steps are performed on the computer or on another programmable device, thereby generating computer-implemented processing. Therefore, instructions executed on the computer or another programmable device provide steps to implement a specific function in one or more processes in flowcharts or in one or more blocks in block diagrams.
    [0089] In the typical configuration, the computing device includes one or more processors (CPU), an input / output interface, a network interface and a memory.
    [0090] The memory can include a form of volatile memory, a random access memory (RAM) and / or a non-volatile memory, etc. in a computer-readable medium, such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
    [0091] The computer-readable medium includes volatile and non-volatile media
    Petition 870190062179, of 07/04/2019, p. 30/52
    24/34 volatile, removable and non-removable, and can store information using any method or technology. The information can be a computer-readable instruction, a data structure, a program module or other data. Examples of computer storage media include but are not limited to a phase change random access memory (PRAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a memory random access (RAM) of another type, a read-only memory (ROM), a programmable and electronically erasable read-only memory (EEPROM), a flash memory or other memory technology, a compact disk read memory (CD-ROM) , a versatile digital disc (DVD) or other optical storage, magnetic tape, magnetic disk storage, other magnetic storage device or any other medium without transmission. The computer's storage medium can be used to store information that can be accessed by the computing device. As described in this specification, the computer-readable medium does not include transient means (transient means), for example, a modulated data signal and a vehicle.
    [0092] It should also be noted that the terms "include", "contain" or any other variant are intended to include non-exclusive inclusion, so that a process, method, article or device that includes a series of elements not only does it include these same elements, but it also includes other elements that are not expressly listed, or it also includes elements inherent to such a process, method, article or device. An element preceded by "includes a ...", without further restrictions, does not prevent the existence of additional identical elements in the process, method, article or device that includes the element.
    [0093] The present invention can be described in contexts
    Petition 870190062179, of 07/04/2019, p. 31/52
    25/34 common computer-executable instructions executed by a computer, such as a program module. Generally, the program module includes a routine, program, object, component, data structure, etc., performing a specific task or implementing a specific abstract data type. The present application can also be practiced in distributed computing environments. In these distributed computing environments, tasks are performed by remote processing devices that are connected via a communication network. In distributed computing environments, the program module can be located on local and remote computer storage media that includes storage devices.
    [0094] The embodiments in this specification are all progressively described, for equal or similar parts in the embodiments, reference can be made to these embodiments, and each embodiment is focused on a difference from other embodiments. In particular, an embodiment of the system is basically similar to an embodiment of the method and is therefore briefly described; for related parties, reference can be made to partial descriptions in the method's realization.
    [0095] The foregoing descriptions are merely embodiments of this application and are not intended to limit this application. For a person skilled in the art, the present invention may have several modifications and alterations. Any modifications, equivalent replacements, improvements, etc. made within the scope and principle of this application, must be within the scope of protection of this application.
    [0096] Figure 6 is a flow chart that illustrates an example of a method (600) implemented by computer to determine the authenticity of the user with automatic face detection, according to a form of
    Petition 870190062179, of 07/04/2019, p. 32/52
    26/34 realization of the present invention. For clarity of presentation, the description that follows generally describes method (600) in the context of the other figures in this description. However, it will be understood that the method (600) can be carried out, for example, by any system, environment, software and hardware, or a combination of systems, environments, software and hardware, as appropriate. In some implementations, several steps of the method (600) can be performed in parallel, in combination, in loops or in any order.
    [0097] In (602), a first model of deep learning is trained to classify general facial images. General facial images are classified into at least live facial images and non-live facial images. In some embodiments, live facial images are considered positive samples and non-live facial images are considered negative samples. In some embodiments, the first deep learning model is a classification model and general facial images are used as inputs to the first deep learning model. Training the first deep learning model improves the accuracy of classification in relation to general facial images.
