SynMorph
收藏资源简介:
SynMorph是由挪威科技大学和达姆施塔特应用科学大学联合创建的高质量合成面部变形数据集,旨在解决面部识别系统在面对变形攻击时的脆弱性问题。该数据集包含2450个身份,总计超过50万张图像,包括10万张变形图像和非变形图像,分辨率为1024×1024。数据集通过StyleGAN2模型生成,并经过严格的质量筛选和属性编辑,以确保图像的真实性和多样性。SynMorph数据集不仅支持单图像变形攻击检测(S-MAD),还支持差分图像变形攻击检测(D-MAD),适用于开发和测试面部变形攻击检测算法,旨在提高面部识别系统的安全性和鲁棒性。
SynMorph is a high-quality synthetic facial morphing dataset jointly created by the Norwegian University of Science and Technology and Darmstadt University of Applied Sciences, aiming to address the vulnerability of facial recognition systems against morphing attacks. This dataset includes 2450 identities, with a total of over 500,000 images, among which there are 100,000 morphing and non-morphing images, with a resolution of 1024×1024. Generated via the StyleGAN2 model, the dataset has undergone rigorous quality filtering and attribute editing to ensure the authenticity and diversity of the images. The SynMorph dataset supports both single-image morphing attack detection (S-MAD) and differential image morphing attack detection (D-MAD), and is applicable to the development and testing of facial morphing attack detection algorithms, with the goal of improving the security and robustness of facial recognition systems.

- 1SynMorph: Generating Synthetic Face Morphing Dataset with Mated Samples挪威科技大学 (NTNU), 挪威 ‡达姆施塔特应用科学大学 (HDA), 德国 · 2024年



