CN107563357B

CN107563357B - Live-broadcast clothing dressing recommendation method and device based on scene segmentation and computing equipment

Info

Publication number: CN107563357B
Application number: CN201710908984.7A
Authority: CN
Inventors: 张蕊; 颜水成; 唐胜; 程斌
Original assignee: Beijing Qihoo Technology Co Ltd
Current assignee: Beijing Qihoo Technology Co Ltd
Priority date: 2017-09-29
Filing date: 2017-09-29
Publication date: 2021-06-04
Anticipated expiration: 2037-09-29
Also published as: CN107563357A

Abstract

The invention discloses a live-broadcast clothing dressing recommendation method and device based on scene segmentation, a computing device and a computer storage medium, wherein the method comprises the following steps: selecting a frame image containing a specific object in a video shot and/or recorded by image acquisition equipment; inputting the frame image into a scene segmentation network to obtain a scene segmentation result corresponding to the frame image; determining contour information of a limb area of a specific object according to a scene segmentation result corresponding to the frame image; identifying the clothes worn by the specific object according to the contour information of the limb area, and determining the clothes style type of the specific object; determining dress of the clothes to be recommended according to the style type of the clothes; and recommending dress of the clothes to be recommended to the user. According to the technical scheme, the scene segmentation result corresponding to the frame image can be quickly and accurately obtained, the clothing style type of a person in the frame image can be more accurately identified based on the scene segmentation result, and live clothing dressing recommendation is carried out by combining the clothing style type.

Description

Live-broadcast clothing dressing recommendation method and device based on scene segmentation and computing equipment

Technical Field

The invention relates to the technical field of image processing, in particular to a live dress dressing recommendation method and device based on scene segmentation, a computing device and a computer storage medium.

Background

With the continuous development of internet technology, people have increasingly rich daily entertainment activities, for example, more and more users enjoy watching video programs or audio programs provided by a main broadcast on line through a live application program. In order to enrich the live content and increase the interest of live broadcasting, various dresses such as live dress are provided for the anchor. However, in the prior art, when live dress is recommended to a user such as a host, the style of dress worn by the user is not well combined, and the recommended live dress may not meet the preference of the user, which results in poor recommendation effect.

Disclosure of Invention

In view of the above, the present invention has been made to provide a live dress-up recommendation method, apparatus, computing device and computer storage medium based on scene segmentation that overcome or at least partially address the above-mentioned problems.

According to an aspect of the present invention, there is provided a live dress-up recommendation method based on scene segmentation, the method being performed based on a trained scene segmentation network, the method comprising:

selecting a frame image containing a specific object in a video shot and/or recorded by image acquisition equipment;

inputting the frame image into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer; the scale regression layer is a middle convolution layer of the scene segmentation network;

obtaining a scene segmentation result corresponding to the frame image;

determining contour information of a limb area of a specific object according to a scene segmentation result corresponding to the frame image;

identifying the clothes worn by the specific object according to the contour information of the limb area, and determining the clothes style type of the specific object;

determining dress of clothes to be recommended according to the style type of the clothes of the specific object;

and recommending dress of the clothes to be recommended to the user.

Further, performing convolution operation on the convolutional layer by using the second convolution block, and obtaining an output result of the convolutional layer further includes:

sampling from the second volume block by using a linear interpolation method to obtain a characteristic vector to form a third volume block;

and performing convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain an output result of the convolution layer.

Further, the samples used for training the scene segmentation network include: the method comprises the steps of storing a plurality of sample images and annotation scene segmentation results corresponding to the sample images in a sample library.

Further, the training process of the scene segmentation network is completed through multiple iterations; in an iteration process, a sample image and an annotated scene segmentation result corresponding to the sample image are extracted from a sample library, and training of a scene segmentation network is achieved by using the sample image and the annotated scene segmentation result.

Further, the training process of the scene segmentation network is completed through multiple iterations; wherein, the one-time iteration process comprises the following steps:

inputting the sample image into a scene segmentation network to obtain a sample scene segmentation result corresponding to the sample image;

and obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and realizing the training of the scene segmentation network by using the scene segmentation network loss function.

Further, the training step of the scene segmentation network comprises:

extracting a sample image and an annotation scene segmentation result corresponding to the sample image from a sample library;

inputting a sample image into a scene segmentation network for training, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient or an initial scale coefficient output by a scale regression layer in the last iteration process to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer;

obtaining a sample scene segmentation result corresponding to the sample image;

obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and updating the weight parameters of the scene segmentation network according to the scene segmentation network loss function;

and iteratively executing the training step of the scene segmentation network until a preset convergence condition is met.

Further, the predetermined convergence condition includes: the iteration times reach the preset iteration times; and/or the output value of the scene segmentation network loss function is smaller than a preset threshold value.

Further, the scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer.

Further, the method further comprises: when the training of the scene segmentation network is started, the weight parameters of the scale regression layer are initialized.

Further, selecting a frame image containing a specific object in a video captured and/or recorded by the image capturing device further comprises:

and selecting frame images containing specific objects in the video shot and/or recorded by the image acquisition equipment at preset time intervals.

Further, according to the clothing style type of the specific object, determining to-be-recommended clothing dress further comprises:

and searching the live dress matched with the style type of the dress of the specific object from the live dress, and determining the searched matched live dress as the dress to be recommended.

According to another aspect of the present invention, there is provided a live dress making recommendation apparatus based on scene segmentation, the apparatus operating based on a trained scene segmentation network, the apparatus comprising:

the selection module is suitable for selecting frame images containing specific objects in videos shot and/or recorded by the image acquisition equipment;

the segmentation module is suitable for inputting the frame image into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer; the scale regression layer is a middle convolution layer of the scene segmentation network;

the generating module is suitable for obtaining a scene segmentation result corresponding to the frame image;

the first determining module is suitable for determining the outline information of the limb area of the specific object according to the scene segmentation result corresponding to the frame image;

the identification module is suitable for identifying the clothes worn by the specific object according to the contour information of the limb area and determining the clothes style type of the specific object;

the second determining module is suitable for determining dress of clothes to be recommended according to the clothes style type of the specific object;

and the recommending module is suitable for recommending the dress of the clothes to be recommended to the user.

