• Journal of the Semiconductor & Display Technology
• /
• v.20 no.3
• /
• pp.120-124
• /
• 2021

Recently, human attribute recognition (HAR) attracts a lot of attention due to its wide application in video surveillance systems. Recent deep-learning-based solutions for HAR require time-consuming training processes. In this paper, we propose a post-processing technique that utilizes the historical video frames to improve prediction results without invoking re-training or modifying existing deep-learning-based classifiers. Experiment results on a large-scale benchmark dataset show the effectiveness of our proposed method.

• ETRI Journal
• /
• v.32 no.6
• /
• pp.950-960
• /
• 2010

In response to increased security concerns, biometrics is becoming more focused on overcoming or complementing conventional knowledge and possession-based authentication. However, biometric authentication requires special care since the loss of biometric data is irrecoverable. In this paper, we present a biometric authentication framework, where several novel techniques are applied to provide security and privacy. First, a biometric template is saved in a transformed form. This makes it possible for a template to be canceled upon its loss while the original biometric information is not revealed. Second, when a user is registered with a server, a biometric template is stored in a special form, named a 'soft vault'. This technique prevents impersonation attacks even if data in a server is disclosed to an attacker. Finally, a one-time template technique is applied in order to prevent replay attacks against templates transmitted over networks. In addition, the whole scheme keeps decision equivalence with conventional face authentication, and thus it does not decrease biometric recognition performance. As a result, the proposed techniques construct a secure face authentication framework in open networks.