WO2022227765A1 - 生成图像修复模型的方法、设备、介质及程序产品 - Google Patents
生成图像修复模型的方法、设备、介质及程序产品 Download PDFInfo
- Publication number
- WO2022227765A1 WO2022227765A1 PCT/CN2022/075070 CN2022075070W WO2022227765A1 WO 2022227765 A1 WO2022227765 A1 WO 2022227765A1 CN 2022075070 W CN2022075070 W CN 2022075070W WO 2022227765 A1 WO2022227765 A1 WO 2022227765A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- repaired
- model
- inpainting
- restoration
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 238000012549 training Methods 0.000 claims abstract description 28
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 8
- 230000008439 repair process Effects 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 12
- 239000002131 composite material Substances 0.000 claims description 9
- 230000015572 biosynthetic process Effects 0.000 claims description 8
- 238000003786 synthesis reaction Methods 0.000 claims description 8
- 238000013473 artificial intelligence Methods 0.000 abstract description 9
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000012545 processing Methods 0.000 description 16
- 238000005516 engineering process Methods 0.000 description 15
- 238000004891 communication Methods 0.000 description 10
- 238000001514 detection method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000003062 neural network model Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 230000003993 interaction Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000001815 facial effect Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 241000282412 Homo Species 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- 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 embodiments of the present disclosure relate to the field of computers, in particular to the fields of artificial intelligence such as deep learning and computer vision, and in particular, to a method, device, medium and program product for generating an image inpainting model.
- the image to be repaired is manually repaired by professional software to complete the repair of the image.
- the embodiments of the present disclosure provide a method, device, medium and program product for generating an image restoration model.
- an embodiment of the present disclosure provides a method for generating an image restoration model, including: acquiring a first image and a second image, where the second image is an image after restoring the first image; The image corresponding to the point is synthesized with the first image to obtain a synthesized image; the second image and the synthesized image are used for training to obtain an image restoration model.
- an embodiment of the present disclosure provides an apparatus for generating an image inpainting model, including: an image acquisition module configured to acquire a first image and a second image, wherein the second image is an image after repairing the first image
- the image synthesis module is configured to synthesize the image corresponding to the feature point of the first image and the first image to obtain a synthesized image
- the model training module is configured to utilize the second image and the synthesized image for training to obtain an image restoration model .
- an embodiment of this announcement proposes an image restoration method, which includes: acquiring an image to be restored; inputting the image to be restored into a pre-trained image restoration model to obtain the restored image.
- an embodiment of this announcement proposes an image restoration device, including: an image acquisition module configured to acquire an image to be restored; an image restoration module configured to input the image to be restored into a pre-trained image restoration model, Get the repaired image.
- an embodiment of the present disclosure provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor.
- the at least one processor executes to enable the at least one processor to perform a method as described in the first aspect or the second aspect.
- an embodiment of the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to cause a computer to execute the method described in the first aspect or the second aspect.
- an embodiment of the present disclosure provides a computer program product, including a computer program, which implements the method described in the first aspect or the second aspect when the computer program is executed by a processor.
- the method, device, medium and program product for generating an image restoration model provided by the embodiments of the present disclosure firstly acquire a first image and a second image, wherein the second image is the image after restoration of the first image; The image corresponding to the feature point is synthesized with the first image to obtain a synthesized image; finally, the second image and the synthesized image are used for training to obtain an image restoration model.
- a composite image obtained by synthesizing the first image and images corresponding to the feature points of the objects in the first image can be used for model training with the second image to obtain an image restoration model, so that image restoration can be achieved.
- FIG. 1 is an exemplary system architecture diagram to which the present disclosure may be applied;
- FIG. 2 is a flowchart of one embodiment of a method for generating an image inpainting model according to the present disclosure
- FIG. 3 is a flowchart of another embodiment of a method for generating an image inpainting model according to the present disclosure
- FIG 5 is an application scene diagram of the image restoration method according to the present disclosure.
- FIG. 6 is a schematic structural diagram of an embodiment of an apparatus for generating an image inpainting model according to the present disclosure
- FIG. 7 is a schematic structural diagram of an embodiment of an image restoration apparatus according to the present disclosure.
- FIG. 8 is a block diagram of an electronic device used to implement embodiments of the present disclosure.
- FIG. 1 illustrates an exemplary system architecture 100 of an embodiment of a method of generating an image inpainting model or an apparatus for generating an image inpainting model to which the present disclosure may be applied.
- the system architecture 100 may include terminal devices 101 , 102 , and 103 , a network 104 and a server 105 .
- the network 104 is a medium used to provide a communication link between the terminal devices 101 , 102 , 103 and the server 105 .
- the network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
- the user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send video frames and the like.
- Various client applications and intelligent interactive applications, such as image processing applications, etc., may be installed on the terminal devices 101 , 102 and 103 .
- the terminal devices 101, 102, and 103 may be hardware or software.
- the terminal devices may be electronic products that perform human-computer interaction with the user through one or more methods such as keyboards, touch pads, touch screens, remote controls, voice interaction or handwriting devices, for example PC (Personal Computer, personal computer), mobile phone, smart phone, PDA (Personal Digital Assistant, personal digital assistant), wearable device, PPC (Pocket PC, handheld computer), tablet computer, smart car machine, smart TV, smart speakers , tablets, laptops, desktops, and more.
- the terminal devices 101, 102, and 103 are software, they can be installed in the above-mentioned electronic devices. It can be implemented as a plurality of software or software modules, and can also be implemented as a single software or software module. There is no specific limitation here.
- the server 105 may provide various services. For example, the server 105 may analyze and process the videos displayed on the terminal devices 101, 102, and 103, and generate processing results.
- the server 105 may be hardware or software.
- the server 105 can be implemented as a distributed server cluster composed of multiple servers, or can be implemented as a single server.
- the server 105 is software, it can be implemented as a plurality of software or software modules (for example, for providing distributed services), or can be implemented as a single software or software module. There is no specific limitation here.
- the method for generating an image restoration model provided by the embodiments of the present disclosure is generally performed by the server 105 , and accordingly, the apparatus for generating an image restoration model is generally set in the server 105 .
- terminal devices, networks and servers in FIG. 1 are merely illustrative. There can be any number of terminal devices, networks and servers according to implementation needs.
- the method for generating an image inpainting model may include the following steps:
- Step 201 Acquire a first image and a second image, where the second image is an image after repairing the first image.
- the execution subject (for example, the terminal devices 101, 102, and 103 shown in FIG. 1 ) of the method for generating an image restoration model may acquire the first image and the second image through a photographing device, and the photographing device may be a camera or its external camera; or, the execution body of the method for generating an image restoration model (for example, the server 105 shown in FIG. 1 ) obtains the first image and second image.
- the above-mentioned first image may be one or several frames of images to be repaired in a certain image to be repaired or a video stream, the first image may include one or more areas to be repaired, and the second image may be the image in the repaired first image. image of the area to be repaired.
- acquiring the first image and the second image may include: acquiring the second image; and generating the first image according to the second image.
- generating the first image according to the second image may include:
- a mask of the same size is applied to the second image to obtain the first image.
- the second image is multiplied by a binary mask to obtain the first image.
- the first image may be an image obtained by adding noise to the second image.
- the first images can be obtained by processing the second images, so as to increase the training samples for training the image restoration model, thereby improving the image restoration accuracy of the image restoration model.
- the method for generating an image repair model may further include: determining an area to be repaired of the first image.
- determining the to-be-repaired area of the first image may include: using a model to identify the first image to determine the to-be-repaired area of the first image; or determining the to-be-repaired area of the first image by manual annotation Repair area.
- the model mainly uses artificial intelligence (Artificial Intelligence, AI), that is, a neural network model.
- AI Artificial Intelligence
- the neural network model can specifically identify the area to be repaired in the first image based on algorithms such as target detection algorithms, such as R-FCN, Faster R-CNN, SSD, YOLO V3 and other algorithms, these neural network models can be obtained by training the initial neural network model by marking the area to be repaired in the first image.
- the second image may be a repaired image of the first image.
- image inpainting refers to image inpainting, which refers to repairing and reconstructing damaged images or removing redundant objects in images.
- the image restoration technology in the embodiment of the present disclosure is a kind of image processing technology.
- the image restoration technology aims to restore the missing or occluded parts of the image according to the image context.
