CN108564102A - Image clustering evaluation of result method and apparatus - Google Patents
Image clustering evaluation of result method and apparatus Download PDFInfo
- Publication number
- CN108564102A CN108564102A CN201810007796.1A CN201810007796A CN108564102A CN 108564102 A CN108564102 A CN 108564102A CN 201810007796 A CN201810007796 A CN 201810007796A CN 108564102 A CN108564102 A CN 108564102A
- Authority
- CN
- China
- Prior art keywords
- clustering result
- result evaluation
- image cluster
- cluster
- target image
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present application discloses image clustering evaluation of result method and apparatus.One specific implementation mode of this method includes：Each image in the target image cluster obtained to cluster carries out feature extraction；The corresponding Eigen Covariance matrix of target image cluster is generated based on the feature extracted；Covariance matrix is inputted in cluster result evaluation model, the cluster result evaluation information of target image cluster is obtained, wherein sample image cluster training of the cluster result evaluation model based on marked cluster result evaluation information obtains.The embodiment realizes the assessment of reliable image clustering effect.
Description
Technical Field
The embodiment of the application relates to the technical field of computers, in particular to the technical field of image processing, and particularly relates to an image clustering result evaluation method and device.
Background
Clustering is one method of data mining. In image processing techniques, image clustering is a process of dividing a plurality of images into a plurality of classes composed of similar images based on image features. Image clustering plays an important role in the fields of image segmentation, target tracking and the like.
In a large-scale image clustering scene, errors may occur in a clustering process, and if images which do not belong to the same class are classified into the same image cluster, the center of the image cluster is shifted, so that subsequent clustering results are more and more inaccurate, and therefore, the accuracy of the clustering results needs to be evaluated, and the wrong image clustering results need to be corrected in time.
Disclosure of Invention
The embodiment of the application provides an image clustering result evaluation method and device.
In a first aspect, an embodiment of the present application provides an image clustering result evaluation method, including: extracting the characteristics of each image in the target image cluster obtained by clustering; generating a feature covariance matrix corresponding to the target image cluster based on the extracted features; inputting the covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster; and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
In some embodiments, the above-mentioned clustering result evaluation model is trained as follows: obtaining a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster; extracting the characteristics of each sample image cluster; generating a sample feature covariance matrix of each sample image cluster based on the extracted features; and training to obtain a clustering result evaluation model based on a marking result of clustering result evaluation information of the sample image cluster and a preset loss function by adopting a deep learning method, wherein the preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and a prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
In some embodiments, the obtaining of the clustering result evaluation model by using the deep learning method and training based on the labeling result of the clustering result evaluation information of the sample image cluster and a preset loss function with the sample feature covariance matrix as the input of the neural network corresponding to the clustering result evaluation model includes: determining an initial value of a parameter of a neural network corresponding to the clustering result evaluation model, and executing a comparison step; the comparison step comprises: inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of clustering result evaluation information of a corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition or not; if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step; and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
In some embodiments, the performing feature extraction on each image in the clustered target image cluster includes: extracting the characteristics of each image in the target image cluster by adopting a convolutional neural network to obtain a characteristic image set of each image in the target image cluster; the generating of the feature covariance matrix corresponding to the target image cluster based on the extracted features includes: constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster; and calculating a covariance matrix of the feature matrix of the target image cluster.
In some embodiments, the above method further comprises: and reserving or/deleting the target image cluster according to the clustering result evaluation information of the target image cluster.
In a second aspect, an embodiment of the present application provides an image clustering result evaluation device, including: the extraction unit is used for extracting the characteristics of each image in the target image cluster obtained by clustering; a generating unit, configured to generate a feature covariance matrix corresponding to the target image cluster based on the extracted features; the evaluation unit is used for inputting the characteristic covariance matrix into the clustering result evaluation model to obtain clustering result evaluation information of the target image cluster; and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
In some embodiments, the above-mentioned clustering result evaluation model is trained as follows: obtaining a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster; extracting the characteristics of each sample image cluster; generating a sample feature covariance matrix of each sample image cluster based on the extracted features; and training to obtain a clustering result evaluation model based on a marking result of clustering result evaluation information of the sample image cluster and a preset loss function by adopting a deep learning method, wherein the preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and a prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
In some embodiments, the clustering result evaluation model is obtained by training a sample feature covariance matrix as an input of a neural network corresponding to the clustering result evaluation model by using a deep learning method based on a labeling result of clustering result evaluation information of a sample image cluster and a preset loss function in the following manner: determining an initial value of a parameter of a neural network corresponding to the clustering result evaluation model, and executing a comparison step; the comparison step comprises: inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of clustering result evaluation information of a corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition or not; if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step; and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
In some embodiments, the extracting unit is further configured to perform feature extraction on each image in the clustered target image cluster as follows: extracting the characteristics of each image in the target image cluster by adopting a convolutional neural network to obtain a characteristic image set of each image in the target image cluster; the generating unit is further configured to generate a feature covariance matrix corresponding to the target image cluster as follows: constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster; and calculating a covariance matrix of the feature matrix of the target image cluster.
