CN113435522A - Image classification method, device, equipment and storage medium - Google Patents
Image classification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113435522A CN113435522A CN202110739194.7A CN202110739194A CN113435522A CN 113435522 A CN113435522 A CN 113435522A CN 202110739194 A CN202110739194 A CN 202110739194A CN 113435522 A CN113435522 A CN 113435522A
- Authority
- CN
- China
- Prior art keywords
- sample
- image
- preset
- model
- result
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000013145 classification model Methods 0.000 claims abstract description 54
- 238000002372 labelling Methods 0.000 claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 49
- 238000007781 pre-processing Methods 0.000 claims abstract description 19
- 238000000605 extraction Methods 0.000 claims description 38
- 238000011176 pooling Methods 0.000 claims description 34
- 238000012545 processing Methods 0.000 claims description 34
- 230000015654 memory Effects 0.000 claims description 23
- 239000013598 vector Substances 0.000 claims description 20
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000011248 coating agent Substances 0.000 description 25
- 238000000576 coating method Methods 0.000 description 25
- 230000006870 function Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000005070 sampling Methods 0.000 description 4
- 238000013500 data storage Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to artificial intelligence and provides an image classification method, device, equipment and storage medium. The method includes the steps of calculating a first loss value of a preset learner according to a labeling result and an output result of the preset learner on a labeled image, adjusting network parameters according to a labeled sample until convergence to obtain an initial model, predicting an unlabeled sample based on the initial model to obtain a prediction result and confidence, selecting a target sample from the unlabeled sample, cutting the initial model to obtain a student model, training the student model based on the labeled sample and the target sample until convergence to obtain a classification model, obtaining an object image according to a preset object, preprocessing the object image to obtain object features, and inputting the object features into the classification model to obtain an object type. The method and the device can improve the accuracy of the target type. Furthermore, the invention also relates to a blockchain technique, the target type being storable in a blockchain.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to an image classification method, device, equipment and storage medium.
Background
In the existing tongue coating classification mode, the prediction accuracy of the tongue coating classification model generated by training is low due to the small number of labeled samples. In order to improve the accuracy of the tongue coating classification model, the labeled sample is generally subjected to up-down sampling processing, and because the data subjected to the up-sampling processing is highly similar to the labeled sample in characteristics and the characteristics in the data subjected to the down-sampling processing are reduced, the tongue coating classification model generated by training the data subjected to the up-down sampling processing is reduced in classification performance and is not beneficial to tongue coating type identification.
Disclosure of Invention
In view of the above, it is desirable to provide an image classification method, apparatus, device and storage medium, which can improve the accuracy of the target type.
In one aspect, the present invention provides an image classification method, including:
acquiring an image sample according to a preset object, wherein the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
adjusting network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
selecting a sample with the confidence coefficient larger than a preset threshold value from the unlabeled samples as a target sample;
cutting the initial model to obtain a student model, and training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
acquiring an object image according to the preset object, and preprocessing the object image to obtain object characteristics;
and inputting the object features into the classification model to obtain the target type of the object image.
According to a preferred embodiment of the present invention, the calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image includes:
calculating the effective sample amount and the sample quantity of the labeling result on each preset category according to the labeling samples, and calculating the total amount of the labeling samples to obtain the total amount of the samples;
calculating the loss weight of each preset category according to the number of each sample and the effective sample amount:
wherein alpha isiIs the loss weight, E, of each sample in the ith predetermined classniMeans an effective sample size, n, in the ith predetermined categoryiThe number of samples in the ith preset category is referred to, and beta is a preset parameter;
calculating the first loss value according to each loss weight, the labeling result and the output result:
wherein L is the first loss value, M is the total amount of the samples, M is any one of the samples in the ith preset category, and ymIs the labeling result, p, of the mth sample in the ith preset categorymRefers to the output result of the m-th sample.
According to a preferred embodiment of the present invention, the unlabeled sample includes a training image, the initial model includes a convolutional network, a pooling network, a plurality of feature extraction networks, and a classification network, and predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence of the prediction result includes:
performing convolution processing on the training image based on the convolution network to obtain a convolution result;
performing pooling processing on the convolution result based on the pooling network to obtain a pooling result;
performing feature analysis on the pooling result based on the plurality of feature extraction networks to obtain feature vectors;
processing the feature vector based on the classification network to obtain a classification vector;
extracting the element with the largest value from the classification vector as the confidence coefficient, and determining the dimension of the element as a target dimension;
and obtaining a result corresponding to the target dimension from the classification network as the prediction result.
According to a preferred embodiment of the present invention, said clipping said initial model to obtain a student model comprises:
extracting any network in the plurality of feature extraction networks as a target extraction network;
and splicing the convolution network, the pooling network, the target extraction network and the classification network to obtain the student model.
According to a preferred embodiment of the present invention, the training the student model based on the labeled sample and the target sample until a second loss value of the student model converges to obtain a classification model includes:
predicting the marked sample and the target sample based on the student model to obtain model output;
generating the second loss value according to the model output, the prediction result and the labeling result;
obtaining model parameters of the student model;
and adjusting the model parameters according to the prediction result and the labeling result until the second loss value is converged to obtain the classification model.
According to a preferred embodiment of the present invention, the acquiring an object image according to the preset object includes:
acquiring an object identification code of the preset object;
acquiring a storage path of the preset object according to the object identification code;
and acquiring information corresponding to a preset label from the storage path as the object image, wherein the preset label is used for indicating an image which is not classified.
According to a preferred embodiment of the present invention, the preprocessing the object image to obtain the object feature includes:
obtaining an object segmentation model according to the preset object;
inputting the object image into the segmentation model to obtain an object region;
performing expansion processing on the object area according to preset parameters until the area shape corresponding to the object area after the expansion processing is a preset shape to obtain a target area;
and acquiring pixel information of the target area, and performing normalization processing on the pixel information to obtain the object characteristics.