Further, the segmentation module is further adapted to:

Further, the apparatus further comprises: a scene segmentation network training module; the training process of the scene segmentation network is completed through multiple iterations;

the scene segmentation network training module is adapted to: in an iteration process, a sample image and an annotated scene segmentation result corresponding to the sample image are extracted from a sample library, and training of a scene segmentation network is achieved by using the sample image and the annotated scene segmentation result.

the scene segmentation network training module is adapted to: in the one-time iteration process, inputting a sample image into a scene segmentation network to obtain a sample scene segmentation result corresponding to the sample image;

Further, the apparatus further comprises: a scene segmentation network training module;

the scene segmentation network training module comprises:

the extraction unit is suitable for extracting a sample image and an annotation scene segmentation result corresponding to the sample image from a sample library;

the training unit is suitable for inputting a sample image into a scene segmentation network for training, wherein at least one layer of convolution layer in the scene segmentation network performs scaling processing on a first convolution block of the convolution layer by using a scale coefficient or an initial scale coefficient output by a scale regression layer in the last iteration process to obtain a second convolution block, and then performs convolution operation on the convolution layer by using the second convolution block to obtain an output result of the convolution layer;

the acquisition unit is suitable for acquiring a sample scene segmentation result corresponding to a sample image;

the updating unit is suitable for obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and updating the weight parameters of the scene segmentation network according to the scene segmentation network loss function;

and the scene segmentation network training module is operated iteratively until a preset convergence condition is met.

Further, the scene segmentation network training module is further adapted to: when the training of the scene segmentation network is started, the weight parameters of the scale regression layer are initialized.

Further, the selection module is further adapted to:

Further, the second determination module is further adapted to:

According to yet another aspect of the present invention, there is provided a computing device comprising: the processor, the memory and the communication interface complete mutual communication through the communication bus;

the memory is used for storing at least one executable instruction, and the executable instruction enables the processor to execute the operation corresponding to the live dress decorating recommendation method based on scene segmentation.

According to still another aspect of the present invention, a computer storage medium is provided, where at least one executable instruction is stored in the storage medium, and the executable instruction causes a processor to perform an operation corresponding to the above live dress decorating recommendation method based on scene segmentation.

According to the technical scheme provided by the invention, a frame image containing a specific object in a video shot and/or recorded by an image acquisition device is selected, the frame image is input into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is utilized to carry out scaling processing on a first convolution block of the convolution layer by utilizing a scale coefficient output by a scale regression layer to obtain a second convolution block, then the convolution operation of the convolution layer is carried out by utilizing the second convolution block to obtain an output result of the convolution layer, then a scene segmentation result corresponding to the frame image is obtained, the contour information of a limb area of the specific object is determined according to the scene segmentation result corresponding to the frame image, the clothing worn by the specific object is identified according to the contour information of the limb area, the clothing style type of the specific object is determined, according to the clothing style type of the specific object, and determining dress of the clothes to be recommended, and recommending the dress of the clothes to be recommended to the user. According to the technical scheme provided by the invention, the convolution block is zoomed according to the scale coefficient, the adaptive zooming to the receptive field is realized, the scene segmentation result corresponding to the frame image in the video can be quickly and accurately obtained by utilizing the trained scene segmentation network, the accuracy and the processing efficiency of image scene segmentation are effectively improved, the dress style type of a person in the frame image can be more accurately identified based on the obtained scene segmentation result, and the live dress is recommended by combining the dress style type of the person, so that the recommended live dress can accord with the preference of a user, the live dress recommendation effect is effectively improved, and the live dress recommendation mode is optimized.

The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:

fig. 1 shows a flow diagram of a live dress-up recommendation method based on scene segmentation according to an embodiment of the present invention;

FIG. 2 is a flow diagram illustrating a method for training a scene segmentation network according to an embodiment of the invention;

fig. 3 is a flowchart illustrating a live dress-up recommendation method based on scene segmentation according to another embodiment of the present invention;

fig. 4 is a block diagram illustrating a structure of a live dress preparation recommending apparatus based on scene segmentation according to an embodiment of the present invention;

fig. 5 is a block diagram illustrating a structure of a live dress preparation recommending apparatus based on scene segmentation according to another embodiment of the present invention;

FIG. 6 shows a schematic structural diagram of a computing device according to an embodiment of the invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

In the prior art, image scene segmentation processing methods are mainly based on a full convolution neural network in deep learning, and these processing methods utilize the idea of transfer learning to transfer a network obtained by pre-training on a large-scale classification data set to an image segmentation data set for training, so as to obtain a segmentation network for scene segmentation, and then use the segmentation network to perform scene segmentation on an image. The network architecture used by the segmentation network obtained in the prior art directly utilizes an image classification network, and the size of a convolution block in a convolution layer is fixed and invariable, so that the size of a receptive field is fixed and invariable, wherein the receptive field refers to a region of an input image corresponding to a certain node of an output characteristic diagram, and the receptive field with the fixed size is only suitable for capturing targets with the fixed size and scale. However, for image scene segmentation, objects with different sizes are often contained in the scene, and problems often occur when processing too large and too small objects by using a segmentation network with a fixed-size receptive field, for example, for small objects, the receptive field captures too much background around the object, thereby confusing the object with the background, resulting in the object being missed and misjudged as the background; for a larger target, the receptive field can only capture a part of the target, so that the target class judgment is biased, resulting in a discontinuous segmentation result. Therefore, the image scene segmentation processing method in the prior art has the problem of low accuracy of image scene segmentation.

The trained scene segmentation network is adopted in the invention, and can scale the convolution block of the convolution layer by utilizing the scale coefficient output by the scale regression layer in the network, so that the input image can be more accurately subjected to scene segmentation, the accuracy of image scene segmentation is effectively improved, and the dress style type of a person in the image can be more accurately identified based on the obtained scene segmentation result, so that the live dress dressing recommendation can be carried out by combining the dress style types.

Fig. 1 is a flow chart of a scene segmentation-based live dress-up recommendation method, which is executed based on a trained scene segmentation network, according to an embodiment of the present invention, and as shown in fig. 1, the method includes the following steps:

step S100, selecting a frame image containing a specific object in a video shot and/or recorded by an image acquisition device.

In this embodiment, the image capturing device is described by taking a mobile terminal as an example. Because the recommendation is carried out according to the clothes of the specific object, the frame image is selected only by selecting the frame image containing the specific object when the camera of the mobile terminal shoots and/or records the video.

Step S101, inputting a frame image into a scene segmentation network.

The frame image includes a specific object such as a human body. In order to accurately identify the dress style type of a person in a frame image, a scene segmentation network is required to perform scene segmentation on the frame image. The scene segmentation network is trained, and the trained scene segmentation network can scale the convolution blocks of the convolution layer by utilizing the scale coefficient output by the scale regression layer in the network, so that the scene segmentation can be carried out on the input frame image more accurately. Specifically, the samples used for training the scene segmentation network include: the method comprises the steps of storing a plurality of sample images and annotation scene segmentation results corresponding to the sample images in a sample library. And the marked scene segmentation result is a segmentation result obtained by artificially segmenting and marking each scene in the sample image.

The training process of the scene segmentation network is completed through multiple iterations. Optionally, in an iteration process, the sample image and the annotated scene segmentation result corresponding to the sample image are extracted from the sample library, and the training of the scene segmentation network is achieved by using the sample image and the annotated scene segmentation result.

Optionally, the one-iteration process comprises: inputting the sample image into a scene segmentation network to obtain a sample scene segmentation result corresponding to the sample image; and obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and realizing the training of the scene segmentation network by using the scene segmentation network loss function.

Step S102, at least one layer of convolution layer in the scene segmentation network utilizes the scale coefficient output by the scale regression layer to carry out scaling processing on the first convolution block of the convolution layer, and a second convolution block is obtained.

The skilled person can select which layer or layers of convolution blocks of convolution layers are scaled according to actual needs, and this is not limited here. For the convenience of distinction, the convolution block to be scaled is referred to as a first convolution block, and the scaled convolution block is referred to as a second convolution block. If the scaling processing is performed on the first convolution block of a certain layer of convolution layer in the scene segmentation network, then, in the convolution layer, the scaling processing is performed on the first convolution block of the convolution layer by using the scale coefficient output by the scale regression layer, so as to obtain a second convolution block.

The scale regression layer is an intermediate convolution layer of the scene segmentation network, the intermediate convolution layer refers to one or more convolution layers in the scene segmentation network, and a person skilled in the art can select an appropriate one or more convolution layers in the scene segmentation network as the scale regression layer according to actual needs, which is not limited herein. In the invention, the characteristic diagram output by the scale regression layer is called a scale coefficient characteristic diagram, and the scale coefficient is a characteristic vector in the scale coefficient characteristic diagram output by the scale regression layer. The method and the device zoom the convolution block according to the scale coefficient, thereby realizing the self-adaptive zooming of the receptive field, more accurately carrying out scene segmentation on the input frame image and effectively improving the accuracy of image scene segmentation.

Step S103, the convolution operation of the convolution layer is carried out by utilizing the second convolution block, and the output result of the convolution layer is obtained.

After the second convolution block is obtained, the convolution operation of the convolution layer can be performed by using the second convolution block to obtain an output result of the convolution layer.

And step S104, obtaining a scene segmentation result corresponding to the frame image.

After obtaining the output result of the convolutional layer in step S103, if there are other convolutional layers after the convolutional layer in the scene segmentation network, the subsequent convolution operation is performed using the output result of the convolutional layer as the input of the subsequent convolutional layer. After convolution operation of all convolution layers in the scene segmentation network, a scene segmentation result corresponding to the frame image is obtained.

In step S105, contour information of the limb region of the specific object is determined from the scene segmentation result corresponding to the frame image.

After the scene segmentation result corresponding to the frame image is obtained, the contour information of the limb area of the specific object can be determined according to the scene segmentation result corresponding to the frame image. When the specific object is a human body, the contour information of the limb area of the human body can be determined according to the scene segmentation result so as to identify the clothing worn by the specific object in the following. Specifically, the limb area includes a head area, an upper body area, a lower body area, a foot area, and the like.

And S106, identifying the clothes worn by the specific object according to the contour information of the limb area, and determining the clothes style type of the specific object.

After the contour information of the limb area of the specific object is determined, the clothes worn by the specific object can be identified according to the contour information of the limb area, and the clothes style type of the specific object is determined. For example, when the specific object is a human body, a hat, a jacket, a shirt, a skirt, shoes, and the like worn by the person are recognized based on the contour information of the head region, the contour information of the upper body region, the contour information of the lower body region, and the contour information of the foot region in the limb region, and the style of the clothing corresponding to the clothing is determined based on the recognition result.

And S107, determining dress of the clothes to be recommended according to the style type of the clothes of the specific object.

In a specific embodiment, the specific object is a human body, in order to facilitate the user to dress in the live broadcast process, live broadcast dress of various dress style types has been set for the user in advance, and then after the dress style type of the specific object is determined, dress to be recommended can be determined from various live broadcast dress according to the dress style type. Specifically, the clothing style types may include a rui style, a hip-hop style, a joker style, a gentlewoman style, a korean style, a national style, an european and american style, a college style, a commuting style, a neutral style, a hip-hop style, a garden style, a punk style, an OL (office family female) style, a lolita style, a street style, a conciseness style, a boshemia style, and the like. Assuming that the clothing style type of the character is determined to be the college style in step S106, live dress fitting the college style is determined as dress to be recommended in step S107.

And step S108, recommending dress to be recommended to the user.

After determining the dress of the clothes to be recommended, the dress of the clothes to be recommended can be recommended to the user by sending recommendation information or showing the dress of the clothes to be recommended and the like.

According to the live dress-up recommendation method based on scene segmentation provided by the embodiment, a frame image containing a specific object in a video shot by an image acquisition device is selected, and the frame image is input into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network performs scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, then the convolution operation of the convolution layer is performed by using the second convolution block to obtain an output result of the convolution layer, then a scene segmentation result corresponding to the frame image is obtained, contour information of a limb area of the specific object is determined according to the scene segmentation result corresponding to the frame image, a dress style worn by the specific object is identified according to the contour information of the limb area, a dress style type of the specific object is determined, and according to the dress style type of the specific object, and determining dress of the clothes to be recommended, and recommending the dress of the clothes to be recommended to the user. According to the technical scheme provided by the invention, the convolution block is zoomed according to the scale coefficient, the adaptive zooming to the receptive field is realized, the scene segmentation result corresponding to the frame image in the video can be quickly and accurately obtained by utilizing the trained scene segmentation network, the accuracy and the processing efficiency of image scene segmentation are effectively improved, the dress style type of a person in the frame image can be more accurately identified based on the obtained scene segmentation result, and the live dress is recommended by combining the dress style type of the person, so that the recommended live dress can accord with the preference of a user, the live dress recommendation effect is effectively improved, and the live dress recommendation mode is optimized.

Fig. 2 is a flowchart illustrating a training method of a scene segmentation network according to an embodiment of the present invention, and as shown in fig. 2, the training step of the scene segmentation network includes the following steps:

step S200, extracting a sample image and an annotation scene segmentation result corresponding to the sample image from a sample library.

The sample library not only stores the sample images, but also stores the segmentation results of the labeled scenes corresponding to the sample images. The number of the sample images stored in the sample library can be set by a person skilled in the art according to actual needs, and is not limited herein. In step S200, a sample image is extracted from the sample library, and an annotation scene segmentation result corresponding to the sample image is extracted.

Step S201, inputting the sample image into the scene segmentation network for training.

After the sample images are extracted, the sample images are input into a scene segmentation network for training.

Step S202, at least one layer of convolution layer in the scene segmentation network utilizes the scale coefficient or the initial scale coefficient output by the scale regression layer in the last iteration process to carry out scaling processing on the first convolution block of the convolution layer, and a second convolution block is obtained.

The skilled person can select which layer or layers of convolution blocks of convolution layers are scaled according to actual needs, and this is not limited here. If the scaling processing is performed on the first convolution block of a certain convolution layer in the scene segmentation network, then, on the convolution layer, the scaling processing is performed on the first convolution block of the convolution layer by using the scale coefficient or the initial scale coefficient output by the scale regression layer in the last iteration process to obtain a second convolution block.

Specifically, in order to train the scene segmentation network effectively, when the training of the scene segmentation network starts, the weight parameters of the scale regression layer may be initialized. The person skilled in the art can set the specific initialized weight parameters according to the actual needs, which is not limited herein. The initial scale coefficient is the feature vector in the scale coefficient feature map output by the scale regression layer after initialization processing.

Step S203, the convolution operation of the convolution layer is carried out by utilizing the second convolution block, and the output result of the convolution layer is obtained.

After the second convolution block is obtained, the convolution operation of the convolution layer can be performed by using the second convolution block to obtain an output result of the convolution layer. Since the second convolution block is obtained by scaling the first convolution block, the coordinates corresponding to the feature vectors in the second convolution block may not be integers, and therefore, the feature vectors corresponding to the non-integer coordinates may be obtained by using a preset calculation method. The skilled person can set the preset calculation method according to the actual needs, and the method is not limited herein. For example, the preset calculation method may be a linear interpolation method, and specifically, a feature vector is sampled from the second convolution block by using the linear interpolation method to form a third convolution block, and then convolution operation is performed according to the third convolution block and a convolution kernel of the convolution layer to obtain an output result of the convolution layer.

After obtaining the output result of the convolutional layer, if there are other convolutional layers after the convolutional layer in the scene segmentation network, the subsequent convolution operation is performed by using the output result of the convolutional layer as the input of the subsequent convolutional layer. After convolution operation of all convolution layers in the scene segmentation network, a scene segmentation result corresponding to the sample image is obtained.

Step S204, a sample scene segmentation result corresponding to the sample image is obtained.

And acquiring a sample scene segmentation result which is obtained by the scene segmentation network and corresponds to the sample image.

Step S205, a scene segmentation network loss function is obtained according to the segmentation loss between the sample scene segmentation result and the labeling scene segmentation result, and the weight parameters of the scene segmentation network are updated according to the scene segmentation network loss function.

Wherein, those skilled in the art may set the specific content of the scene segmentation network loss function according to actual needs, which is not limited herein. And performing back propagation (back propagation) operation according to the loss function of the scene segmentation network, and updating the weight parameters of the scene segmentation network according to the operation result.

And step S206, iteratively executing the training step of the scene segmentation network until a preset convergence condition is met.

Wherein, those skilled in the art can set the predetermined convergence condition according to the actual requirement, and the present disclosure is not limited herein. For example, the predetermined convergence condition may include: the iteration times reach the preset iteration times; and/or the output value of the scene segmentation network loss function is smaller than a preset threshold value. Specifically, whether the predetermined convergence condition is satisfied may be determined by determining whether the iteration count reaches a preset iteration count, or may be determined according to whether an output value of the scene segmentation network loss function is smaller than a preset threshold. In step S206, the training step of the scene segmentation network is iteratively performed until a predetermined convergence condition is satisfied, thereby obtaining a trained scene segmentation network.

In a specific training process, for example, a first convolution block of a convolution layer in a scene segmentation network needs to be scaled, and the convolution layer is assumed to be called convolution layer J, and an input feature map of the convolution layer J is

Wherein H_AFor the height parameter of the input profile, W_AFor the width parameter of the input feature map, C_AThe number of channels of the input feature map is obtained; the output characteristic diagram of the convolution layer J is

Wherein H_BFor the height parameter of the output profile, W_BAs a width parameter of the output feature map, C_BThe number of channels of the output characteristic diagram; the scale coefficient characteristic diagram output by the scale regression layer is

Wherein H_SIs the height parameter of the scale factor profile, W_SThe width parameter of the scale factor feature map is that the number of channels of the scale factor feature map is 1, specifically, H_S＝H_BAnd W is_S＝W_B。

In the scene segmentation network, a common 3 × 3 convolutional layer can be selected as a scale regression layer, and an output feature map with the number of channels corresponding to the scale regression layer being 1 is a scale coefficient feature map. In order to effectively train the scene segmentation network and prevent the scene segmentation network from collapsing in the training process, it is necessary to initialize the weight parameters of the scale regression layer when the training of the scene segmentation network is started. Wherein the initialized weight parameter of the scale regression layer is

Wherein, w₀A convolution kernel initialized for the scale regression layer, a being any position in the convolution kernel, b₀Is the initialized bias term. In the initialization process of the weight parameters of the scale regression layer, the convolution kernel is set to satisfy the random coefficient σ of the gaussian distribution and its value is small, close to 0, and the bias term is set to 1, and therefore, the scale regression layer subjected to the initialization process outputs all the coefficients close to 1The value, namely the initial scale coefficient is close to 1, and then after the initial scale coefficient is applied to the convolution layer J, the difference between the obtained output result and the standard convolution result is not large, so that a stable training process is provided, and the scene segmentation network is effectively prevented from collapsing in the training process.

