Facebook to open source AI image recognition software to public.
Facebook will be opening up its image recognition artificial intelligence research to the public.
The Menlo Park, Calif.-based tech behemoth is releasing years’ worth of research and code on “machine vision,” or computer image recognition and understanding, the company announced on Thursday.
The goal is to help increasingly advance the field of machine vision as the social network expands on users’ interest in sharing and interacting with all photos and videos. The company will continue to publish its latest results and update open source tools made available.
Facebook hopes to use AI to eventually search for specific images without an explicit tag on each photo, and enable those with vision loss to understand an image from a system describing it to them.
Like other big tech companies, particularly Google, Amazon, Microsoft and Apple, Facebook is racing to have an edge in artificial intelligence, seen as key to making its ecosystem of services, which include Facebook Messenger, Instagram and WhatsApp, more intertwined with its users’ daily lives.
Founder Mark Zuckerberg’s 10-year vision for the company relies on major technological breakthroughs in artificial intelligence, augmented and virtual reality, and global internet connectivity.
It’s also been working on ways to integrate AI to make using the service easier for those with disabilities, such as vision impairment.
Image recognition is notoriously tricky, and the nascent field has experienced some bad stumbles. Last year, Google apologized after its new Photos application, which used machine learning to identify people and places on its own, identified black people as gorillas.
Facebook is no stranger to open source, or releasing software online to be redistributed or modified. On its website, the company has a variety of open sourced software for its Android and iOS apps, web service and the backend infrastructure.