In some embodiments, the above apparatus further comprises: and the processing unit is used for reserving or deleting the target image cluster according to the clustering result evaluation information of the target image cluster.
In a third aspect, an embodiment of the present application provides an electronic device, including: one or more processors; a storage device for storing one or more programs, which when executed by one or more processors, cause the one or more processors to implement the image clustering result evaluation method as provided in the first aspect.
In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, where the program, when executed by a processor, implements the image clustering result evaluation method provided in the first aspect.
According to the image clustering result evaluation method and device of the embodiment, each image in a target image cluster obtained through clustering is subjected to feature extraction, and then a feature covariance matrix corresponding to the target image cluster is generated based on the extracted features; and then inputting the covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster, wherein the clustering result evaluation model is obtained based on sample image cluster training marked with the clustering result evaluation information, so that reliable evaluation of the image clustering effect is realized.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:
FIG. 1 is an exemplary system architecture diagram in which the present application may be applied;
FIG. 2 is a flow diagram of one embodiment of an image clustering result evaluation method according to the present application;
FIG. 3 is a schematic diagram of an application scenario of the image clustering result evaluation method according to the present application;
fig. 4 is a schematic structural diagram of an image clustering result evaluation apparatus according to an embodiment of the present application;
FIG. 5 is a schematic block diagram of a computer system suitable for use in implementing an electronic device according to embodiments of the present application.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the related invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
Fig. 1 shows an exemplary system architecture 100 to which an embodiment of the image clustering result evaluating method or the image clustering result evaluating apparatus of the present application may be applied.
As shown in fig. 1, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The terminal devices 101, 102, 103 may interact with the server 105 via the network 104 to receive or transmit data. The terminal devices 101, 102, 103 may be devices with image capturing functions, such as a surveillance camera, a smart phone, a tablet computer, a personal computer, etc. The terminal devices 101, 102, and 103 may have a network interface, and may receive an image acquisition request sent by the server 105, or may upload a captured image to the server 105 through the network interface for processing in response to a request of the user 110.
The server 105 may be a server that provides various services, such as a server that performs image clustering on images uploaded by the terminal devices 101, 102, 103. After the terminal devices 101, 102 upload images, the server 105 may perform feature extraction, clustering, object recognition, and the like on the uploaded images, and return the processing results to the terminal devices 101, 102, 103.
It should be noted that the image clustering result evaluation method provided in the embodiment of the present application may be executed by the terminal devices 101, 102, and 103 or the server 105, and accordingly, the image clustering result evaluation device may be disposed in the terminal devices 101, 102, and 103 or the server 105.
It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation. For example, the server may be a clustered server, including multiple servers with different processes deployed.
With continued reference to FIG. 2, a flow 200 of one embodiment of an image clustering result evaluation method according to the present application is shown. The image clustering result evaluation method comprises the following steps:
step 201, performing feature extraction on each image in the target image cluster obtained by clustering.
In this embodiment, an electronic device (for example, a terminal device or a server shown in fig. 1) on which the above-described image clustering result evaluation method operates may acquire image clusters obtained by clustering images, where each image cluster may include a plurality of images, for example, may include a plurality of face images. Here, the target image cluster may be an image cluster to be evaluated, and may be any one of a plurality of image clusters obtained by clustering.
In an actual scene, after the images are clustered to obtain image clusters, each image cluster can be sequentially used as a target image cluster, and the clustering result is evaluated. The selected image cluster may also be determined to be a target image cluster in response to receiving an instruction to select an image cluster to be evaluated.
Various feature extraction methods can be employed to perform feature extraction on each image in the target image cluster. For example, statistical features of the image, including area, gray value, histogram statistical features, etc., may be extracted; the image features can be extracted by algorithms such as edge detection, corner detection, Scale Invariant Feature Transform (SIFT), principal component analysis and the like. The extracted features can be expressed in the form of arrays, vectors or matrixes.
Optionally, each image in the target image cluster is subjected to feature extraction by using the same feature extraction method, so that the dimensions of the extracted features of each image are the same.
Step 202, generating a feature covariance matrix corresponding to the target image cluster based on the extracted features.
After the features of each image are extracted, a feature covariance matrix corresponding to the target image cluster can be constructed based on the features of all the images in the target image cluster. Specifically, the covariance between all the features of the target image cluster may be calculated, and each covariance is used as an element in the covariance matrix.
In some optional implementation manners of this embodiment, the step 201 of performing feature extraction on each image in the clustered target image cluster may include: and (3) performing feature extraction on each image in the target image cluster by adopting a convolutional neural network to obtain a feature map set of each image in the target image cluster. At this time, step 202 may be performed as follows: constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster; and calculating a covariance matrix of the feature matrix of the target image cluster. That is, a convolutional neural network may be used as a feature extraction method, and a plurality of feature maps of each extracted image may be combined into a feature map set. Then, feature map sets of each image in the same target image cluster can be combined together to form a feature map set of the target image cluster, and a feature matrix is constructed based on the feature map set of the target image cluster. The feature matrix can be formed by sequentially combining the pixel gray level matrix of each feature map in the feature map set of the target image cluster along the row direction, the column direction or a preset combination mode of the matrix. The covariance matrix of the feature matrix may then be calculated.
