In this work, we investigate biometrics applied on 2D faces in order to secure areas requiring high security level. Based on emerging deep learning methods (more precisely transfer learning) as well as two classical m...
详细信息
ISBN:
(数字)9781510621886
ISBN:
(纸本)9781510621886
In this work, we investigate biometrics applied on 2D faces in order to secure areas requiring high security level. Based on emerging deep learning methods (more precisely transfer learning) as well as two classical machine learning techniques (Support Vector Machines and Random Forest), different approaches have been used to perform person authentication. Preprocessing filtering steps of input images have been included before features extraction and selection. The goal has been to compare those in terms of processing time, storage size and authentication accuracy according to the number of input images (for the learning task) and preprocessing tasks. We focus on data-related aspects to store biometric informations on a low storage capacity remote card (10Ko), not only in a high security context but also in terms of privacy control. The proposed solutions guarantee users the control of their own biometrics data. The study highlights the impact of preprocessing to perform real-time computation, preserving a relevant accuracy while reducing the amount of biometric data. Considering application constraints, this study concludes with a discussion dealing with the tradeoff of the available resources aginst the required performances to determine the most appropriate method.
Designing embedded systems is a challenging task during which wrong choices can lead to extremely costly re-design loops, especially when these wrong choices are made during the algorithm specification and the mapping...
详细信息
Designing embedded systems is a challenging task during which wrong choices can lead to extremely costly re-design loops, especially when these wrong choices are made during the algorithm specification and the mapping...
详细信息
Designing embedded systems is a challenging task during which wrong choices can lead to extremely costly re-design loops, especially when these wrong choices are made during the algorithm specification and the mapping over the selected architecture. In this paper we propose a high-level approach for design space exploration, using a usual standard language as input. More precisely we present the two first steps of the Design Trotter framework: (i) the specification step and its underlying internal model (HCDFG: Hierarchical and Control Data Flow Graph) and (ii) the characterization step which takes place very early in the design flow. Indeed, once transformed into our internal representation, the specification is rapidly and automatically characterized and explored at the algorithmic level. The framework provides the designer with metrics so that he can evaluate, very early in the design process, the impact of algorithmic choices on resource requirements in terms of processing, control, memory bandwidth and potential parallelism at different levels of granularity. The overall aim of our approach is to improve the algorithm/architecture matching that sorely influences the implementation efficiency in terms of silicon area, performances and energy consumption. We give examples which illustrate how designers can refer to the outcomes of the Design Trotter framework in order to select or build suitable architectures for specific applications.
暂无评论