SOTAMD

SOTAMD数据集是由挪威科技大学等研究机构共同创建的一个大型隔离数据集，旨在推动面部变形攻击检测技术的发展。该数据集包含来自150个不同种族、年龄和性别的受试者的面部图像，共计5748张变形面部图像。数据集中的图像经过精心预选和后处理，以移除变形过程中产生的伪影，确保图像质量符合ICAO标准，模拟真实的自动化边境控制（ABC）场景。此外，还提供了一个在线评估平台，允许研究人员上传SDK并测试其MAD算法的性能，从而促进算法的标准化和可重复性测试，以及对算法的鲁棒性进行基准测试。

The SOTAMD dataset is a large-scale isolated dataset co-created by research institutions including the Norwegian University of Science and Technology, aiming to advance the development of facial spoofing attack detection technologies. This dataset contains facial images from 150 subjects with diverse ethnicities, ages and genders, totaling 5748 altered facial images. The images in the dataset have been carefully pre-selected and post-processed to remove artifacts generated during the alteration process, ensuring that the image quality complies with ICAO standards and simulates real-world automated border control (ABC) scenarios. In addition, an online evaluation platform is provided, allowing researchers to upload SDKs and test the performance of their MAD algorithms, thereby promoting the standardization and reproducibility testing of the algorithms, as well as benchmarking the robustness of the algorithms.