WO2020232977A1 - 神经网络训练方法及装置以及图像处理方法及装置 - Google Patents

神经网络训练方法及装置以及图像处理方法及装置 Download PDF

Info

Publication number
WO2020232977A1
WO2020232977A1 PCT/CN2019/114470 CN2019114470W WO2020232977A1 WO 2020232977 A1 WO2020232977 A1 WO 2020232977A1 CN 2019114470 W CN2019114470 W CN 2019114470W WO 2020232977 A1 WO2020232977 A1 WO 2020232977A1
Authority
WO
WIPO (PCT)
Prior art keywords
state
feature
neural network
target image
category
Prior art date
Application number
PCT/CN2019/114470
Other languages
English (en)
French (fr)
Inventor
韩江帆
罗平
王晓刚
Original Assignee
北京市商汤科技开发有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京市商汤科技开发有限公司 filed Critical 北京市商汤科技开发有限公司
Priority to JP2021538254A priority Critical patent/JP2022516518A/ja
Priority to SG11202106979WA priority patent/SG11202106979WA/en
Publication of WO2020232977A1 publication Critical patent/WO2020232977A1/zh
Priority to US17/364,731 priority patent/US20210326708A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133—Distances to prototypes
    • G06F18/24137—Distances to cluster centroïds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • 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
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods

