HRT Transgender Dataset

The Transgender dataset is the first known dataset that includes images of a subject taken during the period of gender transformation. The images are collected from Youtube videos that are compilations of images taken during various time periods of the gender transformation. In one case it was a picture a day for three years and in others it was a picture a week or a random sampling over a year or more. The transgender dataset includes videos of 38 subjects whose images span at least a year of their life showing the gender transformation variations on their face. The dataset includes an average of 278 images per subject that are taken under real-world conditions, and hence, include variations in pose, illumination, expression, and occlusion.

Transgender dataset is freely distributed and can be obtained by sending a request to Dr. Karl Ricanek Jr. (ricanekk@uncw.edu) quoting the purpose of use along with the signed license agreement form provided below.

HRT-Transgender Database Release Agreement

Any work that uses or incorporates the database must acknowledge the authors by including the following reference:

Gayathri Mahalingam and Karl Ricanek Jr., Is the Eye Region More Reliable Than the Face? A Preliminary Study of Face-based Recognition on a Transgender Dataset, In Proc. of IEEE Intl. Conf. on Biometrics: Theory, Applications, and Systems, 2013.

Gayathri Mahalingam, Karl Ricanek Jr., and Midori M. Albert, “investigating the Periocular-Based Face Recognition Across Gender Transformation”, IEEE Trans. on Information Forensics and Security, vol. 9, no. 12, pp. 2180-2192, 2014.

Comments are closed.