For convolutional layer J, assume that the convolutional kernel of convolutional layer J is

Is biased to

The input characteristic diagram of the convolution layer J is

The output characteristic diagram of the convolution layer J is

The first volume block of the convolution layer J is X^tFor the first rolling block X^tThe second volume block obtained after scaling is Y^tWhere, in general, k is 1. At any position t in the output feature map B, the corresponding feature vector is

Feature vector B^tCorresponding to the second volume block Y in the input feature map A by the feature vector^tInner product with convolution kernel K, where position

First volume block X^tIs to input a (p) in the feature map A^t,q^t) A central square area with a side length fixed at 2kd +1, wherein,

is the coefficient of expansion of the convolution,

and

are the coordinates in the input feature map a. First volume block X^tWherein (2K +1) × (2K +1) feature vectors are uniformly selected to be multiplied by a convolution kernel K, and specifically, the coordinates of the feature vectors are

Wherein the content of the first and second substances,

suppose s^tIs a feature vector B in the scale coefficient feature map corresponding to a position t in the output feature map B^tScale factor of, s^tThe position in the scale coefficient feature map is also t, and the feature vector B^tThe positions in the output feature map B are the same.

Using a scale factor s^tFor the first convolution block X of convolution layer J^tScaling to obtain a second convolution block Y^tSecond rolling block Y^tIs to input a (p) in the feature map A^t,q^t) A square area as a center, the side length of which is determined according to a scale factor s^tIs changed into

Second rolling block Y^tWherein (2K +1) × (2K +1) feature vectors are uniformly selected to be multiplied by a convolution kernel K, and specifically, the coordinates of the feature vectors are

Wherein the scale factor s^tIs a real number value, then the coordinates of the feature vector x'_ijAnd y'_ijMay not be an integer. In the invention, the non-integer coordinates are obtained by using a linear interpolation methodA feature vector. From the second volume block Y using a linear interpolation method^tThe feature vector is obtained by middle sampling to form a third volume block Z^tThen for the third volume block Z^tEach feature vector of

The specific calculation formula of (2) is:

wherein the content of the first and second substances,

if (x'_ij,y'_ij) Beyond the range of the input feature map a, the corresponding feature vector will be set to 0 as a pad. Suppose that

Is a convolution vector where the convolution kernel K is multiplied by the corresponding feature vector and the output channel is c, where,

then the element-wise multiplication process for all channels in the convolution operation can be used with

Expressed by matrix multiplication, the forward propagation (forward propagation) process is

In the back propagation process, let us assume that from B^tThe gradient g (B) conveyed^t) Gradient of

g(b)＝g(B^t)

Wherein g (·) represents a gradient function (·)^TRepresenting a matrix transposition. It is worth noting that in calculating the gradient, the final gradient of the convolution kernel K and the bias B is the sum of the gradients obtained from all positions in the output feature map B. For a linear interpolation process, the corresponding eigenvector has a partial derivative of

Corresponding to the partial derivative of the coordinates as

Corresponding to

Partial derivatives of and above

The formulas are similar and are not described in detail here.

Since the coordinates are determined by the scale factor s^tCalculated, then the partial derivative of the coordinate corresponding to the scale coefficient is

Based on the above partial derivatives, the gradients of the scale factor feature map S and the input feature map a can be obtained by the following formula:

therefore, the convolution process forms an overall derivable calculation process, and therefore, the weight parameters of each convolution layer and the weight parameters of the scale regression layer in the scene segmentation network can be trained in an end-to-end mode. In addition, the gradient of the scale factor can be calculated by the gradient transmitted from the next layer, so the scale factor is automatically and implicitly obtained. In a specific implementation process, both the forward propagation process and the backward propagation process can be operated in parallel on a Graphics Processing Unit (GPU), and the calculation efficiency is high.

According to the scene segmentation network training method provided by the embodiment, the scene segmentation network for scaling the convolution block according to the scale coefficient can be trained, the self-adaptive scaling of the receptive field is realized, the corresponding scene segmentation result can be quickly obtained by using the scene segmentation network, and the accuracy and the processing efficiency of image scene segmentation are effectively improved.

Fig. 3 is a flowchart illustrating a live dress-up recommendation method based on scene segmentation according to another embodiment of the present invention, which is executed based on a trained scene segmentation network, as shown in fig. 3, and includes the following steps:

and step S300, selecting frame images containing specific objects in the videos shot and/or recorded by the image acquisition equipment at preset time intervals.

In order to recommend live dress more accurately considering that a particular object may replace a dress worn by the particular object during the process of capturing a video, in step S300, frame images including the particular object in the video captured and/or recorded by the image capturing device may be selected at preset time intervals. The skilled person can set the preset time interval according to actual needs, and the preset time interval is not limited herein. For example, the preset time interval may be set to 5 minutes.

In step S301, a frame image is input into a scene segmentation network.

The scene segmentation network is trained, and the trained scene segmentation network can scale the convolution blocks of the convolution layer by utilizing the scale coefficient output by the scale regression layer in the network, so that the scene segmentation can be carried out on the input frame image more accurately.

Step S302, at least one layer of convolution layer in the scene segmentation network utilizes the scale coefficient output by the scale regression layer to carry out scaling processing on the first convolution block of the convolution layer, and a second convolution block is obtained.