Definitions

  • the present disclosure relates to the field of computer technology, in particular to a neural network training method and device, and an image processing method and device.
  • machine learning especially deep learning
  • machine learning relies heavily on large-scale accurately labeled data sets.
  • the present disclosure proposes a technical solution for neural network training and image processing.
  • a neural network training method which includes: classifying a target image in a training set through a neural network to obtain a predicted classification result of the target image; according to the predicted classification result, the The initial category label and the corrected category label of the target image are used to train the neural network.
  • the neural network includes a feature extraction network and a classification network, and the neural network includes N training states, where N is an integer greater than 1, wherein the neural network compares the training set
  • Performing classification processing on the target image to obtain the predicted classification result of the target image includes: performing feature extraction on the target image through the feature extraction network of the i-th state to obtain the first feature of the i-th state of the target image.
  • the i state is one of the N training states, and 0 ⁇ i ⁇ N; the first feature of the i-th state of the target image is classified by the classification network of the i-th state to obtain the first feature of the target image The predicted classification result of the i state.
  • the training of the neural network according to the predicted classification result, the initial category label and the corrected category label of the target image includes: the predicted classification result according to the i-th state, the The initial category label of the target image and the correction category label of the i-th state determine the overall loss of the i-th state of the neural network; according to the overall loss of the i-th state, adjust the network parameters of the i-th state of the neural network to obtain Neural network in state i+1.
  • the method further includes: performing feature extraction on multiple sample images of the k-th category in the training set through a feature extraction network of the i-th state to obtain the i-th state of the multiple sample images
  • the k-th category is one of the K categories of the sample images in the training set, and K is an integer greater than 1; for the i-th state of the multiple sample images of the k-th category Perform clustering processing on the second feature of the k category to determine the class prototype feature of the i-th state of the k category; according to the class prototype feature of the i-th state of the K categories and the first feature of the i-th state of the target image , Determine the correction category label of the i-th state of the target image.
  • the correction category label of the i-th state of the target image is determined based on the prototype features of the i-th state of the K categories and the first characteristic of the i-th state of the target image , Including: respectively acquiring the first feature similarity between the first feature of the i-th state of the target image and the prototype features of the i-th state of the K categories; according to the corresponding value corresponding to the maximum value of the first feature similarity
  • the category to which the prototype feature belongs determines the correction category label of the i-th state of the target image.
  • the class prototype feature of the i-th state of each category includes multiple class prototype features, wherein the first feature of the i-th state of the target image and the K categories of the
  • the first feature similarity between the class prototype features of the i-th state includes: acquiring the second feature similarity between the first feature of the i-th state and the plurality of class prototype features of the i-th state of the k-th category Degree; according to the second feature similarity, the first feature similarity between the first feature of the i-th state and the class prototype feature of the i-th state of the k-th category is determined.
  • the class prototype feature of the i-th state of the k-th category includes the class center of the second feature of the i-th state of the multiple sample images of the k-th category.
  • the overall loss of the i-th state of the neural network is determined according to the predicted classification result of the i-th state, the initial category label of the target image, and the corrected category label of the i-th state, It includes: determining the first loss of the i-th state of the neural network according to the predicted classification result of the i-th state and the initial category label of the target image; according to the predicted classification result of the i-th state and the target The correction category label of the i-th state of the image determines the second loss of the i-th state of the neural network; determines the neural network according to the first loss of the i-th state and the second loss of the i-th state The overall loss of the i-th state.
  • an image processing method comprising: inputting an image to be processed into a neural network for classification processing to obtain an image classification result, wherein the neural network includes training according to the above method.
  • Neural network is provided.
  • a neural network training device which includes: a predictive classification module for classifying target images in a training set through a neural network to obtain a predictive classification result of the target image; network training The module is used to train the neural network according to the predicted classification result, the initial category label and the corrected category label of the target image.
  • the neural network includes a feature extraction network and a classification network, and the neural network includes N training states, where N is an integer greater than 1, wherein the predictive classification module includes: feature extraction A sub-module for feature extraction of the target image through the feature extraction network of the i-th state to obtain the first feature of the i-th state of the target image, where the i-th state is one of the N training states, And 0 ⁇ i ⁇ N; the result determination sub-module is used to classify the first feature of the i-th state of the target image through the i-th state classification network to obtain the predicted classification result of the i-th state of the target image .
  • the predictive classification module includes: feature extraction A sub-module for feature extraction of the target image through the feature extraction network of the i-th state to obtain the first feature of the i-th state of the target image, where the i-th state is one of the N training states, And 0 ⁇ i ⁇ N; the result determination sub-module is used to classify the first feature of the i-
  • the network training module includes: a loss determination module, configured to determine the target image according to the predicted classification result of the i-th state, the initial category label of the target image, and the correction category label of the i-th state The overall loss of the i-th state of the neural network; a parameter adjustment module for adjusting the network parameters of the i-th state neural network according to the overall loss of the i-th state to obtain the i+1-th state neural network.
  • the device further includes: a sample feature extraction module, configured to perform feature extraction on multiple sample images of the k-th category in the training set through the feature extraction network in the i-th state to obtain the multiple The second feature of the i-th state of the sample images, the k-th category is one of the K categories of the sample images in the training set, and K is an integer greater than 1;
  • the clustering module is used to compare the Perform clustering processing on the second feature of the i-th state of the multiple sample images of the k-th category to determine the class prototype feature of the i-th state of the k-th category;
  • the prototype feature of the i state and the first feature of the i state of the target image determine the correction category label of the i state of the target image.
  • the label determination module includes: a similarity acquisition sub-module, configured to respectively acquire the first feature of the i-th state of the target image and the prototype features of the i-th state of the K categories The first feature similarity between the two; the label determination sub-module is used to determine the correction category label of the i-th state of the target image according to the category to which the prototype feature corresponding to the maximum value of the first feature similarity belongs.
  • the class prototype feature of the i-th state of each category includes multiple class prototype features, wherein the similarity acquisition sub-module is used to: obtain the first feature of the i-th state and The second feature similarity between the multiple prototype features of the i-th state of the k-th category; according to the second feature similarity, the first feature of the i-th state and the i-th category of the k-th category are determined The first feature similarity between the prototype features of the state.
  • the class prototype feature of the i-th state of the k-th category includes the class center of the second feature of the i-th state of the multiple sample images of the k-th category.
  • the loss determination module includes: a first loss determination submodule, configured to determine the first loss determination sub-module of the neural network according to the predicted classification result of the i-th state and the initial category label of the target image The first loss of the i state; the second loss determination sub-module is used to determine the i state of the neural network according to the predicted classification result of the i state and the correction category label of the i state of the target image The second loss; the overall loss determination sub-module, which is used to determine the overall loss of the i-th state of the neural network according to the first loss of the i-th state and the second loss of the i-th state.
  • an image processing device includes: an image classification module for inputting an image to be processed into a neural network for classification processing to obtain an image classification result, wherein the neural network Including the neural network trained according to the above device.
  • an electronic device including: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to call the instructions stored in the memory to execute The above method.
  • a computer-readable storage medium having computer program instructions stored thereon, and the computer program instructions implement the foregoing method when executed by a processor.
  • a computer program including computer readable code, and when the computer readable code runs in an electronic device, a processor in the electronic device executes the above method.
  • the training process of the neural network can be jointly monitored by the initial category label and the corrected category label of the target image, and the optimization direction of the neural network can be jointly determined, thereby simplifying the training process and network structure.
  • Fig. 1 shows a flowchart of a neural network training method according to an embodiment of the present disclosure.
  • Fig. 2 shows a schematic diagram of an application example of a neural network training method according to an embodiment of the present disclosure.
  • Fig. 3 shows a block diagram of a neural network training device according to an embodiment of the present disclosure.
  • Fig. 4 shows a block diagram of an electronic device according to an embodiment of the present disclosure.
  • Fig. 5 shows a block diagram of an electronic device according to an embodiment of the present disclosure.
  • Fig. 1 shows a flowchart of a neural network training method according to an embodiment of the present disclosure.
  • the neural network training method includes:
  • step S11 the target image in the training set is classified through the neural network to obtain the predicted classification result of the target image
  • step S12 the neural network is trained according to the predicted classification result, the initial category label and the corrected category label of the target image.
  • the neural network training method can be executed by electronic equipment such as a terminal device or a server.
  • the terminal device can be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone,
  • UE user equipment
  • PDA personal digital assistant
  • the method can be implemented by a processor calling computer-readable instructions stored in a memory.
  • the method can be executed by a server.
  • the training set may include a large number of sample images that are not accurately labeled. These sample images belong to different image categories.
  • the image categories are, for example, face categories (such as faces of different customers) and animal categories (such as Cats, dogs, etc.), clothing categories (such as tops, pants, etc.). This disclosure does not limit the source of the sample image and its specific category.
  • each sample image has an initial category label (noise label), which is used to label the category to which the sample image belongs.
  • initial category label due to the inaccurate labeling, the initial category label of a certain number of sample images may exist error.
  • the present disclosure does not limit the noise distribution of the initial category tags.
  • the neural network to be trained may be, for example, a deep convolutional network, and the present disclosure does not limit the specific network type of the neural network.
  • the target image in the training set can be input into the neural network to be trained for classification processing in step S11 to obtain the predicted classification result of the target image.
  • the target image may be one or more of the sample images, for example, multiple sample images of the same training batch.
  • the predicted classification result may include the predicted category to which the target image belongs.
  • the neural network can be trained in step S12 according to the predicted classification result, the initial category label and the corrected category label of the target image.
  • the correction category tag is used to correct the category of the target image.
  • the network loss of the neural network can be determined according to the predicted classification result, the initial category label and the corrected category label, and the network parameters of the neural network can be adjusted inversely according to the network loss. After many adjustments, a neural network that meets the training conditions (such as network convergence) is finally obtained.
  • the training process of the neural network can be jointly monitored by the initial category label and the corrected category label of the target image, and the optimization direction of the neural network can be jointly determined, thereby simplifying the training process and network structure.
  • the neural network may include a feature extraction network and a classification network.
  • the feature extraction network is used to extract features of the target image
  • the classification network is used to classify the target image according to the extracted features to obtain the predicted classification result of the target image.
  • the feature extraction network may include, for example, multiple convolutional layers
  • the classification network may include, for example, a fully connected layer and a softmax layer.
  • the present disclosure does not limit the specific types and number of network layers of the feature extraction network and the classification network.
  • step S11 may include:
  • the i-th state is one of the N training states, and 0 ⁇ i ⁇ N;
  • the target image can be input into the feature extraction network of the i-th state for feature extraction, and the first feature of the i-th state of the target image can be output; the first feature of the i-th state can be input into the classification network of the i-th state for classification, The predicted classification result of the i-th state of the target image is output.
  • the output result of the neural network in the i-th state can be obtained, so as to train the neural network based on the result.
  • the method further includes:
  • the corrected category label of the i-th state of the target image is determined.
  • the sample images in the training set may include K categories, and K is an integer greater than one.
  • the feature extraction network can be used as a feature extractor to extract features of sample images of various categories.
  • a part of sample images for example, M sample images, M is an integer greater than 1 can be selected from the sample images of the kth category for feature extraction, In order to reduce computing costs.
  • feature extraction can also be performed on all sample images of the k-th category, which is not limited in the present disclosure.
  • M sample images can be randomly selected from the sample images of the k-th category, or M sample images can be selected in other ways (for example, according to parameters such as image clarity). No restrictions.
  • the M sample images of the k-th category can be input into the feature extraction network of the i-th state for feature extraction, and the second feature of the i-th state of the M sample images (M ); Then, clustering can be performed on the M second features of the i-th state to determine the class prototype features of the i-th state of the k-th category.
  • the M second features can be clustered by means of density peak clustering, K-means clustering, spectral clustering, etc.
  • the present disclosure does not limit the manner of clustering .
  • the class prototype feature of the i-th state of the k-th category includes the class center of the second feature of the i-th state of the plurality of sample images of the k-th category. That is, the cluster center of the M second feature clusters of the i-th state can be used as the class prototype feature of the i-th state of the k-th category.
  • there may be multiple prototype features that is, multiple prototype features are selected from M second features.
  • the second feature of the p images (p ⁇ M) with the highest density value can be selected as the prototype feature, or based on the density value and the similarity measurement between the features.
  • Comprehensive considerations to select class prototype features Those skilled in the art can select prototype features according to actual conditions, which are not limited in the present disclosure.
  • the prototype features of the class can be used to represent the features that should be extracted from the samples in each class for comparison with the features of the target image.
  • part of the sample images can be selected from the sample images of the K categories, and the selected images can be input into the feature extraction network to obtain the second feature.
  • Cluster the second features of each category to obtain the prototype features of each category that is, obtain the prototype features of the i-th state of the K categories.
  • the correction category label of the i-th state of the target image can be determined based on the class prototype features of the i-th state of the K categories and the first feature of the i-th state of the target image.
  • the category label of the target image can be corrected to provide additional supervision signals for training the neural network.
  • the step of determining the correction category label of the i-th state of the target image according to the class prototype features of the i-th state of the K categories and the first feature of the i-th state of the target image Can include:
  • the first feature similarity between the first feature of the i-th state of the target image and the prototype features of the i-th state of the K categories can be calculated respectively.
  • the first feature similarity may be, for example, the cosine similarity or Euclidean distance between features, which is not limited in the present disclosure.
  • the maximum value of the first feature similarity of the K categories may be determined, and the category to which the class prototype feature corresponding to the maximum value belongs is determined as the corrected category label of the i-th state of the target image. That is, the label corresponding to the category feature prototype with the greatest similarity is selected to assign a new label to the sample.
  • the class label of the target image can be corrected by the class prototype feature, and the accuracy of the corrected class label can be improved; when the training of the neural network is supervised by the corrected class label, the training effect of the network can be improved.
  • the class prototype feature of the i-th state of each category includes multiple class prototype features, wherein the first feature of the i-th state of the target image and the K categories of the
  • the step of the first feature similarity between the prototype features of the i-th state may include:
  • the first feature similarity between the first feature of the i-th state and the prototype feature of the i-th state of the k-th category is determined.
  • the second feature between the first feature of the i-th state and the multiple prototype features of the i-th state of the k-th category can be calculated respectively.
  • the feature similarity, and then the first feature similarity is determined according to the multiple second feature similarities.
  • the average value of the similarities of multiple second features can be determined as the first feature similarity, or an appropriate similarity value can be selected from the multiple second feature similarities as the first
  • the feature similarity is not limited in this disclosure.
  • step S12 may include:
  • the network parameters of the i-th state neural network are adjusted to obtain the i+1-th state neural network.
  • the difference between the predicted classification result of the i-th state obtained in step S11 and the initial class label of the target image and the corrected class label of the i-th state can be used to calculate the first state of the neural network.
  • the network parameters of the neural network of the i-th state can be adjusted according to the total loss of the N-1th state to obtain the neural network of the Nth state (network convergence) . Therefore, the neural network of the Nth state can be determined as the trained neural network, and the entire training process of the neural network can be completed.
  • the training process of the neural network can be completed in multiple cycles to obtain a high-precision neural network.
  • the predictive classification result of the i-th state, the initial category label of the target image, and the corrected category label of the i-th state determine the overall loss of the i-th state of the neural network Steps can include:
  • the overall loss of the i-th state of the neural network is determined.