In another aspect, the present invention further provides an image classification apparatus, including:
the device comprises an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used for acquiring an image sample according to a preset object, the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
the calculation unit is used for calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
the adjusting unit is used for adjusting the network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
the prediction unit is used for predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
the selecting unit is used for selecting the sample with the confidence coefficient larger than a preset threshold value from the unmarked samples as a target sample;
the training unit is used for cutting the initial model to obtain a student model, training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
the preprocessing unit is used for acquiring an object image according to the preset object and preprocessing the object image to obtain object characteristics;
and the input unit is used for inputting the object characteristics into the classification model to obtain the target type of the object image.
In another aspect, the present invention further provides an electronic device, including:
a memory storing computer readable instructions; and
a processor executing computer readable instructions stored in the memory to implement the image classification method.
In another aspect, the present invention also provides a computer-readable storage medium, in which computer-readable instructions are stored, and the computer-readable instructions are executed by a processor in an electronic device to implement the image classification method.
According to the technical scheme, the unlabeled samples are analyzed through the initial model generated by the training of the labeled samples, the prediction result and the confidence coefficient of the unlabeled samples can be generated, so that the samples suitable for analyzing the student model can be selected, the unlabeled samples can be fully utilized, the classification accuracy of the classification model is improved, and meanwhile, the classification efficiency of the student model can be improved through the cutting of the initial model, and the classification efficiency of the classification model is improved. In addition, the first loss value and the second loss value are calculated through the loss weight values of the effective sample size on a plurality of categories, so that the problem of classification caused by unbalanced distribution of different categories can be solved, and the classification performance of the initial model and the classification model is effectively improved.
Drawings
FIG. 1 is a flowchart of an image classification method according to a preferred embodiment of the present invention.
FIG. 2 is a functional block diagram of an image classifying apparatus according to a preferred embodiment of the present invention.
FIG. 3 is a schematic structural diagram of an electronic device implementing the image classification method according to the preferred embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a flow chart of a preferred embodiment of the image classification method according to the present invention. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs.
The image classification method is applied to one or more electronic devices, which are devices capable of automatically performing numerical calculation and/or information processing according to computer readable instructions set or stored in advance, and the hardware thereof includes, but is not limited to, a microprocessor, an Application Specific Integrated Circuit (ASIC), a Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), an embedded device, and the like.
The electronic device may be any electronic product capable of performing human-computer interaction with a user, for example, a Personal computer, a tablet computer, a smart phone, a Personal Digital Assistant (PDA), a game machine, an interactive Internet Protocol Television (IPTV), a smart wearable device, and the like.
The electronic device may include a network device and/or a user device. Wherein the network device includes, but is not limited to, a single network electronic device, an electronic device group consisting of a plurality of network electronic devices, or a Cloud Computing (Cloud Computing) based Cloud consisting of a large number of hosts or network electronic devices.
The network in which the electronic device is located includes, but is not limited to: the internet, a wide area Network, a metropolitan area Network, a local area Network, a Virtual Private Network (VPN), and the like.
S10, obtaining an image sample according to the preset object, wherein the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result.
In at least one embodiment of the present invention, the preset objects may be various organs in a human body, for example, the preset objects may be tongue fur.
The images in the labeled image and the unlabeled sample both comprise the preset object.
The annotation result may be the type identification of the preset object in the annotation image by any user, for example, when the preset object is a tongue coating, the annotation result may be a coating, a slippery coating, or a dry coating.
The unlabeled sample includes a training image.
And S11, calculating a first loss value of the preset learner according to the labeling result and the output result of the preset learner on the labeling image.
In at least one embodiment of the invention, the pre-set learner is an initially configured network.
The output result refers to the prediction information of the pre-set learner on the labeled image.
The first loss value refers to the degree of loss of the annotation image predicted by the preset learner.
In at least one embodiment of the present invention, the calculating, by the electronic device, a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image includes:
calculating the effective sample amount and the sample quantity of the labeling result on each preset category according to the labeling samples, and calculating the total amount of the labeling samples to obtain the total amount of the samples;
calculating the loss weight of each preset category according to the number of each sample and the effective sample amount:
wherein alpha isiIs the loss weight, E, of each sample in the ith predetermined classniMeans an effective sample size, n, in the ith predetermined categoryiThe number of samples in the ith preset category is referred to, and beta is a preset parameter;
calculating the first loss value according to each loss weight, the labeling result and the output result:
wherein L is the first loss value, M is the total amount of the samples, M is any one of the samples in the ith preset category, and ymIs the labeling result, p, of the mth sample in the ith preset categorymRefers to the output result of the m-th sample.
The loss weight of each preset category can be accurately determined through the effective sample amount of each preset category, so that the first loss value can be accurately generated according to the loss weight.
And S12, adjusting the network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model.
In at least one embodiment of the present invention, the network parameters may include initialization configuration parameters in the preset learner.
The initial model is a preset learner when the first loss value converges. The initial model comprises a convolution network, a pooling network, a plurality of feature extraction networks and a classification network.
In at least one embodiment of the present invention, the electronic device adjusts the network parameters in the preset learner according to the labeled sample until the first loss value converges, and obtaining the initial model includes:
acquiring configuration parameters on the convolutional network, the pooling network, the plurality of feature extraction networks and the classification network as the network parameters;
and adjusting the network parameters according to the labeled samples until the first loss value is converged, stopping adjusting the network parameters, and determining the adjusted preset learner as the initial model.
S13, predicting the unlabeled sample based on the initial model to obtain the prediction result of the unlabeled sample and the confidence of the prediction result.
In at least one embodiment of the present invention, the prediction result refers to a result of the initial model predicting the training image in the unlabeled sample, and the confidence level refers to a probability of classifying the training image as the prediction result.
The output result of the initial model for the unlabeled sample is shown in the form of a soft label, for example, the output result of the sample a is [0.1, 0.85, 0.05], and then the category corresponding to the dimension where 0.85 is located is the prediction result, and 0.85 is the confidence coefficient of the sample a.
