A framework for dynamic restructuring of semantic video analysis systems based on learning attention control

摘要: Abstract Current semantic video analysis systems are usually hierarchical and consist of some levels to overcome gaps between low-level features high-level concepts. In these systems, features, descriptors, objects or concepts extracted in each level therefore, total computational complexity such is huge. this paper, we present a new general framework impose attention control on system using Q-learning. Thus, our proposed restructures given dynamically direct the blocks extracting most informative features/concepts reduces system. other words, directs flow processing actively learning method. The evaluated for event detection broadcast soccer videos limited numbers training samples. Experiments show that able learn how restructure initial structure reach final goal with less complexity.