- the image restoration task requires the restoration of the image as a whole as natural as possible and as close to the original image as possible. ground close. Through image inpainting technology, some noise, scratches, deletions and occlusions in the image can be removed to improve the image quality.
- Step 202 Combine the image corresponding to the feature point of the first image with the first image to obtain a combined image.
- the above-mentioned execution main body may synthesize the image corresponding to the feature point of the first image and the first image to obtain the synthesized image.
- target detection may be performed on the first image; then, the object in the first image may be determined; then, feature point detection may be performed on the object in the first image to obtain the feature points of the object;
- An image is segmented to obtain an image corresponding to the feature point; then the image corresponding to the feature point is synthesized with the first image to obtain a synthesized image; for example, based on the number of channels of the image corresponding to the feature point and the number of channels of the first image synthesizing to obtain a composite image; or, splicing the target feature point in the image corresponding to the specific point with the target feature point in the first image, wherein the target feature point in the image corresponding to the feature point and the target feature point in the first image
- the positions of the feature points are the same.
- the above-mentioned feature points can be used to characterize the feature of the object, and the target feature point can be one or more features among all the features that characterize the object.
- the above-mentioned objects may be objects in the first image, for example, a human face, a car, a background, a text, and the like.
- the first image may be an image containing a human face; after performing target detection on the first image, it is determined that the category of the object in the first image is a human face, and the type of the object in the first image is determined as a human face.
- the key points of the face such as facial features (ie, eyes, eyebrows, mouth, nose, etc.), contours, etc.
- the key points of the face are segmented to obtain the image corresponding to the key points of the face; after that, the image corresponding to the key points of the face is synthesized with the feature points in the same position in the first image to obtain a synthesized image, for example, the left eye (ie, the image corresponding to the key points of the face) is stitched with the left eye in the first image.
- performing target detection on the first image may include: using an image recognition model to perform target detection on the first image, and obtaining the category of the target object and the position of the target object in the first image.
- the above-mentioned image recognition model can take the sample image in the training sample in the training sample set as input, and the label corresponding to the input sample image (for example, the position of the object in the sample image in the sample image, and the class label of the object) as output,
- a neural network is trained to obtain an object recognition model.
- the target recognition model may be used to determine the position and/or category of the object in the first image.
- synthesizing the image corresponding to the feature points of the first image and the first image to obtain a composite image may include: combining the target area to be repaired in the first image The image corresponding to the feature point is synthesized with the first image to obtain a synthesized image.
- Step 203 using the second image and the synthesized image for training to obtain an image inpainting model.
- the above-mentioned execution subject may use the second image and the synthesized image for training to obtain an image inpainting model.
- the above-mentioned execution subject may use the synthetic image as the input of the image inpainting model, and use the second image as the output of the image inpainting model, and train the initial model to obtain the image inpainting model.
- the above-mentioned execution subject can use the synthetic image and the second image to train an initial model to obtain an image restoration model.
- the execution subject can use the synthetic image as the input of the image inpainting model, and use the input corresponding second image as the desired output to obtain the image inpainting model.
- the above-mentioned initial model can be a neural network model in the existing technology or future development technology.
- the neural network model can include any one of the following: a generative adversarial network (GAN), a cycle generative adversarial model (Cycle GAN) ), Pix2pixGAN, Dual Learning Generative Adversarial Model (Dual GAN), Disco GAN, Deep Convolutional Generative Adversarial Model (DCGAN).
- GAN can include generator and discriminator. The discriminator is used to distinguish the first image and the second image. Under the supervision of the discriminator, the generator will try its best to generate results close to the real photo to confuse the discriminator and reduce the loss, so that we may get an image that can automatically repair the first image. A model of the image (ie, the image of the defective area).
- the above generator can be a convolutional neural network (for example, various convolutional neural network structures including convolutional layers, pooling layers, de-pooling layers, and deconvolutional layers, which can be down-sampling and up-sampling in sequence. Sampling);
- the above-mentioned discriminator can also be a convolutional neural network (for example, various convolutional neural network structures including a fully-connected layer, wherein the above-mentioned fully-connected layer can implement a classification function).
- the above-mentioned discriminator may also be other model structures that can be used to implement the classification function, such as a Support Vector Machine (SVM).
- SVM Support Vector Machine
- a first image and a second image are obtained first, wherein the second image is the image after restoration of the first image; and then the image corresponding to the feature points of the first image is One image is synthesized to obtain a synthesized image; finally, the second image and the synthesized image are used for training to obtain an image restoration model.
- a composite image obtained by synthesizing the first image and images corresponding to the feature points of the objects in the first image can be used for model training with the second image to obtain an image restoration model, so that image restoration can be achieved.
- combining the image corresponding to the feature points of the first image with the first image to obtain a combined image includes: based on the number of channels of the image corresponding to the feature points of the first image and The number of channels of the first image is combined to obtain a combined image.
- the execution subject may obtain the composite image according to the sum of the number of channels of the image corresponding to the feature points of the first image and the number of channels of the first image.
- a composite image may be obtained based on the number of channels of the image corresponding to the feature points and the number of channels of the first image.
- the feature points of the first image may include the feature points of the first target area to be repaired in the first image.
- the method for generating an image repair model may further include:
- the number of channels of the first image is combined to obtain a combined image.
- the above-mentioned first target area to be repaired may be one or more areas to be repaired in the first image.
- the feature points of the first target area to be repaired may be all the feature points of the first target area to be repaired; the feature points of the first target area to be repaired may also be the more critical ones in the first target area to be repaired. Feature points, such as facial features, facial contours, etc.
- image synthesis can be performed on the feature points of the first target area to be repaired and the first image, and when the synthesized image is obtained, other feature points (for example, the feature points of the first target area to be repaired) can be reduced and synthesized The consumption of resources caused by features other than .
- the image inpainting model is a generative adversarial model, where the generative adversarial model may include a discriminator and a generator.
- the generative adversarial model may include a generator G and a discriminator D.
- the generator G can be used to adjust the resolution of the input image (eg, a composite image) and output the adjusted image
- the discriminator D can be used to determine whether the input image is the image output by the generator G.
- the generative adversarial model trains the generator G and the discriminator D simultaneously through the continuous confrontation process.
- the training process is a process of cross-optimizing the generator G and the discriminator D.
- the generator G is trained to generate fake images to deceive the discriminator D, and the discriminator D is trained to distinguish the real images from those generated by the generator G. fake image.
- the generator G is used to generate the initial repaired image based on the synthetic image; after that, the discriminator D determines whether the initial repaired image is consistent with the real image (the repaired image, that is, the second image); if they are inconsistent, continue to adjust the generative expression
- the parameters of the adversarial model are not adjusted until the initial inpainting image is consistent with the real image, and the final model is determined as the image inpainting model.
- image restoration can be implemented based on a generative adversarial model including a discriminator and a generator.
- FIG. 3 shows a flow 300 of another embodiment of a method for generating an image inpainting model according to the present disclosure.
- the method for generating an image inpainting model may include the following steps:
- Step 301 Acquire a first image and a second image, wherein the second image is an image after repairing the first image.
- Step 302 combining the number of channels of the image corresponding to the feature points of the first image and the number of channels of the first image to obtain a combined image.
- the execution body of the method for generating an image restoration model may be based on the number of channels of the image corresponding to the feature points of the first image
- the number of channels of the images is combined to obtain a combined image, where the number of channels of the combined image is the sum of the number of channels of the image corresponding to the feature point and the number of channels of the first image.
- the above-mentioned number of channels can be used to represent features of multiple dimensions of the image, and the number of channels of the first image can be acquired together with the acquisition of the first image.
- Step 303 using the second image and the synthesized image for training to obtain an image inpainting model.
- steps 301 and 303 have been described in detail in steps 201 and 203 in the embodiment shown in FIG. 2 respectively, and details are not repeated here.
- the method for generating an image inpainting model in this embodiment highlights the step of synthesizing images. Therefore, the solution described in this embodiment is based on combining the number of channels of the image corresponding to the feature points of the first image and the number of channels of the first image to obtain a combined image.
- FIG. 4 shows a process 400 of an embodiment of an image inpainting method according to the present disclosure.
- the image restoration method may include the following steps:
- Step 401 acquiring an image to be repaired.
