Making a machine that sees like us

Zygmunt Pizlo, Yunfeng Li, Tadamasa Sawada, Robert M. Steinman

Chapter 1

Demo 1.1. Animations of the deformations of the 3D shapes shown in Figure 1.8.

Demo 1.2. Recovering 3D Shapes from single 2D images.

Demo 1.3. Ames's Chair.

Chapter 2

Demo 2.1. Three types of constraints.

Demo 2.2. A one-parameter family of recovered 3D shapes.

Demo 2.3. Eight symmetrical 3D shapes are recovered from the same line-drawing.

Demo 2.4. The recovery of a skeleton.

Chapter 3

Demo 3.1. Orthographic projection.

Demo 3.2. Perspective projection.

Demo 3.3. The image of a symmetrical 3D shape that could be perceived as asymmetrical.

Chapter 4

Demo 4.1. Recovery of symmetry from a pair of arbitrary curves.

Demo 4.2. The role of degenerate views in the recovery of symmetry.

Demo 4.3. Why the a priori planarity constraint can eliminate degenerate views.

Demo 4.4. Why having a view, which is not degenerate, is not enough for symmetry perception.

Demo 4.5. Recovery of an indoor scene.

Chapter 5

Demo 5.1. Shape before disparity.

Demo 5.2. How the change of depth-order works.

Chapter 6

Demo 6.1. Detecting and tracking moving people.

Demo 6.2. How the 3D shape of furniture is recovered.

Demo 6.3. How our machine solves a 50-City Travel Salesman Problem.