Specifically, assume that the target image cluster includes n images X1、X2、…、XnEach image XiIs characterized in thatiCharacteristic fiMay be a D-dimensional feature, fi∈RDThen, the feature matrix S corresponding to the target image cluster is:
wherein f isijIs characterized byiJ ═ 1,2, …, D.
Then, a covariance matrix of a feature matrix S corresponding to the target image cluster, that is, a feature covariance matrix E corresponding to the image cluster, can be calculated, as shown in formula (2):
wherein,
in this way, a characteristic covariance matrix E of size D × D can be obtained.
If the similarity between the images in the image cluster is high and the clustering precision is high, the obtained characteristic covariance matrix E presents regular distribution; otherwise, if the similarity between the images in the image cluster is low and the clustering precision is low, the distribution regularity of the obtained feature covariance matrix E is poor.
And 203, inputting the characteristic covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster.
In this embodiment, the feature covariance matrix may be input to a trained clustering result evaluation model for processing. Here, the clustering result evaluation model is a model for evaluating the clustering result of the image cluster, and can determine whether the clustering result is accurate.
In this embodiment, the clustering result evaluation model may be constructed based on a convolutional neural network, and the convolutional neural network may include at least one convolutional layer and may further include at least one downsampling layer. Each convolution layer includes a convolution kernel, and can perform convolution operation on an input image by using the convolution kernel to remove redundant image information and extract image features. The extracted image features may be a feature map, and the feature map may be converted into a feature vector including at least one-dimensional features for representation.
The clustering result evaluation model can be obtained by adopting a supervised machine learning method for training. The method can be specifically obtained by training sample image clusters based on labeled clustering result evaluation information. The sample image cluster to which the clustering result evaluation information is labeled may include a plurality of sample image clusters, and a labeling result of the clustering result evaluation information of each sample image cluster. When a sample image cluster is constructed, some sample images can be selected, negative samples are formed by random classification, positive samples are formed by classification according to the similarity degree between the images, and clustering result evaluation information of the negative samples and the positive samples is respectively marked. In the training process, the clustering result evaluation model can be used for predicting the clustering result evaluation information of the sample image cluster, the prediction result is compared with the marking result, and the parameters of the clustering result evaluation model are continuously adjusted, so that the prediction result is approximate to the marking result. In this way, the clustering result evaluation model can learn the logic of artificially labeling the clustering result evaluation information.
In an actual scene, the clustering result evaluation information of the sample image cluster can be marked according to the requirement on clustering precision in application. In general, the clustering results of sample image clusters can be labeled as "accurate" and "inaccurate," respectively labeled with numerical or character labels. In some scenes, evaluation information of the clustering result of the sample image cluster can be divided into a plurality of levels according to the accuracy degree during marking, for example, the constructed sample image clusters a, b and c respectively contain 100 images, the similarity degree between 100 images in the sample image cluster a is higher, the similarity degree between 5 images in the sample image cluster b and other 95 images is lower, and the similarity degree between 20 images in the sample image cluster c and other 80 images is lower; the clustering results of the sample image clusters a, b, c can be labeled as "accurate", "good", "inaccurate", respectively. In this way, the trained image clustering result evaluation model can evaluate the input target image cluster according to the hierarchy adopted in marking.
In the image clustering result evaluation method provided by the embodiment of the application, each image in a target image cluster obtained by clustering is subjected to feature extraction, and then a feature covariance matrix corresponding to the target image cluster is generated based on the extracted features; and then inputting the covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster, wherein the clustering result evaluation model is obtained based on sample image cluster training marked with the clustering result evaluation information, so that the evaluation of the image clustering effect is realized, and the reliability of the evaluation of the clustering effect can be improved.
In some optional implementations of the present embodiment, the clustering result evaluation model may be trained as follows:
firstly, a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster are obtained. Some pictures can be selected from image databases such as an existing network image library and a monitoring image library to construct a sample image cluster. It should be noted that the constructed sample image clusters at least include image clusters with accurate clustering results, that is, at least one sample image cluster needs to be constructed, and the similarity between images in the sample image cluster is high. For example, in a face image clustering scenario, at least one sample image cluster containing multiple different face images of the same user needs to be constructed.
The above-mentioned evaluation information of the clustering result of the sample image cluster may be marked manually, and may be represented by a numeral or a symbol label, for example, the marking result of the evaluation information of the clustering result of the sample image cluster that is clustered accurately may be represented by a label "1", and the marking result of the evaluation information of the clustering result of the sample image cluster that is clustered inaccurately may be represented by a label "0". Thus, after training is completed, the clustering result evaluation model may output a corresponding label "1" or "0" to indicate a clustering result.
Then, feature extraction is performed on each sample image cluster.
