Activity identification from animal GPS tracks with spatial temporal clustering method DDB-SMoT
Abstract
With various GPS devices or services growing rapidly, large amount of GPS track- ing data has been collected, both for human beings and wild animals. However, the raw GPS data cannot directly provide us with any valuable information because of the semantic gap between it and the raw GPS trajectory data. As a result, algorithms are needed to extract the semantic information from raw GPS data. To solve this problem, this project implements two software tools and a web application. Semantic Analysis Software provides semantic analysis based on stops in the trajectory detected by DDB-SMoT (Direction and Distance Based -- Stop and Move of Trajectory) and POI (Point of Interest) list to output a list of activities in order to explain the meaning of the given trajectory. Trajectory Generator Software generates labeled trajectory based on the stop and move model to evaluate the performance of stops detection algorithms. Semantic Analysis Web Application displays the semantic enrichment process step by step on Google Map use bear and deer GPS trajectories provided by MDC (Missouri Department of Conservation). Through experiments, the DDB-SMoT algorithm has an overall accuracy of 91.18% when detecting stops and movement points in animal trajectory generated by the trajectory generator. Because lack of a rich animal POI dataset and activity ground truth, the verification of the semantic analysis process will leave as future work.
Degree
M.S.
Thesis Department
Rights
OpenAccess.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.