  • the first loss of the i-th state of the neural network can be determined according to the difference between the predicted classification result of the i-th state and the initial category label; according to the predicted classification result of the i-th state and the corrected category label of the i-th state The difference between determines the second loss of the i-th state of the neural network.
  • the first loss and the second loss may be, for example, a cross-entropy loss function, and the present disclosure does not limit the specific type of the loss function.
  • the weighted sum of the first loss and the second loss may be determined as the overall loss of the neural network.
  • Those skilled in the art can set the weights of the first loss and the second loss according to the actual situation, and this disclosure does not limit this.
  • the total loss L total can be expressed as:
  • the first loss and the second loss can be determined respectively by the initial category label and the corrected category label, and then the overall loss of the neural network can be determined, thereby realizing the common supervision of the two supervision signals and improving the network training effect.
  • Fig. 2 shows a schematic diagram of an application example of a neural network training method according to an embodiment of the present disclosure. As shown in Figure 2, the application example can be divided into two parts: training phase 21 and label correction phase 22.
  • the target image x may include multiple sample images of one training batch.
  • the target image x can be input to the feature extraction network 211 (including multiple convolutional layers) for processing, and the target image x can be output
  • the first feature input the first feature into the classification network 212 (including the fully connected layer and the softmax layer) for processing, and output the predicted classification result 213 (F( ⁇ ,x)) of the target image x; according to the predicted classification result 213 and
  • the initial category label y, the first loss L(F( ⁇ ,x),y) can be determined; according to the predicted classification result 213 and the corrected category label Determinable second loss
  • the weighted summation of the first loss and the second loss according to the weights 1- ⁇ and ⁇ can obtain the total loss L total .
  • the feature extraction network 211 in this state can be reused, or the network parameters of the feature extraction network 211 in this state can be copied to obtain the feature extraction network 221 of the label correction stage 22.
  • Randomly select M sample images 222 from the sample images of the k-th category in the training set for example, multiple sample images of the category "pants" in FIG. 2
  • the feature set of the selected sample image of the k-th category is output.
  • sample images can be randomly selected from all K categories of sample images, and a feature set 223 of the selected sample images including K categories can be obtained.
  • the feature sets of the selected sample images of each category can be clustered separately, and the prototype features of the cluster can be selected according to the clustering results.
  • the feature corresponding to the cluster center is determined as the prototype feature, or according to The preset rules select p class prototype features. In this way, the prototype features 224 of each category can be obtained.
  • the target image x can be input to the feature extraction network 221 for processing, and the first feature G(x) of the target image x can be output, or the first feature obtained in the training phase 21 can be directly called. Then, the feature similarity between the first feature G(x) of the target image x and the prototype features of each category is respectively calculated; the category of the prototype feature corresponding to the maximum value of the feature similarity is determined as the target image x Calibration category label This completes the process of label correction. Calibration category label It can be input into the training phase 21 as an additional supervision signal in the training phase.
  • the network parameters of the neural network can be adjusted inversely according to the total loss, thereby obtaining the neural network of the next state.
  • the above-mentioned training phase and label correction phase are performed alternately until the network is trained to convergence, and a trainable neural network is obtained.
  • a self-correction stage is added in the network training process to realize the re-correction of the noise data label, and the corrected label is used as a part of the supervision signal, and the original noise label is combined with the supervision network
  • the training process can improve the generalization ability of neural networks after learning in inaccurately labeled data sets.
  • the prototype features of multiple categories can be extracted, and the data distribution in the category can be better expressed through end-to-end self-learning
  • the framework solves the current difficult problem of network training under real noise data sets, and simplifies the training process and network design.
  • the embodiments according to the present disclosure can be applied to computer vision and other fields to realize model training under noisy data.
  • an image processing method which includes: inputting an image to be processed into a neural network for classification processing to obtain an image classification result, wherein the neural network includes the method training described above The resulting neural network.
  • the present disclosure also provides neural network training devices and image processing devices, electronic equipment, computer-readable storage media, and programs, all of which can be used to implement any neural network training method and image processing method provided in the present disclosure.
  • the scheme and description and refer to the corresponding record in the method section will not be repeated.
  • Fig. 3 shows a block diagram of a neural network training device according to an embodiment of the present disclosure.
  • a neural network training device includes: a predictive classification module 31, which is used to classify target images in the training set through a neural network to obtain a predictive classification result of the target image; a network training module 32 uses Training the neural network according to the predicted classification result, the initial category label and the corrected category label of the target image.
  • the neural network includes a feature extraction network and a classification network, and the neural network includes N training states, where N is an integer greater than 1, wherein the predictive classification module includes: feature extraction A sub-module for feature extraction of the target image through the feature extraction network of the i-th state to obtain the first feature of the i-th state of the target image, where the i-th state is one of the N training states, And 0 ⁇ i ⁇ N; the result determination sub-module is used to classify the first feature of the i-th state of the target image through the i-th state classification network to obtain the predicted classification result of the i-th state of the target image .
  • the predictive classification module includes: feature extraction A sub-module for feature extraction of the target image through the feature extraction network of the i-th state to obtain the first feature of the i-th state of the target image, where the i-th state is one of the N training states, And 0 ⁇ i ⁇ N; the result determination sub-module is used to classify the first feature of the i-
  • the network training module includes: a loss determination module, configured to determine the target image according to the predicted classification result of the i-th state, the initial category label of the target image, and the correction category label of the i-th state The overall loss of the i-th state of the neural network; a parameter adjustment module for adjusting the network parameters of the i-th state of the neural network according to the overall loss of the i-th state to obtain the i+1-th state of the neural network.
  • the device further includes: a sample feature extraction module, configured to perform feature extraction on multiple sample images of the k-th category in the training set through the feature extraction network in the i-th state to obtain the multiple The second feature of the i-th state of the sample images, the k-th category is one of the K categories of the sample images in the training set, and K is an integer greater than 1;
  • the clustering module is used to compare the Perform clustering processing on the second feature of the i-th state of the multiple sample images of the k-th category to determine the class prototype feature of the i-th state of the k-th category;
  • the prototype feature of the i state and the first feature of the i state of the target image determine the correction category label of the i state of the target image.
  • the label determination module includes: a similarity acquisition sub-module, configured to respectively acquire the first feature of the i-th state of the target image and the prototype features of the i-th state of the K categories The first feature similarity between the two; the label determination sub-module is used to determine the correction category label of the i-th state of the target image according to the category to which the prototype feature corresponding to the maximum value of the first feature similarity belongs.
  • the class prototype feature of the i-th state of each category includes multiple class prototype features, wherein the similarity acquisition sub-module is used to: obtain the first feature of the i-th state and The second feature similarity between the multiple prototype features of the i-th state of the k-th category; according to the second feature similarity, the first feature of the i-th state and the i-th category of the k-th category are determined The first feature similarity between the prototype features of the state.
  • the class prototype feature of the i-th state of the k-th category includes the class center of the second feature of the i-th state of the multiple sample images of the k-th category.
  • the loss determination module includes: a first loss determination submodule, configured to determine the first loss determination sub-module of the neural network according to the predicted classification result of the i-th state and the initial category label of the target image The first loss of the i state; the second loss determination sub-module is used to determine the i state of the neural network according to the predicted classification result of the i state and the correction category label of the i state of the target image The second loss; the overall loss determination sub-module, which is used to determine the overall loss of the i-th state of the neural network according to the first loss of the i-th state and the second loss of the i-th state.
  • an image processing device includes: an image classification module for inputting an image to be processed into a neural network for classification processing to obtain an image classification result, wherein the neural network Including the neural network trained according to the above device.
  • the functions or modules contained in the device provided in the embodiments of the present disclosure can be used to execute the methods described in the above method embodiments.
  • the functions or modules contained in the device provided in the embodiments of the present disclosure can be used to execute the methods described in the above method embodiments.
  • the embodiments of the present disclosure also provide a computer-readable storage medium on which computer program instructions are stored, and the computer program instructions implement the above-mentioned method when executed by a processor.
  • the computer-readable storage medium may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium.
  • An embodiment of the present disclosure also proposes an electronic device including: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to call the instructions stored in the memory to execute the above method.
  • the embodiment of the present disclosure also proposes a computer program, the computer program includes computer readable code, and when the computer readable code runs in an electronic device, a processor in the electronic device executes the above method.
  • the electronic device can be provided as a terminal, server or other form of device.
  • FIG. 4 shows a block diagram of an electronic device 800 according to an embodiment of the present disclosure.
  • the electronic device 800 may be a mobile phone, a computer, a digital broadcasting terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, and other terminals.
  • the electronic device 800 may include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input/output (I/O) interface 812, and a sensor component 814 , And communication component 816.
  • the processing component 802 generally controls the overall operations of the electronic device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations.
  • the processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the foregoing method.
  • the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components.
  • the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.
  • the memory 804 is configured to store various types of data to support operations in the electronic device 800. Examples of these data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, pictures, videos, etc.
  • the memory 804 can be implemented by any type of volatile or nonvolatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic Disk or Optical Disk.
  • SRAM static random access memory
  • EEPROM electrically erasable programmable read-only memory
  • EPROM erasable Programmable Read Only Memory
  • PROM Programmable Read Only Memory
  • ROM Read Only Memory
  • Magnetic Memory Flash Memory
  • Magnetic Disk Magnetic Disk or Optical Disk.
  • the power supply component 806 provides power for various components of the electronic device 800.
  • the power supply component 806 may include a power management system, one or more power supplies, and other components associated with the generation, management, and distribution of power for the electronic device 800.
  • the multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user.
  • the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user.
  • the touch panel includes one or more touch sensors to sense touch, sliding, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure related to the touch or slide operation.
  • the multimedia component 808 includes a front camera and/or a rear camera. When the electronic device 800 is in an operation mode, such as a shooting mode or a video mode, the front camera and/or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capabilities.
  • the audio component 810 is configured to output and/or input audio signals.
  • the audio component 810 includes a microphone (MIC).
  • the microphone is configured to receive external audio signals.
  • the received audio signal may be further stored in the memory 804 or transmitted via the communication component 816.
  • the audio component 810 further includes a speaker for outputting audio signals.
  • the I/O interface 812 provides an interface between the processing component 802 and a peripheral interface module.
  • the peripheral interface module may be a keyboard, a click wheel, a button, and the like. These buttons may include but are not limited to: home button, volume button, start button, and lock button.
  • the sensor component 814 includes one or more sensors for providing the electronic device 800 with various aspects of state evaluation.
  • the sensor component 814 can detect the on/off status of the electronic device 800 and the relative positioning of the components.
  • the component is the display and the keypad of the electronic device 800.
  • the sensor component 814 can also detect the electronic device 800 or the electronic device 800.
  • the position of the component changes, the presence or absence of contact between the user and the electronic device 800, the orientation or acceleration/deceleration of the electronic device 800, and the temperature change of the electronic device 800.
  • the sensor component 814 may include a proximity sensor configured to detect the presence of nearby objects when there is no physical contact.
  • the sensor component 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications.
  • the sensor component 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.
  • the communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices.
  • the electronic device 800 can access a wireless network based on a communication standard, such as WiFi, 2G, or 3G, or a combination thereof.
  • the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel.
  • the communication component 816 further includes a near field communication (NFC) module to facilitate short-range communication.
  • the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.
  • RFID radio frequency identification
  • IrDA infrared data association
  • UWB ultra-wideband
  • Bluetooth Bluetooth
  • the electronic device 800 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field A programmable gate array (FPGA), controller, microcontroller, microprocessor, or other electronic components are implemented to implement the above methods.
  • ASIC application specific integrated circuits
  • DSP digital signal processors
  • DSPD digital signal processing devices
  • PLD programmable logic devices
  • FPGA field A programmable gate array
  • controller microcontroller, microprocessor, or other electronic components are implemented to implement the above methods.
  • a non-volatile computer-readable storage medium such as a memory 804 including computer program instructions, which can be executed by the processor 820 of the electronic device 800 to complete the foregoing method.
  • FIG. 5 shows a block diagram of an electronic device 1900 according to an embodiment of the present disclosure.
  • the electronic device 1900 may be provided as a server. 5
  • the electronic device 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by the memory 1932, for storing instructions executable by the processing component 1922, such as application programs.
  • the application program stored in the memory 1932 may include one or more modules each corresponding to a set of instructions.
  • the processing component 1922 is configured to execute instructions to perform the above-described methods.
  • the electronic device 1900 may also include a power supply component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input output (I/O) interface 1958 .
  • the electronic device 1900 can operate based on an operating system stored in the memory 1932, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or the like.
  • a non-volatile computer-readable storage medium such as the memory 1932 including computer program instructions, which can be executed by the processing component 1922 of the electronic device 1900 to complete the foregoing method.
  • the present disclosure may be a system, method, and/or computer program product.
  • the computer program product may include a computer-readable storage medium loaded with computer-readable program instructions for enabling a processor to implement various aspects of the present disclosure.
  • the computer-readable storage medium may be a tangible device that can hold and store instructions used by the instruction execution device.
  • the computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing.
  • Computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) Or flash memory), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, mechanical encoding device, such as a printer with instructions stored thereon
  • RAM random access memory
  • ROM read-only memory
  • EPROM erasable programmable read-only memory
  • flash memory flash memory
  • SRAM static random access memory
  • CD-ROM compact disk read-only memory
  • DVD digital versatile disk
  • memory stick floppy disk
  • mechanical encoding device such as a printer with instructions stored thereon
  • the computer-readable storage medium used here is not interpreted as a transient signal itself, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (for example, light pulses through fiber optic cables), or through wires Transmission of electrical signals.
  • the computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to various computing/processing devices, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and/or a wireless network.
  • the network may include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers.
  • the network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network, and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing/processing device .
  • the computer program instructions used to perform the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, status setting data, or in one or more programming languages.
  • Source code or object code written in any combination, the programming language includes object-oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as "C" language or similar programming languages.
  • Computer-readable program instructions can be executed entirely on the user's computer, partly on the user's computer, executed as a stand-alone software package, partly on the user's computer and partly executed on a remote computer, or entirely on the remote computer or server carried out.
  • the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to access the Internet connection).
  • LAN local area network
  • WAN wide area network
  • an electronic circuit such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), can be customized by using the status information of the computer-readable program instructions.
  • the computer-readable program instructions are executed to realize various aspects of the present disclosure.
  • These computer-readable program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine such that when these instructions are executed by the processor of the computer or other programmable data processing device , A device that implements the functions/actions specified in one or more blocks in the flowchart and/or block diagram is produced. It is also possible to store these computer-readable program instructions in a computer-readable storage medium. These instructions make computers, programmable data processing apparatuses, and/or other devices work in a specific manner, so that the computer-readable medium storing instructions includes An article of manufacture, which includes instructions for implementing various aspects of the functions/actions specified in one or more blocks in the flowchart and/or block diagram.
  • each block in the flowchart or block diagram may represent a module, program segment, or part of an instruction, and the module, program segment, or part of an instruction contains one or more functions for implementing the specified logical function.
  • Executable instructions may also occur in a different order from the order marked in the drawings. For example, two consecutive blocks can actually be executed in parallel, or they can sometimes be executed in the reverse order, depending on the functions involved.
  • each block in the block diagram and/or flowchart, and the combination of the blocks in the block diagram and/or flowchart can be implemented by a dedicated hardware-based system that performs the specified functions or actions Or it can be realized by a combination of dedicated hardware and computer instructions.

