Facebook has always been good at recognizing objects and people in images. Object recognition is one of the most challenging areas of image processing and machine learning today. As Artificial Intelligence continues to become more sophisticated by the day, several enhancements in machine learning algorithms are slowly turning the complex object recognition task into a manageable problem. Facebook is well on its way to achieving higher recognition rates in images. So far, it hasn’t been able to precisely identify what’s happening in a photo. This is all set to change very soon.
Facebook users will be able to search photos based on content
New developments at taking place at Menlo Park. Facebook’s new computer vision engine – called Lumos – is all set to improve user experience by leaps and bounds. With the enhanced artificial intelligence software, Facebook users will be able to search photos based on the content. So if you are someone who’s accustomed to searching images by tags, location or date taken, you will be surprised when the new feature rolls in. Let’s say you wish to find a photo of your friend eating a delicious dish. Just searching for the name of the dish will return the photo you are looking for. Cool, isn’t it?
Facebook’s automatic alt text feature to be greatly improved
According to reports, the upcoming upgrade will also improve Facebook’s automatic alt text feature. For the visually impaired, Facebook’s automatic alt text feature helps experience images like never before by providing a vivid description of the image. Until now, this feature had some very basic stuff – describing, for example, a specific concert with ideas like “contains a person”, “guitar” etc. With the next-generation computer vision engine Lumos, Facebook will be able to provide additional information, describing the image in more vivid details – for example, “a photo with a person playing a guitar”.
Also Check: How to Save Videos From Facebook
How does Lumos work?
Lumos relies on what are called “neural networks”. As Artificial Intelligence makes inroads into our day-to-day lives, neural networks’ adoption is increasing at a staggering rate. Neural networks work like the human brain – they need to be trained before they can identify images. For example, presenting thousands of images of a man’s face to a neural network will enable it to identify and recognize whether or not a given sample photo has a man’s face in it. Facebook’s Lumos works on the same principle but at a much more complex and advanced level. It works by analyzing troves of images shared on the social media platform every day. The more images it takes in to analyze, the better it gets at recognizing.
You may find this interesting to read: New Snapchat Lenses To Bring Augmented Reality For Real-World Object Recognition