The images in the sample image cluster can be feature extracted by adopting a plurality of image feature extraction methods. For example, principal component analysis, scale invariant feature transformation, and the like may be used to extract features of each image, or a neural network may be used to perform feature extraction.
Optionally, the same feature extraction method may be used to extract features of each image in the sample image cluster and the target image cluster, so as to avoid that differences between types or attributes of features extracted by different feature extraction methods (for example, the extracted feature types may be texture features, edge features, gray scale distribution features, and the like) affect the accuracy of the evaluation result of the image clustering result.
And then, generating a sample feature covariance matrix of each sample image cluster based on the extracted features.
Then, the covariance matrix of the matrix formed by the features of the sample image cluster can be calculated by the same method as the calculation method of the feature covariance matrix corresponding to the target image cluster, that is, the sample feature covariance matrix of the sample image cluster can be calculated. The specific calculation method refers to the above formulas (1), (2), and (3), and is not described herein again.
And finally, a deep learning method is adopted, the sample characteristic covariance matrix is used as the input of a neural network corresponding to the clustering result evaluation model, and the clustering result evaluation model is obtained based on the marking result of the clustering result evaluation information of the sample image cluster and a preset loss function training. The preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and the prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
A neural network of a clustering result evaluation model, which may be a convolutional neural network, including a plurality of convolutional layers and downsampling layers, may be constructed. And then inputting the sample characteristic covariance matrix into a convolutional neural network to obtain a prediction result of the evaluation information of the clustering result of the sample image cluster. And then constructing a loss function based on the difference between the output prediction result and the marking result, then minimizing the value of the loss function by repeatedly executing and adjusting the parameters of the clustering result evaluation model, inputting the sample characteristic covariance matrix into the clustering result evaluation model to obtain a new prediction result, and taking the parameters used when the value of the loss function is minimum as the parameters of the trained clustering result evaluation model.
In a further implementation manner, the step of obtaining the clustering result evaluation model by using the deep learning method and using the sample feature covariance matrix as the input of the neural network corresponding to the clustering result evaluation model, and training based on the labeled result of the clustering result evaluation information of the sample image cluster and a preset loss function may be performed in the following manner:
firstly, determining an initial value of a parameter of a neural network corresponding to a clustering result evaluation model, and executing a comparison step. The comparison step comprises: and inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of the clustering result evaluation information of the corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition.
Here, the loss function may represent a difference between a prediction result and a labeling result of the clustering result evaluation information of the sample image cluster. The prediction result of the clustering result evaluation information can have probability distribution, and then a loss function can be constructed by adopting cross entropy based on a softmax function.
After the prediction result is obtained by predicting the evaluation information of the clustering result of the sample image cluster by using the neural network of the clustering result evaluation model in which the parameter value is determined, the value of the loss function can be calculated. Then, it may be determined whether the value of the loss function satisfies a preset convergence condition, where the preset convergence condition may be that the value of the loss function is smaller than a first threshold, or that a difference between values of the loss function obtained in the last several comparison steps is smaller than a second threshold.
Then, whether to update the parameters can be determined according to the judgment result of the comparison step. Specifically, if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step; and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
Specifically, if the judgment result of the comparison step is that the value of the loss function does not satisfy the preset convergence condition, calculating the gradient of the loss function to the parameter in the neural network corresponding to the clustering result evaluation model, and updating each parameter in the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method. And then, executing the comparison step based on the clustering result evaluation model after the parameters are updated.
If the judgment result of the comparison step is that the value of the loss function meets the preset convergence condition, it can be determined that the current clustering result evaluation model realizes the logic learning of the clustering result evaluation of the sample image cluster, and at this time, the current parameter can be output as the parameter of the trained clustering result evaluation model.
The clustering result evaluation model obtained through the training in the mode makes full use of the consistency of the features of the image clusters formed by similar images and the reflection of the difference of the features of the image clusters formed by dissimilar images in the corresponding feature covariance, and continuously iteratively updates the parameters of the neural network, so that the training efficiency can be effectively improved, and the prediction effect of the model can be improved.
In some embodiments, the image clustering result evaluation method may further include: and reserving or/deleting the target image cluster according to the clustering result evaluation information of the target image cluster. If the clustering result evaluation information of the target image cluster indicates that the clustering effect of the target image cluster is poor, deleting the target image cluster and clustering the images in the target image cluster again; if the clustering result evaluation information of the target image cluster indicates that the clustering effect of the target image cluster is good, the target image cluster can be reserved. Due to the fact that inaccurate clustering results can cause deviation of clustering centers, the method can avoid the fact that larger errors are generated in subsequent further clustering based on the clustering results due to the deviation of the clustering centers.
Please refer to fig. 3, which shows a schematic diagram of an application scenario of the image clustering result evaluation method according to the present application.