Landscapes

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

Abstract

本公开涉及一种神经网络训练方法及装置以及图像处理方法及装置。该训练方法包括：通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。本公开实施例可通过初始和校正类别标签共同监督神经网络的训练过程，简化训练过程和网络结构。

Description

神经网络训练方法及装置以及图像处理方法及装置
本申请要求在2019年5月21日提交中国专利局、申请号为201910426010.4、发明名称为“神经网络训练方法及装置以及图像处理方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本公开涉及计算机技术领域，尤其涉及一种神经网络训练方法及装置以及图像处理方法及装置。
背景技术
随着人工智能技术的不断发展，机器学习(尤其是深度学习)在计算机视觉等多个领域都取得了很好的效果。目前的机器学习(深度学习)对大规模的精确标注的数据集有着很强的依赖。
发明内容
本公开提出了一种神经网络训练及图像处理技术方案。
根据本公开的一方面，提供了一种神经网络训练方法，包括：通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
在一种可能的实现方式中，所述神经网络包括特征提取网络和分类网络，并且所述神经网络包括N个训练状态，N为大于1的整数，其中，所述通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果，包括：通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
在一种可能的实现方式中，所述根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络，包括：根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
在一种可能的实现方式中，所述方法还包括：通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；根据K个类别的第i状态的类原型特征以及所述目标图像的第 i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，所述根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签，包括：分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，每个类别的第i状态的类原型特征包括多个类原型特征，其中，所述分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度，包括：获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
在一种可能的实现方式中，所述第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。
在一种可能的实现方式中，所述根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失，包括：根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
根据本公开的另一方面，提供了一种图像处理方法，所述方法包括：将待处理图像输入神经网络中进行分类处理，得到图像分类结果，其中，所述神经网络包括根据上述方法训练得到的神经网络。
根据本公开的另一方面，提供了一种神经网络训练装置，包括：预测分类模块，用于通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；网络训练模块，用于根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
在一种可能的实现方式中，所述神经网络包括特征提取网络和分类网络，并且所述神经网络包括N个训练状态，N为大于1的整数，其中，所述预测分类模块包括：特征提取子模块，用于通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；结果确定子模块，用于通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
在一种可能的实现方式中，所述网络训练模块包括：损失确定模块，用于根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；参数调整模块，用于根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
在一种可能的实现方式中，所述装置还包括：样本特征提取模块，用于通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；聚类模块，用于对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；标签确定模块，用于根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，所述标签确定模块包括：相似度获取子模块，用于分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；标签确定子模块，用于根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，每个类别的第i状态的类原型特征包括多个类原型特征，其中，所述相似度获取子模块用于：获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
在一种可能的实现方式中，所述第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。
在一种可能的实现方式中，损失确定模块包括：第一损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；第二损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；总体损失确定子模块，用于根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
根据本公开的另一方面，提供了一种图像处理装置，所述装置包括：图像分类模块，用于将待处理图像输入神经网络中进行分类处理，得到图像分类结果，其中，所述神经网络包括根据上述装置训练得到的神经网络。
根据本公开的另一方面，提供了一种电子设备，包括：处理器；用于存储处理器可执行指令的存储器；其中，所述处理器被配置为调用所述存储器存储的指令，以执行上述方法。
根据本公开的另一方面，提供了一种计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令被处理器执行时实现上述方法。
根据本公开的一方面，提供了一种计算机程序，所述计算机程序包括计算机可读代码，当所述计算机可读代码在电子设备中运行时，所述电子设备中的处理器执行上述方法。
根据本公开的实施例，能够通过目标图像的初始类别标签和校正类别标签共同监督神经网络的训练过程，共同决定神经网络的优化方向，从而简化训练过程和网络结构。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，而非限制本公开。根据下面参考附图对示例性实施例的详细说明，本公开的其它特征及方面将变得清楚。
附图说明
此处的附图被并入说明书中并构成本说明书的一部分，这些附图示出了符合本公开的实施例，并与说明书一起用于说明本公开的技术方案。
图1示出根据本公开实施例的神经网络训练方法的流程图。
图2示出根据本公开实施例的神经网络训练方法的应用示例的示意图。
图3示出根据本公开实施例的神经网络训练装置的框图。
图4示出根据本公开实施例的一种电子设备的框图。
图5示出根据本公开实施例的一种电子设备的框图。
具体实施方式
以下将参考附图详细说明本公开的各种示例性实施例、特征和方面。附图中相同的附图标记表示功能相同或相似的元件。尽管在附图中示出了实施例的各种方面，但是除非特别指出，不必按比例绘制附图。
在这里专用的词“示例性”意为“用作例子、实施例或说明性”。这里作为“示例性”所说明的任何实施例不必解释为优于或好于其它实施例。
本文中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中术语“至少一种”表示多种中的任意一种或多种中的至少两种的任意组合，例如，包括A、B、C中的至少一种，可以表示包括从A、B和C构成的集合中选择的任意一个或多个元素。
另外，为了更好地说明本公开，在下文的具体实施方式中给出了众多的具体细节。本领域技术人员应当理解，没有某些具体细节，本公开同样可以实施。在一些实例中，对于本领域技术人员熟知的方法、手段、元件和电路未作详细描述，以便于凸显本公开的主旨。
图1示出根据本公开实施例的神经网络训练方法的流程图，如图1所示，所述神经网络训练方法包括：
在步骤S11中，通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；
在步骤S12中，根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
在一种可能的实现方式中，所述神经网络训练方法可以由终端设备或服务器等电子设备执行，终端设备可以为用户设备(User Equipment，UE)、移动设备、用户终端、终 端、蜂窝电话、无绳电话、个人数字处理(Personal Digital Assistant，PDA)、手持设备、计算设备、车载设备、可穿戴设备等，所述方法可以通过处理器调用存储器中存储的计算机可读指令的方式来实现。或者，可通过服务器执行所述方法。
在一种可能的实现方式中，训练集中可包括未精确标注的大量样本图像，这些样本图像属于不同的图像类别，图像的类别例如为人脸类别(例如不同顾客的人脸)、动物类别(例如猫、狗等)、服装类别(例如上衣、裤子等)。本公开对样本图像的来源及其具体类别不作限制。
在一种可能的实现方式中，每个样本图像具有初始类别标签(噪声标签)，用于标注该样本图像所属的类别，但由于未精确标注，导致一定数量的样本图像的初始类别标签可能存在错误。本公开对初始类别标签的噪声分布情况不作限制。
在一种可能的实现方式中，待训练的神经网络可例如为深度卷积网络，本公开对神经网络的具体网络类型不作限制。
在神经网络训练期间，可在步骤S11中将训练集中的目标图像输入到待训练的神经网络中进行分类处理，得到目标图像的预测分类结果。其中，目标图像可以是样本图像中的一个或多个，例如同一训练批次的多个样本图像。预测分类结果可包括目标图像所属的预测类别。
在得到目标图像的预测分类结果后，可在步骤S12中根据预测分类结果、目标图像的初始类别标签及校正类别标签，训练神经网络。其中，校正类别标签用于对目标图像的类别进行校正。也就是说，可根据预测分类结果、初始类别标签及校正类别标签确定神经网络的网络损失，根据该网络损失反向调整神经网络的网络参数。经多次调整后，最终得到满足训练条件(例如网络收敛)的神经网络。
根据本公开的实施例，能够通过目标图像的初始类别标签和校正类别标签共同监督神经网络的训练过程，共同决定神经网络的优化方向，从而简化训练过程和网络结构。
在一种可能的实现方式中，该神经网络可包括特征提取网络和分类网络。特征提取网络用于对目标图像进行特征提取，分类网络用于根据提取到的特征对目标图像进行分类，得到目标图像的预测分类结果。其中，特征提取网络可例如包括多个卷积层，分类网络可例如包括全连接层和softmax层等。本公开对特征提取网络和分类网络的网络层的具体类型及数量不作限制。
在训练神经网络的过程中，会多次调整神经网络的网络参数。对当前状态的神经网络进行调整后，可得到下一个状态的神经网络。可设定神经网络包括N个训练状态，N为大于1的整数。这样，对于当前的第i状态的神经网络，步骤S11可包括：
通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；
通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
也就是说，可将目标图像输入第i状态的特征提取网络进行特征提取，输出目标图像 的第i状态的第一特征；将第i状态的第一特征输入第i状态的分类网络进行分类，输出目标图像的第i状态的预测分类结果。
通过这种方式，可以得到第i状态的神经网络的输出结果，以便根据该结果训练神经网络。
在一种可能的实现方式中，所述方法还包括：
通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；
对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；
根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