    [0098] In some embodiments, a particular general facial image includes a first label indicating whether the specific general facial image corresponding to the first label is a live facial image. In some embodiments, training for the first deep learning model includes: 1) inserting the particular general facial image into the first deep learning model to generate a first prediction result, based on a facial image structure feature particular general, whether the general facial image in particular is a live facial image and 2) adjust the first deep learning model based on the first prediction result and the first label. From (602), the
    Petition 870190062179, of 07/04/2019, p. 33/52
    27/34 method (600) proceeds to (604).
    [0099] In (604), the cropped facial images are extracted from the general facial images. In some embodiments, a given cropped facial image includes a second label, and the second label indicates whether the specific cropped facial image corresponding to the second label is a live facial image. In some embodiments, training the second deep learning model based on the cropped facial image includes: 1) obtaining the specific cropped facial image; 2) insert the specific cropped facial image into the second deep learning model to generate a second prediction result, based on an image material characteristic of the specific cropped facial image, whether a specific cropped facial image is a live facial image ; and 3) adjust the second deep learning model based on the second prediction result and the second label. From (604), method (600) proceeds to (606).
    [00100] In (606), a second model of deep learning is trained based on the cut out facial images. From (606), method (600) proceeds to (608).
    [00101] In (608), an automatic face detection is performed based on the first trained deep learning model and the second trained deep learning model. In some embodiments, the first deep learning model and the second deep learning model are based on a convolutional neural network and in which a convolution nucleus of a convolutional unit in a convolutional neural network of the first deep learning model is larger than a convolution nucleus of a convolution unit in a convolutional neural network of the second model of deep learning. After (608), method (600) stops.
    Petition 870190062179, of 07/04/2019, p. 34/52
    28/34 [00102] In some embodiments, automatic face detection includes: 1) obtaining a general facial image; 2) introduction of the general facial image in the first deep learning model trained to obtain the first corresponding prediction data; 3) obtaining a facial image cut out from the general facial image; 4) introduction of the cropped facial image in the second model of deep learning trained to obtain the corresponding second prediction data; and 5) make a joint decision based on the first prediction data and the second prediction data to obtain an automatic face detection result. From (608), method (600) proceeds to (610).
    [00103] The forms of realization of the subject matter described in this specification can be implemented in order to achieve particular advantages or technical effects. Automatic face detection described can be used to improve authentication processes and ensure data security. For example, the method described can be used to distinguish between live and non-live human face images to help prevent fraud and malicious behavior in relation to secure data. The described method can be incorporated into computing devices (such as mobile computing devices and digital imaging devices).
    [00104] The result of automatic face detection can be displayed in a graphical user interface. Based on the result of automatic face detection, a determination of whether to perform subsequent actions (for example, unlocking protected data, operating a software application, storing data, sending data over a network or displaying data in a graphical user interface) ).
    [00105] The described methodology allows the improvement of several transactions of mobile computing devices and the general security of transaction / data. Participants in transactions that use
    Petition 870190062179, of 07/04/2019, p. 35/52
    29/34 mobile computing can rest assured that the facial images used to unlock a mobile computing device or to authorize a transaction are valid and that they will not be victims of fraud.
    [00106] The described methodology can guarantee the efficient use of computer resources (for example, processing cycles, network bandwidth and memory usage), through efficient verification of data / transactions. At least, these actions can minimize or prevent wasted computer resources available to multiple parties in mobile computing transactions, preventing unwanted / fraudulent transactions. Instead of users having to verify data with additional searches or transactions, transactions can be considered valid.
    [00107] In some embodiments, a graphical user interface can be analyzed to ensure that graphic elements used in automatic detection operations (for example, scanning and verifying the automatic detection of a human face with a mobile computing device) can be positioned on graphical user interfaces to be less intrusive to a user (for example, obscuring the least amount of data and avoid covering any element of the critical or frequently used graphical user interface).