As shown in fig. 3, terminal device a may send the image collection to server B and request server B to cluster the image collection. And the server B obtains an intermediate clustering result comprising cluster 1, cluster 2, cluster 3, … and cluster m in the process of clustering the image set. Then, the server can evaluate the clustering result of each cluster in the preliminary clustering result, specifically, can extract the features of the image in each cluster, generate a feature covariance matrix of each cluster based on the extracted features, and then input the feature covariance matrix into a trained clustering result evaluation model to obtain the clustering result evaluation information of each cluster, wherein the clustering result evaluation information of the cluster 2 is 'unclean' and indicates that the clustering result of the cluster 2 is inaccurate; the cluster result evaluation information of the clusters 1, 3, … and m is "clean", which indicates that the cluster results of the clusters 1, 3, … and m are accurate. At this point, cluster 2 may be deleted and the images in cluster 2 may be re-clustered. As can be seen from fig. 3, the image clustering result evaluation method of the present embodiment can be applied to all the images in the image set when clustering is completed and executed at any time before all the images are clustered, and can improve the reliability of the clustering result.
With further reference to fig. 4, as an implementation of the methods shown in the above-mentioned figures, the present application provides an embodiment of an image clustering result evaluation apparatus, which corresponds to the method embodiment shown in fig. 2, and which is specifically applicable to various electronic devices.
As shown in fig. 4, the image clustering result evaluating apparatus 400 of the present embodiment includes: an extraction unit 401, a generation unit 402 and an evaluation unit 403. The extracting unit 401 may be configured to perform feature extraction on each image in the clustered target image cluster; the generating unit 402 may be configured to generate a feature covariance matrix corresponding to the target image cluster based on the extracted features; the evaluation unit 403 may be configured to input the feature covariance matrix into the clustering result evaluation model to obtain clustering result evaluation information of the target image cluster; and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
In this embodiment, the extraction unit 401 may acquire image clusters obtained by clustering images, and extract the features of each image in the specified target image cluster by using an image feature extraction method. The target image cluster may be any one of a plurality of image clusters obtained by clustering.
After the extracting unit 401 extracts the features of each image, the generating unit 402 may construct a feature matrix corresponding to the target image cluster based on the features of all images in the target image cluster, specifically, may represent the features of each image in the target image cluster by vectors or matrices, combine the features of each image into the feature matrix of the target image, and then may calculate a covariance matrix of the feature matrix, i.e., obtain a feature covariance matrix.
The evaluation unit 403 may input the feature covariance matrix generated by the generation unit 402 into the trained clustering result evaluation model for evaluation, so as to obtain evaluation information of the clustering result of the target image cluster.
In some embodiments, the above-mentioned clustering result evaluation model may be trained as follows: obtaining a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster; extracting the characteristics of each sample image cluster; generating a sample feature covariance matrix of each sample image cluster based on the extracted features; and training to obtain a clustering result evaluation model based on a marking result of clustering result evaluation information of the sample image cluster and a preset loss function by adopting a deep learning method, wherein the preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and a prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
In a further embodiment, the clustering result evaluation model may be obtained by training a sample feature covariance matrix as an input of a neural network corresponding to the clustering result evaluation model by using a deep learning method, based on a labeling result of the clustering result evaluation information of the sample image cluster and a preset loss function, further in the following manner: determining an initial value of a parameter of a neural network corresponding to the clustering result evaluation model, and executing a comparison step; the comparison step comprises: inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of clustering result evaluation information of a corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition or not; if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step; and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
In some embodiments, the extracting unit may be further configured to perform feature extraction on each image in the clustered target image cluster as follows: extracting the characteristics of each image in the target image cluster by adopting a convolutional neural network to obtain a characteristic image set of each image in the target image cluster; the generating unit may further generate the feature covariance matrix corresponding to the target image cluster as follows: constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster; and calculating a covariance matrix of the feature matrix of the target image cluster.
In some embodiments, the apparatus 400 may further include: and the processing unit is used for reserving or deleting the target image cluster according to the clustering result evaluation information of the target image cluster.
It should be understood that the units recited in the apparatus 400 correspond to the various steps in the method described with reference to fig. 2. Thus, the operations and features described above for the method are equally applicable to the apparatus 400 and the units included therein, and are not described in detail here.
In the image clustering result evaluation device 400 according to the embodiment of the present application, the extraction unit extracts the features of each image in the target image cluster obtained by clustering, and the generation unit generates the feature covariance matrix corresponding to the target image cluster based on the extracted features; and then the evaluation unit inputs the covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster, wherein the clustering result evaluation model is obtained based on sample image cluster training marked with the clustering result evaluation information, and reliable evaluation of the image clustering effect is realized.
Referring now to FIG. 5, shown is a block diagram of a computer system 500 suitable for use in implementing the electronic device of an embodiment of the present application. The electronic device shown in fig. 5 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present application.
As shown in fig. 5, the computer system 500 includes a Central Processing Unit (CPU)501 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)502 or a program loaded from a storage section 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data necessary for the operation of the system 500 are also stored. The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
The following components are connected to the I/O interface 505: an input portion 506 including a keyboard, a mouse, and the like; an output portion 507 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 508 including a hard disk and the like; and a communication section 509 including a network interface card such as a LAN card, a modem, or the like. The communication section 509 performs communication processing via a network such as the internet. The driver 510 is also connected to the I/O interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 510 as necessary, so that a computer program read out therefrom is mounted into the storage section 508 as necessary.
