Facebook’s new AI feature can replace your closed eyes with open ones in an image

Artificial Intelligence (AI) is considered to be the next big thing. With all tech giants and device manufacturer's focus on AI enabled device can certainly prove that AI will take over the traditional devices, giving users an ultimate user experience. The tech giants such as Google, Apple, Microsoft, and Facebook are also substantially investing in the AI for manufacturing their products. Now, Facebook has released an AI engine which utilizes its own plethora of photos with people blinking.

The social media giant has come up with a new AI which can open closed eyes in the photos. Yes, you read that correct, the company's new AI can replace the closed eyes in the pictures using the huge repository of images that the platform has.

Facebook's new AI engine is primarily a machine learning system which can make the closed eyes appear as they are open. In order to do so, the AI uses Generative Adversarial Network (, which fools the system into thinking that it is creating real images. The Generative Adversarial Network comprises of two parts:

1. The first part has the ability to recognize faces

2. The second part produces images constantly which keeps growing in realism that is based on the repeated feedback from the first part

3. Facebook's AI engine is essentially a machine learning system that can make closed eyes look as if they're open. It does so with a Generative Adversarial Network, that fools the system into thinking that its creations are real.

Facebook took a different approach to test the feature. It fed the GAN (Generative Adversarial Network) with "exemplar" data that shows the target person with their eyes open. Using this data, GAN learns how a person should look with their eyes open and also learns their eye shape, color, etc.

The researchers quoted that, "Exemplar GANs provide a useful solution for image generation or in-painting when a region of that image has some sort of identifying the feature. They provide superior perceptual results because they incorporate identifying information stored in reference images or perceptual.

The AI-feature is said to be working fine with some small glitches which are expected to be resolved over time.