In this embodiment, the unmarked samples are presented in the form of soft labels, so that guidance for images outside the domain can be improved.
In at least one embodiment of the present invention, the predicting, by the electronic device, the unlabeled sample based on the initial model, and obtaining the prediction result of the unlabeled sample and the confidence of the prediction result includes:
performing convolution processing on the training image based on the convolution network to obtain a convolution result;
performing pooling processing on the convolution result based on the pooling network to obtain a pooling result;
performing feature analysis on the pooling result based on the plurality of feature extraction networks to obtain feature vectors;
processing the feature vector based on the classification network to obtain a classification vector;
extracting the element with the largest value from the classification vector as the confidence coefficient, and determining the dimension of the element as a target dimension;
and obtaining a result corresponding to the target dimension from the classification network as the prediction result.
The classification vector is used to indicate the prediction result and the confidence level, for example, the preset object is a tongue coating, and the classification vector includes information of three dimensions because there are three categories corresponding to the tongue coating.
The classification network stores the mapping relation between the dimension and the category.
The convolutional network can not only reduce the number of features in the training image, but also fuse the channel features of the training image on each channel, the pooling network can ensure the scale of the features in the pooling result, and the plurality of feature extraction networks can extract high-level features in the pooling result, so that the prediction result and the confidence coefficient can be accurately determined.
And S14, selecting the sample with the confidence coefficient larger than a preset threshold value from the unlabeled samples as a target sample.
In at least one embodiment of the present invention, the target sample refers to an unlabeled sample of which the confidence is greater than a preset threshold.
The preset threshold is a value set according to the classification accuracy of the classification model, and for example, the preset threshold may be set to 0.8.
And S15, cutting the initial model to obtain a student model, and training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model.
In at least one embodiment of the present invention, the student model refers to a model that compresses the number of networks in the initial model.
The classification model is obtained by adjusting model parameters in the student model.
In at least one embodiment of the present invention, the electronic device clipping the initial model to obtain a student model includes:
extracting any network in the plurality of feature extraction networks as a target extraction network;
and splicing the convolution network, the pooling network, the target extraction network and the classification network to obtain the student model.
Wherein the target extraction network may be any one of the plurality of feature extraction networks.
Through the above embodiment, since the network formats in the plurality of feature extraction networks are all the same, compressing the plurality of feature extraction networks into the target extraction network does not lose the performance of the plurality of feature extraction networks, and meanwhile, by clipping the initial model, the training efficiency and the prediction efficiency of the student model can be improved.
In at least one embodiment of the present invention, the electronic device trains the student model based on the labeled sample and the target sample until a second loss value of the student model converges, and obtaining a classification model includes:
predicting the marked sample and the target sample based on the student model to obtain model output;
generating the second loss value according to the model output, the prediction result and the labeling result;
obtaining model parameters of the student model;
and adjusting the model parameters according to the prediction result and the labeling result until the second loss value is converged to obtain the classification model.
Wherein the model parameters include configuration parameters of the convolutional network, the pooling network, the target extraction network, and the classification network.
The classification accuracy of the classification model can be ensured by adjusting the model parameters through the samples with the confidence degrees larger than the preset threshold value and the labeled samples.
Specifically, the manner in which the electronic device predicts the labeled samples and the target samples based on the student model is similar to the manner in which the electronic device predicts the unlabeled samples based on the initial model, and details of the method are omitted in the present invention.
In this embodiment, the electronic device determines the classification model obtained by training as the student model, and performs iterative training on the student model until the number of iterations reaches a preset requirement, so as to obtain a trained classification model for predicting subsequent data.
By self-training the student model, a large amount of unlabeled data is fully utilized in a semi-supervised mode, and the generalization capability and robustness of the classification model are improved.
And S16, acquiring an object image according to the preset object, and preprocessing the object image to obtain object characteristics.
In at least one embodiment of the present invention, the object image refers to an image that needs to be classified. The object image comprises the preset object.
The object features can characterize features of the preset object in the object image.
In at least one embodiment of the present invention, the acquiring, by the electronic device, an object image according to the preset object includes:
acquiring an object identification code of the preset object;
acquiring a storage path of the preset object according to the object identification code;
and acquiring information corresponding to a preset label from the storage path as the object image, wherein the preset label is used for indicating an image which is not classified.
Wherein the object identification code is capable of being used to uniquely indicate the preset object.
By the embodiment, the preset object can be ensured to be contained in the obtained object image, and the classification efficiency of the preset object can be improved because the object analysis is not required to be carried out on the image which does not contain the preset object.
In at least one embodiment of the present invention, the electronic device performs preprocessing on the object image to obtain the object feature, including:
obtaining an object segmentation model according to the preset object;
inputting the object image into the segmentation model to obtain an object region;
performing expansion processing on the object area according to preset parameters until the area shape corresponding to the object area after the expansion processing is a preset shape to obtain a target area;
and acquiring pixel information of the target area, and performing normalization processing on the pixel information to obtain the object characteristics.
Wherein the object segmentation model is capable of segmenting the preset object from an image. The training mode of the object segmentation model belongs to the prior art, and the invention is not repeated herein.
The preset shape refers to a shape with the best prediction effect in the classification model. Typically, the preset shape is set to be a rectangle of 512 pixels by 512 pixels.
The target area is expanded through the preset shape, the scale of the target area can be ensured, the accuracy of the target type is ensured, and further, the generation efficiency of the target characteristics can be improved through normalization processing of the pixel information, so that the generation efficiency of the target type is improved.
S17, inputting the object features into the classification model to obtain the target type of the object image.
In at least one embodiment of the present invention, the target type refers to a type to which the preset object belongs in the object image, for example, if the preset object is a tongue coating, the target type may be a tongue coating, a slippery coating, and a dry coating.
It is emphasized that the object type may also be stored in a node of a block chain in order to further ensure privacy and security of the object type.
In at least one embodiment of the present invention, a manner of predicting the object features by the classification model is the same as a manner of predicting the labeling samples and the target samples by the student model, which is not described in detail herein.
