Method, apparatus, electronic device and computer readable storage medium for image searching

Abstract

A method and apparatus for image searching based on artificial intelligent (AI) are provided. The method includes obtaining first feature information by extracting features from an image based on a first neural network, obtaining second feature information corresponding to a target area of a query image by processing the first feature information based on a second neural network and at least two filters having different sizes, and identifying an image corresponding to the query image according to the second feature information.

What is claimed is: 
     
         1 . A method, comprising:
 obtaining feature information including boundary information of at least two target areas in an input image by inputting the input image to a first neural network;   obtaining target feature vectors representing the at least two target areas by processing the feature information based on at least two second neural networks for the at least two target areas, wherein the two target areas in the input image are associated with each other based on position information of the at least two target areas extracted from the input image; and   searching a database storing a plurality of images for an image corresponding to the input image by comparing the target feature vectors with feature vectors of the plurality of images.   
     
     
         2 . The method of  claim 1 , wherein the comparing the target feature vectors with the feature vectors of the plurality of images comprises calculating distances between the target feature vectors and the feature vectors of the plurality of images. 
     
     
         3 . The method of  claim 2 , further comprising:
 sorting the calculated distances to identify a smallest distance.   
     
     
         4 . The method of  claim 2 , wherein the distances between the target feature vectors and the feature vectors of the plurality of images are cosine distances between the target feature vectors and the feature vectors of the plurality of images. 
     
     
         5 . The method of  claim 1 , wherein the plurality of images stored in the database comprise a picture taken by a user. 
     
     
         6 . The method of  claim 1 , further comprising, before the obtaining the feature information, the obtaining the target feature vectors, and searching the database:
 obtaining the feature vectors of the plurality of images based on feature information obtained for each of the plurality of images.   
     
     
         7 . The method of  claim 1 , further comprising:
 selecting a subject related to the at least two target areas when the input image includes a plurality of subjects.   
     
     
         8 . The method of  claim 7 , wherein the selecting the subject comprises receiving a selection of the subject among the plurality of subjects in the input image from a user. 
     
     
         9 . The method of  claim 1 , wherein the at least two second neural networks are separate networks for separately extracting each target feature vector of each of the at least two target areas. 
     
     
         10 . A computer program product comprising a non-transitory computer-readable recording medium having recorded thereon a plurality of instructions, which when executed by a computer, instruct the computer to:
 obtain feature information including boundary information of at least two target areas in an input image by inputting the input image to a first neural network,   obtain target feature vectors representing the at least two target areas by processing the feature information based on at least two second neural networks for the at least two target areas, wherein the two target areas in the input image are associated with each other based on position information of the at least two target areas extracted from the input image, and   search a database storing a plurality of images for an image corresponding to the input image by comparing the target feature vectors with feature vectors of the plurality of images.   
     
     
         11 . An electronic device comprising:
 a memory storing instructions; and   at least one processor configured to execute the instructions to:   obtain feature information including boundary information of at least two target areas in an input image by inputting the input image to a first neural network,   obtain target feature vectors representing the at least two target areas by processing the feature information based on at least two second neural networks for the at least two target areas, wherein the two target areas in the input image are associated with each other based on position information of the at least two target areas extracted from the input image, and   search a database storing a plurality of images for an image corresponding to the input image by comparing the target feature vectors with feature vectors of the plurality of images.   
     
     
         12 . The electronic device of  claim 11 , wherein the processor is further configured to execute the instructions to calculate distances between the target feature vectors and the feature vectors of the plurality of images. 
     
     
         13 . The electronic device of  claim 12 , wherein the processor is further configured to execute the instructions to sort the calculated distances to identify a smallest distance. 
     
     
         14 . The electronic device of  claim 12 , wherein the distances between the target feature vectors and the feature vectors of the plurality of images are cosine distances between the target feature vectors and the feature vectors of the plurality of images. 
     
     
         15 . The electronic device of  claim 11 , wherein the plurality of images stored in the database comprise a picture taken by a user. 
     
     
         16 . The electronic device of  claim 11 , wherein the processor is further configured to execute the instructions to, before the obtaining the feature information, the obtaining the target feature vectors, and searching the database, obtain the feature vectors of the plurality of images based on feature information obtained for each of the plurality of images. 
     
     
         17 . The electronic device of  claim 11 , wherein the processor is further configured to execute the instructions to select a subject related to the at least two target areas when the input image include a plurality of subjects. 
     
     
         18 . The electronic device of  claim 17 , wherein the processor is further configured to execute the instructions to receive a selection of the subject among the plurality of subjects in the input image from a user. 
     
     
         19 . The electronic device of  claim 11 , wherein the at least two second neural networks are separate networks for separately extracting each target feature vector of each of the at least two target areas. 
     
     
         20 . The method of  claim 1 , wherein the feature information includes basic structure information of boundaries, intersections, and shape information of a target feature. 
     
     
         21 . The method of  claim 1 , wherein the feature information includes information reflecting spatial relationship between the two target areas. 
     
     
         22 . The method of  claim 1 , wherein the obtaining of the feature information is based on multilayer feature extraction.