Method and apparatus for obtaining semantic label of digital image
View Patent ↗The present application discloses a method and apparatus for obtaining a semantic label of a digital image. An implementation of the method includes: obtaining the digital image; looking up a semantic label model corresponding to the digital image, the semantic label model being used for representing correlation between digital images and semantic labels, and a semantic label being used for literally describing a digital image; and introducing the digital image into the semantic label model to obtain full-image recognition information and local recognition information corresponding to the digital image, and combining the full-image recognition information and the local recognition information to form a semantic label, the full-image recognition information being a summarized description of the digital image, and the local recognition information being a detailed description of the digital image. According to the implementation, the digital image is obtained first, then a semantic label model corresponding to the digital image is looked up, and a semantic label is obtained by using the semantic label model, which may improve the accuracy of obtaining the semantic label corresponding to the digital image.
1. A method for obtaining a semantic label of a digital image, comprising:
obtaining the digital image;
looking up a semantic label model corresponding to the digital image, the semantic label model being used for representing correlation between digital images and semantic labels, and a semantic label being used for literally describing a digital image; and
introducing the digital image into the semantic label model to obtain full-image recognition information and local recognition information corresponding to the digital image, and combining the full-image recognition information and the local recognition information to form a semantic label, the full-image recognition information being a summarized description of the digital image, and the local recognition information being a detailed description of the digital image;
the method further comprising:
establishing the semantic label model, wherein the establishing the semantic label model comprises:
extracting digital images and semantic labels from a digital image set and a semantic label set, respectively;
classifying the digital images into a set of at least one class of digital images according to the class information, the class information comprising at least one of numeral, character, person, animal and food;
classifying the semantic labels into a set of at least one class of semantic labels according to the class information; and
obtaining at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process;
wherein the obtaining of at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process comprises:
carrying out granularity recognition on the class of digital images to obtain granularity information corresponding to the class of digital images, the granularity information being sub-classification information of the class information;
looking up a class of semantic labels corresponding to the granularity information; and
obtaining the semantic label model corresponding to the class information by training based on the granularity information and the class of semantic labels corresponding to the granularity information by using a machine learning process.
2. The method according to claim 1 , wherein the looking up of a semantic label model corresponding to the digital image comprises:
carrying out class analysis on the digital image to determine class information of the digital image, the class information comprising at least one of numeral, character, person, animal and food; and
looking up the semantic label model corresponding to the class information.
3. The method according to claim 1 , wherein the classifying of the digital images into a set of at least one class of digital images according to the class information comprises:
recognizing the digital images to obtain the class information.
4. The method according to claim 1 , wherein the carrying out of granularity recognition on the class of digital images to obtain granularity information corresponding to the class of digital images comprises:
carrying out full-image recognition on the class of digital images to obtain full-image recognition information;
determining an attention area from the class of digital images, the attention area being an area in which the granularity recognition is carried out on the class of digital images;
carrying out the granularity recognition on the images in the attention area to obtain the local recognition information; and
combining the full-image recognition information and the local recognition information to form the granularity information.
5. An apparatus for obtaining a semantic label of a digital image, comprising:
at least one processor; and
a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
obtaining the digital image;
looking up a semantic label model corresponding to the digital image, the semantic label model being used for representing correlation between digital images and semantic labels, and a semantic label being used for literally describing a digital image; and
introducing the digital image into the semantic label model to obtain full-image recognition information and local recognition information corresponding to the digital image, and combining the full-image recognition information and the local recognition information to form a semantic label, the full-image recognition information being a summarized description of the digital image, and the local recognition information being a detailed description of the digital image;
wherein the operations further comprise:
establishing the semantic label model, the establishing the semantic label model comprising:
extracting digital images and semantic labels from a digital image set and a semantic label set, respectively;
classifying the digital images into a set of at least one class of digital images according to the class information, the class information comprising at least one of numeral, character, person, animal and food;
classifying the semantic labels into a set of at least one class of semantic labels according to the class information; and
obtaining at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process;
wherein the obtaining of at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process comprises:
carrying out granularity recognition on the class of digital images to obtain granularity information corresponding to the class of digital images, the granularity information being sub-classification information of the class information;
looking up a class of semantic labels corresponding to the granularity information; and
obtaining the semantic label model corresponding to the class information by training based on the granularity information and the class of semantic labels corresponding to the granularity information by using a machine learning process.
6. The apparatus according to claim 5 , wherein the looking up of a semantic label model corresponding to the digital image comprises:
carrying out class analysis on the digital image to determine class information of the digital image, the class information comprising at least one of numeral, character, person, animal and food; and
looking up the semantic label model corresponding to the class information.
7. The apparatus according to claim 5 , wherein the classifying of the digital images into a set of at least one class of digital images according to the class information comprises:
recognizing the digital images to obtain the class information.
8. The apparatus according to claim 5 , wherein the carrying out of granularity recognition on the class of digital images to obtain granularity information corresponding to the class of digital images comprises:
carrying out full-image recognition on the class of digital images to obtain full-image recognition information;
determining an attention area from the class of digital images, the attention area being an area in which the granularity recognition is carried out on the class of digital images;
carrying out the granularity recognition on the images in the attention area to obtain the local recognition information; and
combining the full-image recognition information and the local recognition information to form the granularity information.
9. A non-transitory storage medium storing one or more programs, the one or more programs when executed by an apparatus, causing the apparatus to perform operations, the operations comprising:
obtaining the digital image;
looking up a semantic label model corresponding to the digital image, the semantic label model being used for representing correlation between digital images and semantic labels, and a semantic label being used for literally describing a digital image; and
introducing the digital image into the semantic label model to obtain full-image recognition information and local recognition information corresponding to the digital image, and combining the full-image recognition information and the local recognition information to form a semantic label, the full-image recognition information being a summarized description of the digital image, and the local recognition information being a detailed description of the digital image;
wherein the operations further comprise:
establishing the semantic label model, the establishing the semantic label model comprising:
extracting digital images and semantic labels from a digital image set and a semantic label set, respectively;
classifying the digital images into a set of at least one class of digital images according to the class information, the class information comprising at least one of numeral, character, person, animal and food;
classifying the semantic labels into a set of at least one class of semantic labels according to the class information; and
obtaining at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process;
wherein the obtaining of at least one semantic label model corresponding to the class information by training based on the class of digital images and a class of semantic labels associated with the class of digital images by using a machine learning process comprises:
carrying out granularity recognition on the class of digital images to obtain granularity information corresponding to the class of digital images, the granularity information being sub-classification information of the class information;
looking up a class of semantic labels corresponding to the granularity information; and
obtaining the semantic label model corresponding to the class information by training based on the granularity information and the class of semantic labels corresponding to the granularity information by using a machine learning process.