• Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.

Discussion Real Tone from Google Pixel, introducing equity into smartphone photography algorithms.

VHS

Be kind, rewind.
Banned
Pronouns
He/Him/His
Event had some usual cool tech stuff as we expect, but I wanted to highlight this new camera inclusivity that Google is bringing to the tech space with the Pixel 6 series.



Video goes over the issues, but tech in general favors people with light and especially white skin over other shades of skin tone. This was found for example by Twitter in their photo cropping algorithm where it would consistently favor the white person over a black person, and a woman over a man.

This isn't just Twitter though this issue is rampant across tech, I'm sure we are all familiar with the first facial recognition laptops not recognizing black people as human for example.


Often this comes from an unconscious bias due to the lack of non white people working in tech, and I really applaud Google leveraging their Machine learning and just overall platform to address this. The video goes into it but big changes were made to their photo data sets to represent more than just light skinned people, and to fix autoexposure settings which would lead to washed out, ashy or even green skintone for black people as well as many other changes.

Hoping these changes will kickstart the rest of the techspace into a more inclusive space. In the video go to around 2:50 for some before and after shots, if google updates to add a page I will update it here as well.
 
Very cool! Glad to see tech companies reexamining this kind of stuff. Hopefully this leads to a better understanding of how algorithms should be used, that they are only as good as the people who design them and are not objective perfect black boxes.
 
Very cool! Glad to see tech companies reexamining this kind of stuff. Hopefully this leads to a better understanding of how algorithms should be used, that they are only as good as the people who design them and are not objective perfect black boxes.
Agreed completely, also welcome to this site from the other site!
 


Back
Top Bottom