    [00108] Embodiments and operations described in this specification can be implemented in digital electronic circuits, or in computer software, firmware or hardware, including the structures disclosed in this specification or in combinations of one or more of them. Operations can be implemented as operations performed by a data processing device on data stored on one or more storage devices that are readable by computer or received from other sources. A data processing device, computer, or computing device can cover devices,
    Petition 870190062179, of 07/04/2019, p. 36/52
    30/34 data processing devices and machines, including, for example, a programmable processor, a computer, a system on a chip or several, or combinations, of the foregoing. The device may include logic circuits for special purposes, for example, a central processing unit (CPU), an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). The device may also include code that creates an execution environment for the computer program in question, for example, code that constitutes processor firmware, a protocol stack, a database management system, an operating system (for example, example, an operating system or a combination of operating systems), a cross-platform runtime environment, a virtual machine, or a combination of one or more of them. The device and the execution environment can carry out various infrastructures of different computing models, such as Web services, distributed computing and grid computing infrastructures.
    [00109] A computer program (also known, for example, as a program, software, software application, software module, software unit, script or code) can be written in any form of programming language, including compiled languages or interpreted, declarative or procedural languages. and can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object or other unit suitable for use in a computing environment. A program can be stored in a part of a file that contains other programs or data (for example, one or more scripts stored in a markup language document), in a single file dedicated to the program in question or in several coordinated files ( for example, files that store one or more modules, subprograms, or parts of code). A computer program can be
    Petition 870190062179, of 07/04/2019, p. 37/52
    31/34 executed on one computer or on multiple computers located in one location or distributed in several locations and interconnected by a communication network.
    [00110] The processors for executing a computer program include, for example, microprocessors of general and special purpose, and any one or more processors of any type of digital computer. Generally, a processor will receive instructions and data from either a read-only memory or a random access memory or both. The essential elements of a computer are a processor to perform actions according to instructions and one or more memory devices to store instructions and data. Generally, a computer will also include, or be operationally attached to receive data or transfer data to, or both, one or more mass storage devices for data storage. A computer can be incorporated into another device, for example, a mobile device, a personal digital assistant (PDA), a game console, a Global Positioning System (GPS) receiver or a portable storage device. Suitable devices for storing computer program instructions and data include non-volatile memory, media and memory devices, including, for example, semiconductor memory devices, magnetic disks and magneto-optical disks. The processor and memory can be supplemented by, or incorporated into special purpose logic circuits.
    [00111] Mobile devices may include cell phones, user equipment (EU), cell phones (for example, smartphones), tablets, wearable devices (for example, smart watches and smart glasses), devices implanted in the human body (for example , biosensors, cochlear implants) or other types of mobile devices. The
    Petition 870190062179, of 07/04/2019, p. 38/52
    32/34 mobile devices can communicate wirelessly (for example, using radio frequency (RF) signals) with various communication networks (described below). Mobile devices can include sensors to determine the characteristics of the mobile device’s current environment. Sensors can include cameras, microphones, proximity sensors, GPS sensors, motion sensors, accelerometers, ambient light sensors, humidity sensors, gyroscopes, compasses, barometers, fingerprint sensors, facial recognition systems, RF sensors (for example, Wi-Fi and cellular radios), thermal sensors or other types of sensors. For example, cameras can include a forward or backward-facing camera with movable or fixed lenses, a flash, an image sensor and an image processor. The camera can be a megapixel camera capable of capturing details for facial and / or iris recognition. The camera, together with a data processor and authentication information stored in memory or accessed remotely, can form a facial recognition system. The facial recognition system or one or more sensors, for example, microphones, motion sensors, accelerometers, GPS sensors or RF sensors, can be used for user authentication.
    [00112] To provide interaction with a user, the embodiments can be implemented on a computer that has a display device and an input device, for example, a liquid crystal display (LCD) or organic light-emitting diodes (OLED) / virtual reality (VR) Z augmented reality (AR) display to display information to the user and a touch screen, keyboard and a pointing device by which the user can provide input to the computer. Other types of devices can also be used to provide interaction with a user; for example, the feedback provided to the
    Petition 870190062179, of 07/04/2019, p. 39/52
    33/34 user can be any form of sensory feedback, for example, visual feedback, auditory feedback or tactile feedback; and user input can be received in any form, including acoustic, speech, or tactile input. Furthermore, a computer can interact with a user by sending documents and receiving documents from a device used by the user; for example, sending web pages to a web browser on a user’s client device in response to requests received from the web browser.
