Al drew caricatures from the extreme snapshot

Computer scientists at Microsoft and City University of Hong Kong developed two separate AI systems, forming an artificial neural network called the Generative Adversarial Network (GAN) capable of cartooning .

From uploaded portrait photos, one of the GAN networks will analyze and magnify certain facial features. The remaining GAN network will add some pen strokes in the style of art commonly found in caricatures, making portrait photos an interesting caricature.

images 1 of Al drew caricatures from the extreme snapshot — *Images 1 of Al drew caricatures from the extreme snapshot*

The interesting thing about these algorithms is that they can work in reverse, converting caricatures into realistic images like photographs.

However, this system still has some limitations. Compared to other facial features, the degree of geometric magnification is expressed more clearly in the overall face shape. The algorithms can only read 33 out of 63 landmarks located on the face block so it is impossible to apply some degree of detailed exaggeration of geometry on the ears, hair, wrinkles . Researchers said they could solve this limitation by adding more landmarks on the face.