We are developing the next-generation of machine vision models that will allow us to improve our general understanding of Visual Art.

Try Similario , our First Painting Similarity Search Demo now.

Our Mission : Research & Development of Abstract Visual Imagery

At Artificio we are developing a machine that can interpret abstract visual art, as we believe this is a necessary step & condition for high abstract visual intelligence of artificial perceptual systems. Both perceptual psychologists, and computer vision scientists have been re-evaluating their research horizon, where computational aesthetics & visual art has now taken the center stage given the saturation of machine vision systems that perform spectacularly well on object recognition tasks, but do not generalize to out of distribution objects, common corruptions, adversarial stimuli and more importantly -- to abstract visual art -- as this is the hardest visual stimuli to interpret for most machine vision systems and humans.


Consequently, if we can create a model that can correctly interpret abstract visual art, these engineered visual systems will also be robust to many different distortions in the wild, and will benefit state-of-the-art computer vision models' accuracy for real-world applications such as robust vision for autonomous driving and medical imaging. Symbiotically, an achievement of this type may also help us be one step closer to reverse-engineering the human visual system as shown in some of our current research in computational perception.


Institutional Partnerships & Collaborators

MIT Center for Brains, Minds and Machines (CBMM)

National Science Foundation (NSF)

NVIDIA

Powered by

NVIDIA

HuggingFace

Gradio

If you are interested in contacting us: Say Hi at info[at]artificio.org or Follow Us: