Colloquium Jelmer Ypma

The scale-invariant feature transform (SIFT) method developed by Lowe, detects
multiple features from images, which can be used for object recognition. Using a
robot I obtained images from objects from different viewpoints. These images contain
an object with a distracting background. During the learning phase, active vision
was used to differentiate between object-features and background-features, such that
only features corresponding to the object are used in the object representation. In
this talk I will explain how the SIFT method works. I will also show the results of
experiments, when using SIFT for active object recognition.