Auflistung nach Autor:in "Matsunami, Tomoaki"
1 - 2 von 2
Treffer pro Seite
Sortieroptionen
- KonferenzbeitragEvaluation on Biometric Accuracy Estimation Using Generalized Pareto (GP) Distribution(BIOSIG 2021 - Proceedings of the 20th International Conference of the Biometrics Special Interest Group, 2021) Yamada, Shigefumi; Matsunami, TomoakiThe accuracy of biometric authentication technology is becoming more sophisticated with its progress. For this reason, a huge number of biometric samples are required for accuracy evaluation, and the increased collection cost is an issue for biometric vendors. This work establishes a biometric accuracy estimation method using an extreme value theory to reduce the collection cost. It also explains the estimation procedure of false match rate using the generalized Pareto distribution and shows results applied to the face, gait, and voice comparison score data with an estimation effect of about 5–10 times. We investigate the criteria for the applicability of extremum statistics through application cases.
- KonferenzbeitragLearning by Environment Cluster s for Face Presentation Attack Detection(BIOSIG 2021 - Proceedings of the 20th International Conference of the Biometrics Special Interest Group, 2021) Matsunami, Tomoaki; Uchida, Hidetsugu; Abe, Narishige; Yamada, ShigefumiFace recognition has been used widely for personal authentication. However, there is a problem that it is vulnerable to a presentation attack in which a counterfeit such as a photo is presented to a camera to impersonate another person. Although various presentation attack detection methods have been proposed, these methods have not been able to sufficiently cope with the diversity of the heterogeneous environments including presentation attack instruments (PAIs) and lighting conditions. In this paper, we propose Learning by Environment Clusters (LEC) which divides training data into some clusters of similar photographic environments and trains bona-fide and attack classification models for each cluster. Experimental results using Replay-Attack, OULU-NPU, and CelebA-Spoof show the EER of the conventional method which trains one classification model from all data was 20.0%, but LEC can achieve 13.8% EER when using binarized statistical image features (BSIFs) and support vector machine used as the classification method