    [00113] Embodiments can be implemented using computing devices interconnected by any form or means of wired or wireless digital data communication (or combination thereof), for example, a communication network. Examples of interconnected devices are a client and a server that are usually remote from each other and that normally interact through a communication network. A customer, for example, a mobile device, can carry out transactions itself, with a server or through a server, for example, carrying out purchase, sale, payment, delivery, shipping or loan transactions, or authorizing the same. Such transactions can be in real time, so that an action and response are temporarily close; for example, an individual perceives the action and the response occurring substantially simultaneously, the time difference for a response following the individual's action is less than 1 millisecond (ms) or less than 1 second (s), or the response is without delay intentional, taking into account system processing limitations.
    [00114] Examples of communication networks include a local area network (LAN), a radio access network (RAN), a metropolitan area network (MAN) and a wide area network (WAN). The communication network can include all or part of the Internet, another communication network or a combination of communication networks. Information can be transmitted on the communication network according to various protocols and
    Petition 870190062179, of 07/04/2019, p. 40/52
    34/34 standards, including Long Term Evolution (LTE), 5G, IEEE 802, Internet Protocol (IP) or other protocols or combinations of protocols. The communication network can transmit voice, video, biometric or authentication data or other information between the connected computing devices.
    [00115] The features described as separate embodiments can be implemented, in combination, in a single embodiment, while the features described as a single embodiment can be implemented in several embodiments, separately or in any sub- suitable combination. The operations described and claimed in a specific order are not to be understood as requiring that the specific order, nor that all illustrated operations be performed (some operations may be optional). As appropriate, multitasking or parallel processing (or a combination of multitasking and parallel processing) can be performed.
    权利要求:
    Claims (4)
    [1]
    1. FACIAL RECOGNITION METHOD to determine whether an image that includes a face is a live image or a non-live image, characterized by the fact that the method comprises:
    train a first model of deep learning by supervised training on a plurality of general facial images (S301), general facial images comprising live facial images collected when photographing a live face and labeled as positive samples and non-live facial images collected while photograph a non-live face that is a face photo or face model and labeled as negative samples;
    train a plurality of second deep learning models by supervised training on a plurality of facial images extracted from general facial images (S302), the second deep learning models comprising a deep eye learning model and a deep nose learning model corresponding to an eye and a nose of the facial region respectively, the extracted facial images comprising live facial images labeled as positive samples and non-live facial images and labeled as negative samples, the first deep learning model and each of second deep learning models are classification models and in which after training, the models classify facial images into a live facial image category or a non-live facial image category;
    perform automatic face detection on a first general facial image using the first trained deep learning model to obtain a first prediction score and the plurality of second deep learning models trained to obtain second prediction scores (S303), comprising:
    Petition 870190062463, of 07/04/2019, p. 7/12
    [2]
    2. METHOD, according to claim 1, characterized by the fact that the first deep learning model and a second deep learning model use different deep learning network structures.
    2/4 obtain the first general facial image collected for automatic face detection, introduce the first general facial image in the first deep learning model trained for processing to obtain the first prediction score, obtain a plurality of extracted facial images cut from of the first general facial image, the facial images extracted comprising an image of the eye image region and an image of the nose image region, and to introduce the facial images extracted in respective second deep-learning models trained for processing, the second models of deep learning comprising the eye deep learning model and the nose deep learning model, to obtain the second prediction scores, generate a prediction score result based on the first prediction score and the second prediction scores, and compare the result of the prediction score with a threshold to determine whether the first general facial image is a live image or an image not live.
    [3]
    3/4 convolution unit.
    4. METHOD according to any of claims 1 to 3, characterized by the fact that generating a prediction score result based on the first prediction score and the second prediction scores comprises generating the prediction score result as a sum of the first prediction score and the second forecast scores.
