Explaining Object Motion Using Answer Set Programming

摘要

Identifying objects and their motion from sequences of digital images is of growing interest due to the increasing application of autonomous mobile systems like autonomous cars or mobile robots. Although the reliability of object recognition has been increased significantly, there are often cases arising where objects are not classified correctly. There might be objects detected in almost all images of a sequence but not all. Hence, there is a need for finding such situations and taking appropriate measures to improve the overall detection performance. In this paper, we contribute to this research direction and discuss the use of logic for identifying motion in sequences of images that can be used for this purpose. In particular, we introduce the application of diagnosis and show an implementation using answer set programming.