Hierarchical ranking of facial attributes

A system, comprising:

• a processing unit, computer readable memory and a computer readable storage medium device;
• first program instructions to, in response to a query comprising a plurality of specified facial attributes that are each discernible from face image data, estimate the locations of a plurality of facial regions that are each relevant to different ones of the specified facial attributes, and wherein the plurality of facial regions are distinct and different from each other;
• second program instructions to extract a plurality of different features from the estimated facial regions of each of a plurality of database facial images;
• third program instructions to:
• rank the database facial images in each of a plurality of base layer rankings by matching feature vectors in a bipartite graph to a base layer ranking sequence as a function of edge weights parameterized by an associated base layer parameter vector, wherein the feature vectors are each a relevance of the features extracted from one of the images to one of the attributes relevant to one of the estimated regions, wherein each of the plurality of base layer region-feature rankings are for different ones of the plurality of extracted features and the plurality of attributes;
• and
• rank the database facial images in each of a plurality of second-layer rankings by defining each of a plurality of multi-dimensional second-layer attribute vectors as bilinear combinations of one of the feature vectors and its associated base layer parameter vector from each of the base-layer rankings for a common one of the attributes, and matching the multi-dimensional second-layer attribute vectors in a bipartite graph to a second-layer ranking sequence as a function of edge weights parameterized by an associated second-layer parameter vector;
• and
• fourth program instructions to rank the images for relevance to the query as a function of fusing the second-layer rankings;
• and
• wherein the first, second, third and fourth program instructions are stored on the computer readable storage medium device for execution by the processing unit via the computer readable memory.

An article of manufacture, comprising:

• a computer readable storage medium device having computer readable program code embodied therewith, the computer readable program code comprising instructions that, when executed by a computer processor, cause the computer processor to:
• in response to a query comprising a plurality of specified facial attributes that are each discernible from face image data, estimate the locations of a plurality of facial regions that are each relevant to different ones of the specified facial attributes, and wherein the plurality of facial regions are distinct and different from each other;
• extract a plurality of different features from the estimated facial regions of each of a plurality of database facial images;
• rank the database facial images in each of a plurality of base layer rankings by matching feature vectors in a bipartite graph to a base layer ranking sequence as a function of edge weights parameterized by an associated base layer parameter vector, wherein the feature vectors are each a relevance of the features extracted from one of the images to one of the attributes relevant to one of the estimated regions, wherein each of the plurality of base layer region-feature rankings are for different ones of the plurality extracted features and the plurality of attributes;
• rank the database facial images in each of a plurality of second-layer rankings by defining each of a plurality of multi-dimensional second-layer attribute vectors as bilinear combinations of one of the feature vectors and its associated base layer parameter vector from each of the base-layer rankings for a common one of the attributes, and matching the multi-dimensional second-layer attribute vectors in a bipartite graph to a second-layer ranking sequence as a function of edge weights parameterized by an associated second-layer parameter vector;
• and
• rank the images for relevance to the query as a function of fusing the second-layer rankings.

A method for providing a service for ranking images with respect to query attributes, the method comprising:

• providing a computer infrastructure that:
• in response to a query comprising a plurality of specified facial attributes that are each discernible from face image data, estimates the locations of a plurality of facial regions that are each relevant to different ones of the specified facial attributes, and wherein the plurality of facial regions are distinct and different from each other;
• extracts a plurality of different features from the estimated facial regions of each of a plurality of database facial images;
• ranks the database facial images in each of a plurality of base layer rankings by matching feature vectors in a bipartite graph to a base layer ranking sequence as a function of edge weights parameterized by an associated base layer parameter vector, wherein the feature vectors are each a relevance of the features extracted from one of the images to one of the attributes relevant to one of the estimated regions, wherein each of the plurality of base layer region-feature rankings are for different ones of the plurality extracted features and the plurality of attributes;
• ranks the database facial images in each of a plurality of second-layer rankings by defining each of a plurality of multi-dimensional second-layer attribute vectors as bilinear combinations of one of the feature vectors and its associated base layer parameter vector from each of the base-layer rankings for a common one of the attributes, and matching the multi-dimensional second-layer attribute vectors in a bipartite graph to a second-layer ranking sequence as a function of edge weights parameterized by an associated second-layer parameter vector;
• and
• ranks the images for relevance to the query as a function of fusing the second-layer rankings.