- the execution body of the image inpainting method may be the same as or different from the execution body of the method for generating an image restoration model. If they are the same, the execution body of the method for generating the image inpainting model can store the model structure information of the trained image inpainting model and the parameter values of the model parameters locally after obtaining the image inpainting model after training. If different, the executor of the method of generating the image inpainting model can send the model structure information of the trained image inpainting model and the parameter values of the model parameters to the executor of the image inpainting method after training the image inpainting model.
- the execution body of the image restoration method can acquire the image to be restored in various ways.
- the image to be repaired may be acquired through a terminal device (eg, the terminal devices 101 , 102 , and 103 shown in FIG. 1 ).
- the above image to be repaired may be an image with an area to be repaired.
- Step 402 Input the image to be repaired into a pre-trained image repair model to obtain a repaired image.
- the above-mentioned execution subject may input the image to be repaired into a pre-trained image repair model to obtain the repaired image.
- the above-mentioned image restoration model may be a model trained by a method for generating an image restoration model, for example, a model obtained by training in the embodiments corresponding to FIG. 2 and FIG. 3 .
- the image to be repaired can be repaired based on a pre-trained image repair model.
- the image repairing method may further include: determining a second target area to be repaired of the image to be repaired; segmenting the second target to be repaired from the image to be repaired region corresponding to the image.
- the second target area to be repaired may be one or more areas to be repaired in the image to be repaired.
- step 402 may include: inputting the image corresponding to the region to be repaired of the second target into a pre-trained image repair model to obtain a repaired image.
- repairing can be performed on the second target area to be repaired in the image to be repaired, so as to reduce the repair operation on the entire image to be repaired, and improve the efficiency of image repair.
- the image repairing method may further include: recognizing the repaired image to obtain a recognition result; As a result, identity authentication is performed.
- face recognition can be performed on the repaired image to obtain a face recognition result; then, based on the face recognition result and the standard image, the identity authentication is performed; if the face recognition result matches the standard image, It is determined that the identity authentication is successful; if the face recognition result does not match the standard image, the identity authentication is confirmed.
- the standard image may be an image pre-uploaded by the user, and whether the user is a legitimate user can be accurately determined through the standard image.
- the user when the user performs identity authentication, because the user is in a situation where it is inconvenient to take pictures (for example, on a fast-moving vehicle), the user may take an image that is not very clear through the terminal device (that is, the image to be repaired). ), at this time, the user needs to perform identity authentication, and can use the image restoration model to restore the captured image; after obtaining the restored image, perform identity authentication based on the restored image, so as to realize identity authentication in scenes that are inconvenient for shooting.
- subsequent operations related to the information of the repaired image may also be performed based on the repaired image. For example, recommendation based on the information of the repaired image (eg, a scene for image search), and resource transfer based on the information of the repaired image.
- the face image to be transferred and the preset face image of the account to be transferred are obtained; the face image to be transferred is input into the image restoration model, Repair the face image to be transferred by the image repair model to obtain the repaired face image; perform face recognition on the repaired face image to obtain the identification result of the face image; if the identification result indicates that the repaired face image If the face image matches the preset face image of the account to be transferred, the resource will be transferred.
- resource transfer may refer to the change of resource ownership; for example, resources are transferred from place A (or device A, or user A) to place B (or device B, or user B)
- the repaired image can be identified to perform identity authentication according to the identification result.
- a terminal device 501 (such as the terminal devices 101 , 102 , and 103 shown in FIG. 1 ) is taken as an example.
- the terminal device first acquires the first image 51 ; Key point detection 52, obtaining the key point (ie, mask) 53 of the first image; after that, input the number of channels of the image corresponding to the key point of the first image and the number of channels of the first image into the pre-trained image restoration model 54 , the inpainting result 55 (eg, the second image) is obtained.
- the present disclosure provides an embodiment of an apparatus for generating an image inpainting model, and the apparatus embodiment corresponds to the method embodiment shown in FIG. 2 .
- the device can be specifically applied to various electronic devices.
- the apparatus 600 for generating an image restoration model in this embodiment may include: an image acquisition module 601 , an image synthesis module 602 and a model training module 603 .
- the image acquisition module 601 is configured to acquire a first image and a second image, wherein the second image is an image after repairing the first image;
- the image synthesis module 602 is configured to The image is synthesized with the first image to obtain a synthesized image;
- the model training module 603 is configured to perform training using the second image and the synthesized image to obtain an image restoration model.
- the specific processing of the image acquisition module 601, the image synthesis module 602, and the model training module 603 and the technical effects brought about by the image acquisition module 601, and the technical effects brought about by them may refer to the corresponding embodiments in FIG. 2, respectively.
- the relevant descriptions of steps 201-203 will not be repeated here.
- the image synthesis module 602 is further configured to: obtain a synthesized image by synthesizing the image channel number corresponding to the feature point of the first image and the channel number of the first image.
- the feature points of the first image are the feature points of the first target area to be repaired in the first image.
- the image inpainting model is a generative confrontation model.
- the present disclosure provides an embodiment of an image restoration apparatus.
- the apparatus embodiment corresponds to the method embodiment shown in FIG. 4 .
- the apparatus may Used in various electronic devices.
- the image restoration apparatus 700 in this embodiment may include: an image acquisition module 701 and an image restoration module 702 .
- the image acquisition module 701 is configured to acquire the image to be repaired;
- the image repair module 702 is configured to input the image to be repaired into a pre-trained image repair model to obtain the repaired image.
- the specific processing of the image acquisition module 701 and the image restoration module 702 and the technical effects brought about by the image restoration device 700 may refer to the relevant descriptions of steps 401-402 in the corresponding embodiment of FIG. 4, respectively. It is not repeated here.
- the image repairing apparatus further includes: an area determination module (not shown in the figure), configured to determine the second target area to be repaired in the image to be repaired; the image repair module 702, is further configured to: input the image corresponding to the area to be repaired of the second target into a pre-trained image repair model to obtain a repaired image.
- an area determination module (not shown in the figure), configured to determine the second target area to be repaired in the image to be repaired
- the image repair module 702 is further configured to: input the image corresponding to the area to be repaired of the second target into a pre-trained image repair model to obtain a repaired image.
- the image repairing apparatus further includes: an image recognition module (not shown in the figure) configured to recognize the repaired image , to obtain the identification result; the identity authentication module (not shown in the figure) is configured to perform identity authentication according to the identification result.
- the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.
- FIG. 8 shows a schematic block diagram of an example electronic device 800 that may be used to implement embodiments of the present disclosure.
- Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.
- Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices.
- the components shown herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the disclosure described and/or claimed herein.
- the device 800 includes a computing unit 801 that can be executed according to a computer program stored in a read only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803 Various appropriate actions and handling. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored.
- the computing unit 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804.
- An input/output (I/O) interface 805 is also connected to bus 804 .
- Various components in the device 800 are connected to the I/O interface 805, including: an input unit 806, such as a keyboard, mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a magnetic disk, an optical disk, etc. ; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, and the like.
- the communication unit 809 allows the device 800 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
- Computing unit 801 may be various general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of computing units 801 include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), various specialized artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any suitable processor, controller, microcontroller, etc.
- the computing unit 801 performs the various methods and processes described above, such as a method of generating an image inpainting model or an image inpainting method.
- a method of generating an image inpainting model or an image inpainting method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 808 .
- part or all of the computer program may be loaded and/or installed on device 800 via ROM 802 and/or communication unit 809.
- the computing unit 801 may be configured to perform a method of generating an image inpainting model or an image inpainting method by any other suitable means (eg, by means of firmware).
- Various implementations of the systems and techniques described herein above may be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips system (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or combinations thereof.
- FPGAs field programmable gate arrays
- ASICs application specific integrated circuits
- ASSPs application specific standard products
- SOC systems on chips system
- CPLD load programmable logic device
- computer hardware firmware, software, and/or combinations thereof.
- These various embodiments may include being implemented in one or more computer programs executable and/or interpretable on a programmable system including at least one programmable processor that
- the processor which may be a special purpose or general-purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device an output device.
- Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, performs the functions/functions specified in the flowcharts and/or block diagrams. Action is implemented.
- the program code may execute entirely on the machine, partly on the machine, partly on the machine and partly on a remote machine as a stand-alone software package or entirely on the remote machine or server.
