Visual Explanation by High-Level Abduction: On Answer-Set Programming Driven Reasoning about Moving Objects
This work addresses visual reasoning challenges for computer vision researchers, but appears incremental as it integrates existing methods like answer-set programming and object tracking.
The paper tackles the problem of robust visual explanation for moving objects in video data by proposing a hybrid architecture combining answer-set programming based abductive reasoning with a visual processing pipeline, and evaluates it on the MOTChallenge benchmark and a Movie Dataset.
We propose a hybrid architecture for systematically computing robust visual explanation(s) encompassing hypothesis formation, belief revision, and default reasoning with video data. The architecture consists of two tightly integrated synergistic components: (1) (functional) answer set programming based abductive reasoning with space-time tracklets as native entities; and (2) a visual processing pipeline for detection based object tracking and motion analysis. We present the formal framework, its general implementation as a (declarative) method in answer set programming, and an example application and evaluation based on two diverse video datasets: the MOTChallenge benchmark developed by the vision community, and a recently developed Movie Dataset.