Facebook is sharing its computer vision artificial intelligence technology with everyone.
While the human eye can identify objects down to the pixel, machine vision is still learning to recognize content. So in hopes of further advancing AI, the social network has released three sets of code—DeepMask (detection), SharpMask (refinement), and MultiPathNet (identification)—on Github, along with related research and demos.
"We've witnessed massive advances in image classification," Facebook research scientist Piotr Dollar wrote in a blog post. "But this is just the beginning of understanding the most relevant visual content of any image or video."
Like differentiating, for instance, between a zebra, giraffe, and human at the zoo.
"As we continue improving these core technologies we'll continue publishing our latest results and updating the open source tools we make available to the community," Dollar said.
The Facebook AI Research (FAIR) team in November achieved "new milestones" in long-term AI research—including a "state-of-the-art system" that distinguishes between objects in a photo 30 percent faster, and using 10 times less training data, than previous industry benchmarks.
But the technology, which will eventually offer visually impaired users rich descriptions of News Feed photos and videos, is still in its infancy. By open-sourcing the project on GitHub, FAIR hopes the community will help solve any problems with the algorithm.
In June, Facebook teased its DeepText AI, which can comprehend "with near-human accuracy" the textual content of several thousand posts per second, in more than 20 languages.