人脸解锁和人脸支付这两年呼声和发展越来越快，但当前仅基于彩色图的人脸识别方式已经难以满足安全性的需要，因此大家纷纷开始探索使用红外图，深度图等方式来提升活体检测和人脸识别的精度。
相关的研究必然需要依赖大量相关数据，但目前相关数据集很少，比如：红外人脸数据集，3D人脸数据集。
在这里整理了几个能找到的近红外人脸数据集和3D深度人脸数据集。
以下三个数据集都是中科院自动化所生物识别与安全技术研究中心CBSR开放的。点击子标题可以进入对应官网，查看详细信息。
197人，共3940张图片，都是近红外人脸图，大小为640x480，示例如下：
Stan Z. Li, RuFeng Chu, ShengCai Liao, Lun Zhang, "Illumination Invariant Face Recognition Using Near-infrared Images," IEEE Transactions on Pattern Analysis and Machine Intelligence(Special issue on Biometrics: Progress and Directions), Vol.29, No.4, April 2007, pp. 627-639.
免费的，如需使用，需要根据网站的要求做申请。
heterogeneous face biometrics (HFB)异质面部生物识别
所谓异质，是指不同类型的图像，主要是可见光下的彩色图/灰度图Visual (VIS)，近红外图near infrared (NIR)，热感红外图 thermal infrared (TIR)，3D深度图。
普通的人脸识别一般是用同种类型的图像做比对，而异质面部生物识别，卤煮理解就是给你一张可见光面部图像，一张近红外面部图像，依然可以比对出是否视同一张脸。
这个数据集包含 可见光图像visual (VIS), 近红外图像 near infrared (NIR) 和3D人脸图像 three-dimensional (3D) face images。
100人（57男，43女），每人4VIS（640x480）+4NIR（640x480）+2or1 3-Dfaces
This release of HFB database Ver.1 includes the following:
- The raw images, including (1) the VIS and (2) the NIR images of size 640x480 in the JPEG format, and (3) the 3D faces with wrl format;
- The processed 3D faces: Face regions not belonging to the face are discarded and then the 3D data is processed by removing high noise and filling holes using an interpolation algorithm. Finally, the 3D points are sampled to form the depth images. The sizes vary from image to image.
- The eye coordinates of the VIS, NIR and 3D depth images, manually labeled.
- Cropped versions of the raw VIS, NIR and depth images. The crop was made in two sizes, 32x32 and 128x128, and is done based on the eye coordinates.
Stan Z. Li, Zhen Lei, Meng Ao, “The HFB Face Database for Heterogeneous Face Biometrics Research”. In 6th IEEE Workshop on Object Tracking and Classification Beyond and in the Visible Spectrum (OTCBVS, in conjunction with CVPR 2009). Miami, Florida, June, 2009.
免费的，如需使用，需要根据网站的要求做申请。
这个可以说是2的加强版，the NIR-VIS 2.0 database consists of 725 subjects in total. There are 1-22 VIS and 5-50 NIR face images per subject. Figure 1 shows some face images of a subject in the database.
Figure 1. VIS and NIR face images, with variations in resolution, lighting conditions, pose and age, of one subject in the NIR-VIS 2.0 database.
The NIR-VIS 2.0 database includes the following contents:
- The raw images, including the VIS images in JPEG format and the NIR images in BMP format. Their resolutions are both 640X480.
- The eye coordinates of the VIS, NIR images. They are automatically labeled by an eye detector, and several error coordinates are corrected manually.
- Cropped versions of the raw VIS, NIR images. The resolution is 128X128, and the process is done based on the eye coordinates.
- Protocols for performance evaluation. The protocols include two views: View1 for algorithm development,View2 for performance reporting.
Stan Z. Li, Dong Yi, Zhen Lei, Shengcai Liao, “The CASIA NIR-VIS 2.0 Face Database”. In 9th IEEE Workshop on Perception Beyond the Visible Spectrum (PBVS, in conjunction with CVPR 2013). Portland, Oregon, June, 2013.
免费的，如需使用，需要根据网站的要求做申请。