Automatic Prediction of Human Attractiveness

We apply computer vision methods to the task of automatically predicting human attractiveness from frontal face images. A dataset of thousands of images and corresponding scores was obtained from a popular website that asks viewers to rate the attractiveness of the people appearing in the images. Using a combination of radial basis functions and specialized feature detectors, we achieved moderate success in estimating female attractiveness. However, male attractiveness proved more difficult to predict. We believe significant improvements are possible through future development of new feature types.

Ryan White, Ashley Eden, Michael Maire "Automatic Prediction of Human Attractiveness", CS 280 class report, December 2003.

Ryan White "Statistical Learning of Human Attractiveness", CS281B class report, May 2004.                    


If you use this data in a research publication, please cite our CS280 class project. Data is in MATLAB format. download here [66 MB]

Back to research page