In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 509, and/or installed from the removable medium 511. The computer program performs the above-described functions defined in the method of the present application when executed by the Central Processing Unit (CPU) 501. It should be noted that the computer readable medium of the present application can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In this application, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, or the like, as well as conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in the embodiments of the present application may be implemented by software or hardware. The described units may also be provided in a processor, and may be described as: a processor includes an extraction unit, a generation unit, and an evaluation unit. The names of these units do not in some cases constitute a limitation to the unit itself, and for example, the extraction unit may also be described as a "unit that performs feature extraction on each image in the clustered target image cluster".
As another aspect, the present application also provides a computer-readable medium, which may be contained in the apparatus described in the above embodiments; or may be present separately and not assembled into the device. The computer readable medium carries one or more programs which, when executed by the apparatus, cause the apparatus to: extracting the characteristics of each image in the target image cluster obtained by clustering; generating a feature covariance matrix corresponding to the target image cluster based on the extracted features; inputting the covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster; and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the invention herein disclosed is not limited to the particular combination of features described above, but also encompasses other arrangements formed by any combination of the above features or their equivalents without departing from the spirit of the invention. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.
Claims (12)
1. An image clustering result evaluation method comprises the following steps:
extracting the characteristics of each image in the target image cluster obtained by clustering;
generating a feature covariance matrix corresponding to the target image cluster based on the extracted features;
inputting the characteristic covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster;
and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
2. The method of claim 1, wherein the clustering result evaluation model is trained as follows:
obtaining a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster;
extracting the characteristics of each sample image cluster;
generating a sample feature covariance matrix of each sample image cluster based on the extracted features;
and training to obtain the clustering result evaluation model based on a marking result of clustering result evaluation information of the sample image cluster and a preset loss function by adopting a deep learning method, wherein the preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and a prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
3. The method according to claim 2, wherein the using a deep learning method, taking the sample feature covariance matrix as an input of a neural network corresponding to the clustering result evaluation model, and training based on a labeled result of the clustering result evaluation information of the sample image cluster and a preset loss function to obtain the clustering result evaluation model comprises:
determining an initial value of a parameter of a neural network corresponding to the clustering result evaluation model, and executing a comparison step;
the step of aligning comprises: inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of clustering result evaluation information of a corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition;
if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step;
and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
4. The method of claim 1, wherein the feature extraction of each image in the clustered target image cluster comprises:
performing feature extraction on each image in the target image cluster by adopting a convolutional neural network to obtain a feature map set of each image in the target image cluster;
the generating of the feature covariance matrix corresponding to the target image cluster based on the extracted features includes:
constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster;
and calculating a covariance matrix of the feature matrix of the target image cluster.
5. The method of any of claims 1-4, wherein the method further comprises:
and reserving or/deleting the target image cluster according to the clustering result evaluation information of the target image cluster.
6. An image clustering result evaluation apparatus comprising:
the extraction unit is used for extracting the characteristics of each image in the target image cluster obtained by clustering;
a generating unit, configured to generate a feature covariance matrix corresponding to the target image cluster based on the extracted features;
the evaluation unit is used for inputting the characteristic covariance matrix into a clustering result evaluation model to obtain clustering result evaluation information of the target image cluster;
and the clustering result evaluation model is obtained based on sample image cluster training of marked clustering result evaluation information.
7. The apparatus of claim 6, wherein the clustering result evaluation model is trained as follows:
obtaining a plurality of sample image clusters and a marking result of clustering result evaluation information of each sample image cluster;
extracting the characteristics of each sample image cluster;
generating a sample feature covariance matrix of each sample image cluster based on the extracted features;
and training to obtain the clustering result evaluation model based on a marking result of clustering result evaluation information of the sample image cluster and a preset loss function by adopting a deep learning method, wherein the preset loss function is constructed based on the difference between the marking result of the clustering result evaluation information of the sample image cluster and a prediction result of the evaluation information of the clustering result of the sample image cluster output by the neural network corresponding to the clustering result evaluation model.
8. The apparatus according to claim 7, wherein the clustering result evaluation model is obtained by training, by using a deep learning method, the sample feature covariance matrix as an input of a neural network corresponding to the clustering result evaluation model, based on a labeling result of the clustering result evaluation information of the sample image cluster and a preset loss function, as follows:
determining an initial value of a parameter of a neural network corresponding to the clustering result evaluation model, and executing a comparison step;
the step of aligning comprises: inputting the sample characteristic covariance matrix into a neural network corresponding to the clustering result evaluation model to obtain a prediction result of clustering result evaluation information of a corresponding sample image cluster, and judging whether the value of the loss function meets a preset convergence condition;
if the judgment result of the comparison step is negative, updating parameters corresponding to the neural network corresponding to the clustering result evaluation model by adopting a gradient descent method based on the loss function, and executing the comparison step;
and if the judgment result of the comparison step is yes, outputting the parameters of the neural network corresponding to the clustering result evaluation model.
