IBM hopes 1 million faces will help fight bias in facial recognition
IBM thinks the data being used to train facial recognition systems isn't diverse enough.
The tech giant released a trove of data containing 1 million images of faces taken from a Flickr dataset with 100 million photos and videos.
The images are annotated with tags related to features including craniofacial measurements, facial symmetry, age and gender.
Researchers at the company hope that these specific details will help developers train their artificial intelligence-powered facial recognition systems to identify faces more fairly and accurately.
And then the police adopted the new facial recognition algorithms and everyone lived happily ever after.
IBM blog post. Also at TechCrunch and VentureBeat.
(Score: 3, Insightful) by bradley13 on Thursday January 31 2019, @07:33PM
Without addressing the specifics you are referring to, let me just point out: Something is not "bias" if its true. Too many people - on all sides of the political spectrum - want reality to conform to their preconceptions. You could create the most perfect training set, and all that will accomplish is for everyone to declare it biased.
It's not clear from that very brief article, but it would be interesting to know what metadata is being included. They do say "tags related to features including craniofacial measurements, facial symmetry, age and gender" - but that is clearly not the complete list. Regardless, having a wide range of faces will be hugely helpful for many applications.
Note that this won't help with the "gorilla" problem, because they're only working within the range of faces - they are not training to determine if something is a face.
Everyone is somebody else's weirdo.