Machine Learning and Friends Lunch

Learning Static Object Segmentation from Motion Segmentation

Abstract

Motion segmentation is a rich source of training data for learning to
segment objects by their static image properties. Background subtraction
can distinguish between moving objects and their backgrounds, and the
techniques of statistical machine learning can be used to capture
information about objects' shape, size, color, brightness, and texture
properties. Presented with a new, static image, the trained model can
infer the proper segmentation of the objects present in a scene. The
algorithm presented in this work uses the techniques of Markov random
field modeling and belief propagation inference, and outperforms a
standard segmentation algorithm on the object segmentation task.

Back to ML Lunch home