    5. METHOD, according to any one of claims 1 to 3, characterized by the fact that the first deep learning model and the second deep learning model are based on a convolutional neural network; and a convolution nucleus of a convolution unit in a convolutional neural network on which the first deep learning model is based is relatively large so that the first deep learning model extracts a structural feature from a general facial image, and a convolution nucleus of a convolution unit in a convolutional neural network on which the second model of deep learning is based is relatively small, so that the second model of deep learning extracts a characteristic of image material from the extracted facial image.
    6. METHOD according to any one of claims 1 to 5, characterized by the fact that the prediction scores are all a probability value and a Boolean value.
    7. APPARATUS, characterized by the fact that it comprises a plurality of modules configured to execute the method, as defined in any one of claims 1 to 6.
    8. ELECTRONIC DEVICE, characterized by the fact that it comprises:
    Petition 870190062463, of 07/04/2019, p. 9/12
    3. METHOD, according to claim 2, characterized by the fact that the different deep learning network structures comprise at least one of the following parameters: a number of hidden variable layers, a type of hidden variable layer, a number of nodes neuronal cells, or a size of a convolution nucleus of a
    Petition 870190062463, of 07/04/2019, p. 12/12
    [4]
    4/4 at least one processor; and a memory communicatively connected to at least one processor, in which the memory stores an instruction that can be executed by at least one processor, and the instruction is executed by at least one processor, to allow the at least one processor to perform the method as defined in any one of claims 1 to 6.
    类似技术:
    公开号 | 公开日 | 专利标题
    BR112019009219A2|2019-08-13|method for facial, handset and electronic device recognition
    De Marsico et al.2014|Firme: Face and iris recognition for mobile engagement
    JP2018160237A|2018-10-11|Facial verification method and apparatus
    CN108804884B|2020-08-07|Identity authentication method, identity authentication device and computer storage medium
    US10691929B2|2020-06-23|Method and apparatus for verifying certificates and identities
    US10691794B2|2020-06-23|Method, apparatus, and electronic device for detecting model security
    US10896318B2|2021-01-19|Occlusion detection for facial recognition processes
    US10824849B2|2020-11-03|Method, apparatus, and system for resource transfer
    CN106169075A|2016-11-30|Auth method and device
    TW201903644A|2019-01-16|Facial recognition method and apparatus and imposter recognition method and apparatus
    WO2021238956A1|2021-12-02|Identity verification method, apparatus and device based on privacy protection
    GB2560340A|2018-09-12|Verification method and system
    WO2020134238A1|2020-07-02|Living body detection method and apparatus, and storage medium
    WO2019161730A1|2019-08-29|Living body detection method, apparatus and device
    CN110287672A|2019-09-27|Verification method and device, electronic equipment and storage medium
    WO2017101071A1|2017-06-22|Facial gesture captcha
    OA19067A|2019-12-27|Face liveness detection method and apparatus, and electronic device.