The skilled person can select which layer or layers of convolution blocks of convolution layers are scaled according to actual needs, and this is not limited here. The scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer, and in step S302, the scale coefficient is used to perform scaling processing on the first convolution block of the convolution layer to obtain a second convolution block.

Step S303, a linear interpolation method is used to sample feature vectors from the second convolution block to form a third convolution block.

Since the second convolution block is obtained by scaling the first convolution block, the coordinates corresponding to the feature vector in the second convolution block may not be integers, and therefore, the feature vector corresponding to the non-integer coordinates may be obtained by using a linear interpolation method. And sampling from the second volume block by using a linear interpolation method to obtain a characteristic vector, and then forming a third volume block according to the characteristic vector obtained by sampling. Assume the second volume block is Y^tThe third volume block is Z^tThen for the third volume block Z^tEach feature vector of

The specific calculation formula of (2) is:

wherein the content of the first and second substances,

d is the coefficient of expansion of the convolution, s^tIs a scale factor, and in general, k is 1.

Step S304, performing convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain the output result of the convolution layer.

After the third convolution block is obtained, performing convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain an output result of the convolution layer.

In step S305, a scene segmentation result corresponding to the frame image is obtained.

After obtaining the output result of the convolutional layer in step S304, if there are other convolutional layers after the convolutional layer in the scene segmentation network, the subsequent convolution operation is performed using the output result of the convolutional layer as the input of the subsequent convolutional layer. After convolution operation of all convolution layers in the scene segmentation network, a scene segmentation result corresponding to the frame image is obtained.

Step S306 determines contour information of the limb region of the specific object based on the scene segmentation result corresponding to the frame image.

After the scene segmentation result corresponding to the frame image is obtained in step S305, the contour information of the limb area of the specific object can be determined based on the scene segmentation result corresponding to the frame image. When the specific object is a human body, the contour information of the limb area of the human body can be determined according to the scene segmentation result so as to identify the clothing worn by the specific object in the following.

Step S307, according to the contour information of the limb area, the clothing worn by the specific object is identified, and the clothing style type of the specific object is determined.

The specific object is a human body, and specifically, the limb area includes a head area, an upper body area, a lower body area, a foot area, and the like. The method can respectively identify the hat, the coat, the underclothes or the skirt worn by the person, the shoes and the like according to the contour information of the head area, the contour information of the upper body area, the contour information of the lower body area and the contour information of the foot area in the limb area, and determine the style type of the clothes corresponding to the clothes according to the identification result.

Step S308, searching live dress matched with the style type of the clothes of the specific object from the live dress, and determining the searched live dress matched as dress to be recommended.

The specific object is a human body, in order to facilitate the user to dress in the live broadcast process, live broadcast dress of various dress style types is set for the user in advance, then after the dress style type of the specific object is determined, live broadcast dress matched with the dress style type of the specific object can be searched from the live broadcast dress, and then the searched live broadcast dress is determined as dress to be recommended. For example, it is determined in step S307 that the clothing style type of the person is a garden style, then in step S308, a live clothing dress matching the garden style is searched for from the live clothing dresses, that is, a live clothing dress conforming to the garden style is searched for, and the found live clothing dress matching the garden style is determined as a clothing dress to be recommended.

Step S309, recommend to the user to dress up the clothing to be recommended.

Specifically, dress of the clothing to be recommended may be recommended to the user by sending recommendation information or presenting dress of the clothing to be recommended, or the like.

According to the live broadcast clothing dressing recommendation method based on scene segmentation provided by the embodiment, the convolution block is scaled according to the scale coefficient, so that the self-adaptive scaling of the receptive field is realized, the convolution block after the scaling processing is further processed by using a linear interpolation method, and the problem of selecting the eigenvector of which the coordinate is a non-integer in the convolution block after the scaling processing is solved; and the trained scene segmentation network can be used for quickly and accurately obtaining the scene segmentation result corresponding to the frame image in the video, so that the accuracy rate and the processing efficiency of image scene segmentation are effectively improved, the dress style types of figures in the frame image can be more accurately identified based on the obtained scene segmentation result, and the live dress is recommended by combining the dress style types, so that the recommended live dress can accord with the preference of a user, the live dress recommendation effect is effectively improved, and the live dress recommendation mode is optimized.

Fig. 4 is a block diagram illustrating a structure of a scene segmentation-based live dress decorating recommendation apparatus operating based on a trained scene segmentation network according to an embodiment of the present invention, as shown in fig. 4, the apparatus includes: a selecting module 410, a segmenting module 420, a generating module 430, a first determining module 440, an identifying module 450, a second determining module 460, and a recommending module 470.

The selection module 410 is adapted to: and selecting frame images containing specific objects in the videos shot and/or recorded by the image acquisition equipment.

The segmentation module 420 is adapted to: inputting the frame image into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer.

The scene segmentation network is trained, and specifically, samples used for training the scene segmentation network include: the method comprises the steps of storing a plurality of sample images and annotation scene segmentation results corresponding to the sample images in a sample library. The scale regression layer is a middle convolution layer of the scene segmentation network. One skilled in the art can select one or more convolution layers in the scene segmentation network as a scale regression layer according to actual needs, which is not limited herein. And the scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer.

The generation module 430 is adapted to: and obtaining a scene segmentation result corresponding to the frame image.

The first determination module 440 is adapted to: and determining the outline information of the limb area of the specific object according to the scene segmentation result corresponding to the frame image.

The identification module 450 is adapted to: according to the outline information of the limb area, clothes worn by the specific object are identified, and the clothes style type of the specific object is determined.

The second determination module 460 is adapted to: and determining dress of the clothes to be recommended according to the style type of the clothes of the specific object.

The second determination module 460 may determine to-be-recommended dress from various live dress dresses according to the dress style type determined by the identification module 450.

The recommendation module 470 is adapted to: and recommending dress of the clothes to be recommended to the user.

The recommending module 470 may recommend the dress of the clothing to be recommended to the user by sending recommendation information or showing the dress of the clothing to be recommended.

