Whether you are a professional level photographer or a beginner or a person with just photography as a hobby, at some point in our lives we all feel the need for photo editors. To make the pictures perfect in all aspects people use various photo editors. We live in an era of digitalization and Artificial Intelligence.
Over the past couple of years, technology changed a lot and along with that, there has been a gradual evolution in the category of photo editing software. With new AI capabilities, it is now easy to manipulate snaps according to your desires.
Usually, editing applications follow user instructions but this software might decide on its own to draw or not draw anything if the object is impossible to be placed at a location. This software helps us to understand how GANs learn to represent the visual world.
The team used a two-part network to train the model. They used generations to create samples of realistic images and discriminators were used to identify the differences between generated and real-life photos. The input data from discriminators are used to create better realistic images. This procedure helps to test other GANs and to improve them as well. This tool will help researchers understand the kind of mistakes AI can make and weed out fake images.
Read next: Now AI Would Be Able To Recognize Emotions through Body Postures
Over the past couple of years, technology changed a lot and along with that, there has been a gradual evolution in the category of photo editing software. With new AI capabilities, it is now easy to manipulate snaps according to your desires.
GAN (Generative Adversarial Network) Paint Studio
The photo editing software evolved a lot with large help from machine learning. A new software made by researchers from MIT and IBM made the lives of the people easy by introducing web-based tool that lets you edit and remove objects from a set of pictures with just a click. With this new GAN software, you can manipulate different parts without ruining the original feel and look of a photo. If you add more clouds or grass to the scene, the picture will look the same as real without any hint of manipulation in it.Special features of this tool
This photo editing tool has a special built-in feature which helps it to not perform any action if it thinks the specific manipulation wouldn’t fit the image.
Usually, editing applications follow user instructions but this software might decide on its own to draw or not draw anything if the object is impossible to be placed at a location. This software helps us to understand how GANs learn to represent the visual world.
Testing the software
The software was tested by the researchers by removing certain objects in the imageries and analyzing the artifacts around the removed objects. They used this information to train the algorithm to create realistic modifications when an object is removed or added.The team used a two-part network to train the model. They used generations to create samples of realistic images and discriminators were used to identify the differences between generated and real-life photos. The input data from discriminators are used to create better realistic images. This procedure helps to test other GANs and to improve them as well. This tool will help researchers understand the kind of mistakes AI can make and weed out fake images.
Read next: Now AI Would Be Able To Recognize Emotions through Body Postures