    TWI753271B|2022-01-21|Resource transfer method, device and system
    EP3869448A1|2021-08-25|Iris authentication device, iris authentication method, and recording medium
    WO2019000957A1|2019-01-03|Iris recognition security system and method
    Dixit et al.2018|SIFRS: Spoof Invariant Facial Recognition System |
    Smith2015|The use of 3D sensor for computer authentication by way of facial recognition for the eyeglasses wearing persons
    同族专利:
    公开号 | 公开日
    KR102142232B1|2020-08-07|
    PL3523754T3|2021-10-04|
    CN107358157A|2017-11-17|
    CN107358157B|2020-10-02|
    US20180357501A1|2018-12-13|
    JP2020504360A|2020-02-06|
    AU2018280235A1|2019-05-23|
    AU2020201662B2|2020-04-16|
    SG10202005728SA|2020-07-29|
    AU2020201662A1|2020-03-26|
    EP3523754B1|2021-04-14|
    CA3043230A1|2018-12-13|
    CN113095124A|2021-07-09|
    CA3043230C|2020-06-23|
    TW201903652A|2019-01-16|
    ES2878374T3|2021-11-18|
    US10671870B2|2020-06-02|
    TWI714834B|2021-01-01|
    ZA201902833B|2021-08-25|
    MX2019005352A|2019-08-05|
    AU2018280235B2|2020-01-23|
    KR20190072563A|2019-06-25|
    JP6732317B2|2020-07-29|
    EP3872699A1|2021-09-01|
    WO2018226990A1|2018-12-13|
    EP3523754A1|2019-08-14|
    PH12019501009A1|2019-12-02|
    RU2714096C1|2020-02-11|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US6959109B2|2002-06-20|2005-10-25|Identix Incorporated|System and method for pose-angle estimation|
    US20080260212A1|2007-01-12|2008-10-23|Moskal Michael D|System for indicating deceit and verity|
    CN100514353C|2007-11-26|2009-07-15|清华大学|Living body detecting method and system based on human face physiologic moving|
    JP2010035048A|2008-07-30|2010-02-12|Fujifilm Corp|Imaging apparatus and imaging method|
    US8675926B2|2010-06-08|2014-03-18|Microsoft Corporation|Distinguishing live faces from flat surfaces|
    TWI556194B|2012-06-29|2016-11-01|希科母股份有限公司|Object detection device, object detection method and computer program for object detection|
    CN104166835A|2013-05-17|2014-11-26|诺基亚公司|Method and device for identifying living user|
    CN103440479B|2013-08-29|2016-12-28|湖北微模式科技发展有限公司|A kind of method and system for detecting living body human face|
    US10095917B2|2013-11-04|2018-10-09|Facebook, Inc.|Systems and methods for facial representation|
    KR102257897B1|2014-05-09|2021-05-28|삼성전자주식회사|Apparatus and method for liveness test,and apparatus and method for image processing|
    US9396537B2|2014-09-09|2016-07-19|EyeVerify, Inc.|Systems and methods for liveness analysis|
    CN105844202A|2015-01-12|2016-08-10|芋头科技（杭州）有限公司|Image recognition system and method|
    CN105117695B|2015-08-18|2017-11-24|北京旷视科技有限公司|In vivo detection equipment and biopsy method|
    CN105740779B|2016-01-25|2020-11-13|北京眼神智能科技有限公司|Method and device for detecting living human face|
    CN105740780B|2016-01-25|2020-07-28|北京眼神智能科技有限公司|Method and device for detecting living human face|
    CN105956572A|2016-05-15|2016-09-21|北京工业大学|In vivo face detection method based on convolutional neural network|
    CN106203373B|2016-07-19|2019-04-16|中山大学|A kind of human face in-vivo detection method based on deep vision bag of words|
    US10289823B2|2016-07-22|2019-05-14|Nec Corporation|Camera system for traffic enforcement|
    CN106599772B|2016-10-31|2020-04-28|北京旷视科技有限公司|Living body verification method and device and identity authentication method and device|
    US11003988B2|2016-11-23|2021-05-11|General Electric Company|Hardware system design improvement using deep learning algorithms|
    CN106778525B|2016-11-25|2021-08-10|北京旷视科技有限公司|Identity authentication method and device|
    CN106709458A|2016-12-27|2017-05-24|深圳市捷顺科技实业股份有限公司|Human face living body detection method and device|
    CN106780482A|2017-01-08|2017-05-31|广东工业大学|A kind of classification method of medical image|
    KR20180109171A|2017-03-27|2018-10-08|삼성전자주식회사|Liveness test method and apparatus for|