举例来说，训练集中的样本图像可包括K个类别，K为大于1的整数。可以以特征提取网络作为特征提取器，提取各个类别的样本图像的特征。对于K个类别中的第k个类别(1≤k≤K)，可以从第k个类别的样本图像中选取部分样本图像(例如M个样本图像，M为大于1的整数)进行特征提取，以便降低计算成本。应当理解，也可以对第k个类别的全部样本图像进行特征提取，本公开对此不作限制。
在一种可能的实现方式中，可从第k个类别的样本图像中随机选取M个样本图像，也可以采用其它方式(例如根据图像清晰度等参数)选取M个样本图像，本公开对此不作限制。
在一种可能的实现方式中，可以将第k个类别的M个样本图像分别输入第i状态的特征提取网络中进行特征提取，输出M个样本图像的第i状态的第二特征(M个)；然后，可对第i状态的M个第二特征进行聚类处理，以便确定第k个类别的第i状态的类原型特征。
在一种可能的实现方式中，可采用密度峰值聚类、K均值(K-means)聚类、谱聚类等方式对M个第二特征进行聚类，本公开对聚类的方式不作限制。
在一种可能的实现方式中，第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。也即，可将对第i状态的M个第二特征聚类的类中心作为第k个类别的第i状态的类原型特征。
在一种可能的实现方式中，类原型特征可以为多个，也即从M个第二特征中选择多个类原型特征。例如，在采用密度峰值聚类的方式时，可选取密度值最高的p个图像(p<M)的第二特征作为类原型特征，也可根据密度值和特征之间相似性测度等参数的综合考量来选取类原型特征。本领域技术人员可根据实际情况选取类原型特征，本公开对此不作限制。
通过这种方式，可以通过类原型特征来代表每一类中的样本应该提取出的特征，以便与目标图像的特征进行比对。
在一种可能的实现方式中，可从K类别的样本图像中分别选取部分样本图像，将选中 的图像分别输入特征提取网络中得到第二特征。分别对各个类别的第二特征聚类，获取各个类别的类原型特征，也即得到K个类别的第i状态的类原型特征。进而，可根据K个类别的第i状态的类原型特征以及目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
通过这种方式，可以对目标图像的类别标签进行校正，为训练神经网络提供额外的监督信号。
在一种可能的实现方式中，根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签的步骤，可包括：
分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；
根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
举例来说，如果目标图像属于某个类别，则该目标图像的特征与该类别中的样本应该提取出的特征(类原型特征)相似度较高。因此，可以分别计算目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度。该第一特征相似度可例如为特征之间的余弦相似度或欧氏距离等，本公开对此不作限制。
在一种可能的实现方式中，可确定K个类别的第一特征相似度中的最大值，将该最大值对应的类原型特征所属的类别确定为目标图像的第i状态的校正类别标签。也即，选择相似度最大的类别特征原型所对应的标签给该样本赋予新的标签。
通过这种方式，可以通过类原型特征对目标图像的类别标签进行校正，提高校正的类别标签的准确性；在采用校正类别标签监督神经网络的训练时，能够提高网络的训练效果。
在一种可能的实现方式中，每个类别的第i状态的类原型特征包括多个类原型特征，其中，所述分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度的步骤，可包括：
获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；
根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
举例来说，类原型特征可以为多个，以便更准确地代表每一类中的样本应该提取出的特征。在该情况下，对于K个类别的任意一个类别(第k个类别)，可分别计算第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度，再根据多个第二特征相似度确定第一特征相似度。
在一种可能的实现方式中，可例如将多个第二特征相似度的平均值确定为第一特征相似度，也可以从多个第二特征相似度中选取适当的相似度值作为第一特征相似度，本公开对此不作限制。
通过这种方式，可进一步提高目标图像的特征与类原型特征之间的相似度计算的准确性。
在一种可能的实现方式中，在确定出目标图像的第i状态的校正类别标签后，可根据该校正类别标签训练神经网络。其中，步骤S12可包括：
根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；
根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
举例来说，对于当前的第i状态，可根据步骤S11中得到的第i状态的预测分类结果与目标图像的初始类别标签及第i状态的校正类别标签之间的差异，计算神经网络的第i状态的总体损失；进而根据该总体损失反向调整第i状态的神经网络的网络参数，从而得到下一个训练状态(第i+1状态)的神经网络。
在一种可能的实现方式中，在第一次训练之前，神经网络为初始状态(i＝0)，可仅采用初始类别标签去监督网络的训练。也即，根据初始状态的预测分类结果和初始类别标签来确定神经网络的总体损失，进而反向调整网络参数，得到下一训练状态(i＝1)的神经网络。
在一种可能的实现方式中，当i＝N-1时，可根据第N-1状态的总体损失，调整第i状态的神经网络的网络参数，得到第N状态的神经网络(网络收敛)。从而，可将第N状态的神经网络确定为已训练的神经网络，完成神经网络的整个训练过程。
通过这种方式，可以多次循环完成神经网络的训练过程，得到高精度的神经网络。
在一种可能的实现方式中，所述根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失的步骤，可包括：
根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；
根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；
根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
举例来说，可根据第i状态的预测分类结果和初始类别标签之间的差异，确定神经网络的第i状态的第一损失；根据第i状态的预测分类结果和第i状态的校正类别标签之间的差异，确定神经网络的第i状态的第二损失。其中，第一损失和第二损失可例如为交叉熵损失函数，本公开对损失函数的具体类型不作限制。
在一种可能的实现方式中，可将第一损失与第二损失的加权和确定为神经网络的总体损失。本领域技术人员可根据实际情况设定第一损失和第二损失的权重，本公开对此不作限制。
在一种可能的实现方式中，总体损失L total可表示为：
Figure PCTCN2019114470-appb-000001
在公式(1)中，x可表示目标图像；θ可表示神经网络的网络参数；F(θ,x)可表示预测分类结果；y可表示初始类别标签；
Figure PCTCN2019114470-appb-000002
可表示校正类别标签；L(F(θ,x),y)可表示第一损失；
Figure PCTCN2019114470-appb-000003
可表示第二损失；α可表示第二损失的权重。
通过这种方式，可通过初始类别标签及校正类别标签分别确定第一损失和第二损失，进而确定神经网络的总体损失，从而实现两个监督信号的共同监督，提高网络训练效果。
图2示出根据本公开实施例的神经网络训练方法的应用示例的示意图。如图2所示，该应用示例可分为训练阶段21和标签校正阶段22两个部分。
在该应用示例中，目标图像x可包括一个训练批次的多个样本图像。在神经网络训练过程中的任意一个中间状态(例如第i状态)下，对于训练阶段21，可将目标图像x输入到特征提取网络211(包括多个卷积层)中处理，输出目标图像x的第一特征；将第一特征输入到分类网络212(包括全连接层和softmax层)中处理，输出目标图像x的预测分类结果213(F(θ,x))；根据预测分类结果213和初始类别标签y，可确定第一损失L(F(θ,x),y)；根据预测分类结果213和校正类别标签
Figure PCTCN2019114470-appb-000004
可确定第二损失
Figure PCTCN2019114470-appb-000005
根据权重1-α和α对第一损失和第二损失进行加权求和，可得到总体损失L total。
在该应用示例中，对于标签校正阶段22，可复用该状态下的特征提取网络211，或复制该状态下特征提取网络211的网络参数，得到标签校正阶段22的特征提取网络221。从训练集中第k个类别的样本图像中随机选取M个样本图像222(例如图2中的类别为“裤子”的多个样本图像)，并将选中的M个样本图像222分别输入特征提取网络221中处理，输出第k个类别的选中的样本图像的特征集。这样，可以从所有的K个类别的样本图像中随机选取样本图像，得到包括K个类别的选中的样本图像的特征集223。
在该应用示例中，可以对每个类别的选中的样本图像的特征集分别进行聚类处理，并根据聚类结果选取类原型特征，例如将类中心对应的特征确定为类原型特征，或根据预设的规则选取p个类原型特征。这样，可得到各个类别的类原型特征224。
在该应用示例中，可以将目标图像x输入到特征提取网络221中处理，输出目标图像x的第一特征G(x)，也可以直接调用训练阶段21中得到的第一特征。然后，分别计算目标图像x的第一特征G(x)与各个类别的类原型特征之间的特征相似度；将与特征相似度的最大值对应的类原型特征的类别确定为目标图像x的校正类别标签
Figure PCTCN2019114470-appb-000006
从而完成标签校正的过程。校正类别标签
Figure PCTCN2019114470-appb-000007
可输入到训练阶段21中作为训练阶段的额外监督信号。
在该应用示例中，对于训练阶段21，在根据预测分类结果213、初始类别标签y、校 正类别标签
Figure PCTCN2019114470-appb-000008
确定总体损失L total后，可根据总体损失反向调整神经网络的网络参数，从而得到下一个状态的神经网络。
上述的训练阶段和标签校正阶段交替进行，直到网络训练到收敛，得到可训练后的神经网络。
根据本公开实施例的神经网络训练方法，在网络训练过程中加入自我校正的阶段，实现噪声数据标签的重新校正，并把校正之后的标签作为监督信号的一部分，与原来的噪声标签联合监督网络的训练过程，能够提升神经网络在非准确标注的数据集中学习之后的泛化能力。
根据本公开的实施例，不需要预先假定噪声分布，不需要额外的监督数据及辅助网络，能够提取出多个类别的原型特征，更好的表达类别中的数据分布，通过端到端的自我学习框架来解决当前在真实噪声数据集下网络训练困难的问题，简化了训练过程和网络设计。根据本公开的实施例能够应用于计算机视觉等领域，实现在噪声数据下模型的训练。
根据本公开的实施例，还提供了一种图像处理方法，该方法包括：将待处理图像输入神经网络中进行分类处理，得到图像分类结果，其中，所述神经网络包括如上所述的方法训练得到的神经网络。通过这种方式，可以以规模较小的单个网络实现高性能的图像处理。
可以理解，本公开提及的上述各个方法实施例，在不违背原理逻辑的情况下，均可以彼此相互结合形成结合后的实施例，限于篇幅，本公开不再赘述。本领域技术人员可以理解，在具体实施方式的上述方法中，各步骤的具体执行顺序应当以其功能和可能的内在逻辑确定。
此外，本公开还提供了神经网络训练装置及图像处理装置、电子设备、计算机可读存储介质、程序，上述均可用来实现本公开提供的任一种神经网络训练方法及图像处理方法，相应技术方案和描述和参见方法部分的相应记载，不再赘述。
图3示出根据本公开实施例的神经网络训练装置的框图。根据本公开的另一方面，提供了一种神经网络训练装置。如图3所示，所述神经网络训练装置包括：预测分类模块31，用于通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；网络训练模块32，用于根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
在一种可能的实现方式中，所述神经网络包括特征提取网络和分类网络，并且所述神经网络包括N个训练状态，N为大于1的整数，其中，所述预测分类模块包括：特征提取子模块，用于通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；结果确定子模块，用于通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
在一种可能的实现方式中，所述网络训练模块包括：损失确定模块，用于根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；参数调整模块，用于根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