- a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with the instruction execution system, apparatus or device.
- the machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
- Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any suitable combination of the foregoing.
- machine-readable storage media would include one or more wire-based electrical connections, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), fiber optics, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.
- RAM random access memory
- ROM read only memory
- EPROM or flash memory erasable programmable read only memory
- CD-ROM compact disk read only memory
- magnetic storage or any suitable combination of the foregoing.
- the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or trackball) through which a user can provide input to the computer.
- a display device eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor
- a keyboard and pointing device eg, a mouse or trackball
- Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); and can be in any form (including acoustic input, voice input, or tactile input) to receive input from the user.
- the systems and techniques described herein may be implemented on a computing system that includes back-end components (eg, as a data server), or a computing system that includes middleware components (eg, an application server), or a computing system that includes front-end components (eg, a user computer having a graphical user interface or web browser through which a user may interact with implementations of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system.
- the components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include: Local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
- a computer system can include clients and servers.
- Clients and servers are generally remote from each other and usually interact through a communication network.
- the relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other.
- Artificial intelligence is a discipline that studies computers to simulate certain thinking processes and intelligent behaviors of humans (such as learning, reasoning, thinking, planning, etc.).
- hardware-level technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural speech processing technology, and machine learning/depth Learning, big data processing technology, knowledge graph technology and other major directions.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Human Computer Interaction (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
Abstract
一种生成图像修复模型的方法、设备、介质及程序产品,涉及深度学习和计算机视觉等人工智能领域。方法包括:获取第一图像和第二图像,其中,第二图像为修复第一图像得到的图像(201);将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像(202);利用第二图像和合成图像进行训练,得到图像修复模型(203)。
Description
本专利申请要求于2021年4月29日提交的、申请号为202110475219.7、发明名称为“生成图像修复模型的方法、设备、介质及程序产品”的中国专利申请的优先权,该申请的全文以引用的方式并入本申请中。
本公开实施例涉及计算机领域,具体涉及深度学习和计算机视觉等人工智能领域,尤其涉及一种生成图像修复模型的方法、设备、介质及程序产品。
在那个数码相机及数字存储设备并未普及的年代,人们会在拍照之后冲洗出来保存记录美好瞬间,但是由于相纸本身的缺点,在保存的过程中很容易出现划痕、褪色、污点等,严重影响照片的视觉质量。
目前,由人工通过专业软件对待修复图像进行修复,以完成对图像的修复。
发明内容
本公开实施例提出了一种生成图像修复模型的方法、设备、介质及程序产品。
第一方面,本公开实施例提出了一种生成图像修复模型的方法,包括:获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像;将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像;利用第二图像和合成图像进行训练,得到图像修复模型。
第二方面,本公开实施例提出了一种生成图像修复模型的装置,包括:图像获取模块,被配置成获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像;图像合成模块,被配置成将第一图像的特征 点对应的图像与第一图像进行合成,得到合成图像;模型训练模块,被配置成利用第二图像和合成图像进行训练,得到图像修复模型。
第三方面,本公告实施例提出了一种图像修复方法,包括:获取待修复图像;将待修复图像输入预先训练的图像修复模型中,得到修复图像。
第四方面,本公告实施例提出了一种图像修复装置,包括:图像获取模块,被配置成获取待修复图像;图像修复模块,被配置成将待修复图像输入预先训练的图像修复模型中,得到修复图像。
第五方面,本公开实施例提出了一种电子设备,包括:至少一个处理器;以及与至少一个处理器通信连接的存储器;其中,存储器存储有可被至少一个处理器执行的指令,指令被至少一个处理器执行,以使至少一个处理器能够执行如第一方面或第二方面描述的方法。
第六方面,本公开实施例提出了一种存储有计算机指令的非瞬时计算机可读存储介质,计算机指令用于使计算机执行如第一方面或第二方面描述的方法。
第七方面,本公开实施例提出了一种计算机程序产品,包括计算机程序,计算机程序在被处理器执行时实现如第一方面或第二方面描述的方法。
本公开实施例提供的生成图像修复模型的方法、设备、介质及程序产品,首先获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像;然后将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像;最后利用第二图像和合成图像进行训练,得到图像修复模型。可以通过由第一图像和第一图像中的对象的特征点对应的图像合成得到的合成图像,与第二图像进行模型训练,以得到图像修复模型,从而能够实现对图像的修复。
应当理解,本部分所描述的内容并非旨在标识本公开的实施例的关键或重要特征,也不用于限制本公开的范围。本公开的其它特征将通过以下的说明书而变得容易理解。
通过阅读参照以下附图所作的对非限制性实施例所作的详细描述, 本公开的其它特征、目的和优点将会变得更明显。附图用于更好地理解本方案,不构成对本公开的限定。其中:
图1是本公开可以应用于其中的示例性系统架构图;
图2是根据本公开的生成图像修复模型的方法的一个实施例的流程图;
图3是根据本公开的生成图像修复模型的方法的另一个实施例的流程图;
图4是根据本公开的图像修复方法的一个实施例的流程图;
图5是根据本公开的图像修复方法的一个应用场景图;
图6是根据本公开的生成图像修复模型的装置的一个实施例的结构示意图;
图7是根据本公开的图像修复装置的一个实施例的结构示意图;
图8是用来实现本公开实施例的电子设备的框图。
以下结合附图对本公开的示范性实施例做出说明,其中包括本公开实施例的各种细节以助于理解,应当将它们认为仅仅是示范性的。因此,本领域普通技术人员应当认识到,可以对这里描述的实施例做出各种改变和修改,而不会背离本公开的范围和精神。同样,为了清楚和简明,以下的描述中省略了对公知功能和结构的描述。
需要说明的是,在不冲突的情况下,本公开中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本公开。
