RETRIEVAL SYSTEM AND METHOD LEVERAGING CATEGORY-LEVEL LABELS
First Claim
Patent Images
1. A retrieval method comprising:
- providing a projection for embedding an original image representation in an embedding space, the original image representation being based on features extracted from the image, the projection having been learned from category-labeled training data to optimize a classification rate on the training data; and
with a processor, for each of plurality of database images, computing a comparison measure between a query image and the database image, the comparison measure being computed in the embedding space, respective original image representations of the query image and the database image being embedded in the embedding space with the projection; and
providing for retrieving at least one of the database images based on the comparison.
1 Assignment
0 Petitions
Accused Products
Abstract
An instance-level retrieval method and system are provided. A representation of a query image is embedded in a multi-dimensional space using a learned projection. The projection is learned using category-labeled training data to optimize a classification rate on the training data. The joint learning of the projection and the classifiers improves the computation of similarity/distance between images by embedding them in a subspace where the similarity computation outputs more accurate results. An input query image can thus be used to retrieve similar instances in a database by computing the comparison measure in the embedding space.
91 Citations
28 Claims
-
1. A retrieval method comprising:
-
providing a projection for embedding an original image representation in an embedding space, the original image representation being based on features extracted from the image, the projection having been learned from category-labeled training data to optimize a classification rate on the training data; and with a processor, for each of plurality of database images, computing a comparison measure between a query image and the database image, the comparison measure being computed in the embedding space, respective original image representations of the query image and the database image being embedded in the embedding space with the projection; and providing for retrieving at least one of the database images based on the comparison. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21)
-
-
22. A retrieval system comprising:
-
memory which stores; a projection matrix for embedding image features in an embedding space, the projection matrix having been learned from category-labeled training data to optimize a classification rate on the training data; and instructions for computing a comparison between a query image and a database image whose respective features are embedded in the embedding space with the projection matrix; and a processor in communication with the memory which implements the instructions. - View Dependent Claims (23, 24, 25)
-
-
26. A method of generating a retrieval system, comprising:
-
providing a feature-based representation and a category label for each of a set of training images, each of the category labels corresponding to a respective one of a set of categories; jointly learning a projection and set of classifiers based on the feature-based representations and category labels, the learning optimizing a classification of the training images by the set of classifiers in an embedding space into which the feature-based representations are embedded with the projection, the set of classifiers including a classifier for each of the categories; and storing the projection for embedding a query image and database images into the embedding space. - View Dependent Claims (27, 28)
-
Specification