The Hong Kong Polytechnic University Disguise and Makeup Faces Database

The accuracy of automated human face recognition algorithms can significantly degrade while recognizing same subjects under make-up and disguised appearances. Increasing constraints on enhanced security and surveillance requires enhanced accuracy from face recognition algorithms for faces under disguise and/or makeup.  

Brief Description

This database consists of 2460 different face images from 410 different subjects. These images have been acquired under real environment which primarily focuses on make-up and disguises covariates. This database also provides ground truth (eyeglass, goggle, mustache, and beard) to evaluate capability of advanced algorithms in automatically oredicting them for advanced surveillance capabilities. The origin or the source of these images is detailed in a seperate text file which is available along with ground truth and detected/segmented face images.

The people in the database are acquired from the publicly accessible websites. Majority of them are celebrities from the movie, music, sports and also politics. Celebrities from the Europe and US are the majority, along with some from Asia, Africa, Pacific and Middle East.

Each person in the database has six images, the first image is selected to represent clean impression, i.e., without any makeup and disguises, and is from the frontal pose. The rest of the five face images are faces with different levels of makeup or disguise creating accessories. The naming convention of the subjects in the database is detailed in the enclosed readme.txt file and all the images are in JPG format.

Sample Images

Description: F:\FaceAll_cropped - Copy\sub0543.jpg

Description: F:\FaceAll_cropped - Copy\sub0545.jpg

Description: F:\FaceAll_cropped - Copy\sub0544.jpg

Description: F:\FaceAll_cropped - Copy\sub0425.jpg

Description: F:\FaceAll_cropped - Copy\sub0423.jpg

Description: F:\FaceAll_cropped - Copy\sub0422.jpg

Description: F:\FaceAll_cropped - Copy\sub0204.jpg

Description: F:\FaceAll_cropped - Copy\sub0202.jpg

Description: F:\FaceAll_cropped - Copy\sub0205.jpg

Description: F:\FaceAll_cropped - Copy\sub4035.jpg

Description: F:\FaceAll_cropped - Copy\sub4036.jpg

Description: F:\FaceAll_cropped - Copy\sub4031.jpg

*Ground Truth included: Gender, Ethnicity, Skin Color, Hat, Hair Style, Glasses and Beard    

Download and Copyright

The database is being made available for the researchers from 2016 onwards. Interested researchers can download this database in this link.  Any commercial use/distribution of this database is prohibited. All the technical reports and papers that report experimental results from this database should provide due acknowledgement and reference to [1]. 

 

References

 

[1] Tsung Ying Wang and Ajay Kumar, "Recognizing Human Faces under Disguise and MakeupProc. ISBA 2016, pp. 1-7, Sendai, Japan, March Dec. 2016.  

Back to Databases Page