African 25
Caucasian 25
All are one-to-one correspondence. African/Asian/Caucasian are 1:1:1.
For each ID, there are 4 (1 reality +3 attack) X 8（8 capturing devices）X 26 (26 types of lighting conditions) = 832 videos.
For this dataset, there are 75 subjects X 832 videos/subject = 62,400 videos.
Surfing Tech has a variety of anti-fraud datasets:
6000 attack dataset, 3000 ID Screen arrack dataset, 100 ID 3D mask arrack dataset,
Anti-spoofing dataset, Hi-Fi Challenge dataset, and 3D mask attack dataset.
Some of these datasets were used in the 2019 CVPR challenge,
while others were used in the 2020 CVPR challenge, and some in the 2021 ICCV challenge
Our dataset boasts a wide array of multiracial facial images and videos, captured under different lighting conditions
It encompasses multiple expressions, postures, and facial occlusion data.
Furthermore, we ensure the quality of our data by providing high-resolution imagery, even under different lighting conditions.
Diversity is at the core of our dataset. It features individuals from various age groups and maintains a balanced gender ratio. More importantly, our dataset is globally inclusive, capturing the unique facial features of various ethnicities, including Africans, Asians, Caucasians, and Southeast Asians, ensuring a comprehensive representation for your AI applications.