图1示出了可以应用本公开的生成图像修复模型的方法或生成图像修复模型的装置的实施例的示例性系统架构100。
如图1所示,系统架构100可以包括终端设备101、102、103,网络104和服务器105。网络104用以在终端设备101、102、103和服务器105之间提供通信链路的介质。网络104可以包括各种连接类型,例如有线、无线通信链路或者光纤电缆等等。
用户可以使用终端设备101、102、103通过网络104与服务器105 交互,以接收或发送视频帧等。终端设备101、102、103上可以安装有各种客户端应用、智能交互应用,例如图像处理应用等等。
终端设备101、102、103可以是硬件,也可以是软件。当终端设备101、102、103为硬件时,终端设备可以为与用户通过键盘、触摸板、触摸屏、遥控器、语音交互或手写设备等一种或多种方式进行人机交互的电子产品,例如PC(Personal Computer,个人计算机)、手机、智能手机、PDA(Personal Digital Assistant,个人数字助手)、可穿戴设备、PPC(Pocket PC,掌上电脑)、平板电脑、智能车机、智能电视、智能音箱、平板电脑、膝上型便携计算机和台式计算机等等。当终端设备101、102、103为软件时,可以安装在上述电子设备中。其可以实现成多个软件或软件模块,也可以实现成单个软件或软件模块。在此不做具体限定。
服务器105可以提供各种服务。例如,服务器105可以对终端设备101、102、103上显示的视频进行分析和处理,并生成处理结果。
需要说明的是,服务器105可以是硬件,也可以是软件。当服务器105为硬件时,可以实现成多个服务器组成的分布式服务器集群,也可以实现成单个服务器。当服务器105为软件时,可以实现成多个软件或软件模块(例如用来提供分布式服务),也可以实现成单个软件或软件模块。在此不做具体限定。
需要说明的是,本公开实施例所提供的生成图像修复模型的方法一般由服务器105执行,相应地,生成图像修复模型的装置一般设置于服务器105中。
应该理解,图1中的终端设备、网络和服务器的数目仅仅是示意性的。根据实现需要,可以具有任意数目的终端设备、网络和服务器。
继续参考图2,其示出了根据本公开的生成图像修复模型的方法的一个实施例的流程200。该生成图像修复模型的方法可以包括以下步骤:
步骤201,获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像。
在本实施例中,生成图像修复模型的方法的执行主体(例如图1所 示的终端设备101、102、103)可以通过拍摄装置获取第一图像和第二图像,拍摄装置可以为终端设备的摄像头或其外部的摄像头;或,生成图像修复模型的方法的执行主体(例如图1所示的服务器105)从终端设备(例如图1所示的终端设备101、102、103)中获取第一图像和第二图像。上述第一图像可以为某张待修复的图像或视频流中的一帧或几帧待修复图像,该第一图像可以包括一个或多个待修复区域,第二图像可以为修复第一图像中的待修复区域得到的图像。
在本实施例中,获取第一图像和第二图像可以包括:获取第二图像;并根据第二图像生成第一图像。
其中,根据第二图像生成第一图像可以包括:
(1)利用预设的掩膜图像对第二图像进行破损处理,生成第一图像,其中,预设的掩膜图像可以是随机生成的各种噪声。
在一个示例中,给第二图像上打上大小相同的掩膜,以得到第一图像。
(2)第二图像通过乘以一个二值掩码得到第一图像。
(3)第一图像可以为第二图像中加入噪声后得到的图像。
本实施例中,在第一图像的数量较少时,可以通过对第二图像进行处理得到第一图像,以增加训练图像修复模型的训练样本,进而提高了图像修复模型的图像修复精度。
在本实施例中,在得到第一图像之后,该生成图像修复模型的方法还可以包括:确定第一图像的待修复区域。
对应地,在该示例中,确定第一图像的待修复区域可以包括:利用模型对第一图像进行识别,以确定第一图像的待修复区域;或由人工标注的方式确定第一图像的待修复区域。
其中,模型主要通过人工智能(Artificial Intelligence,AI),即神经网络模型,神经网络模型具体可基于目标检测算法等算法来识别第一图像的待修复区域,例如R-FCN、Faster R-CNN、SSD、YOLO V3等算法,这些神经网络模型可通过标注出第一图像的待修复区域训练初始神经网络模型得到。
在这里,第二图像可以为对第一图像进行修复图像。
需要说明的是,图像修复是指图像修复是指对受到损坏的图像进行 修复重建或者去除图像中的多余物体。
本公开实施例中的图像修复技术作为图像处理技术的一种,图像修复技术旨在根据图像上下文对图像丢失或遮挡部分进行修复,图像修复任务要求修复图像整体尽可能自然并与原图尽可能地接近。通过图像修复技术,可以去除图像中的一些噪声、划痕、缺失以及遮挡等,以提高图像质量。
步骤202,将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像。
在本实施例中,上述执行主体可以将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像。
具体地,可以先对第一图像进行目标检测;之后,确定第一图像中的对象;之后,对第一图像中的对象进行特征点检测,得到对象的特征点;之后,将特征点从第一图像中分割出来,得到特征点对应的图像;之后将特征点对应的图像与第一图像进行合成,得到合成图像;例如,基于特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像;或,将特定点对应的图像中的目标特征点与第一图像中的目标特征点进行拼接,其中,特征点对应的图像中的目标特征点与第一图像中的目标特征点的位置相同。上述特征点可以用于表征对象的特征,该目标特征点可以为表征对象的所有特征中的一个或多个特征。上述对象可以为第一图像中的目标,例如,人脸、汽车、背景、文字等等。
在一个具体的示例中,第一图像可以为包含人脸的图像;在对第一图像进行目标检测;之后,确定第一图像中的对象的类别为人脸,以及人脸在第一图像中的位置;之后,对第一图像中人脸进行关键点检测,得到人脸的关键点,例如五官(即,眼睛、眉毛、嘴巴、鼻子等)、轮廓等;之后,将第一图像中的人脸的关键点进行分割,得到人脸的关键点对应的图像;之后,将人脸的关键点对应的图像与第一图像中位置相同的特征点进行合成,得到合成图像,例如,将左眼(即人脸的关键点对应的图像)与第一图像中左眼进行拼接。
对应地,在该示例中,对第一图像进行目标检测,可以包括:利用图像识别模型对第一图像进行目标检测,获取目标对象的类别和目标对象在第一图像中的位置。上述图像识别模型可以以训练样本集中的训练样本中的样本图像作为输入,输入的样本图像对应的标签(例如,样本图像中对象的在样本图像中的位置,以及对象的类别标签)作为输出,对神经网络进行训练,以得到目标识别模型。其中,目标识别模型可以用于确定对象在第一图像中的位置和/或类别。
在步骤201中确定第一图像的待修复区域之后,该将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像,可以包括:将第一图像中的目标待修复区域的特征点对应的图像与第一图像进行合成,得到合成图像。
步骤203,利用第二图像和合成图像进行训练,得到图像修复模型。
在本实施例中,上述执行主体可以利用第二图像和合成图像进行训练,得到图像修复模型。
具体地,上述执行主体可以将合成图像作为图像修复模型的输入,将第二图像作为图像修复模型的输出,训练初始模型,得到图像修复模型。
本实施例中,上述执行主体在得到合成图像,以及第二图像后,可以利用合成图像和第二图像训练初始模型,得到图像修复模型。在训练时,执行主体可以将合成图像作为图像修复模型的输入,以及将所输入对应的第二图像,作为期望输出,得到图像修复模型。上述初始模型可以为现有技术或未来发展技术中的神经网络模型,例如,神经网络模型可以包括以下任意一项:生成式对抗模型(Generative Adversarial Networks,GAN)、循环生成式对抗模型(Cycle GAN)、Pix2pixGAN、对偶学习的生成式对抗模型(Dual GAN)、Disco GAN、深度卷积生成式对抗模型(DCGAN)。其中,GAN可以包括生成器和判别器。判别器用于区别第一图像和第二图像,在判别器的监督下,生成器就会尽力去生成接近真实照片的结果来迷惑判别器,减少损失, 这样我们就可能得到一个可以自动修复第一图像(即,存在缺陷区域的图像)的模型了。
需要说明的是,上述生成器可以是卷积神经网络(例如包含卷积层、池化层、反池化层、反卷积层的各种卷积神经网络结构,可以依次进行降采样和上采样);上述判别器也可以是卷积神经网络(例如包含全连接层的各种卷积神经网络结构,其中,上述全连接层可以实现分类功能)。此外,上述判别器也可以是可以用于实现分类功能的其他模型结构,例如支持向量机(Support Vector Machine,SVM)。
本公开实施例提供的生成图像修复模型的方法,首先获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像;然后将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像;最后利用第二图像和合成图像进行训练,得到图像修复模型。可以通过由第一图像和第一图像中的对象的特征点对应的图像合成得到的合成图像,与第二图像进行模型训练,以得到图像修复模型,从而能够实现对图像的修复。
在本实施例的一些可选的实现方式中,将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像,包括:基于第一图像的特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像。
在本实现方式中,上述执行主体可以根据第一图像的特征点对应的图像的通道数与第一图像的通道数的和,得到合成图像。
在本实现方式中,可以基于特征点对应的图像的通道数和第一图像的通道数进行合成,以得到合成图像。
在本实施例的一些可选的实现方式中,若第一图像的特征点可以包括第一图像中的第一目标待修复区域的特征点。
在本实现方式中,在得到第一图像的第一目标待修复区域之后,该生成图像修复模型的方法还可以包括:
基于第一图像中的第一目标待修复区域的特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像。上述第一目标待修复区域可以为第一图像中的一个或多个待修复区域。
需要说明的是,第一目标待修复区域的特征点可以为第一目标待修复区域的所有特征点;第一目标待修复区域的特征点还可以为第一目标待修复区域中的比较关键的特征点,例如人脸的五官、人脸轮廓等。
在本实现方式中,可以针对第一目标待修复区域的特征点与第一图像进行图像合成,在得到合成图像的同时,可以减少合成其他特征点(例如,第一目标待修复区域的特征点之外的特征)带来的资源的消耗。
