| 摘要 |
Technology is disclosed for discovering décor harmonized with a décor style (“the technology”). The décor includes décor items, e.g. artworks, paintings, pictures, artifacts, architectural pieces, arrangement of artworks, color selection, room décor, rugs, mats, furnishings, household items, fashion, clothes, jewelry, car interiors, garden arrangements etc. The technology facilitates analyzing user input to identify a décor style from a décor style dictionary, obtaining décor that harmonizes with décor style, and presenting a representation of the décor to the user. The décor style dictionary includes décor styles that are generated based on an analysis of content, including images and description of décor, from a plurality of sources. The décor styles can be based on a number of concepts, including a theme of the décor, a color/color palette, a mood of the person, a fashion era, a type of architecture, etc. The technology facilitates presentation of discovered décor using computer generated imagery techniques. |
| 主权项 |
1. A method comprising:
receiving, at a computer system and from a user, user input including data indicative of a particular décor style; identifying, at the computer system and using a decor style classifier dictionary database, a décor style classifier whose signature matches with the user input indicative of the particular décor style, the décor style classifier dictionary database storing a plurality of décor style classifiers corresponding to a plurality of décor styles, the décor style classifier dictionary database created by:
analyzing a plurality of images and description indicative of the plurality of décor styles received from a plurality of sources to determine a first plurality of features of the images and a first plurality of keywords from the description that are indicative of the plurality of decor styles,categorizing the first plurality of features and the first plurality of keywords into the plurality of décor styles,generating, by the computer system, a plurality of signatures for the plurality of décor styles, wherein a signature of the signatures includes a subset of the first plurality of features and a subset of the first plurality of keywords that are indicative of the particular décor style, andassociating the signature with the décor style classifier; and determining, by the computer system and based on the signature of the décor style classifier, a subset of the plurality of images representing the plurality of decors of the particular décor style. |
