WMCA (Wide Multi Channel Presentation Attack)

Wide Multi Channel Presentation Attack (WMCA) 数据库包含来自 72 个不同身份的真实和演示攻击的 1941 个短视频记录。数据是从几个通道记录的，包括颜色、深度、红外线和热。
具有灰度、深度、红外和热的数据集的预处理版本以及仅具有 RGB 通道的数据集的预处理版本可供下载。
此外，还提供了真实记录的脉冲读数数据。
还为参考出版物中使用的部分数据提供了一些通道的预处理图像。
WMCA 数据库是在 Idiap 在“IARPA BATL”和“H2020 TESLA”项目的框架内制作的，旨在研究人脸识别系统的演示攻击检测 (PAD) 方法。

The Wide Multi Channel Presentation Attack (WMCA) database comprises 1941 short video recordings of genuine samples and presentation attacks from 72 unique individuals. Data was captured via multiple channels, including color, depth, infrared, and thermal. Preprocessed variants of the dataset with grayscale, depth, infrared, and thermal channels, as well as the preprocessed version of the dataset containing only the RGB channel, are available for download. Additionally, pulse reading data from genuine recording sessions is provided. Preprocessed images of selected channels are also offered for a subset of the data used in the reference publications. The WMCA database was developed at Idiap within the framework of both the "IARPA BATL" and "H2020 TESLA" projects, with the goal of investigating presentation attack detection (PAD) methods for facial recognition systems.

AI搜集汇总

背景与挑战

背景概述

WMCA数据集是一个用于人脸呈现攻击检测（PAD）的多通道视频数据库，包含来自72个身份的1941个短视频，涵盖真实和攻击样本，数据通过颜色、深度、红外和热等多个通道记录。该数据集由Idiap研究所在相关项目支持下制作，旨在促进人脸识别系统的安全研究，并提供预处理版本以方便使用。