قالب وردپرس درنا توس
Home / Technology / Facebook trains AI to ‘see’ using 1 billion public Instagram photos

Facebook trains AI to ‘see’ using 1 billion public Instagram photos



A person who uses Instagram.

Lorenzo Di Cola | NurPhoto via Getty Images

Pugs, ferraris, mountains, brunch, beaches and babies – Instagram is full of them. In fact, it has become one of the largest image databases on the planet in the last decade, and the company̵

7;s owner, Facebook, uses this treasure to teach machines what is in an image.

Facebook announced on Thursday that they had built an artificial intelligence program that can “see” what it is looking at. It did so by feeding it over 1 billion public photos from Instagram.

The “computer vision” program, nicknamed SEER, surpassed existing AI models in an object recognition test, Facebook said.

It achieved a “classification accuracy score” of 84.2% when it tested a test provided by ImageNet, a large visual database designed for use in research on visual object recognition software. Basically, it tests whether an AI program can identify what is in an image.

New approach

While many AI models are trained on carefully marked datasets, Facebook said that SEER learned to identify objects in photos by analyzing random, unmarked and uncurated Instagram photos. This AI technique is known as self-monitored learning (SEER is a game on SElf-supERvised).

“The future of AI lies in creating systems that can learn directly from the information they receive – whether it is text, images or other types of data – without relying on carefully curated and marked datasets to teach them to recognize objects in an image, interpret a block of text or perform some of the countless other tasks we ask it to do, “Facebook researchers wrote in a blog post.

“SEER’s performance demonstrates that self-monitored learning can excel at real-time computer vision tasks,” they added. “This is a breakthrough that will eventually pave the way for more flexible, accurate and customizable computer vision models in the future.”

Although this is only a research project, a spokesman for Facebook said that the potential use was relatively broad. They include improved auto-generated text to describe images to visually impaired people, better automatic categorization of items sold on the Facebook Marketplace, and better systems to keep malicious images away from the Facebook platform, the company said.

Privacy issue?

However, many Instagram users may be surprised to hear that their photos are used to train Facebook AI systems.

“We inform Instagram account holders in our data policies that we use the information we have to support research and innovation, including technological developments such as this,” Priya Goyal, software engineer at Facebook AI Research, told CNBC.

Facebook said it would open source code for the software so other researchers could experiment with it.

“While we share the details of our research and create an open-source library that allows other researchers to use self-directed learning to train models on uncurated images, we do not share the images or SEER mode,” Goyal said.

Other major technology companies, including Google and Microsoft, are also trying to push the boundaries of data visibility. Last summer, Google published SimCLRv2 computer vision model, while OpenAI published iGPT 2.


Source link