An Efficient Face Photo Clustering From User Feedback through Query Generation

Computer Engineering

In the area of the big data, when a large no. of images has been stored, it is the very difficult procedure to every image with every potential tag. By presenting the greatness of information, its uninterrupted cause, and the bit of one to one face that perform, again and again, it is impossible to tag a large for the total grouping of faces. Automatic face clustering is the key component, which aims to group faces referring to the same people together, is a key component for the face tagging and image management. To improve the clustering recall without reducing the high precision, we use the heterogeneous context information to iteratively merge the clusters referring the same entities. A data-driven Gaussian procedure model of the visual aspect of faces where Queries are refined on a probabilistic database to return the conditional answers of user’s questions. The query-driven active learning scheme will select questions to return to users for feedback, which will be computed to modify the query answers. The experiments show the excellent effectiveness of the system of many real world face realization tasks.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.