在本实施例的一些可选的实现方式中,图像修复模型为生成式对抗模型,其中,生成式对抗模型可以包括判别器和生成器。
在本实现方式中,生成式对抗模型可以包括生成器G和判别器D。上述生成器G可用于对所输入的图像(例如,合成图像)进行分辨率调整并输出调整后的图像,上述判别器D用于确定所输入的图像是否为生成器G所输出的图像。生成式对抗模型通过不断的对抗过程,同时训练生成器G和判别器D。训练过程中是对生成器G和判别器D交叉优化的过程,生成器G被训练来生成假图像去欺骗判别器D,而判别器D被训练去区分出是真实图像还是生成器G产生的假图像。其中,生成器G用于基于合成图像生成初始的修复图像;之后,由判别器D判断初始的修复图像与真实图像(修复图像,即第二图像)是否一致;如果不一致,则继续调整生成式对抗模型的参数,直至初始的修复图像与真实图像一致,才停止对模型参数的调整,并将最终的模型确定为图像修复模型。
在本实现方式中,可以基于包括判别器和生成器的生成式对抗模型,实现对图像的修复。
进一步参考图3,图3示出了根据本公开的生成图像修复模型的方法的另一个实施例的流程300。该生成图像修复模型的方法可以包括以下步骤:
步骤301,获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像。
步骤302,基于第一图像的特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像。
在本实施例中,生成图像修复模型的方法的执行主体(例如图1所示的终端设备101、102、103或服务器105)可以基于第一图像的特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像,该合成图像的通道数为特征点对应的图像的通道数与第一图像的通道数的和。上述通道数可以用于表征图像多个维度的特征,该第一图像的通道数可以在获取第一图像时一并获取。
步骤303,利用第二图像和合成图像进行训练,得到图像修复模型。
在本实施例中,步骤301和303的具体操作分别已在图2所示的实施例中步骤201和203进行了详细的介绍,在此不再赘述。
从图3中可以看出,与图2对应的实施例相比,本实施例中的生成图像修复模型的方法突出了合成图像的步骤。由此,本实施例描述的方案基于第一图像的特征点对应的图像的通道数与第一图像的通道数进行合成,以得到合成图像。
进一步参考图4,图4示出了根据本公开的一种图像修复方法的一个实施例的流程400。该一种图像修复方法可以包括以下步骤:
步骤401,获取待修复图像。
在本实施例中,图像修复方法的执行主体可以与生成图像修复模型的方法的执行主体相同或者不同。如果相同,则生成图像修复模型的方法的执行主体可以在训练得到图像修复模型后将训练好的图像修复模型的模型结构信息和模型参数的参数值存储在本地。如果不同,则生成图像修复模型的方法的执行主体可以在训练得到图像修复模型后将训练好的图像修复模型的模型结构信息和模型参数的参数值发送给图像修复方法的执行主体。
在本实施例中,图像修复方法的执行主体可以通过多种方式来获取待修复图像。例如,可以通过终端设备(例如图1所示的终端设备101、102、103)来获取待修复图像。上述待修复图像可以为存在待修复区域的图像。
步骤402,将待修复图像输入预先训练的图像修复模型中,得到修复图像。
在本实施例中,上述执行主体可以将待修复图像输入预先训练的图像修复模型中,得到修复图像。上述图像修复模型可以为由生成图像修复模型的方法训练得到的模型,例如图2和图3对应的实施例训练得到的模型。
本公开实施例提供的方法,基于预先训练的图像修复模型可以实现对待修复图像的修复。
在本实施例的一些可选的实现方式中,在执行步骤402之前,该图像修复方法还可以包括:确定待修复图像的第二目标待修复区域;从待修复图像中分割第二目标待修复区域对应的图像。
需要说明的是,确定待修复图像中的第二目标待修复区域的描述可以参照确定第一图像中的待修复区域的描述。其中,第二目标待修复区域可以为待修复图像中的一个或多个待修复区域。
在确第二目标待修复区域之后,步骤402可以包括:将第二目标待修复区域对应的图像输入预先训练的图像修复模型中,得到修复图像。
在本实现方式中,可以针对待修复图像中的第二目标待修复区域进行修复,以减少对整个待修复图像的修复操作,提升了图像修复的效率。
在本实施例的一些可选的实现方式中,若所述待修复图像为人脸图像,在得到修复图像之后,该图像修复方法还可以包括:对修复后图像进行识别,得到识别结果;根据识别结果,进行身份认证。
在本实现方式中,可以对修复图像进行人脸识别,得到人脸识别结果;之后,基于人脸识别结果与标准图像进行匹配,进行身份认证;如果人脸识别结果与标准图像匹配上了,则确定身份认证成功;如果人脸识别结果与标准图像不匹配,则确认身份认证识别。其中,标准图像可以为用户预先上传的图像,通过该标准图像可以准确地确定用户是否为合法用户。
需要说明的是,在用户进行身份认证时,由于用户处于不方便拍摄(例如,快速行驶的车辆上)的情况下,用户通过终端设备可能拍 摄到一个不是很清晰的图像(即为待修复图像),此时用户需要进行身份认证,可以通过图像修复模型对拍摄的图像进行修复;在得到修复图像之后,基于修复图像进行身份认证,以实现在不方便拍摄的场景下进行身份认证。
在本实现方式中,在对用户进行身份认证之后,还可以基于该修复图像进行与修复图像的信息相关的后续操作。例如,基于修复图像的信息进行推荐(例如,进行图像搜索的场景)、基于修复图像的信息进行资源转移。
在一个具体的示例中,获取到待资源转移的人脸图像和待资源转移的账户预设的人脸图像(即,标准图像);将待资源转移的人脸图像输入到图像修复模型中,通过图像修复模型对待资源转移的人脸图像进行修复,得到修复后的人脸图像;对修复后的人脸图像进行人脸识别,得到人脸图像的身份识别结果;若身份识别结果表示修复后的人脸图像与待资源转移的账户预设的人脸图像匹配,则进行资源转移。
需要说明的是,资源转移可以指资源的所属发生了变化;例如资源从A地(或A设备、或A用户)转移至B地(或B设备、或B用户)
在本实现方式中,在通过图像修复模型对待修复图像进行修复,得到修复图像之后,可以对修复图像进行识别,以根据识别结果进行身份认证。
为了便于理解,下面提供可以实现本公开实施例的图像修复方法的应用场景。如图5所示,以人脸图像为例,终端设备501(例如图1所示的终端设备101、102、103)为例,终端设备先获取第一图像51;之后,对第一图像进行关键点检测52,得到第一图像的关键点(即,mask)53;之后,将第一图像的关键点对应的图像的通道数与第一图像的通道数输入预先训练的图像修复模型54中,得到修复结果55(例如,第二图像)。
进一步参考图6,作为对上述各图所示方法的实现,本公开提供了一种生成图像修复模型的装置的一个实施例,该装置实施例与图2 所示的方法实施例相对应,该装置具体可以应用于各种电子设备中。
如图6所示,本实施例的生成图像修复模型的装置600可以包括:图像获取模块601、图像合成模块602和模型训练模块603。其中,图像获取模块601,被配置成获取第一图像和第二图像,其中,第二图像为修复第一图像后的图像;图像合成模块602,被配置成将第一图像的特征点对应的图像与第一图像进行合成,得到合成图像;模型训练模块603,被配置成利用第二图像和合成图像进行训练,得到图像修复模型。
在本实施例中,生成图像修复模型的装置600中:图像获取模块601、图像合成模块602和模型训练模块603的具体处理及其所带来的技术效果可分别参考图2对应实施例中的步骤201-203的相关说明,在此不再赘述。
在本实施例的一些可选的实现方式中,图像合成模块602,进一步被配置成:基于第一图像的特征点对应的图像的通道数与第一图像的通道数进行合成,得到合成图像。
在本实施例的一些可选的实现方式中,第一图像的特征点为第一图像中的第一目标待修复区域的特征点。
在本实施例的一些可选的实现方式中,图像修复模型为生成式对抗模型。
进一步参考图7,作为对上述各图所示方法的实现,本公开提供了一种图像修复装置的一个实施例,该装置实施例与图4所示的方法实施例相对应,该装置具体可以应用于各种电子设备中。
如图7所示,本实施例的图像修复装置700可以包括:图像获取模块701和图像修复模块702。其中,图像获取模块701,被配置成获取待修复图像;图像修复模块702,被配置成将待修复图像输入预先训练的图像修复模型中,得到修复图像。
在本实施例中,图像修复装置700中:图像获取模块701和图像修复模块702的具体处理及其所带来的技术效果可分别参考图4对应实施例中的步骤401-402的相关说明,在此不再赘述。
在本实施例的一些可选的实现方式中,该图像修复装置还包括:区 域确定模块(图中未示出),被配置成确定待修复图像中的第二目标待修复区域;图像修复模块702,进一步被配置成:将第二目标待修复区域对应的图像输入预先训练的图像修复模型中,得到修复图像。
在本实施例的一些可选的实现方式中,若待修复图像为待修复人脸图像,该图像修复装置还包括:图像识别模块(图中未示出),被配置成对修复图像进行识别,得到识别结果;身份认证模块(图中未示出),被配置成根据识别结果,进行身份认证。
根据本公开的实施例,本公开还提供了一种电子设备、一种可读存储介质和一种计算机程序产品。
图8示出了可以用来实施本公开的实施例的示例电子设备800的示意性框图。电子设备旨在表示各种形式的数字计算机,诸如,膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算机。电子设备还可以表示各种形式的移动装置,诸如,个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例,并且不意在限制本文中描述的和/或者要求的本公开的实现。
如图8所示,设备800包括计算单元801,其可以根据存储在只读存储器(ROM)802中的计算机程序或者从存储单元808加载到随机访问存储器(RAM)803中的计算机程序,来执行各种适当的动作和处理。在RAM 803中,还可存储设备800操作所需的各种程序和数据。计算单元801、ROM 802以及RAM 803通过总线804彼此相连。输入/输出(I/O)接口805也连接至总线804。
设备800中的多个部件连接至I/O接口805,包括:输入单元806,例如键盘、鼠标等;输出单元807,例如各种类型的显示器、扬声器等;存储单元808,例如磁盘、光盘等;以及通信单元809,例如网卡、调制解调器、无线通信收发机等。通信单元809允许设备800通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。