According to the technical scheme, the unlabeled samples are analyzed through the initial model generated by the training of the labeled samples, the prediction result and the confidence coefficient of the unlabeled samples can be generated, so that the samples suitable for analyzing the student model can be selected, the unlabeled samples can be fully utilized, the classification accuracy of the classification model is improved, and meanwhile, the classification efficiency of the student model can be improved through the cutting of the initial model, and the classification efficiency of the classification model is improved. In addition, the first loss value and the second loss value are calculated through the loss weight values of the effective sample size on a plurality of categories, so that the problem of classification caused by unbalanced distribution of different categories can be solved, and the classification performance of the initial model and the classification model is effectively improved.
Fig. 2 is a functional block diagram of an image classifying apparatus according to a preferred embodiment of the present invention. The image classification device 11 includes an obtaining unit 110, a calculating unit 111, an adjusting unit 112, a predicting unit 113, a selecting unit 114, a training unit 115, a preprocessing unit 116, and an input unit 117. The module/unit referred to herein is a series of computer readable instruction segments that can be accessed by the processor 13 and perform a fixed function and that are stored in the memory 12. In the present embodiment, the functions of the modules/units will be described in detail in the following embodiments.
The obtaining unit 110 obtains an image sample according to a preset object, where the image sample includes an annotated sample and an unlabeled sample, and the annotated sample includes an annotated image and an annotated result.
In at least one embodiment of the present invention, the preset objects may be various organs in a human body, for example, the preset objects may be tongue fur.
The images in the labeled image and the unlabeled sample both comprise the preset object.
The annotation result may be the type identification of the preset object in the annotation image by any user, for example, when the preset object is a tongue coating, the annotation result may be a coating, a slippery coating, or a dry coating.
The unlabeled sample includes a training image.
The calculating unit 111 calculates a first loss value of the preset learner according to the labeling result and an output result of the preset learner on the labeled image.
In at least one embodiment of the invention, the pre-set learner is an initially configured network.
The output result refers to the prediction information of the pre-set learner on the labeled image.
The first loss value refers to the degree of loss of the annotation image predicted by the preset learner.
In at least one embodiment of the present invention, the calculating unit 111 calculates the first loss value of the preset learner according to the annotation result and the output result of the preset learner on the annotation image, including:
calculating the effective sample amount and the sample quantity of the labeling result on each preset category according to the labeling samples, and calculating the total amount of the labeling samples to obtain the total amount of the samples;
calculating the loss weight of each preset category according to the number of each sample and the effective sample amount:
wherein alpha isiIs the loss weight, E, of each sample in the ith predetermined classniMeans an effective sample size, n, in the ith predetermined categoryiThe number of samples in the ith preset category is referred to, and beta is a preset parameter;
calculating the first loss value according to each loss weight, the labeling result and the output result:
wherein L is the first loss value, M is the total amount of the samples, M is any one of the samples in the ith preset category, and ymIs the labeling result, p, of the mth sample in the ith preset categorymRefers to the output result of the m-th sample.
The loss weight of each preset category can be accurately determined through the effective sample amount of each preset category, so that the first loss value can be accurately generated according to the loss weight.
The adjusting unit 112 adjusts the network parameters in the preset learner according to the labeled sample until the first loss value converges, so as to obtain an initial model.
In at least one embodiment of the present invention, the network parameters may include initialization configuration parameters in the preset learner.
The initial model is a preset learner when the first loss value converges. The initial model comprises a convolution network, a pooling network, a plurality of feature extraction networks and a classification network.
In at least one embodiment of the present invention, the adjusting unit 112 adjusts the network parameters in the preset learner according to the labeled sample until the first loss value converges, and obtaining the initial model includes:
acquiring configuration parameters on the convolutional network, the pooling network, the plurality of feature extraction networks and the classification network as the network parameters;
and adjusting the network parameters according to the labeled samples until the first loss value is converged, stopping adjusting the network parameters, and determining the adjusted preset learner as the initial model.
The prediction unit 113 predicts the unlabeled sample based on the initial model, and obtains a prediction result of the unlabeled sample and a confidence of the prediction result.
In at least one embodiment of the present invention, the prediction result refers to a result of the initial model predicting the training image in the unlabeled sample, and the confidence level refers to a probability of classifying the training image as the prediction result.
The output result of the initial model for the unlabeled sample is shown in the form of a soft label, for example, the output result of the sample a is [0.1, 0.85, 0.05], and then the category corresponding to the dimension where 0.85 is located is the prediction result, and 0.85 is the confidence coefficient of the sample a.
In this embodiment, the unmarked samples are presented in the form of soft labels, so that guidance for images outside the domain can be improved.
In at least one embodiment of the present invention, the predicting unit 113 predicts the unlabeled sample based on the initial model, and obtaining the prediction result of the unlabeled sample and the confidence of the prediction result includes:
performing convolution processing on the training image based on the convolution network to obtain a convolution result;
performing pooling processing on the convolution result based on the pooling network to obtain a pooling result;
performing feature analysis on the pooling result based on the plurality of feature extraction networks to obtain feature vectors;
processing the feature vector based on the classification network to obtain a classification vector;
extracting the element with the largest value from the classification vector as the confidence coefficient, and determining the dimension of the element as a target dimension;
and obtaining a result corresponding to the target dimension from the classification network as the prediction result.
The classification vector is used to indicate the prediction result and the confidence level, for example, the preset object is a tongue coating, and the classification vector includes information of three dimensions because there are three categories corresponding to the tongue coating.
The classification network stores the mapping relation between the dimension and the category.
The convolutional network can not only reduce the number of features in the training image, but also fuse the channel features of the training image on each channel, the pooling network can ensure the scale of the features in the pooling result, and the plurality of feature extraction networks can extract high-level features in the pooling result, so that the prediction result and the confidence coefficient can be accurately determined.
The selecting unit 114 selects the sample with the confidence degree greater than a preset threshold value from the unlabeled samples as a target sample.
