System and method for learning scene embeddings via visual semantics and application thereof
The present teaching relates to method, system, and programming for responding to an image related query. Information related to each of a plurality of images is received, wherein the information represents concepts co-existing in the image. Visual semantics for each of the plurality of images are created based on the information related thereto. Representations of scenes of the plurality of images are obtained via machine learning, based on the visual semantics of the plurality of images, wherein the representations capture concepts associated with the scenes.
1. A method, implemented on a machine having at least one processor, storage, and a communication platform for responding to an image related query for an image, comprising:
receiving the image related query with an associated annotation;
establishing, based on the annotation, visual semantics of the image, wherein the visual semantics of the image comprises a hierarchy of abstraction including a first level of abstraction naming a type of the image abstracted based on what are observed in an entire scene of the image and a second level of abstraction with a category name for each group of objects observed in the image, wherein the first level of abstraction does not directly describe any object in the image; and
providing, based on the visual semantics, a response to the image related query.
2. The method of claim 1 , wherein the annotation indicates concepts co-existing in the image.
3. The method of claim 1 , further comprising:
obtaining, based on the visual semantics of the image, a representation of the scene of the image, wherein the representation conceptually summarizes the scene based on spatial relationships among concepts co-existing in the image.
4. The method of claim 3 , wherein the response is provided based on the representation of the scene of the image.
5. The method of claim 1 , wherein the hierarchy further includes a third level of abstraction on each instance of each category of each object in the image.
6. The method of claim 5 , wherein the first level is a top level of the hierarchy, the second level is an intermediate level of the of the hierarchy, and the third level is a low level of the of the hierarchy.
7. The method of claim 1 , wherein the visual semantics of the image further comprises an identifier of the image providing a context of the visual semantics.
8. A non-transitory, computer-readable medium having information recorded thereon for responding to an image related query for an image, when read by at least one processor, effectuate operations comprising:
receiving the image related query with an associated annotation;
establishing, based on the annotation, visual semantics of the image, wherein the visual semantics of the image comprises a hierarchy of abstraction including a first level of abstraction naming a type of the image abstracted based on what are observed in an entire scene of the image and a second level of abstraction with a category name for each group of objects observed in the image, wherein the first level of abstraction does not directly describe any object in the image; and
providing, based on the visual semantics, a response to the image related query.
9. The medium of claim 8 , wherein the annotation indicates concepts co-existing in the image.
10. The medium of claim 8 , wherein the operations further comprise:
obtaining, based on the visual semantics of the image, a representation of the scene of the image, wherein the representation conceptually summarizes the scene based on spatial relationships among concepts co-existing in the image.
11. The medium of claim 10 , wherein the response is provided based on the representation of the scene of the image.
12. The medium of claim 8 , wherein the hierarchy further includes a third level of abstraction on each instance of each category of each object in the image.
13. The medium of claim 12 , wherein the first level is a top level of the hierarchy, the second level is an intermediate level of the of the hierarchy, and the third level is a low level of the of the hierarchy.
14. The medium of claim 8 , wherein the visual semantics of the image further comprises an identifier of the image providing a context of the visual semantics.
15. A system for online user profiling, the system comprising:
memory storing computer program instructions; and
one or more processors that, in response to executing the computer program instructions, effectuate operations comprising:
receiving the image related query with an associated annotation;
establishing, based on the annotation, visual semantics of the image, wherein the visual semantics of the image comprises a hierarchy of abstraction including a first level of abstraction naming a type of the image abstracted based on what are observed in an entire scene of the image and a second level of abstraction with a category name for each group of objects observed in the image, wherein the first level of abstraction does not directly describe any object in the image; and
providing, based on the visual semantics, a response to the image related query.
16. The system of claim 15 , wherein the annotation indicates concepts co-existing in the image.
17. The system of claim 15 , wherein the operations further comprise:
obtaining, based on the visual semantics of the image, a representation of the scene of the image, wherein the representation conceptually summarizes the scene based on spatial relationships among concepts co-existing in the image.
18. The system of claim 17 , wherein the response is provided based on the representation of the scene of the image.
19. The system of claim 15 , wherein the hierarchy further includes a third level of abstraction on each instance of each category of each object in the image.
20. The system of claim 19 , wherein the first level is a top level of the hierarchy, the second level is an intermediate level of the of the hierarchy, and the third level is a low level of the of the hierarchy.