计算单元801可以是各种具有处理和计算能力的通用和/或专用处 理组件。计算单元801的一些示例包括但不限于中央处理单元(CPU)、图形处理单元(GPU)、各种专用的人工智能(AI)计算芯片、各种运行机器学习模型算法的计算单元、数字信号处理器(DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元801执行上文所描述的各个方法和处理,例如生成图像修复模型的方法或图像修复方法。例如,在一些实施例中,生成图像修复模型的方法或图像修复方法可被实现为计算机软件程序,其被有形地包含于机器可读介质,例如存储单元808。在一些实施例中,计算机程序的部分或者全部可以经由ROM 802和/或通信单元809而被载入和/或安装到设备800上。当计算机程序加载到RAM 803并由计算单元801执行时,可以执行上文描述的生成图像修复模型的方法或图像修复方法的一个或多个步骤。备选地,在其他实施例中,计算单元801可以通过其他任何适当的方式(例如,借助于固件)而被配置为执行生成图像修复模型的方法或图像修复方法。
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、芯片上系统的系统(SOC)、负载可编程逻辑设备(CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括:实施在一个或者多个计算机程序中,该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释,该可编程处理器可以是专用或者通用可编程处理器,可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令,并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。
用于实施本公开的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器,使得程序代码当由处理器或控制器执行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行,作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远 程机器或服务器上执行。
在本公开的上下文中,机器可读介质可以是有形的介质,其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备,或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。
为了提供与用户的交互,可以在计算机上实施此处描述的系统和技术,该计算机具有:用于向用户显示信息的显示装置(例如,CRT(阴极射线管)或者LCD(液晶显示器)监视器);以及键盘和指向装置(例如,鼠标或者轨迹球),用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互;例如,提供给用户的反馈可以是任何形式的传感反馈(例如,视觉反馈、听觉反馈、或者触觉反馈);并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如,作为数据服务器)、或者包括中间件部件的计算系统(例如,应用服务器)、或者包括前端部件的计算系统(例如,具有图形用户界面或者网络浏览器的用户计算机,用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如,通信网络)来将系统的部件相互连接。通信网络的示例包括:局域网(LAN)、广域网(WAN)和互联网。
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并 且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。
人工智能是研究计算机来模拟人的某些思维过程和智能行为(如学习、推理、思考、规划等)的学科,既有硬件层面的技术也有软件层面的技术。人工智能硬件技术一般包括如传感器、专用人工智能芯片、云计算、分布式存储、大数据处理等技术;人工智能软件技术主要包括计算机视觉技术、语音识别技术、自然语音处理技术以及机器学习/深度学习、大数据处理技术、知识图谱技术等几大方向。
应该理解,可以使用上面所示的各种形式的流程,重新排序、增加或删除步骤。例如,本发公开中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行,只要能够实现本公开提及的技术方案所期望的结果,本文在此不进行限制。
上述具体实施方式,并不构成对本公开保护范围的限制。本领域技术人员应该明白的是,根据设计要求和其他因素,可以进行各种修改、组合、子组合和替代。任何在本公开的精神和原则之内所作的修改、等同替换和改进等,均应包含在本公开保护范围之内。
Claims (17)
- 一种生成图像修复模型的方法,包括:获取第一图像和第二图像,其中,第二图像为修复第一图像得到的图像;将所述第一图像的特征点对应的图像与所述第一图像进行合成,得到合成图像;利用所述第二图像和所述合成图像进行训练,得到图像修复模型。
- 根据权利要求1所述的方法,其中,所述将所述第一图像的特征点对应的图像与所述第一图像进行合成,得到合成图像,包括:基于所述第一图像的特征点对应的图像的通道数与所述第一图像的通道数进行合成,得到合成图像。
- 根据权利要求1或2所述的方法,其中,所述第一图像的特征点为所述第一图像中的第一目标待修复区域的特征点。
- 根据权利要求1-3任一项所述的方法,其中,所述图像修复模型为生成式对抗模型。
- 一种图像修复方法,包括:获取待修复图像;将所述待修复图像输入如权利要求1-4任意一项所述的图像修复模型中,得到修复图像。
- 根据权利要求5所述的方法,所述方法还包括:确定所述待修复图像中的第二目标待修复区域;所述将所述待修复图像输入如权利要求1-4任意一项所述的图像修复模型中,得到修复图像,包括:将所述第二目标待修复区域对应的图像输入如权利要求1-4任意一项所述的图像修复模型中,得到修复图像。
- 根据权利要求5或6所述的方法,其中,若所述待修复图像为待修复人脸图像,所述方法还包括:对所述修复图像进行识别,得到识别结果;根据所述识别结果,进行身份认证。
- 一种生成图像修复模型的装置,包括:图像获取模块,被配置成获取第一图像和第二图像,其中,第二图像为修复第一图像得到的图像;图像合成模块,被配置成将所述第一图像的特征点对应的图像与所述第一图像进行合成,得到合成图像;模型训练模块,被配置成利用所述第二图像和所述合成图像进行训练,得到图像修复模型。
- 根据权利要求8所述的装置,其中,所述图像合成模块,进一步被配置成:基于所述第一图像的特征点对应的图像的通道数与所述第一图像的通道数进行合成,得到合成图像。
- 根据权利要求8或9所述的装置,其中,所述第一图像的特征点为所述第一图像中的第一目标待修复区域的特征点。
- 根据权利要求8-10任一项所述的装置,其中,所述图像修复模型为生成式对抗模型。
- 一种图像修复装置,包括:图像获取模块,被配置成获取待修复图像;图像修复模块,被配置成将所述待修复图像输入如权利要求1-4任意 一项所述的图像修复模型中,得到修复图像。
- 根据权利要求12所述的装置,所述装置还包括:区域确定模块,被配置成确定所述待修复图像中的第二目标待修复区域;所述图像修复模块,进一步被配置成:将所述第二目标待修复区域对应的图像输入如权利要求1-4任意一项所述的图像修复模型中,得到修复图像。
- 根据权利要求12或13所述的装置,其中,若所述待修复图像为待修复人脸图像,所述装置还包括:图像识别模块,被配置成对所述修复图像进行识别,得到识别结果;身份认证模块,被配置成根据所述识别结果,进行身份认证。
- 一种电子设备,包括:至少一个处理器;以及与所述至少一个处理器通信连接的存储器;其中,所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器能够执行权利要求1-7中任一项所述的方法。
- 一种存储有计算机指令的非瞬时计算机可读存储介质,所述计算机指令用于使所述计算机执行权利要求1-7中任一项所述的方法。
- 一种计算机程序产品,包括计算机程序,所述计算机程序在被处理器执行时实现根据权利要求1-7中任一项所述的方法。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2022565694A JP2023526899A (ja) | 2021-04-29 | 2022-01-29 | 画像修復モデルを生成するための方法、デバイス、媒体及びプログラム製品 |
US17/963,384 US20230036338A1 (en) | 2021-04-29 | 2022-10-11 | Method and apparatus for generating image restoration model, medium and program product |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110475219.7 | 2021-04-29 | ||
CN202110475219.7A CN113177892A (zh) | 2021-04-29 | 2021-04-29 | 生成图像修复模型的方法、设备、介质及程序产品 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US17/963,384 Continuation US20230036338A1 (en) | 2021-04-29 | 2022-10-11 | Method and apparatus for generating image restoration model, medium and program product |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022227765A1 true WO2022227765A1 (zh) | 2022-11-03 |
Family
ID=76925328
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/075070 WO2022227765A1 (zh) | 2021-04-29 | 2022-01-29 | 生成图像修复模型的方法、设备、介质及程序产品 |
Country Status (4)
Country | Link |
---|---|
US (1) | US20230036338A1 (zh) |
JP (1) | JP2023526899A (zh) |
CN (1) | CN113177892A (zh) |