In at least one embodiment of the present invention, the target sample refers to an unlabeled sample of which the confidence is greater than a preset threshold.
The preset threshold is a value set according to the classification accuracy of the classification model, and for example, the preset threshold may be set to 0.8.
The training unit 115 cuts the initial model to obtain a student model, and trains the student model based on the labeled sample and the target sample until a second loss value of the student model converges to obtain a classification model.
In at least one embodiment of the present invention, the student model refers to a model that compresses the number of networks in the initial model.
The classification model is obtained by adjusting model parameters in the student model.
In at least one embodiment of the present invention, the training unit 115 clipping the initial model to obtain a student model comprises:
extracting any network in the plurality of feature extraction networks as a target extraction network;
and splicing the convolution network, the pooling network, the target extraction network and the classification network to obtain the student model.
Wherein the target extraction network may be any one of the plurality of feature extraction networks.
Through the above embodiment, since the network formats in the plurality of feature extraction networks are all the same, compressing the plurality of feature extraction networks into the target extraction network does not lose the performance of the plurality of feature extraction networks, and meanwhile, by clipping the initial model, the training efficiency and the prediction efficiency of the student model can be improved.
In at least one embodiment of the present invention, the training unit 115 trains the student model based on the labeled sample and the target sample until a second loss value of the student model converges, and obtaining a classification model includes:
predicting the marked sample and the target sample based on the student model to obtain model output;
generating the second loss value according to the model output, the prediction result and the labeling result;
obtaining model parameters of the student model;
and adjusting the model parameters according to the prediction result and the labeling result until the second loss value is converged to obtain the classification model.
Wherein the model parameters include configuration parameters of the convolutional network, the pooling network, the target extraction network, and the classification network.
The classification accuracy of the classification model can be ensured by adjusting the model parameters through the samples with the confidence degrees larger than the preset threshold value and the labeled samples.
Specifically, the way that the training unit 115 predicts the labeled samples and the target samples based on the student model is similar to the way that the prediction unit 113 predicts the unlabeled samples based on the initial model, which is not described in detail herein.
In this embodiment, the electronic device determines the classification model obtained by training as the student model, and performs iterative training on the student model until the number of iterations reaches a preset requirement, so as to obtain a trained classification model for predicting subsequent data.
By self-training the student model, a large amount of unlabeled data is fully utilized in a semi-supervised mode, and the generalization capability and robustness of the classification model are improved.
The preprocessing unit 116 obtains an object image according to the preset object, and preprocesses the object image to obtain object characteristics.
In at least one embodiment of the present invention, the object image refers to an image that needs to be classified. The object image comprises the preset object.
The object features can characterize features of the preset object in the object image.
In at least one embodiment of the present invention, the pre-processing unit 116 acquiring the object image according to the preset object includes:
acquiring an object identification code of the preset object;
acquiring a storage path of the preset object according to the object identification code;
and acquiring information corresponding to a preset label from the storage path as the object image, wherein the preset label is used for indicating an image which is not classified.
Wherein the object identification code is capable of being used to uniquely indicate the preset object.
By the embodiment, the preset object can be ensured to be contained in the obtained object image, and the classification efficiency of the preset object can be improved because the object analysis is not required to be carried out on the image which does not contain the preset object.
In at least one embodiment of the present invention, the preprocessing unit 116 performs preprocessing on the object image to obtain the object feature includes:
obtaining an object segmentation model according to the preset object;
inputting the object image into the segmentation model to obtain an object region;
performing expansion processing on the object area according to preset parameters until the area shape corresponding to the object area after the expansion processing is a preset shape to obtain a target area;
and acquiring pixel information of the target area, and performing normalization processing on the pixel information to obtain the object characteristics.
Wherein the object segmentation model is capable of segmenting the preset object from an image. The training mode of the object segmentation model belongs to the prior art, and the invention is not repeated herein.
The preset shape refers to a shape with the best prediction effect in the classification model. Typically, the preset shape is set to be a rectangle of 512 pixels by 512 pixels.
The target area is expanded through the preset shape, the scale of the target area can be ensured, the accuracy of the target type is ensured, and further, the generation efficiency of the target characteristics can be improved through normalization processing of the pixel information, so that the generation efficiency of the target type is improved.
The input unit 117 inputs the object features into the classification model, and obtains the target type of the object image.
In at least one embodiment of the present invention, the target type refers to a type to which the preset object belongs in the object image, for example, if the preset object is a tongue coating, the target type may be a tongue coating, a slippery coating, and a dry coating.
It is emphasized that the object type may also be stored in a node of a block chain in order to further ensure privacy and security of the object type.
In at least one embodiment of the present invention, a manner of predicting the object features by the classification model is the same as a manner of predicting the labeling samples and the target samples by the student model, which is not described in detail herein.
According to the technical scheme, the unlabeled samples are analyzed through the initial model generated by the training of the labeled samples, the prediction result and the confidence coefficient of the unlabeled samples can be generated, so that the samples suitable for analyzing the student model can be selected, the unlabeled samples can be fully utilized, the classification accuracy of the classification model is improved, and meanwhile, the classification efficiency of the student model can be improved through the cutting of the initial model, and the classification efficiency of the classification model is improved. In addition, the first loss value and the second loss value are calculated through the loss weight values of the effective sample size on a plurality of categories, so that the problem of classification caused by unbalanced distribution of different categories can be solved, and the classification performance of the initial model and the classification model is effectively improved.
Fig. 3 is a schematic structural diagram of an electronic device implementing the image classification method according to the preferred embodiment of the invention.
In one embodiment of the present invention, the electronic device 1 includes, but is not limited to, a memory 12, a processor 13, and computer readable instructions, such as an image classification program, stored in the memory 12 and executable on the processor 13.
It will be appreciated by a person skilled in the art that the schematic diagram is only an example of the electronic device 1 and does not constitute a limitation of the electronic device 1, and that it may comprise more or less components than shown, or some components may be combined, or different components, e.g. the electronic device 1 may further comprise an input output device, a network access device, a bus, etc.
