Our workforce has labeled nearly fifty thousand images of faces, classifying them by age, gender, hair color, beard and mustache color (if present), and glasses.
Each image has been independently labeled by two of our labelers, and then manually reviewed by one of our expert reviewers. Our labelers combined classified a total of 46064 faces, 20662 of which were men, and 24890 women. The images labeled are those in the FFHQ Dataset.
Among the images, 2003 of them had one additional face in the background (which is not the object of the classification).
The download folder contains the annotations themselves, plus a Python script to download the assets to your machine. Included is also a short README on how to use the script.
You are free to use the annotations in your project, with the only condition that you must mention Ango AI and link to our website somewhere in the project description or acknowledgements.