A customer uses Machine Box to wrangle profile photos
I wanted to share this great use case a customer of ours is solving using Facebox from Machine Box.
Imagine you have millions of profile photos from people’s social media accounts and you want to pick the one that best represents who they are. This particular startup wants to make sure that the photos they use to represent people in their network are high quality.
So they use Facebox to scan every photo and make sure that;
A) The image has a picture of a human face in it (no more dog photos guys).
B) There is only one face detected, so group photos get thrown out.
Fortunately, Facebox has an API that allows you to disable features such as recognizing people when you make the request to keep the process as lean as possible. This allows the customer to keep compute costs really low, but also process a large amount of images fast enough for their needs. Thanks to Docker, Facebox can scale as wide as it needs to be, as the amount of requests start to increase.
A great use case that uses Facebox in its simplest form; detecting faces.
