Someone using Instagram.
Lorenzo Di Cola | NurPhoto through Getty Images
Pugs, Ferraris, mountains, brunches, beaches, and babies – Instagram is full of them. In fact, it has grown into one of the largest image databases on the planet over the past decade and the company’s owner, Facebook, is using this funding to teach tools what which is in a picture.
Facebook announced on Thursday that it has launched a fake information program that sees “what it looks like.” It did this by feeding over 1 billion public images from Instagram.
The “computer vision” program, nicknamed SEER, outperformed the AI models in the substance recognition test, Facebook said.
He achieved a “classification accuracy score” of 84.2% when he attempted a test provided by ImageNet, which is a large visual database designed for use in visual content recognition software research. Basically, it tests whether an AI program can recognize what is in a picture.
A new approach
While many AI models are trained on datasets with careful labels, Facebook said SEER has learned how to recognize objects in photos by analyzing random, unlabeled and Instagram images. non-submerged. This AI method is called self-directed learning (SEER is a SElf-supERvised play).
“The future of AI lies in creating systems that can learn directly from whatever information is given to them – whether it be text, images, or other types of data – without rely on carefully stored and named data sets to learn how to identify objects in a picture, interpret a text block, or perform any of the countless other tasks that require us on, “Facebook researchers wrote in a blog post.
“SEER’s performance shows that self-directed learning can be excellent on computer vision tasks in real-world situations,” they said. “This is a breakdown that ultimately clears the way for more flexible, accurate and flexible computer vision models in the future. ”
While this is just a research project, a Facebook spokesperson said the potential uses were quite widespread. They include enhanced text that has been automatically generated for reporting images to visually impaired people, improved automated classification of items sold on Facebook Marketplace, and better systems to keep harmful images away from the Facebook platform, the company said.
Privacy issue?
But many Instagram users may be surprised to hear that their images are being used to train Facebook AI systems.
“We inform Instagram account holders in our data policy that we will use the information we hold to support research and innovation including the advancement of such technology,” Priya Goyal, software engineer at Facebook AI Research, told CNBC.
Facebook said it will open some of its software so other researchers can try it out.
“While we are sharing details of our research and creating an open library that will allow other researchers to use self-directed learning to train models on submerged images, we are not share the images or SEER mode, ”Goyal said.
Other major tech companies including Google and Microsoft are also trying to push the boundaries of computer vision. Last summer, Google unveiled the SimCLRv2 computer vision module, while OpenAI unveiled iGPT 2.