The famous Mona Lisa painting by Leonardo da Vinci has been brought to life by AI researchers at Samsung’s AI research laboratory in Moscow. Using just a single photo, the researchers have created a video in which the model in the portrait moves her head, eyes, and mouth.
Samsung's algorithms were trained on a public database of 7,000 images of celebrities gathered from YouTube, with the AI system mapping facial features and movements on to a photo to bring it to life in what the Samsung team call "realistic neural talking heads”.
It is made possible by frontloading the facial landmark identification process with a huge amount of data, making the model highly efficient at finding the parts of the target face that correspond to the source and can now do it with one image. This is why it has been possible to take a picture of the Mona Lisa, though researchers also used a picture of Einstein and Marilyn Monroe, to make it move and speak like a real person. The rise of convincing deepfake technology, such as this recent “single-shot learning” has huge potential for misuse.