According to the live dress up recommendation device based on scene segmentation provided by the embodiment, convolution blocks can be zoomed according to scale coefficients, adaptive zooming of a receptive field is achieved, a scene segmentation result corresponding to a frame image in a video can be obtained quickly and accurately by utilizing a trained scene segmentation network, accuracy and processing efficiency of image scene segmentation are effectively improved, dress style types of people in the frame image can be identified more accurately based on the obtained scene segmentation result, live dress up recommendation is carried out by combining the dress style types, the recommended live dress up can accord with user preferences, live dress up recommendation effects are effectively improved, and live dress up recommendation modes are optimized.

Fig. 5 is a block diagram illustrating a structure of a scene segmentation-based live dress-up recommendation apparatus according to another embodiment of the present invention, which operates based on a trained scene segmentation network, as shown in fig. 5, and includes: a selecting module 510, a scene segmentation network training module 520, a segmentation module 530, a generating module 540, a first determining module 550, a recognition module 560, a second determining module 570, and a recommending module 580.

The selecting module 510 is adapted to: and selecting frame images containing specific objects in the video shot and/or recorded by the image acquisition equipment at preset time intervals.

The training process of the scene segmentation network is completed through multiple iterations. The scene segmentation network training module 520 is adapted to: in an iteration process, a sample image and an annotated scene segmentation result corresponding to the sample image are extracted from a sample library, and training of a scene segmentation network is achieved by using the sample image and the annotated scene segmentation result.

Optionally, the scene segmentation network training module 520 is adapted to: in the one-time iteration process, inputting a sample image into a scene segmentation network to obtain a sample scene segmentation result corresponding to the sample image; and obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and realizing the training of the scene segmentation network by using the scene segmentation network loss function.

In a particular embodiment, the scene segmentation network training module 520 may include: an extraction unit 521, a training unit 522, an acquisition unit 523, and an update unit 524.

In particular, the extraction unit 521 is adapted to: and extracting a sample image and an annotation scene segmentation result corresponding to the sample image from the sample library.

The training unit 522 is adapted to: inputting a sample image into a scene segmentation network for training, wherein at least one layer of convolution layer in the scene segmentation network performs scaling processing on a first convolution block of the convolution layer by using a scale coefficient or an initial scale coefficient output by a scale regression layer in the last iteration process to obtain a second convolution block, and then performs convolution operation on the convolution layer by using the second convolution block to obtain an output result of the convolution layer.

The scale regression layer is a middle convolution layer of the scene segmentation network, and the scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer.

Optionally, the training unit 522 is further adapted to: sampling from the second volume block by using a linear interpolation method to obtain a characteristic vector to form a third volume block; and performing convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain an output result of the convolution layer.

The obtaining unit 523 is adapted to: and acquiring a sample scene segmentation result corresponding to the sample image.

The update unit 524 is adapted to: and obtaining a scene segmentation network loss function according to the segmentation loss between the sample scene segmentation result and the labeled scene segmentation result, and updating the weight parameters of the scene segmentation network according to the scene segmentation network loss function.

The scene segmentation network training module 520 runs iteratively until a predetermined convergence condition is met.

Wherein, those skilled in the art can set the predetermined convergence condition according to the actual requirement, and the present disclosure is not limited herein. For example, the predetermined convergence condition may include: the iteration times reach the preset iteration times; and/or the output value of the scene segmentation network loss function is smaller than a preset threshold value. Specifically, whether the predetermined convergence condition is satisfied may be determined by determining whether the iteration count reaches a preset iteration count, or may be determined according to whether an output value of the scene segmentation network loss function is smaller than a preset threshold.

Optionally, the scene segmentation network training module 520 is further adapted to: when the training of the scene segmentation network is started, the weight parameters of the scale regression layer are initialized.

The segmentation module 530 is adapted to: inputting the frame image into a scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, and then a linear interpolation method is used for sampling from the second convolution block to obtain a feature vector to form a third convolution block; and performing convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain an output result of the convolution layer.

The generating module 540 is adapted to: and obtaining a scene segmentation result corresponding to the frame image.

The first determination module 550 is adapted to: and determining the outline information of the limb area of the specific object according to the scene segmentation result corresponding to the frame image.

The identification module 560 is adapted to: according to the outline information of the limb area, clothes worn by the specific object are identified, and the clothes style type of the specific object is determined.

The second determination module 570 is adapted to: and searching the live dress matched with the style type of the dress of the specific object from the live dress, and determining the searched matched live dress as the dress to be recommended.

The recommendation module 580 is adapted to: and recommending dress of the clothes to be recommended to the user.

According to the live-broadcast clothing decorating recommendation device based on scene segmentation provided by the embodiment, the convolution block is scaled according to the scale coefficient, so that the self-adaptive scaling of the receptive field is realized, the convolution block after the scaling processing is further processed by using a linear interpolation method, and the problem of selecting the eigenvector of which the coordinate is a non-integer in the convolution block after the scaling processing is solved; and the trained scene segmentation network can be used for quickly and accurately obtaining the scene segmentation result corresponding to the frame image in the video, so that the accuracy rate and the processing efficiency of image scene segmentation are effectively improved, the dress style types of figures in the frame image can be more accurately identified based on the obtained scene segmentation result, and the live dress is recommended by combining the dress style types, so that the recommended live dress can accord with the preference of a user, the live dress recommendation effect is effectively improved, and the live dress recommendation mode is optimized.

The invention also provides a nonvolatile computer storage medium, and the computer storage medium stores at least one executable instruction which can execute the live dress decorating recommendation method based on scene segmentation in any method embodiment.

Fig. 6 is a schematic structural diagram of a computing device according to an embodiment of the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the computing device.

As shown in fig. 6, the computing device may include: a processor (processor)602, a communication Interface 604, a memory 606, and a communication bus 608.

Wherein:

the processor 602, communication interface 604, and memory 606 communicate with one another via a communication bus 608.

A communication interface 604 for communicating with network elements of other devices, such as clients or other servers.

The processor 602 is configured to execute the program 610, and may specifically execute relevant steps in the above live dress grooming recommendation method embodiment based on scene segmentation.

In particular, program 610 may include program code comprising computer operating instructions.

The processor 602 may be a central processing unit CPU or an application Specific Integrated circuit asic or one or more Integrated circuits configured to implement embodiments of the present invention. The computing device includes one or more processors, which may be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.