在一种可能的实现方式中，所述装置还包括：样本特征提取模块，用于通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；聚类模块，用于对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；标签确定模块，用于根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，所述标签确定模块包括：相似度获取子模块，用于分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；标签确定子模块，用于根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
在一种可能的实现方式中，每个类别的第i状态的类原型特征包括多个类原型特征，其中，所述相似度获取子模块用于：获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
在一种可能的实现方式中，所述第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。
在一种可能的实现方式中，损失确定模块包括：第一损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；第二损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；总体损失确定子模块，用于根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
根据本公开的另一方面，提供了一种图像处理装置，所述装置包括：图像分类模块，用于将待处理图像输入神经网络中进行分类处理，得到图像分类结果，其中，所述神经网络包括根据上述装置训练得到的神经网络。
在一些实施例中，本公开实施例提供的装置具有的功能或包含的模块可以用于执行上文方法实施例描述的方法，其具体实现可以参照上文方法实施例的描述，为了简洁，这里不再赘述。
本公开实施例还提出一种计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令被处理器执行时实现上述方法。计算机可读存储介质可以是非易失性计算机可读存储介质或易失性计算机可读存储介质。
本公开实施例还提出一种电子设备，包括：处理器；用于存储处理器可执行指令的存储器；其中，所述处理器被配置为调用所述存储器存储的指令，以执行上述方法。
本公开实施例还提出一种计算机程序，所述计算机程序包括计算机可读代码，当所述计算机可读代码在电子设备中运行时，所述电子设备中的处理器执行上述方法。
电子设备可以被提供为终端、服务器或其它形态的设备。
图4示出根据本公开实施例的一种电子设备800的框图。例如，电子设备800可以是移动电话，计算机，数字广播终端，消息收发设备，游戏控制台，平板设备，医疗设备，健身设备，个人数字助理等终端。
参照图4，电子设备800可以包括以下一个或多个组件：处理组件802，存储器804，电源组件806，多媒体组件808，音频组件810，输入/输出(I/O)的接口812，传感器组件814，以及通信组件816。
处理组件802通常控制电子设备800的整体操作，诸如与显示，电话呼叫，数据通信，相机操作和记录操作相关联的操作。处理组件802可以包括一个或多个处理器820来执行指令，以完成上述的方法的全部或部分步骤。此外，处理组件802可以包括一个或多个模块，便于处理组件802和其他组件之间的交互。例如，处理组件802可以包括多媒体模块，以方便多媒体组件808和处理组件802之间的交互。
存储器804被配置为存储各种类型的数据以支持在电子设备800的操作。这些数据的示例包括用于在电子设备800上操作的任何应用程序或方法的指令，联系人数据，电话簿数据，消息，图片，视频等。存储器804可以由任何类型的易失性或非易失性存储设备或者它们的组合实现，如静态随机存取存储器(SRAM)，电可擦除可编程只读存储器(EEPROM)，可擦除可编程只读存储器(EPROM)，可编程只读存储器(PROM)，只读存储器(ROM)，磁存储器，快闪存储器，磁盘或光盘。
电源组件806为电子设备800的各种组件提供电力。电源组件806可以包括电源管理系统，一个或多个电源，及其他与为电子设备800生成、管理和分配电力相关联的组件。
多媒体组件808包括在所述电子设备800和用户之间的提供一个输出接口的屏幕。在一些实施例中，屏幕可以包括液晶显示器(LCD)和触摸面板(TP)。如果屏幕包括触摸面板，屏幕可以被实现为触摸屏，以接收来自用户的输入信号。触摸面板包括一个或多个触摸传感器以感测触摸、滑动和触摸面板上的手势。所述触摸传感器可以不仅感测触摸或滑动动作的边界，而且还检测与所述触摸或滑动操作相关的持续时间和压力。在一些实施例中，多媒体组件808包括一个前置摄像头和/或后置摄像头。当电子设备800处于操作模式，如拍摄模式或视频模式时，前置摄像头和/或后置摄像头可以接收外部的多媒体数据。每个前置摄像头和后置摄像头可以是一个固定的光学透镜系统或具有焦距和光学变焦能力。
音频组件810被配置为输出和/或输入音频信号。例如，音频组件810包括一个麦克风(MIC)，当电子设备800处于操作模式，如呼叫模式、记录模式和语音识别模式时，麦克风被配置为接收外部音频信号。所接收的音频信号可以被进一步存储在存储器804或经 由通信组件816发送。在一些实施例中，音频组件810还包括一个扬声器，用于输出音频信号。
I/O接口812为处理组件802和外围接口模块之间提供接口，上述外围接口模块可以是键盘，点击轮，按钮等。这些按钮可包括但不限于：主页按钮、音量按钮、启动按钮和锁定按钮。
传感器组件814包括一个或多个传感器，用于为电子设备800提供各个方面的状态评估。例如，传感器组件814可以检测到电子设备800的打开/关闭状态，组件的相对定位，例如所述组件为电子设备800的显示器和小键盘，传感器组件814还可以检测电子设备800或电子设备800一个组件的位置改变，用户与电子设备800接触的存在或不存在，电子设备800方位或加速/减速和电子设备800的温度变化。传感器组件814可以包括接近传感器，被配置用来在没有任何的物理接触时检测附近物体的存在。传感器组件814还可以包括光传感器，如CMOS或CCD图像传感器，用于在成像应用中使用。在一些实施例中，该传感器组件814还可以包括加速度传感器，陀螺仪传感器，磁传感器，压力传感器或温度传感器。
通信组件816被配置为便于电子设备800和其他设备之间有线或无线方式的通信。电子设备800可以接入基于通信标准的无线网络，如WiFi，2G或3G，或它们的组合。在一个示例性实施例中，通信组件816经由广播信道接收来自外部广播管理系统的广播信号或广播相关信息。在一个示例性实施例中，所述通信组件816还包括近场通信(NFC)模块，以促进短程通信。例如，在NFC模块可基于射频识别(RFID)技术，红外数据协会(IrDA)技术，超宽带(UWB)技术，蓝牙(BT)技术和其他技术来实现。
在示例性实施例中，电子设备800可以被一个或多个应用专用集成电路(ASIC)、数字信号处理器(DSP)、数字信号处理设备(DSPD)、可编程逻辑器件(PLD)、现场可编程门阵列(FPGA)、控制器、微控制器、微处理器或其他电子元件实现，用于执行上述方法。
在示例性实施例中，还提供了一种非易失性计算机可读存储介质，例如包括计算机程序指令的存储器804，上述计算机程序指令可由电子设备800的处理器820执行以完成上述方法。
图5示出根据本公开实施例的一种电子设备1900的框图。例如，电子设备1900可以被提供为一服务器。参照图5，电子设备1900包括处理组件1922，其进一步包括一个或多个处理器，以及由存储器1932所代表的存储器资源，用于存储可由处理组件1922的执行的指令，例如应用程序。存储器1932中存储的应用程序可以包括一个或一个以上的每一个对应于一组指令的模块。此外，处理组件1922被配置为执行指令，以执行上述方法。
电子设备1900还可以包括一个电源组件1926被配置为执行电子设备1900的电源管理，一个有线或无线网络接口1950被配置为将电子设备1900连接到网络，和一个输入输出(I/O)接口1958。电子设备1900可以操作基于存储在存储器1932的操作系统，例如Windows ServerTM，Mac OS XTM，UnixTM,LinuxTM，FreeBSDTM或类似。
在示例性实施例中，还提供了一种非易失性计算机可读存储介质，例如包括计算机程序指令的存储器1932，上述计算机程序指令可由电子设备1900的处理组件1922执行以完成上述方法。
本公开可以是系统、方法和/或计算机程序产品。计算机程序产品可以包括计算机可读存储介质，其上载有用于使处理器实现本公开的各个方面的计算机可读程序指令。
计算机可读存储介质可以是可以保持和存储由指令执行设备使用的指令的有形设备。计算机可读存储介质例如可以是――但不限于――电存储设备、磁存储设备、光存储设备、电磁存储设备、半导体存储设备或者上述的任意合适的组合。计算机可读存储介质的更具体的例子(非穷举的列表)包括：便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、静态随机存取存储器(SRAM)、便携式压缩盘只读存储器(CD-ROM)、数字多功能盘(DVD)、记忆棒、软盘、机械编码设备、例如其上存储有指令的打孔卡或凹槽内凸起结构、以及上述的任意合适的组合。这里所使用的计算机可读存储介质不被解释为瞬时信号本身，诸如无线电波或者其他自由传播的电磁波、通过波导或其他传输媒介传播的电磁波(例如，通过光纤电缆的光脉冲)、或者通过电线传输的电信号。
这里所描述的计算机可读程序指令可以从计算机可读存储介质下载到各个计算/处理设备，或者通过网络、例如因特网、局域网、广域网和/或无线网下载到外部计算机或外部存储设备。网络可以包括铜传输电缆、光纤传输、无线传输、路由器、防火墙、交换机、网关计算机和/或边缘服务器。每个计算/处理设备中的网络适配卡或者网络接口从网络接收计算机可读程序指令，并转发该计算机可读程序指令，以供存储在各个计算/处理设备中的计算机可读存储介质中。
用于执行本公开操作的计算机程序指令可以是汇编指令、指令集架构(ISA)指令、机器指令、机器相关指令、微代码、固件指令、状态设置数据、或者以一种或多种编程语言的任意组合编写的源代码或目标代码，所述编程语言包括面向对象的编程语言—诸如Smalltalk、C++等，以及常规的过程式编程语言—诸如“C”语言或类似的编程语言。计算机可读程序指令可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络—包括局域网(LAN)或广域网(WAN)—连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。在一些实施例中，通过利用计算机可读程序指令的状态信息来个性化定制电子电路，例如可编程逻辑电路、现场可编程门阵列(FPGA)或可编程逻辑阵列(PLA)，该电子电路可以执行计算机可读程序指令，从而实现本公开的各个方面。
这里参照根据本公开实施例的方法、装置(系统)和计算机程序产品的流程图和/或框图描述了本公开的各个方面。应当理解，流程图和/或框图的每个方框以及流程图和/或框图中各方框的组合，都可以由计算机可读程序指令实现。
这些计算机可读程序指令可以提供给通用计算机、专用计算机或其它可编程数据处理装置的处理器，从而生产出一种机器，使得这些指令在通过计算机或其它可编程数据处理装置的处理器执行时，产生了实现流程图和/或框图中的一个或多个方框中规定的功能/动作的装置。也可以把这些计算机可读程序指令存储在计算机可读存储介质中，这些指令使得计算机、可编程数据处理装置和/或其他设备以特定方式工作，从而，存储有指令的计算机可读介质则包括一个制造品，其包括实现流程图和/或框图中的一个或多个方框中规定的功能/动作的各个方面的指令。
也可以把计算机可读程序指令加载到计算机、其它可编程数据处理装置、或其它设备上，使得在计算机、其它可编程数据处理装置或其它设备上执行一系列操作步骤，以产生计算机实现的过程，从而使得在计算机、其它可编程数据处理装置、或其它设备上执行的指令实现流程图和/或框图中的一个或多个方框中规定的功能/动作。
附图中的流程图和框图显示了根据本公开的多个实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段或指令的一部分，所述模块、程序段或指令的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个连续的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或动作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
在不违背逻辑的情况下，本公开不同实施例之间可以相互结合，不同实施例描述有所侧重，为侧重描述的部分可以参见其他实施例的记载。
以上已经描述了本公开的各实施例，上述说明是示例性的，并非穷尽性的，并且也不限于所披露的各实施例。在不偏离所说明的各实施例的范围和精神的情况下，对于本技术领域的普通技术人员来说许多修改和变更都是显而易见的。本文中所用术语的选择，旨在最好地解释各实施例的原理、实际应用或对市场中的技术的改进，或者使本技术领域的其它普通技术人员能理解本文披露的各实施例。

