Deep Metric Learning using Similarities from Nonlinear Rank Approximations
This work addresses the challenge of enhancing image similarity search for computer vision applications, representing an incremental advancement in deep metric learning techniques.
The paper tackled the problem of improving retrieval quality in deep metric learning by focusing on the ranking order of feature vectors rather than their actual distances, resulting in significant improvements over existing methods on CUB-200-2011, Cars196, and Stanford Online Products datasets for all embedding sizes.
In recent years, deep metric learning has achieved promising results in learning high dimensional semantic feature embeddings where the spatial relationships of the feature vectors match the visual similarities of the images. Similarity search for images is performed by determining the vectors with the smallest distances to a query vector. However, high retrieval quality does not depend on the actual distances of the feature vectors, but rather on the ranking order of the feature vectors from similar images. In this paper, we introduce a metric learning algorithm that focuses on identifying and modifying those feature vectors that most strongly affect the retrieval quality. We compute normalized approximated ranks and convert them to similarities by applying a nonlinear transfer function. These similarities are used in a newly proposed loss function that better contracts similar and disperses dissimilar samples. Experiments demonstrate significant improvement over existing deep feature embedding methods on the CUB-200-2011, Cars196, and Stanford Online Products data sets for all embedding sizes.