WO (1) | WO2022227765A1 (zh) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115689946A (zh) * | 2022-12-29 | 2023-02-03 | 北京集度科技有限公司 | 图像修复方法、电子设备及计算机程序产品 |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113177892A (zh) * | 2021-04-29 | 2021-07-27 | 北京百度网讯科技有限公司 | 生成图像修复模型的方法、设备、介质及程序产品 |
CN116309160B (zh) * | 2023-03-10 | 2024-04-12 | 北京百度网讯科技有限公司 | 图像分辨率修复方法、装置、设备及存储介质 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109345456A (zh) * | 2018-09-30 | 2019-02-15 | 京东方科技集团股份有限公司 | 生成对抗网络训练方法、图像处理方法、设备及存储介质 |
US20200151466A1 (en) * | 2018-11-08 | 2020-05-14 | Aisin Seiki Kabushiki Kaisha | Periphery monitoring device |
CN111539903A (zh) * | 2020-04-16 | 2020-08-14 | 北京百度网讯科技有限公司 | 训练人脸图像合成模型的方法和装置 |
CN111553858A (zh) * | 2020-04-28 | 2020-08-18 | 四川大学青岛研究院 | 基于生成对抗网络的图像修复方法、系统及其应用 |
CN113177892A (zh) * | 2021-04-29 | 2021-07-27 | 北京百度网讯科技有限公司 | 生成图像修复模型的方法、设备、介质及程序产品 |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108961174A (zh) * | 2018-05-24 | 2018-12-07 | 北京飞搜科技有限公司 | 一种图像修复方法、装置以及电子设备 |
KR102106898B1 (ko) * | 2018-06-18 | 2020-05-06 | 주식회사 쓰임기술 | 얼굴 데이터베이스를 이용한 인물 추적 방법 및 시스템 |
CN110648294B (zh) * | 2019-09-19 | 2022-08-30 | 北京百度网讯科技有限公司 | 图像修复方法、装置及电子设备 |
CN112712472A (zh) * | 2019-10-25 | 2021-04-27 | 北京三星通信技术研究有限公司 | 图像处理方法、装置、电子设备及计算机可读存储介质 |
CN111507914B (zh) * | 2020-04-10 | 2023-08-08 | 北京百度网讯科技有限公司 | 人脸修复模型的训练方法、修复方法、装置、设备和介质 |
CN111612708B (zh) * | 2020-05-06 | 2023-05-12 | 长沙理工大学 | 一种基于对抗生成网络的图像修复方法 |
CN112541864A (zh) * | 2020-09-25 | 2021-03-23 | 中国石油大学(华东) | 一种基于多尺度生成式对抗网络模型的图像修复方法 |
CN112132766A (zh) * | 2020-09-28 | 2020-12-25 | 北京金山云网络技术有限公司 | 图像的修复方法和装置、存储介质、电子装置 |
CN112365412A (zh) * | 2020-10-27 | 2021-02-12 | 天津大学 | 基于动态的面部表情动作单元信息的人脸修复方法 |
CN112541866B (zh) * | 2020-11-24 | 2022-09-13 | 同济大学 | 一种基于进化式生成对抗网络的人脸图像修复模型 |
CN112614066A (zh) * | 2020-12-23 | 2021-04-06 | 文思海辉智科科技有限公司 | 图像修复方法、装置及电子设备 |
-
2021
- 2021-04-29 CN CN202110475219.7A patent/CN113177892A/zh active Pending
-
2022
- 2022-01-29 JP JP2022565694A patent/JP2023526899A/ja active Pending
- 2022-01-29 WO PCT/CN2022/075070 patent/WO2022227765A1/zh unknown
- 2022-10-11 US US17/963,384 patent/US20230036338A1/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109345456A (zh) * | 2018-09-30 | 2019-02-15 | 京东方科技集团股份有限公司 | 生成对抗网络训练方法、图像处理方法、设备及存储介质 |
US20200151466A1 (en) * | 2018-11-08 | 2020-05-14 | Aisin Seiki Kabushiki Kaisha | Periphery monitoring device |
CN111539903A (zh) * | 2020-04-16 | 2020-08-14 | 北京百度网讯科技有限公司 | 训练人脸图像合成模型的方法和装置 |
CN111553858A (zh) * | 2020-04-28 | 2020-08-18 | 四川大学青岛研究院 | 基于生成对抗网络的图像修复方法、系统及其应用 |
CN113177892A (zh) * | 2021-04-29 | 2021-07-27 | 北京百度网讯科技有限公司 | 生成图像修复模型的方法、设备、介质及程序产品 |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115689946A (zh) * | 2022-12-29 | 2023-02-03 | 北京集度科技有限公司 | 图像修复方法、电子设备及计算机程序产品 |
Also Published As
Publication number | Publication date |
---|---|
US20230036338A1 (en) | 2023-02-02 |
CN113177892A (zh) | 2021-07-27 |
JP2023526899A (ja) | 2023-06-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109214343B (zh) | 用于生成人脸关键点检测模型的方法和装置 | |
EP3467707B1 (en) | System and method for deep learning based hand gesture recognition in first person view | |
US12039454B2 (en) | Microexpression-based image recognition method and apparatus, and related device | |
WO2022227765A1 (zh) | 生成图像修复模型的方法、设备、介质及程序产品 | |
US20190392587A1 (en) | System for predicting articulated object feature location | |
WO2023016007A1 (zh) | 人脸识别模型的训练方法、装置及计算机程序产品 | |
WO2020024484A1 (zh) | 用于输出数据的方法和装置 | |
CN113221771B (zh) | 活体人脸识别方法、装置、设备、存储介质及程序产品 | |
JP2022133378A (ja) | 顔生体検出方法、装置、電子機器、及び記憶媒体 | |
CN108388889B (zh) | 用于分析人脸图像的方法和装置 | |
CN110705353A (zh) | 基于注意力机制的遮挡人脸的识别方法和装置 | |
CN113221767B (zh) | 训练活体人脸识别模型、识别活体人脸的方法及相关装置 | |
CN113570689B (zh) | 人像卡通化方法、装置、介质和计算设备 | |
WO2023050868A1 (zh) | 融合模型的训练方法、图像融合方法、装置、设备及介质 | |
WO2022247343A1 (zh) | 识别模型训练方法、识别方法、装置、设备及存储介质 | |
JP2023543964A (ja) | 画像処理方法、画像処理装置、電子機器、記憶媒体およびコンピュータプログラム | |
CN113223125B (zh) | 一种虚拟形象的面部驱动方法、装置、设备和介质 | |
CN114049290A (zh) | 图像处理方法、装置、设备及存储介质 | |
CN112634413B (zh) | 生成模型和生成3d动画的方法、装置、设备和存储介质 | |
US20230115765A1 (en) | Method and apparatus of transferring image, and method and apparatus of training image transfer model | |
US20230139994A1 (en) | Method for recognizing dynamic gesture, device, and storage medium | |
CN112580395A (zh) | 基于深度信息的3d人脸活体识别方法、系统、设备及介质 | |
CN113221766B (zh) | 训练活体人脸识别模型、识别活体人脸的方法及相关装置 | |
CN114038045A (zh) | 一种跨模态人脸识别模型构建方法、装置及电子设备 | |
CN114038067A (zh) | 煤矿人员行为检测方法、设备及存储介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
ENP | Entry into the national phase |
Ref document number: 2022565694 Country of ref document: JP Kind code of ref document: A |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22794246 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |