Scientists at Osaka University have developed a system capable of virtually removing buildings from live images via mobile devices such as cell phones or tablets.
The engine transmits, in real time, the images captured on the device to a remote server, using so-called generative adversarial network (GAN) algorithms.
“The GAN algorithm can produce images that don’t really exist, but are plausible,” explains one of the authors, Tomohiro Fukuda.
This type of artificial intelligence uses two competing neural networks: a generator and a discriminator. The generator is programmed to create increasingly realistic images, while the discriminator is tasked with distinguishing whether the image exists or has been artificially generated.
According to scientists, technology can help in urban renewal and planning processes. That’s because stakeholders in a project that involves demolition of old buildings, for example, could “see” the outcome before committing to it.
“Our method allows users to intuitively understand what the future landscape will look like, which can contribute to reducing the time and cost to form a consensus,” explains first author Takuya Kikuchi.
In order for the entire process to happen in real time, the research team used a “semantic segmentation” scheme on the input image. In this way, the system can sort images pixel by pixel.
Source: CNN Brasil