Claims (21)

  1. 一种神经网络训练方法，其特征在于，包括：
    通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；
    根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
  2. 根据权利要求1所述的方法，其特征在于，所述神经网络包括特征提取网络和分类网络，并且所述神经网络包括N个训练状态，N为大于1的整数，
    其中，所述通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果，包括：
    通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；
    通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
  3. 根据权利要求2所述的方法，其特征在于，所述根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络，包括：
    根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；
    根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
  4. 根据权利要求2或3所述的方法，其特征在于，所述方法还包括：
    通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；
    对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；
    根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
  5. 根据权利要求4所述的方法，其特征在于，所述根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签，包括：
    分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；
    根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
  6. 根据权利要求5所述的方法，其特征在于，每个类别的第i状态的类原型特征包括多个类原型特征，
    其中，所述分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度，包括：
    获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；
    根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
  7. 根据权利要求4-6中任意一项所述的方法，其特征在于，所述第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。
  8. 根据权利要求3-7中任意一项所述的方法，其特征在于，所述根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失，包括：
    根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；
    根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；
    根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
  9. 一种图像处理方法，其特征在于，所述方法包括：
    将待处理图像输入神经网络中进行分类处理，得到图像分类结果，
    其中，所述神经网络包括根据权利要求1-8中任意一项所述的方法训练得到的神经网络。
  10. 一种神经网络训练装置，其特征在于，包括：
    预测分类模块，用于通过神经网络对训练集中的目标图像进行分类处理，得到所述目标图像的预测分类结果；
    网络训练模块，用于根据所述预测分类结果、所述目标图像的初始类别标签及校正类别标签，训练所述神经网络。
  11. 根据权利要求10所述的装置，其特征在于，所述神经网络包括特征提取网络和分类网络，并且所述神经网络包括N个训练状态，N为大于1的整数，其中，所述预测分类模块包括：
    特征提取子模块，用于通过第i状态的特征提取网络对目标图像进行特征提取，得到所述目标图像的第i状态的第一特征，所述第i状态为所述N个训练状态中的一个，且0≤i<N；
    结果确定子模块，用于通过第i状态的分类网络对所述目标图像的第i状态的第一特征进行分类，得到所述目标图像的第i状态的预测分类结果。
  12. 根据权利要求11所述的装置，其特征在于，所述网络训练模块包括：
    损失确定模块，用于根据第i状态的预测分类结果、所述目标图像的初始类别标签及第i状态的校正类别标签，确定所述神经网络的第i状态的总体损失；
    参数调整模块，用于根据所述第i状态的总体损失，调整第i状态的神经网络的网络参数，得到第i+1状态的神经网络。
  13. 根据权利要求11或12所述的装置，其特征在于，所述装置还包括：
    样本特征提取模块，用于通过第i状态的特征提取网络对训练集中第k个类别的多个样本图像进行特征提取，得到所述多个样本图像的第i状态的第二特征，所述第k个类别是所述训练集中的样本图像的K个类别中的一个，K为大于1的整数；
    聚类模块，用于对所述第k个类别的多个样本图像的第i状态的第二特征进行聚类处理，确定所述第k个类别的第i状态的类原型特征；
    标签确定模块，用于根据K个类别的第i状态的类原型特征以及所述目标图像的第i状态的第一特征，确定所述目标图像的第i状态的校正类别标签。
  14. 根据权利要求13所述的装置，其特征在于，所述标签确定模块包括：
    相似度获取子模块，用于分别获取所述目标图像的第i状态的第一特征与K个类别的第i状态的类原型特征之间的第一特征相似度；
    标签确定子模块，用于根据与第一特征相似度的最大值对应的类原型特征所属的类别，确定所述目标图像的第i状态的校正类别标签。
  15. 根据权利要求14所述的装置，其特征在于，每个类别的第i状态的类原型特征包括多个类原型特征，其中，所述相似度获取子模块用于：
    获取所述第i状态的第一特征与第k个类别的第i状态的多个类原型特征之间的第二特征相似度；
    根据所述第二特征相似度，确定所述第i状态的第一特征与第k个类别的第i状态的类原型特征之间的第一特征相似度。
  16. 根据权利要求13-15中任意一项所述的装置，其特征在于，所述第k个类别的第i状态的类原型特征包括所述第k个类别的多个样本图像的第i状态的第二特征的类中心。
  17. 根据权利要求12-16中任意一项所述的装置，其特征在于，损失确定模块包括：
    第一损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的初始类别标签，确定所述神经网络的第i状态的第一损失；
    第二损失确定子模块，用于根据所述第i状态的预测分类结果以及所述目标图像的第i状态的校正类别标签，确定所述神经网络的第i状态的第二损失；
    总体损失确定子模块，用于根据所述第i状态的第一损失和所述第i状态的第二损失，确定所述神经网络的第i状态的总体损失。
  18. 一种图像处理装置，其特征在于，所述装置包括：
    图像分类模块，用于将待处理图像输入神经网络中进行分类处理，得到图像分类结果，其中，所述神经网络包括根据权利要求10-17中任意一项所述的装置训练得到的神经网络。
  19. 一种电子设备，其特征在于，包括：
    处理器；
    用于存储处理器可执行指令的存储器；
    其中，所述处理器被配置为调用所述存储器存储的指令，以执行权利要求1至9中任意一项所述的方法。
  20. 一种计算机可读存储介质，其上存储有计算机程序指令，其特征在于，所述计算机程序指令被处理器执行时实现权利要求1至9中任意一项所述的方法。
  21. 一种计算机程序，其特征在于，所述计算机程序包括计算机可读代码，当所述计算机可读代码在电子设备中运行时，所述电子设备中的处理器执行用于实现权利要求1至9中的任意一项所述的方法。
PCT/CN2019/114470 2019-05-21 2019-10-30 神经网络训练方法及装置以及图像处理方法及装置 WO2020232977A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2021538254A JP2022516518A (ja) 2019-05-21 2019-10-30 ニューラルネットワークをトレーニングする方法及び装置、画像を処理する方法及び装置
SG11202106979WA SG11202106979WA (en) 2019-05-21 2019-10-30 Neural network training method and apparatus, and image processing method and apparatus
US17/364,731 US20210326708A1 (en) 2019-05-21 2021-06-30 Neural network training method and apparatus, and image processing method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910426010.4A CN110210535B (zh) 2019-05-21 2019-05-21 神经网络训练方法及装置以及图像处理方法及装置
CN201910426010.4 2019-05-21

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/364,731 Continuation US20210326708A1 (en) 2019-05-21 2021-06-30 Neural network training method and apparatus, and image processing method and apparatus

Publications (1)

Publication Number Publication Date
WO2020232977A1 true WO2020232977A1 (zh) 2020-11-26

Family

ID=67788041

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/114470 WO2020232977A1 (zh) 2019-05-21 2019-10-30 神经网络训练方法及装置以及图像处理方法及装置

Country Status (6)

Country Link
US (1) US20210326708A1 (zh)
JP (1) JP2022516518A (zh)
CN (1) CN110210535B (zh)
SG (1) SG11202106979WA (zh)
TW (1) TWI759722B (zh)
WO (1) WO2020232977A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112861975A (zh) * 2021-02-10 2021-05-28 北京百度网讯科技有限公司 分类模型的生成方法、分类方法、装置、电子设备与介质
CN113486957A (zh) * 2021-07-07 2021-10-08 西安商汤智能科技有限公司 神经网络训练和图像处理方法及装置
CN115563522A (zh) * 2022-12-02 2023-01-03 湖南工商大学 交通数据的聚类方法、装置、设备及介质