And a memory 606 for storing a program 610. Memory 606 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.

Program 610 may be specifically configured to cause processor 602 to execute a live dress preparation recommendation method based on scene segmentation in any of the above-described method embodiments. For specific implementation of each step in the program 610, reference may be made to corresponding steps and corresponding descriptions in units in the above live dress preparation recommendation embodiment based on scene segmentation, which are not described herein again. It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described devices and modules may refer to the corresponding process descriptions in the foregoing method embodiments, and are not described herein again.

The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.

In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

Furthermore, those skilled in the art will appreciate that while some embodiments described herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.

The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functionality of some or all of the components in accordance with embodiments of the present invention. The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.

Claims

1. A live dress grooming recommendation method based on scene segmentation, the method performed based on a trained scene segmentation network, the method comprising:

inputting the frame image into the scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient output by a scale regression layer to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer; the scale regression layer is a middle convolution layer of the scene segmentation network; the scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer;

obtaining a scene segmentation result corresponding to the frame image;

determining contour information of the limb area of the specific object according to a scene segmentation result corresponding to the frame image;

according to the outline information of the limb area, identifying the clothes worn by the specific object, and determining the clothes style type of the specific object;

determining dress of the clothes to be recommended according to the style type of the clothes of the specific object;

recommending the dress to be recommended to a user;

the training process of the scene segmentation network is completed through multiple iterations; in the one-time iteration process, inputting a sample image into the scene segmentation network for training, wherein at least one layer of convolution layer in the scene segmentation network is subjected to scaling processing on a first convolution block of the convolution layer by using a scale coefficient or an initial scale coefficient output by a scale regression layer in the last iteration process to obtain a second convolution block, and then the second convolution block is used for performing convolution operation on the convolution layer to obtain an output result of the convolution layer;

the performing convolution operation on the convolutional layer by using the second convolution block to obtain an output result of the convolutional layer further includes:

sampling from the second volume block by using a linear interpolation method to obtain a feature vector to form a third volume block;

and carrying out convolution operation according to the third convolution block and the convolution kernel of the convolution layer to obtain an output result of the convolution layer.

2. The method of claim 1, wherein the samples used for the scene segmentation network training comprise: the method comprises the steps of storing a plurality of sample images and annotation scene segmentation results corresponding to the sample images in a sample library.

3. The method of claim 2, wherein a sample image and an annotated scene segmentation result corresponding to the sample image are extracted from the sample library in an iterative process, and the training of the scene segmentation network is achieved by using the sample image and the annotated scene segmentation result.

4. A method according to claim 2 or 3, wherein an iterative process comprises:

5. The method of claim 2 or 3, wherein the training of the scene segmentation network comprises:

extracting a sample image and an annotation scene segmentation result corresponding to the sample image from the sample library;

inputting the sample image into the scene segmentation network for training;

obtaining a sample scene segmentation result corresponding to the sample image;

6. The method of claim 5, wherein the predetermined convergence condition comprises: the iteration times reach the preset iteration times; and/or the output value of the scene segmentation network loss function is smaller than a preset threshold value.

7. The method of claim 1, wherein the method further comprises: and when the scene segmentation network training is started, initializing the weight parameters of the scale regression layer.

8. The method of claim 1, wherein the selecting frame images of the video captured and/or recorded by the image capture device that contain the specific object further comprises:

9. The method of claim 1, wherein the determining to-be-recommended apparel outfit according to the apparel style type of the particular object further comprises:

10. A live dress grooming recommendation device based on scene segmentation, the device operating based on a trained scene segmentation network, the device comprising:

the segmentation module is suitable for inputting the frame image into the scene segmentation network, wherein at least one layer of convolution layer in the scene segmentation network utilizes a scale coefficient output by a scale regression layer to carry out scaling processing on a first convolution block of the convolution layer to obtain a second convolution block, and then utilizes the second convolution block to carry out convolution operation on the convolution layer to obtain an output result of the convolution layer; the scale regression layer is a middle convolution layer of the scene segmentation network; the scale coefficient is a feature vector in a scale coefficient feature map output by the scale regression layer;

the first determining module is suitable for determining the outline information of the limb area of the specific object according to a scene segmentation result corresponding to the frame image;

the second determining module is suitable for determining clothes dressing to be recommended according to the clothes style type of the specific object;

the recommendation module is suitable for recommending the dress to be recommended to a user;

the segmentation module is further adapted to:

11. The apparatus of claim 10, wherein the samples used for the scene segmentation network training comprise: the method comprises the steps of storing a plurality of sample images and annotation scene segmentation results corresponding to the sample images in a sample library.

12. The apparatus of claim 11, wherein the apparatus further comprises: a scene segmentation network training module;

the scene segmentation network training module is adapted to: in an iteration process, a sample image and an annotated scene segmentation result corresponding to the sample image are extracted from the sample library, and the training of a scene segmentation network is realized by using the sample image and the annotated scene segmentation result.

13. The apparatus of claim 11 or 12, wherein the apparatus further comprises: a scene segmentation network training module;

14. The apparatus of claim 11 or 12, wherein the apparatus further comprises: a scene segmentation network training module;

the scene segmentation network training module comprises:

the extraction unit is suitable for extracting a sample image and an annotation scene segmentation result corresponding to the sample image from the sample library;

the training unit is suitable for inputting the sample image into the scene segmentation network for training;

15. The apparatus of claim 14, wherein the predetermined convergence condition comprises: the iteration times reach the preset iteration times; and/or the output value of the scene segmentation network loss function is smaller than a preset threshold value.

16. The apparatus of claim 10, wherein the scene segmentation network training module is further adapted to: and when the scene segmentation network training is started, initializing the weight parameters of the scale regression layer.

17. The apparatus of claim 10, wherein the selection module is further adapted to:

18. The apparatus of claim 10, wherein the second determining module is further adapted to:

19. A computing device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;

the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the scene segmentation based live dress decorating recommendation method according to any one of claims 1-9.

20. A computer storage medium having at least one executable instruction stored therein, the executable instruction causing a processor to perform operations corresponding to the live dress grooming recommendation method based on scene segmentation according to any one of claims 1-9.