Meta-Recognition: The Theory and Practice of Recognition Score Analysis
Walter J. Scheirer, Anderson Rocha, Ross J. Micheals, and Terrance E. Boult
[IEEE T.PAMI, V. 33, Issue 8, August 2011, pages 1689 – 1695]
Abstract. In this paper, we define meta-recognition, a perfor- mance prediction method for recognition algorithms, and exam- ine the theoretical basis for its post-recognition score analysis form through the use of the statistical extreme value theory (EVT). The ability to predict the performance of a recognition system based on its outputs for each match instance is desirable for a number of important reasons, including automatic thresh- old selection for determining matches and non-matches, and automatic algorithm selection or weighting for multi-algorithm fusion. The emerging body of literature on post-recognition score analysis has been largely constrained to biometrics, where the analysis has been shown to successfully complement or replace image quality metrics as a predictor. We develop a new statistical predictor based upon the Weibull distribution, which produces accurate results on a per instance recognition basis across different recognition problems. Experimental results are provided for two different face recognition algorithms, a fingerprint recognition algorithm, a SIFT-based object recognition system, and a content-based image retrieval system.
Index Terms. Meta-Recognition, Performance Modeling, Multi-Algorithm Fusion, Object Recognition, Face Recognition, Fingerprint Recognition, Content-Based Image Retrieval, Similarity Scores, Extreme Value Theory
Click here to download our implementation for improving recognition systems using our proposed Extreme Value Theory Meta-Recognition approach.
Alg. 1 on page 5: Line 3 should read if Inv(δ; W) < s1 then