Our essay discusses an AI process developed for making art (AICAN), and the issues AI creativity raises for understanding art and artists in the 21st century. Backed by our training in computer science (Elgammal) and art history (Mazzone), we argue for the consideration of AICAN’s works as art, relate AICAN works to the contemporary art context, and urge a reconsideration of how we might define human and machine creativity. Our work in developing AI processes for art making, style analysis, and detecting large-scale style patterns in art history has led us to carefully consider the history and dynamics of human art-making and to examine how those patterns can be modeled and taught to the machine. We advocate for a connection between machine creativity and art broadly defined as parallel to but not in conflict with human artists and their emotional and social intentions of art making. Rather, we urge a partnership between human and machine creativity when called for, seeing in this collaboration a means to maximize both partners’ creative strengths.
19 August 2014 in Multimedia Tools and Applications
In this paper we consider modeling data lying on multiple continuous manifolds. In particular, we model the shape manifold of a person performing a motion observed from different view points along a view circle at fixed camera height. We introduce a model that ties together the body configuration (kinematics) manifold and the visual manifold (observations) in a way that facilitates tracking the 3D configuration with continuous relative view variability. The model exploits the low dimensionality nature of both the body configuration manifold and the view manifold where each of them are represented separately.
25 April 2008 in IEEE Transactions on Pattern Analysis and Machine Intelligence
We present a framework for monocular 3D kinematic pose tracking and viewpoint estimation of periodic and quasi-periodic human motions from an uncalibrated camera. The approach we introduce here is based on learning both the visual observation manifold and the kinematic manifold of the motion using a joint representation. We show that the visual manifold of the observed shape of a human performing a periodic motion, observed from different viewpoints, is topologically equivalent to a torus manifold. The approach we introduce here is based on the supervised learning of both the visual and kinematic manifolds. Instead of learning an embedding of the manifold, we learn the geometric deformation between an ideal manifold (conceptual equivalent topological structure) and a twisted version of the manifold (the data). Experimental results show accurate estimation of the 3D body posture and the viewpoint from a single uncalibrated camera.
15 July 2009 in International Journal of Computer Vision