The Processor 13 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, etc. The processor 13 is an operation core and a control center of the electronic device 1, and is connected to each part of the whole electronic device 1 by various interfaces and lines, and executes an operating system of the electronic device 1 and various installed application programs, program codes, and the like.
Illustratively, the computer readable instructions may be partitioned into one or more modules/units that are stored in the memory 12 and executed by the processor 13 to implement the present invention. The one or more modules/units may be a series of computer readable instruction segments capable of performing specific functions, which are used for describing the execution process of the computer readable instructions in the electronic device 1. For example, the computer readable instructions may be partitioned into an acquisition unit 110, a calculation unit 111, an adjustment unit 112, a prediction unit 113, a selection unit 114, a training unit 115, a pre-processing unit 116, and an input unit 117.
The memory 12 may be used for storing the computer readable instructions and/or modules, and the processor 13 implements various functions of the electronic device 1 by executing or executing the computer readable instructions and/or modules stored in the memory 12 and invoking data stored in the memory 12. The memory 12 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data created according to use of the electronic device, and the like. The memory 12 may include non-volatile and volatile memories, such as: a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other storage device.
The memory 12 may be an external memory and/or an internal memory of the electronic device 1. Further, the memory 12 may be a memory having a physical form, such as a memory stick, a TF Card (Trans-flash Card), or the like.
The integrated modules/units of the electronic device 1 may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the flow of the method according to the above embodiments may be implemented by hardware that is configured to be instructed by computer readable instructions, which may be stored in a computer readable storage medium, and when the computer readable instructions are executed by a processor, the steps of the method embodiments may be implemented.
Wherein the computer readable instructions comprise computer readable instruction code which may be in source code form, object code form, an executable file or some intermediate form, and the like. The computer-readable medium may include: any entity or device capable of carrying said computer readable instruction code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM).
The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
In conjunction with fig. 1, the memory 12 in the electronic device 1 stores computer-readable instructions to implement a method for image classification, and the processor 13 can execute the computer-readable instructions to implement:
acquiring an image sample according to a preset object, wherein the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
adjusting network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
selecting a sample with the confidence coefficient larger than a preset threshold value from the unlabeled samples as a target sample;
cutting the initial model to obtain a student model, and training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
acquiring an object image according to the preset object, and preprocessing the object image to obtain object characteristics;
and inputting the object features into the classification model to obtain the target type of the object image.
Specifically, the processor 13 may refer to the description of the relevant steps in the embodiment corresponding to fig. 1 for a specific implementation method of the computer readable instructions, which is not described herein again.
In the embodiments provided in the present invention, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.
The computer readable storage medium has computer readable instructions stored thereon, wherein the computer readable instructions when executed by the processor 13 are configured to implement the steps of:
acquiring an image sample according to a preset object, wherein the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
adjusting network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
selecting a sample with the confidence coefficient larger than a preset threshold value from the unlabeled samples as a target sample;
cutting the initial model to obtain a student model, and training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
acquiring an object image according to the preset object, and preprocessing the object image to obtain object characteristics;
and inputting the object features into the classification model to obtain the target type of the object image.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.
The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. The plurality of units or devices may also be implemented by one unit or device through software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.
Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (10)
1. An image classification method, characterized in that the image classification method comprises:
acquiring an image sample according to a preset object, wherein the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
adjusting network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
selecting a sample with the confidence coefficient larger than a preset threshold value from the unlabeled samples as a target sample;
cutting the initial model to obtain a student model, and training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
acquiring an object image according to the preset object, and preprocessing the object image to obtain object characteristics;
and inputting the object features into the classification model to obtain the target type of the object image.
2. The image classification method according to claim 1, wherein the calculating of the first loss value of the pre-learner from the labeling result and the labeling image of the pre-learner comprises:
calculating the effective sample amount and the sample quantity of the labeling result on each preset category according to the labeling samples, and calculating the total amount of the labeling samples to obtain the total amount of the samples;
calculating the loss weight of each preset category according to the number of each sample and the effective sample amount:
wherein alpha isiIs the loss weight, E, of each sample in the ith predetermined classniMeans an effective sample size, n, in the ith predetermined categoryiThe number of samples in the ith preset category is referred to, and beta is a preset parameter;
calculating the first loss value according to each loss weight, the labeling result and the output result:
wherein L is the first loss value, M is the total amount of the samples, M is any one of the samples in the ith preset category, and ymIs the labeling result, p, of the mth sample in the ith preset categorymRefers to the output result of the m-th sample.
3. The image classification method according to claim 1, wherein the unlabeled sample includes a training image, the initial model includes a convolutional network, a pooling network, a plurality of feature extraction networks, and a classification network, and predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence of the prediction result includes:
performing convolution processing on the training image based on the convolution network to obtain a convolution result;
performing pooling processing on the convolution result based on the pooling network to obtain a pooling result;
performing feature analysis on the pooling result based on the plurality of feature extraction networks to obtain feature vectors;
processing the feature vector based on the classification network to obtain a classification vector;
extracting the element with the largest value from the classification vector as the confidence coefficient, and determining the dimension of the element as a target dimension;
and obtaining a result corresponding to the target dimension from the classification network as the prediction result.
4. The image classification method of claim 3, wherein the cropping the initial model to obtain a student model comprises:
extracting any network in the plurality of feature extraction networks as a target extraction network;
and splicing the convolution network, the pooling network, the target extraction network and the classification network to obtain the student model.
5. The image classification method of claim 1, wherein the training of the student model based on the annotation sample and the target sample until a second loss value of the student model converges to obtain a classification model comprises:
predicting the marked sample and the target sample based on the student model to obtain model output;
generating the second loss value according to the model output, the prediction result and the labeling result;
obtaining model parameters of the student model;
and adjusting the model parameters according to the prediction result and the labeling result until the second loss value is converged to obtain the classification model.
