Betul N. Guner

Comparison with machine classifiers For each image, the aggregate observer model (AO) specifies a probability that the image is an occlusion, much in the same manner as the machine classifiers. In order to see if images which were more likely to be categorized as occlusions by the AO were also more likely to be categorized as such by various machine observers, we developed a simple analysis procedure, which we now describe.