    CN113095124A|2017-06-07|2021-07-09|创新先进技术有限公司|Face living body detection method and device and electronic equipment|
    CN107480725A|2017-08-23|2017-12-15|京东方科技集团股份有限公司|Image-recognizing method, device and computer equipment based on deep learning|WO2017015390A1|2015-07-20|2017-01-26|University Of Maryland, College Park|Deep multi-task learning framework for face detection, landmark localization, pose estimation, and gender recognition|
    CN113095124A|2017-06-07|2021-07-09|创新先进技术有限公司|Face living body detection method and device and electronic equipment|
    US10497122B2|2017-10-11|2019-12-03|Adobe Inc.|Image crop suggestion and evaluation using deep-learning|
    US10516830B2|2017-10-11|2019-12-24|Adobe Inc.|Guided image composition on mobile devices|
    CN107832735A|2017-11-24|2018-03-23|百度在线网络技术（北京）有限公司|Method and apparatus for identifying face|
    CN108280418A|2017-12-12|2018-07-13|北京深醒科技有限公司|The deception recognition methods of face image and device|
    CN108205708A|2018-01-02|2018-06-26|安徽大学|A kind of novel expansible deep learning system and data identification method|
    US11030486B2|2018-04-20|2021-06-08|XNOR.ai, Inc.|Image classification through label progression|
    CN108805024B|2018-04-28|2020-11-24|Oppo广东移动通信有限公司|Image processing method, image processing device, computer-readable storage medium and electronic equipment|
    JP2021530038A|2018-06-29|2021-11-04|バイドゥ ドットコム タイムス テクノロジー （ベイジン） カンパニー リミテッド|Systems and methods for low power real-time object detection|
    CN108985228A|2018-07-16|2018-12-11|百度在线网络技术（北京）有限公司|Information generating method and device applied to terminal device|
    US10699150B2|2018-10-23|2020-06-30|Polarr, Inc.|Machine guided photo and video composition|
    CN109460733A|2018-11-08|2019-03-12|北京智慧眼科技股份有限公司|Recognition of face in-vivo detection method and system based on single camera, storage medium|
    CN109784148A|2018-12-06|2019-05-21|北京飞搜科技有限公司|Biopsy method and device|
    CN109800648B|2018-12-18|2021-09-28|北京英索科技发展有限公司|Face detection and recognition method and device based on face key point correction|
    CN109753938B|2019-01-10|2021-11-05|京东方科技集团股份有限公司|Image recognition method and device, application of image recognition method and device, and training method of neural network|
    CN109961006A|2019-01-30|2019-07-02|东华大学|A kind of low pixel multiple target Face datection and crucial independent positioning method and alignment schemes|
    CN110569826A|2019-09-18|2019-12-13|深圳市捷顺科技实业股份有限公司|Face recognition method, device, equipment and medium|
    CN110728215A|2019-09-26|2020-01-24|杭州艾芯智能科技有限公司|Face living body detection method and device based on infrared image|
    CN111062019A|2019-12-13|2020-04-24|支付宝信息技术有限公司|User attack detection method and device and electronic equipment|
    KR102307671B1|2020-01-02|2021-10-01|주식회사 블루프린트랩|Method for face verification|
    CN111241989A|2020-01-08|2020-06-05|腾讯科技（深圳）有限公司|Image recognition method and device and electronic equipment|
    CN112257685A|2020-12-08|2021-01-22|成都新希望金融信息有限公司|Face copying recognition method and device, electronic equipment and storage medium|
    RU2758966C1|2021-05-13|2021-11-03|Общество с ограниченной ответственностью "ВижнЛабс"|Method for determining face authority by segmentation masks|
    CN113610071B|2021-10-11|2021-12-24|深圳市一心视觉科技有限公司|Face living body detection method and device, electronic equipment and storage medium|
    法律状态:
    2021-04-06| B25A| Requested transfer of rights approved|Owner name: ADVANTAGEOUS NEW TECHNOLOGIES CO., LTD. (KY) |
    2021-04-27| B25A| Requested transfer of rights approved|Owner name: ADVANCED NEW TECHNOLOGIES CO., LTD. (KY) |
    2021-10-05| B350| Update of information on the portal [chapter 15.35 patent gazette]|
    优先权:
    申请号 | 申请日 | 专利标题
    CN201710421333.5A|CN107358157B|2017-06-07|2017-06-07|Face living body detection method and device and electronic equipment|
    PCT/US2018/036505|WO2018226990A1|2017-06-07|2018-06-07|Face liveness detection method and apparatus, and electronic device|
    [返回顶部]