6. The image classification method according to claim 1, wherein the acquiring of the object image according to the preset object includes:
acquiring an object identification code of the preset object;
acquiring a storage path of the preset object according to the object identification code;
and acquiring information corresponding to a preset label from the storage path as the object image, wherein the preset label is used for indicating an image which is not classified.
7. The image classification method according to claim 1, wherein the preprocessing the object image to obtain the object feature comprises:
obtaining an object segmentation model according to the preset object;
inputting the object image into the segmentation model to obtain an object region;
performing expansion processing on the object area according to preset parameters until the area shape corresponding to the object area after the expansion processing is a preset shape to obtain a target area;
and acquiring pixel information of the target area, and performing normalization processing on the pixel information to obtain the object characteristics.
8. An image classification apparatus, characterized by comprising:
the device comprises an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used for acquiring an image sample according to a preset object, the image sample comprises an annotated sample and an unlabelled sample, and the annotated sample comprises an annotated image and an annotated result;
the calculation unit is used for calculating a first loss value of a preset learner according to the labeling result and an output result of the preset learner on the labeled image;
the adjusting unit is used for adjusting the network parameters in the preset learner according to the labeled samples until the first loss value is converged to obtain an initial model;
the prediction unit is used for predicting the unlabeled sample based on the initial model to obtain a prediction result of the unlabeled sample and a confidence coefficient of the prediction result;
the selecting unit is used for selecting the sample with the confidence coefficient larger than a preset threshold value from the unmarked samples as a target sample;
the training unit is used for cutting the initial model to obtain a student model, training the student model based on the labeling sample and the target sample until a second loss value of the student model is converged to obtain a classification model;
the preprocessing unit is used for acquiring an object image according to the preset object and preprocessing the object image to obtain object characteristics;
and the input unit is used for inputting the object characteristics into the classification model to obtain the target type of the object image.
9. An electronic device, characterized in that the electronic device comprises:
a memory storing computer readable instructions; and
a processor executing computer readable instructions stored in the memory to implement the image classification method of any of claims 1 to 7.
10. A computer-readable storage medium characterized by: the computer-readable storage medium has stored therein computer-readable instructions which are executed by a processor in an electronic device to implement the image classification method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110739194.7A CN113435522A (en) | 2021-06-30 | 2021-06-30 | Image classification method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110739194.7A CN113435522A (en) | 2021-06-30 | 2021-06-30 | Image classification method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113435522A true CN113435522A (en) | 2021-09-24 |
Family
ID=77758132
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110739194.7A Pending CN113435522A (en) | 2021-06-30 | 2021-06-30 | Image classification method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113435522A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113627576A (en) * | 2021-10-08 | 2021-11-09 | 平安科技（深圳）有限公司 | Code scanning information detection method, device, equipment and storage medium |
| CN113963220A (en) * | 2021-12-22 | 2022-01-21 | 熵基科技股份有限公司 | Security check image classification model training method, security check image classification method and device |
| CN113971757A (en) * | 2021-09-26 | 2022-01-25 | 阿里巴巴达摩院(杭州)科技有限公司 | Image classification method, computer terminal and storage medium |
| CN114140637A (en) * | 2021-10-21 | 2022-03-04 | 阿里巴巴达摩院(杭州)科技有限公司 | Image classification method, storage medium and electronic device |
| CN114155417A (en) * | 2021-12-13 | 2022-03-08 | 中国科学院空间应用工程与技术中心 | Image target identification method and device, electronic equipment and computer storage medium |
| CN114241243A (en) * | 2021-12-20 | 2022-03-25 | 百度在线网络技术（北京）有限公司 | Training method and device of image classification model, electronic equipment and storage medium |
| CN114417871A (en) * | 2021-12-17 | 2022-04-29 | 北京百度网讯科技有限公司 | Model training and named entity recognition method and device, electronic equipment and medium |
| CN114419375A (en) * | 2022-01-20 | 2022-04-29 | 北京百度网讯科技有限公司 | Image classification method, training method, device, electronic equipment and storage medium |
| CN115049836A (en) * | 2022-08-16 | 2022-09-13 | 平安科技（深圳）有限公司 | Image segmentation method, device, equipment and storage medium |
| CN115482436A (en) * | 2022-09-21 | 2022-12-16 | 北京百度网讯科技有限公司 | Training method and device for image screening model and image screening method |
| CN117853490A (en) * | 2024-03-06 | 2024-04-09 | 阿里巴巴达摩院(杭州)科技有限公司 | Image processing method and training method of image processing model |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875821A (en) * | 2018-06-08 | 2018-11-23 | Oppo广东移动通信有限公司 | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing |
| CN111160406A (en) * | 2019-12-10 | 2020-05-15 | 北京达佳互联信息技术有限公司 | Training method of image classification model, and image classification method and device |
| CN111639710A (en) * | 2020-05-29 | 2020-09-08 | 北京百度网讯科技有限公司 | Image recognition model training method, device, equipment and storage medium |
| CN111666993A (en) * | 2020-05-28 | 2020-09-15 | 平安科技（深圳）有限公司 | Medical image sample screening method and device, computer equipment and storage medium |
| CN112508126A (en) * | 2020-12-22 | 2021-03-16 | 北京百度网讯科技有限公司 | Deep learning model training method and device, electronic equipment and readable storage medium |
| CN112560791A (en) * | 2020-12-28 | 2021-03-26 | 苏州科达科技股份有限公司 | Recognition model training method, recognition method and device and electronic equipment |
