A topological solution to object segmentation and tracking

We address a question at the foundation of natural and artificial vision: how can a visual system segment and track objects? In the real world, objects can undergo drastic changes in appearance due to deformation, perspective change, or dynamic occlusion. For example, an animal moving behind a fence will split into multiple pieces. How can a visual system apply a common label to all the pieces of the same object across space and time? Here, we prove that this can be solved using purely geometric mechanisms and furthermore demonstrate the approach on cluttered synthetic video. By enabling the automatic transformation of visual information from a sensory to symbolic format, the mechanism described here provides a springboard from sensation to intelligent symbolic reasoning. The world is composed of objects, the ground, and the sky. Visual perception of objects requires solving two fundamental challenges: 1) segmenting visual input into discrete units and 2) tracking identities of these u

We address a question at the foundation of natural and artificial vision: how can a visual system segment and track objects? In the real world, objects can undergo drastic changes in appearance due to deformation, perspective change, or dynamic occlusion. For example, an animal moving behind a fence will split into multiple pieces. How can a visual system apply a common label to all the pieces of the same object across space and time? Here, we prove that this can be solved using purely geometric mechanisms and furthermore demonstrate the approach on cluttered synthetic video. By enabling the a