SIDTD

SIDTD数据集是由计算机视觉中心创建的合成ID和旅行文档数据集，旨在辅助训练和评估伪造ID文档检测系统。该数据集包含1222个伪造模板和191个视频，涵盖了十种欧洲国籍的文档。创建过程中使用了Crop & Replace和Inpainting技术生成伪造文档，确保了数据集的高内部变异性。SIDTD数据集主要应用于数字环境中的公民身份验证，特别是用于检测和防止通过伪造ID文档进行的欺诈行为。

The SIDTD dataset is a synthetic ID and travel document dataset created by the Computer Vision Center, aiming to assist in training and evaluating forged ID document detection systems. This dataset contains 1222 forgery templates and 191 videos, covering documents from ten European nationalities. During its creation, Crop & Replace and Inpainting technologies were employed to generate forged documents, ensuring high internal variability of the dataset. The SIDTD dataset is primarily applied to citizen authentication in digital environments, particularly for detecting and preventing fraudulent activities via forged ID documents.

搜集汇总

背景与挑战

背景概述

SIDTD是一个合成身份文档数据集，基于MIDV2020数据集生成，用于训练模型区分真实和伪造的身份文档。数据集结构清晰，支持多种模型验证技术，并已用五种先进深度学习模型评估，展示了在不同任务中的性能表现。