Labeled Faces in the Wild-a (LFW-a)

The "Labeled Faces in the Wild-a" image collection is a database of labeled, face images intended for studying Face Recognition in unconstrained images. It contains the same images available in the original Labeled Faces in the Wild data set, however, here we provide them after alignment using a commercial face alignment software.

Some of our results, published in [1,2,3], were produced using these images. We show this alignment to improve the performance of face recognition algorithms. More information on how these images were aligned may be found in the two papers.

We have maintained the same directory structure as in the original LFW data set, and so these images can be used as direct substitutes for those in the original image set. Note, however, that the images available here are grayscale versions of the originals.

Funneled LFW-a
Funneled LFW-a
Two images aligned using the Funneling technique of Huang, Jain, and Learned-Miller (on the left) and the commercial, fiducial-points based alignment system. Misalignments on the left hand pair are visible when comparing the positions of the mouth and the eyes to the markers. These misalignments are all but removed in the right hand pair. Note: The markers shown here are not part of the alignment system, and meant only to demonstrate misalignments.

Citation: If you find these images useful and use them in your work, please follow these guidlines:

  1. Comply with any instructions specified for the original LFW data set
  2. Cite one (or all) of the papers [1,2,3] below


[1] Lior Wolf, Tal Hassner, and Yaniv Taigman, Effective Face Recognition by Combining Multiple Descriptors and Learned Background Statistics, IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), 33(10), Oct. 2011 (PDF)

[2] Lior Wolf, Tal Hassner and Yaniv Taigman, Similarity Scores based on Background SamplesAsian Conference on Computer Vision (ACCV), Xi' an, Sept 2009 (PDF)

[3] Yaniv Taigman, Lior Wolf and Tal Hassner, Multiple One-Shots for Utilizing Class Label Information, The British Machine Vision Conference (BMVC), London, Sept 2009 (project, PDF)

The LFW-a data set is availabe as one big tar.gz file (94.5 Mb)

Last update 9th of August, 2012
page maintained by Tal Hassner