9. The apparatus according to claim 6, wherein the extracting unit is further configured to perform feature extraction on each image in the clustered target image cluster as follows:
performing feature extraction on each image in the target image cluster by adopting a convolutional neural network to obtain a feature map set of each image in the target image cluster;
the generating unit is further configured to generate a feature covariance matrix corresponding to the target image cluster as follows:
constructing a feature matrix of the target image cluster based on the feature map set of each image in the target image cluster;
and calculating a covariance matrix of the feature matrix of the target image cluster.
10. The apparatus of any of claims 6-9, wherein the apparatus further comprises:
and the processing unit is used for reserving or deleting the target image cluster according to the clustering result evaluation information of the target image cluster.
11. An electronic device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-5.
12. A computer-readable storage medium, on which a computer program is stored, wherein the program, when executed by a processor, implements the method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810007796.1A CN108564102A (en) | 2018-01-04 | 2018-01-04 | Image clustering evaluation of result method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810007796.1A CN108564102A (en) | 2018-01-04 | 2018-01-04 | Image clustering evaluation of result method and apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108564102A true CN108564102A (en) | 2018-09-21 |
Family
ID=63530706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810007796.1A Pending CN108564102A (en) | 2018-01-04 | 2018-01-04 | Image clustering evaluation of result method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108564102A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109615075A (en) * | 2018-12-14 | 2019-04-12 | 大连海事大学 | Resident daily behavior identification method based on multilayer clustering model |
| CN110020022A (en) * | 2019-01-03 | 2019-07-16 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and readable storage medium storing program for executing |
| CN110222590A (en) * | 2019-05-15 | 2019-09-10 | 北京字节跳动网络技术有限公司 | Image difference judgment method, device and electronic equipment |
| CN110400138A (en) * | 2019-07-30 | 2019-11-01 | 中国工商银行股份有限公司 | The method of mobile payment and device of illegal |
| CN110580510A (en) * | 2019-09-12 | 2019-12-17 | 深圳力维智联技术有限公司 | A clustering result evaluation method and system |
| CN110826616A (en) * | 2019-10-31 | 2020-02-21 | Oppo广东移动通信有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN111738319A (en) * | 2020-06-11 | 2020-10-02 | 佳都新太科技股份有限公司 | Clustering result evaluation method and device based on large-scale samples |
| CN112155729A (en) * | 2020-10-15 | 2021-01-01 | 中国科学院合肥物质科学研究院 | Intelligent automatic planning method and system for surgical puncture path and medical system |
| CN112232384A (en) * | 2020-09-27 | 2021-01-15 | 北京迈格威科技有限公司 | Model training method, image feature extraction method, target detection method and device |
| CN112906824A (en) * | 2021-03-29 | 2021-06-04 | 苏州科达科技股份有限公司 | Vehicle clustering method, system, device and storage medium |
| CN113011221A (en) * | 2019-12-19 | 2021-06-22 | 广州极飞科技股份有限公司 | Crop distribution information acquisition method and device and measurement system |
| WO2022121801A1 (en) * | 2020-12-07 | 2022-06-16 | 北京有竹居网络技术有限公司 | Information processing method and apparatus, and electronic device |
| CN117058496A (en) * | 2023-08-15 | 2023-11-14 | 世优（北京）科技有限公司 | Evaluation methods, devices and electronic equipment for image clustering |
| CN118691864A (en) * | 2024-08-23 | 2024-09-24 | 浙江大华技术股份有限公司 | A target clustering method, device, equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101334893A (en) * | 2008-08-01 | 2008-12-31 | 天津大学 | Comprehensive Evaluation Method of Fusion Image Quality Based on Fuzzy Neural Network |
| US20140201208A1 (en) * | 2013-01-15 | 2014-07-17 | Corporation Symantec | Classifying Samples Using Clustering |
| CN106850333A (en) * | 2016-12-23 | 2017-06-13 | 中国科学院信息工程研究所 | A kind of network equipment recognition methods and system based on feedback cluster |
| CN106845331A (en) * | 2016-11-18 | 2017-06-13 | 深圳云天励飞技术有限公司 | A kind of image processing method and terminal |
-
2018
- 2018-01-04 CN CN201810007796.1A patent/CN108564102A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101334893A (en) * | 2008-08-01 | 2008-12-31 | 天津大学 | Comprehensive Evaluation Method of Fusion Image Quality Based on Fuzzy Neural Network |
| US20140201208A1 (en) * | 2013-01-15 | 2014-07-17 | Corporation Symantec | Classifying Samples Using Clustering |
| CN106845331A (en) * | 2016-11-18 | 2017-06-13 | 深圳云天励飞技术有限公司 | A kind of image processing method and terminal |
| CN106850333A (en) * | 2016-12-23 | 2017-06-13 | 中国科学院信息工程研究所 | A kind of network equipment recognition methods and system based on feedback cluster |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109615075A (en) * | 2018-12-14 | 2019-04-12 | 大连海事大学 | Resident daily behavior identification method based on multilayer clustering model |
