Main Citation: Bainbridge, W.A., Isola, P., & Oliva, A. (2013). The intrinsic memorability of face images. Journal of Experimental Psychology: General. Journal of Experimental Psychology: General, 142(4), 1323-1334. .
(Related citation: Khosla, A., Bainbridge, W.A., Torralba, A., & Oliva, A. (2013). Modifying the memorability of face photographs. Proceedings of the International Conference on Computer Vision (ICCV), Sydney, Australia. . Please cite in addition if you use the annotations.)
This database contains 10,168 natural face photographs and several measures for 2,222 of the faces, including memorability scores, computer vision and psychology attributes, and landmark point annotations. The face photographs are JPEGs with 72 pixels/in resolution and 256-pixel height. The attribute data are stored in either MATLAB or Excel files. Landmark annotations are stored in TXT files. Any parts of the database may be used upon citation of the article and acceptance of the license agreement. To obtain the database, fill out the following form to get access information.
*New*: (7/15/15) The database now has manual ground-truth annotations of 77 different landmark points on each of the 2,222 target faces. This is particularly useful for face recognition, manipulation, and active appearance modeling.
*New*: (7/15/15) The psychology attributes now include participant information, so you can now easily study subject-centric (versus item-centric) face and memory effects.
*New*: (4/18/14) The database now additionally includes a software tool that allows you to export custom image sets from the database for your own research, based on our collected attributes and memorability information. (Ex: You can now easily create a stimulus set of faces based on memorability, gender, race, emotion, attractiveness, etc).
How to download:
In order to ensure that these data are used solely for research purposes, we have password protected the download. If you already have the password, click here to access the database. If you do not, fill out this form and we will send you a personalized password. (Note: if this form opens your mail client, just send the e-mail that is created with the relevant information). The password will bring you to a license agreement, and upon agreement you will have access to the database.
This work was funded by National Science Foundation Grant 1016862 and by Google and Xerox Research Awards to Aude Oliva.