Families Citing this family (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110210535B (zh) * 2019-05-21 2021-09-10 北京市商汤科技开发有限公司 神经网络训练方法及装置以及图像处理方法及装置
KR20210017185A (ko) * 2019-08-07 2021-02-17 한국전자통신연구원 심층 신경망을 기반으로 영상의 압축 포아송 잡음을 제거하는 방법 및 장치
CN110647938B (zh) * 2019-09-24 2022-07-15 北京市商汤科技开发有限公司 图像处理方法及相关装置
US11429809B2 (en) 2019-09-24 2022-08-30 Beijing Sensetime Technology Development Co., Ltd Image processing method, image processing device, and storage medium
CN110659625A (zh) * 2019-09-29 2020-01-07 深圳市商汤科技有限公司 物体识别网络的训练方法及装置、电子设备和存储介质
CN110991321B (zh) * 2019-11-29 2023-05-02 北京航空航天大学 一种基于标签更正与加权特征融合的视频行人再识别方法
CN111292329B (zh) * 2020-01-15 2023-06-06 北京字节跳动网络技术有限公司 视频分割网络的训练方法、装置及电子设备
CN111310806B (zh) * 2020-01-22 2024-03-15 北京迈格威科技有限公司 分类网络及图像处理方法、装置、系统和存储介质
CN111368923B (zh) * 2020-03-05 2023-12-19 上海商汤智能科技有限公司 神经网络训练方法及装置、电子设备和存储介质
CN111414921B (zh) * 2020-03-25 2024-03-15 抖音视界有限公司 样本图像处理方法、装置、电子设备及计算机存储介质
CN111461304B (zh) * 2020-03-31 2023-09-15 北京小米松果电子有限公司 分类神经网络的训练方法、文本分类方法、装置及设备
CN111507419B (zh) * 2020-04-22 2022-09-30 腾讯科技（深圳）有限公司 图像分类模型的训练方法及装置
CN111581488B (zh) * 2020-05-14 2023-08-04 上海商汤智能科技有限公司 一种数据处理方法及装置、电子设备和存储介质
CN111553324B (zh) * 2020-05-22 2023-05-23 北京字节跳动网络技术有限公司 人体姿态预测值校正方法、装置，服务器及存储介质
CN111811694B (zh) * 2020-07-13 2021-11-30 广东博智林机器人有限公司 一种温度校准方法、装置、设备及存储介质
CN111898676B (zh) * 2020-07-30 2022-09-20 深圳市商汤科技有限公司 目标检测方法及装置、电子设备和存储介质
CN111984812B (zh) * 2020-08-05 2024-05-03 沈阳东软智能医疗科技研究院有限公司 一种特征提取模型生成方法、图像检索方法、装置及设备
CN112287993B (zh) * 2020-10-26 2022-09-02 推想医疗科技股份有限公司 模型生成方法、图像分类方法、装置、电子设备及介质
CN112541577A (zh) * 2020-12-16 2021-03-23 上海商汤智能科技有限公司 神经网络生成方法及装置、电子设备和存储介质
CN112508130A (zh) * 2020-12-25 2021-03-16 商汤集团有限公司 聚类方法及装置、电子设备和存储介质
CN112598063A (zh) * 2020-12-25 2021-04-02 深圳市商汤科技有限公司 神经网络生成方法及装置、电子设备和存储介质
CN112785565B (zh) * 2021-01-15 2024-01-05 上海商汤智能科技有限公司 目标检测方法及装置、电子设备和存储介质
CN112801116B (zh) * 2021-01-27 2024-05-21 商汤集团有限公司 图像的特征提取方法及装置、电子设备和存储介质
CN113206824B (zh) * 2021-03-23 2022-06-24 中国科学院信息工程研究所 动态网络异常攻击检测方法、装置、电子设备和存储介质
CN113065592A (zh) * 2021-03-31 2021-07-02 上海商汤智能科技有限公司 图像分类方法、装置、电子设备及存储介质
CN113159202B (zh) * 2021-04-28 2023-09-26 平安科技（深圳）有限公司 图像分类方法、装置、电子设备及存储介质
CN114140637B (zh) * 2021-10-21 2023-09-12 阿里巴巴达摩院(杭州)科技有限公司 图像分类方法、存储介质和电子设备
CN113837670A (zh) * 2021-11-26 2021-12-24 北京芯盾时代科技有限公司 风险识别模型训练方法及装置
CN114049502B (zh) * 2021-12-22 2023-04-07 贝壳找房(北京)科技有限公司 神经网络的训练、特征提取、数据处理方法和设备
CN115082748B (zh) * 2022-08-23 2022-11-22 浙江大华技术股份有限公司 分类网络训练和目标重识别方法、装置、终端及存储介质
CN115661619A (zh) * 2022-11-03 2023-01-31 北京安德医智科技有限公司 网络模型训练、超声图像质量评估方法及装置、电子设备
CN116912535B (zh) * 2023-09-08 2023-11-28 中国海洋大学 一种基于相似筛选的无监督目标重识别方法、装置及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108305296A (zh) * 2017-08-30 2018-07-20 深圳市腾讯计算机系统有限公司 图像描述生成方法、模型训练方法、设备和存储介质
CN108614858A (zh) * 2018-03-23 2018-10-02 北京达佳互联信息技术有限公司 图像分类模型优化方法、装置及终端
CN108765340A (zh) * 2018-05-29 2018-11-06 Oppo（重庆）智能科技有限公司 模糊图像处理方法、装置及终端设备
CN110210535A (zh) * 2019-05-21 2019-09-06 北京市商汤科技开发有限公司 神经网络训练方法及装置以及图像处理方法及装置

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5156452B2 (ja) * 2008-03-27 2013-03-06 東京エレクトロン株式会社 欠陥分類方法、プログラム、コンピュータ記憶媒体及び欠陥分類装置
CN102542014B (zh) * 2011-12-16 2013-09-18 华中科技大学 基于内容的图像检索反馈方法
TWI655587B (zh) * 2015-01-22 2019-04-01 美商前進公司 神經網路及神經網路訓練的方法
CN104794489B (zh) * 2015-04-23 2019-03-08 苏州大学 一种基于深度标签预测的诱导式图像分类方法及系统
CN104933588A (zh) * 2015-07-01 2015-09-23 北京京东尚科信息技术有限公司 用于拓展商品品类的数据标注平台及方法
GB201517462D0 (en) * 2015-10-02 2015-11-18 Tractable Ltd Semi-automatic labelling of datasets
CN107729901B (zh) * 2016-08-10 2021-04-27 阿里巴巴集团控股有限公司 图像处理模型的建立方法、装置及图像处理方法及系统
CN106528874B (zh) * 2016-12-08 2019-07-19 重庆邮电大学 基于Spark内存计算大数据平台的CLR多标签数据分类方法
JP2018142097A (ja) * 2017-02-27 2018-09-13 キヤノン株式会社 情報処理装置、情報処理方法及びプログラム
US10534257B2 (en) * 2017-05-01 2020-01-14 Lam Research Corporation Layout pattern proximity correction through edge placement error prediction
CN108021931A (zh) * 2017-11-20 2018-05-11 阿里巴巴集团控股有限公司 一种数据样本标签处理方法及装置
CN108009589A (zh) * 2017-12-12 2018-05-08 腾讯科技（深圳）有限公司 样本数据处理方法、装置和计算机可读存储介质
CN108062576B (zh) * 2018-01-05 2019-05-03 百度在线网络技术（北京）有限公司 用于输出数据的方法和装置
CN108959558B (zh) * 2018-07-03 2021-01-29 百度在线网络技术（北京）有限公司 信息推送方法、装置、计算机设备及存储介质
CN109214436A (zh) * 2018-08-22 2019-01-15 阿里巴巴集团控股有限公司 一种针对目标场景的预测模型训练方法及装置
CN109543713B (zh) * 2018-10-16 2021-03-26 北京奇艺世纪科技有限公司 训练集的修正方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108305296A (zh) * 2017-08-30 2018-07-20 深圳市腾讯计算机系统有限公司 图像描述生成方法、模型训练方法、设备和存储介质
CN108614858A (zh) * 2018-03-23 2018-10-02 北京达佳互联信息技术有限公司 图像分类模型优化方法、装置及终端
CN108765340A (zh) * 2018-05-29 2018-11-06 Oppo（重庆）智能科技有限公司 模糊图像处理方法、装置及终端设备
CN110210535A (zh) * 2019-05-21 2019-09-06 北京市商汤科技开发有限公司 神经网络训练方法及装置以及图像处理方法及装置

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112861975A (zh) * 2021-02-10 2021-05-28 北京百度网讯科技有限公司 分类模型的生成方法、分类方法、装置、电子设备与介质
CN112861975B (zh) * 2021-02-10 2023-09-26 北京百度网讯科技有限公司 分类模型的生成方法、分类方法、装置、电子设备与介质
CN113486957A (zh) * 2021-07-07 2021-10-08 西安商汤智能科技有限公司 神经网络训练和图像处理方法及装置
CN115563522A (zh) * 2022-12-02 2023-01-03 湖南工商大学 交通数据的聚类方法、装置、设备及介质
CN115563522B (zh) * 2022-12-02 2023-04-07 湖南工商大学 交通数据的聚类方法、装置、设备及介质

Also Published As

Publication number Publication date
CN113743535A (zh) 2021-12-03
CN110210535B (zh) 2021-09-10
US20210326708A1 (en) 2021-10-21
CN110210535A (zh) 2019-09-06
TWI759722B (zh) 2022-04-01
JP2022516518A (ja) 2022-02-28
SG11202106979WA (en) 2021-07-29
TW202111609A (zh) 2021-03-16

Similar Documents

Publication Publication Date Title
WO2020232977A1 (zh) 神经网络训练方法及装置以及图像处理方法及装置
TWI749423B (zh) 圖像處理方法及裝置、電子設備和電腦可讀儲存介質
TWI754855B (zh) 人臉圖像識別方法、裝置、電子設備及儲存介質
WO2021155632A1 (zh) 图像处理方法及装置、电子设备和存储介质
TWI728621B (zh) 圖像處理方法及其裝置、電子設備、電腦可讀儲存媒體和電腦程式
CN107491541B (zh) 文本分类方法及装置
TWI736230B (zh) 圖像處理方法、電子設備和儲存媒體
TWI747325B (zh) 目標對象匹配方法及目標對象匹配裝置、電子設備和電腦可讀儲存媒介
WO2021196401A1 (zh) 图像重建方法及装置、电子设备和存储介质
CN109389162B (zh) 样本图像筛选方法和装置、电子设备及存储介质
CN110516745A (zh) 图像识别模型的训练方法、装置及电子设备
US11455491B2 (en) Method and device for training image recognition model, and storage medium
WO2021208667A1 (zh) 图像处理方法及装置、电子设备和存储介质
WO2021027343A1 (zh) 一种人脸图像识别方法及装置、电子设备和存储介质
WO2021031645A1 (zh) 图像处理方法及装置、电子设备和存储介质
CN111259967B (zh) 图像分类及神经网络训练方法、装置、设备及存储介质
WO2019237870A1 (zh) 目标匹配方法及装置、电子设备和存储介质
CN110532956B (zh) 图像处理方法及装置、电子设备和存储介质
WO2020172979A1 (zh) 数据处理方法和装置、电子设备和存储介质
WO2020147414A1 (zh) 网络优化方法及装置、图像处理方法及装置、存储介质
WO2020192113A1 (zh) 图像处理方法及装置、电子设备和存储介质
TWI738349B (zh) 圖像處理方法及圖像處理裝置、電子設備和電腦可讀儲存媒體
CN109670077A (zh) 视频推荐方法、装置和计算机可读存储介质
CN111242303A (zh) 网络训练方法及装置、图像处理方法及装置
CN105824955A (zh) 短信聚类方法及装置

Legal Events

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

Ref document number: 19929297

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2021538254

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19929297

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24.03.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19929297

Country of ref document: EP

Kind code of ref document: A1