Skip to content

Automatic classification of pedestrian trajectories using clustering techniques.

License

Notifications You must be signed in to change notification settings

SasCezar/PedestrianTrajectoryClustering

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

34 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PedestrianTrajectoryClustering

Pedestrian flow analysis aims to provide insight into human movement patterns in buildings or outdoor areas. These analyses provide valuable information to building designers and other decision makers. A substantial bodyof research is based on pedestrian behavior and the interaction of pedestrian swith their environment and other pedestrians.In order to help this analysis we propose a pedestrian trajectories clustering system for lane discovery. Our solution uses the work of Atev et al. on clustering of vehicular trajectories applied to our domain. We evaluate our solution on multiple dataset.

Image 1 Image 2 Image 3

Conclusions

We presented a system for pedestrian lane recognition based on spectral clustering and Hausdorff distance. We performed analysis on two different dataset. The results, even if the quantitative measures were promising, showed us that a system based on whole trajectories does not give enough information on a lane. This type of analysis does not give us information on how much a lane stays alive, how many times a lane is reformed, and so on. Our type of analysis is more suitable for static analyses such as urban and architectural planning. As future work, we aim to perform a time based clustering, and try different distance measures. Another improvement could be performed to the clustering algorithm, which actually does not make use of the direction and of the speed of a pedestrian.

Check the report.pdf file for the full description of the project.

About

Automatic classification of pedestrian trajectories using clustering techniques.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages