The term GAN inversion originates from the fields of Artificial Intelligence and Digital Transformation. It refers to a specific method that allows for the „backward“ decoding of images or other data using so-called Generative Adversarial Networks, or GANs for short.
A GAN is a type of artificial neural network often used to create images – for example, portraits of people who don't actually exist. GAN inversion is about figuring out what „ingredients“ or hidden features a GAN used to create a specific image. This process, therefore, gives us the blueprint or genetic fingerprint of an AI-generated image.
A vivid example: Imagine an AI creating an image of a car. With GAN inversion, an expert can then analyse which features – such as shape, colour, or style – the image was composed of. In practice, this is useful for better understanding AI-generated media, detecting errors, or making targeted adjustments, for example for new designs in the automotive industry.
Deutsch 
English (UK)