| CN112966701A (en) * | 2019-12-12 | 2021-06-15 | 北京沃东天骏信息技术有限公司 | Method and device for classifying objects |
-
2021
- 2021-06-30 CN CN202110739194.7A patent/CN113435522A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875821A (en) * | 2018-06-08 | 2018-11-23 | Oppo广东移动通信有限公司 | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing |
| CN111160406A (en) * | 2019-12-10 | 2020-05-15 | 北京达佳互联信息技术有限公司 | Training method of image classification model, and image classification method and device |
| CN112966701A (en) * | 2019-12-12 | 2021-06-15 | 北京沃东天骏信息技术有限公司 | Method and device for classifying objects |
| CN111666993A (en) * | 2020-05-28 | 2020-09-15 | 平安科技（深圳）有限公司 | Medical image sample screening method and device, computer equipment and storage medium |
| CN111639710A (en) * | 2020-05-29 | 2020-09-08 | 北京百度网讯科技有限公司 | Image recognition model training method, device, equipment and storage medium |
| CN112508126A (en) * | 2020-12-22 | 2021-03-16 | 北京百度网讯科技有限公司 | Deep learning model training method and device, electronic equipment and readable storage medium |
| CN112560791A (en) * | 2020-12-28 | 2021-03-26 | 苏州科达科技股份有限公司 | Recognition model training method, recognition method and device and electronic equipment |
Non-Patent Citations (2)
| Title |
|---|
| 孙其功: "《深度神经网络FPGA设计与实现》", 31 December 2020 * |
| 张国基等: "《生物辨识系统与深度学习》", 31 December 2018 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113971757A (en) * | 2021-09-26 | 2022-01-25 | 阿里巴巴达摩院(杭州)科技有限公司 | Image classification method, computer terminal and storage medium |
| CN113627576A (en) * | 2021-10-08 | 2021-11-09 | 平安科技（深圳）有限公司 | Code scanning information detection method, device, equipment and storage medium |
| CN114140637B (en) * | 2021-10-21 | 2023-09-12 | 阿里巴巴达摩院(杭州)科技有限公司 | Image classification method, storage medium and electronic device |
| CN114140637A (en) * | 2021-10-21 | 2022-03-04 | 阿里巴巴达摩院(杭州)科技有限公司 | Image classification method, storage medium and electronic device |
| CN114155417B (en) * | 2021-12-13 | 2022-07-19 | 中国科学院空间应用工程与技术中心 | Image target identification method and device, electronic equipment and computer storage medium |
| CN114155417A (en) * | 2021-12-13 | 2022-03-08 | 中国科学院空间应用工程与技术中心 | Image target identification method and device, electronic equipment and computer storage medium |
| CN114417871A (en) * | 2021-12-17 | 2022-04-29 | 北京百度网讯科技有限公司 | Model training and named entity recognition method and device, electronic equipment and medium |
| CN114417871B (en) * | 2021-12-17 | 2023-01-31 | 北京百度网讯科技有限公司 | Model training and named entity recognition method, device, electronic equipment and medium |
| CN114241243A (en) * | 2021-12-20 | 2022-03-25 | 百度在线网络技术（北京）有限公司 | Training method and device of image classification model, electronic equipment and storage medium |
| CN113963220A (en) * | 2021-12-22 | 2022-01-21 | 熵基科技股份有限公司 | Security check image classification model training method, security check image classification method and device |
| CN114419375A (en) * | 2022-01-20 | 2022-04-29 | 北京百度网讯科技有限公司 | Image classification method, training method, device, electronic equipment and storage medium |
| CN114419375B (en) * | 2022-01-20 | 2023-08-15 | 北京百度网讯科技有限公司 | Image classification method, training device, electronic equipment and storage medium |
| CN115049836A (en) * | 2022-08-16 | 2022-09-13 | 平安科技（深圳）有限公司 | Image segmentation method, device, equipment and storage medium |
| CN115482436A (en) * | 2022-09-21 | 2022-12-16 | 北京百度网讯科技有限公司 | Training method and device for image screening model and image screening method |
| CN117853490A (en) * | 2024-03-06 | 2024-04-09 | 阿里巴巴达摩院(杭州)科技有限公司 | Image processing method and training method of image processing model |
| CN117853490B (en) * | 2024-03-06 | 2024-05-24 | 阿里巴巴达摩院(杭州)科技有限公司 | Image processing method and training method of image processing model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113435522A (en) | Image classification method, device, equipment and storage medium | |
| AU2019213369B2 (en) | Non-local memory network for semi-supervised video object segmentation | |
| CN111368254B (en) | Multi-view data missing completion method for multi-manifold regularization non-negative matrix factorization | |
| Muhammad et al. | Visual saliency models for summarization of diagnostic hysteroscopy videos in healthcare systems | |
| CN113689436B (en) | Image semantic segmentation method, device, equipment and storage medium | |
| CN110472049B (en) | Disease screening text classification method, computer device and readable storage medium | |
| CN108804591A (en) | A kind of file classification method and device of case history text | |
| CN113762309B (en) | Object matching method, device and equipment | |
| CN113450822B (en) | Voice enhancement method, device, equipment and storage medium | |
| CN114896067B (en) | Automatic generation method and device of task request information, computer equipment and medium | |
| CN113762303A (en) | Image classification method and device, electronic equipment and storage medium | |
| CN113268597A (en) | Text classification method, device, equipment and storage medium | |
| CN113627576B (en) | Code scanning information detection method, device, equipment and storage medium | |
| CN113705468B (en) | Digital image recognition method based on artificial intelligence and related equipment | |
| CN113486848A (en) | Document table identification method, device, equipment and storage medium | |
| CN113269179A (en) | Data processing method, device, equipment and storage medium | |
| US20240233325A9 (en) | Method for classifying images and electronic device | |
| CN114581177B (en) | Product recommendation method, device, equipment and storage medium | |
| CN113343970B (en) | Text image detection method, device, equipment and storage medium | |
| CN113850632B (en) | User category determination method, device, equipment and storage medium | |
| CN112434631B (en) | Target object identification method, target object identification device, electronic equipment and readable storage medium | |
| CN112949305B (en) | Negative feedback information acquisition method, device, equipment and storage medium | |
| CN115762721A (en) | Medical image quality control method and system based on computer vision technology | |
| CN114821062A (en) | Commodity identification method and device based on image segmentation | |
| CN113408265A (en) | Semantic analysis method, device and equipment based on human-computer interaction and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210924 |