Vocabulary tree based place recognition is widely used in topological localization and its various applications have been proposed during the past decade. But the bag-of-words representations from the vocabulary tree which is trained with fixed training data, are difficult to be optimal to dynamic environments. To solve this problem, adaptive vocabulary tree is proposed. However, there has been no comparison considering the adaptive property of conventional vocabulary tree. This paper provides the performance evaluation of the vocabulary tree and the adaptive vocabulary tree in dynamic scenes.
Supplementary notes can be added here, including code and math.