| CN109615075B (en) * | 2018-12-14 | 2022-08-19 | 大连海事大学 | Resident daily behavior identification method based on multilayer clustering model |
| CN110020022A (en) * | 2019-01-03 | 2019-07-16 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and readable storage medium storing program for executing |
| CN110020022B (en) * | 2019-01-03 | 2023-08-01 | 创新先进技术有限公司 | Data processing method, device, equipment and readable storage medium |
| CN110222590B (en) * | 2019-05-15 | 2021-05-25 | 北京字节跳动网络技术有限公司 | Image difference judgment method and device and electronic equipment |
| CN110222590A (en) * | 2019-05-15 | 2019-09-10 | 北京字节跳动网络技术有限公司 | Image difference judgment method, device and electronic equipment |
| CN110400138A (en) * | 2019-07-30 | 2019-11-01 | 中国工商银行股份有限公司 | The method of mobile payment and device of illegal |
| CN110580510A (en) * | 2019-09-12 | 2019-12-17 | 深圳力维智联技术有限公司 | A clustering result evaluation method and system |
| CN110580510B (en) * | 2019-09-12 | 2023-07-25 | 深圳力维智联技术有限公司 | A clustering result evaluation method and system |
| CN110826616B (en) * | 2019-10-31 | 2023-06-30 | Oppo广东移动通信有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN110826616A (en) * | 2019-10-31 | 2020-02-21 | Oppo广东移动通信有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN113011221A (en) * | 2019-12-19 | 2021-06-22 | 广州极飞科技股份有限公司 | Crop distribution information acquisition method and device and measurement system |
| CN111738319A (en) * | 2020-06-11 | 2020-10-02 | 佳都新太科技股份有限公司 | Clustering result evaluation method and device based on large-scale samples |
| CN112232384A (en) * | 2020-09-27 | 2021-01-15 | 北京迈格威科技有限公司 | Model training method, image feature extraction method, target detection method and device |
| CN112155729A (en) * | 2020-10-15 | 2021-01-01 | 中国科学院合肥物质科学研究院 | Intelligent automatic planning method and system for surgical puncture path and medical system |
| WO2022121801A1 (en) * | 2020-12-07 | 2022-06-16 | 北京有竹居网络技术有限公司 | Information processing method and apparatus, and electronic device |
| CN112906824A (en) * | 2021-03-29 | 2021-06-04 | 苏州科达科技股份有限公司 | Vehicle clustering method, system, device and storage medium |
| CN112906824B (en) * | 2021-03-29 | 2022-07-05 | 苏州科达科技股份有限公司 | Vehicle clustering method, system, device and storage medium |
| CN117058496A (en) * | 2023-08-15 | 2023-11-14 | 世优（北京）科技有限公司 | Evaluation methods, devices and electronic equipment for image clustering |
| CN118691864A (en) * | 2024-08-23 | 2024-09-24 | 浙江大华技术股份有限公司 | A target clustering method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108564102A (en) | Image clustering evaluation of result method and apparatus | |
| CN108304835B (en) | character detection method and device | |
| CN108280477B (en) | Method and apparatus for clustering images | |
| US11392792B2 (en) | Method and apparatus for generating vehicle damage information | |
| CN108171203B (en) | Method and device for identifying vehicle | |
| CN108229419B (en) | Method and apparatus for clustering images | |
| US20240257423A1 (en) | Image processing method and apparatus, and computer readable storage medium | |
| EP4220555B1 (en) | Training method and apparatus for image segmentation model, image segmentation method and apparatus, and device | |
| CN108229504B (en) | Image analysis method and device | |
| CN112949710A (en) | Image clustering method and device | |
| CN107944450B (en) | License plate recognition method and device | |
| CN111582409A (en) | Training method of image label classification network, image label classification method and device | |
| CN108304775A (en) | Remote sensing images recognition methods, device, storage medium and electronic equipment | |
| US20240312252A1 (en) | Action recognition method and apparatus | |
| CN114741697B (en) | Malicious code classification method and device, electronic equipment and medium | |
| CN113223011A (en) | Small sample image segmentation method based on guide network and full-connection conditional random field | |
| CN114663751A (en) | Power transmission line defect identification method and system based on incremental learning technology | |
| CN108229680A (en) | Nerve network system, remote sensing images recognition methods, device, equipment and medium | |
| CN114693997B (en) | Image description generation method, device, equipment and medium based on transfer learning | |
| CN110135428B (en) | Image segmentation processing method and device | |
| CN109165572B (en) | Method and apparatus for generating information | |
| CN115761397B (en) | Model training methods, image classification methods, devices, equipment and storage media | |
| CN110263779A (en) | Text filed detection method and device, Method for text detection, computer-readable medium | |
| CN115131626A (en) | Training method and device for image ambiguity recognition model | |
| CN115797291A (en) | Circuit terminal identification method and device, computer equipment 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 |
Application publication date: 20180921 |
|
| RJ01 | Rejection of invention patent application after publication |