Crowdshaping realistic

1. A computer-implemented method for indexing a database of human body shapes, wherein the human body shapes are represented in terms of coefficient vectors in a geometric body space, the method comprising:

predicting, for each human body shape in the database, a vector of word ratings, based on parameters of the human body shape, using a mapping between the geometric body space and a linguistic body space, wherein the linguistic body space is represented in terms of body descriptor words; and

storing the body descriptor words and their predicted rating values in the database with each human body shape.

2. The method according to claim 1 , wherein the mapping was learned based on empirical user ratings.

3. The method of claim 2 , wherein the user ratings indicate a degree to which a given body descriptor applies to a given body shape.

4. The method of claim 3 , wherein the degree is selected from a given Likert scale.

5. The method of claim 4 , wherein the mapping is linear.

6. The method of claim 5 , wherein the mapping has been learned using principal component analyses (PCA).

7. A method for querying a database of human body shapes, wherein the human body shapes are represented in terms of coefficient vectors in a geometric body space, the method comprising:

receiving one or more body descriptor words related to body shape;

matching the one or more received body descriptor words against body descriptor words and their predicted rating values stored in an index of the database with each human body shape;

retrieving, in case of a match, from the database, the coefficient vectors representing one or more body shapes, based on the index; and

outputting body shapes, based on the retrieved coefficient vectors.

8. The method of claim 7 , wherein the database index of descriptor words and body shapes has been generated by:

predicting, for each human body shape in the database, a vector of word ratings, based on parameters of the human body shape, using a mapping between the geometric body space and a linguistic body space, wherein the linguistic body space is represented in terms of body descriptor words; and

storing the body descriptor words and their predicted rating values in the database with each human body shape.

9. A device for querying a database of human body shapes, wherein the human body shapes are represented in terms of coefficient vectors in a geometric body space, the device comprising:

a receiving unit, for receiving one or more body descriptor words related to body shape;

a matching unit, for matching the one or more received body descriptor words against body descriptor words and their predicted rating values stored in an index of the database with each human body shape; and

a retrieving unit, for retrieving, in case of a match, from the database, the coefficient vectors representing one or more body shapes, based on the index; and

an outputting unit, for outputting body shapes, based on the retrieved coefficient vectors.

10. The method according to claim 8 , wherein the mapping was learned based on empirical user ratings.

11. The method of claim 10 , wherein the user ratings indicate a degree to which a given body descriptor applies to a given body shape.

12. The method of claim 11 , wherein the degree is selected from a given Likert scale.

13. The method of claim 12 , wherein the mapping is linear.

14. The method of claim 13 , wherein the mapping has been learned using principal component analyses (PCA).