Multi-label Classification of Moving Object Trajectories based on Frequent Behavior Type Detection

Abstract : This paper proposes a method that analyzes data representing trajectories of moving objects. The method is based on extracting frequent patterns and introduces different types of patterns, for example, latent, emergent, etc. A set of algorithms are introduced to pre-process the data, to extract the frequent patterns and detect the behaviors. These types are used to generate a city map tagged by behaviors. The classification of a given trajectory consists in its projection on this spatial zone. This leads to a multi-label classification which depends on the spatio-temporal granularity. Finally, we discuss the application of our method on real-world data representing taxi trajectories.
Document type :
Conference papers
Complete list of metadatas

Cited literature [5 references]  Display  Hide  Download

https://hal-amu.archives-ouvertes.fr/hal-01627398
Contributor : Nicolas Durand <>
Submitted on : Wednesday, December 12, 2018 - 2:26:14 PM
Last modification on : Thursday, December 13, 2018 - 11:01:57 AM
Long-term archiving on : Wednesday, March 13, 2019 - 4:25:15 PM

File

almuhisen_sfc2017.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01627398, version 1

Collections

Citation

Feda Almuhisen, Nicolas Durand, Mohamed Quafafou. Multi-label Classification of Moving Object Trajectories based on Frequent Behavior Type Detection. 24èmes Rencontres de la Société Francophone de Classification (SFC 2017), Société Francophone de Classification, Jun 2017, Lyon, France. ⟨hal-01627398⟩

Share

Metrics

Record views

264

Files downloads

21