Facebook users upload a lot of photos to the site, and Facebook has a novel way to help visually impared users still get some impact from these images. Facebook uses an automated computer process to provide "tags" (short descriptions) that can be read aloud by some screen-reading software.
The computer vision (CV) platform built as part of Facebook's applied machine learning efforts makes it possible to leverage all that rich content to create better experiences for everybody on Facebook. The platform provides a visual recognition engine whose role is to "see" inside images and videos to understand what they depict. For example, the engine would know if an image contains a cat, features skiing, was taken at the beach, includes the Eiffel Tower, and so on.
|click to enlarge|