One Shot Similarity Metric Learning for Action Recognition
Published in International Workshop on Similarity-Based Pattern Analysis and Recognition (SIMBAD), Venice, Italy, 2011
Recommended citation: Orit Kliper-Gross, Tal Hassner, and Lior Wolf. One Shot Similarity Metric Learning for Action Recognition. International Workshop on Similarity-Based Pattern Analysis and Recognition (SIMBAD), Venice, Italy, 2011.
Most confident OSSML results. The Same/Not-Same labels are the ground truth labels, and the Correct/Incorrect labels indicate whether the method predicted correctly. For example, the top right quadrant displays same-action pairs that were most confidently labeled as not-same.
Abstract
The One-Shot-Similarity (OSS) is a framework for classifierbased similarity functions. It is based on the use of background samples and was shown to excel in tasks ranging from face recognition to document analysis. However, we found that its performance depends on the ability to effectively learn the underlying classifiers, which in turn depends on the underlying metric. In this work we present a metric learning technique that is geared toward improved OSS performance. We test the proposed technique using the recently presented ASLAN action similarity labeling benchmark. Enhanced, state of the art performance is obtained, and the method compares favorably to leading similarity learning techniques