Trajectory Reconnection - Historique des versions

NicolasSaunier le 29 mai 2016 à 21:34

2016-05-29T21:34:28Z

← Version précédente		Version du 29 mai 2016 à 21:34
Ligne 1 :		Ligne 1 :
			== Problem Description ==

	As identified by Paul St-Aubin, there are three cases of split or lost trajectories:		As identified by Paul St-Aubin, there are three cases of split or lost trajectories:

Ligne 4 :		Ligne 6 :
	# objects tracked by different cameras, overlapping or not		# objects tracked by different cameras, overlapping or not
	# objects stopping and lost by the tracker (in the same camera view)		# objects stopping and lost by the tracker (in the same camera view)

			== Potential Methods ==

	Reconnection methods:		Reconnection methods:
Ligne 10 :		Ligne 14 :
	#* similarity: position, time, predicted position, volume, type (if classified), appearance		#* similarity: position, time, predicted position, volume, type (if classified), appearance
	# for features lost because of foreground occlusion (one camera view), they could be searched by optical flow over a set of frames (and interpolated) (the quality control checks should be done through linear prediction).		# for features lost because of foreground occlusion (one camera view), they could be searched by optical flow over a set of frames (and interpolated) (the quality control checks should be done through linear prediction).

			== Implementation Ideas ==

	Regarding the method to store the result of reconnecting trajectories, it depends if the objects are initially stored in the same database or not.		Regarding the method to store the result of reconnecting trajectories, it depends if the objects are initially stored in the same database or not.
Ligne 17 :		Ligne 23 :
	* if stored in different databases, an option would be to merge both in a new database, with renumbering of objects and features		* if stored in different databases, an option would be to merge both in a new database, with renumbering of objects and features

	Potentially interesting papers:		== Potentially interesting papers ==

	* Swaminathan Sankararaman and Pankaj K. Agarwal and Thomas Mølhave and Jiangwei Pan and Arnold P. Boedihardjo (2013). Model-driven matching and segmentation of trajectories. Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems - SIGSPATIAL'13 . Association for Computing Machinery (ACM).		* Swaminathan Sankararaman and Pankaj K. Agarwal and Thomas Mølhave and Jiangwei Pan and Arnold P. Boedihardjo (2013). Model-driven matching and segmentation of trajectories. Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems - SIGSPATIAL'13 . Association for Computing Machinery (ACM).
	* Sheikh, Y. and Shah, M. (2005). Object tracking across multiple independently moving airborne cameras. In Computer Vision, ICCV 2005. 2, 1555-1560. IEEE.		* Sheikh, Y. and Shah, M. (2005). Object tracking across multiple independently moving airborne cameras. In Computer Vision, ICCV 2005. 2, 1555-1560. IEEE.

NicolasSaunier le 29 mai 2016 à 04:29

2016-05-29T04:29:36Z

← Version précédente		Version du 29 mai 2016 à 04:29
Ligne 9 :		Ligne 9 :
	# measure similarity of trajectory end and beginning, stored in a bipartite graph, then run an optimization matching algorithm such as the Hungarian algorithm		# measure similarity of trajectory end and beginning, stored in a bipartite graph, then run an optimization matching algorithm such as the Hungarian algorithm
	#* similarity: position, time, predicted position, volume, type (if classified), appearance		#* similarity: position, time, predicted position, volume, type (if classified), appearance
			# for features lost because of foreground occlusion (one camera view), they could be searched by optical flow over a set of frames (and interpolated) (the quality control checks should be done through linear prediction).

	Regarding the method to store the result of reconnecting trajectories, it depends if the objects are initially stored in the same database or not.		Regarding the method to store the result of reconnecting trajectories, it depends if the objects are initially stored in the same database or not.

NicolasSaunier le 23 mai 2016 à 05:07

2016-05-23T05:07:17Z

← Version précédente		Version du 23 mai 2016 à 05:07
Ligne 15 :		Ligne 15 :
	** if there is a time interval during which there is no feature, an artificial feature corresponding to the interpolated object position should be created and added to the feature group		** if there is a time interval during which there is no feature, an artificial feature corresponding to the interpolated object position should be created and added to the feature group
	* if stored in different databases, an option would be to merge both in a new database, with renumbering of objects and features		* if stored in different databases, an option would be to merge both in a new database, with renumbering of objects and features

			Potentially interesting papers:
			* Swaminathan Sankararaman and Pankaj K. Agarwal and Thomas Mølhave and Jiangwei Pan and Arnold P. Boedihardjo (2013). Model-driven matching and segmentation of trajectories. Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems - SIGSPATIAL'13 . Association for Computing Machinery (ACM).
			* Sheikh, Y. and Shah, M. (2005). Object tracking across multiple independently moving airborne cameras. In Computer Vision, ICCV 2005. 2, 1555-1560. IEEE.

NicolasSaunier le 23 mai 2016 à 04:46

2016-05-23T04:46:28Z

← Version précédente		Version du 23 mai 2016 à 04:46
Ligne 1 :		Ligne 1 :
	As identified by Paul St-Aubin, there are three cases of split trajectories:		As identified by Paul St-Aubin, there are three cases of split or lost trajectories:

			# objects disrupted because of foregroung occlusion, typically by static foreground objects
			# objects tracked by different cameras, overlapping or not
			# objects stopping and lost by the tracker (in the same camera view)

			Reconnection methods:

			# measure similarity of trajectory end and beginning, stored in a bipartite graph, then run an optimization matching algorithm such as the Hungarian algorithm
			#* similarity: position, time, predicted position, volume, type (if classified), appearance

			Regarding the method to store the result of reconnecting trajectories, it depends if the objects are initially stored in the same database or not.

			* if stored in the same database, an option is to connect the features of the two objects to the same object.
			** if there is a time interval during which there is no feature, an artificial feature corresponding to the interpolated object position should be created and added to the feature group
			* if stored in different databases, an option would be to merge both in a new database, with renumbering of objects and features

NicolasSaunier : NicolasSaunier a déplacé la page TrajectoryReconnection vers Trajectory Reconnection

2016-05-22T21:04:34Z

NicolasSaunier a déplacé la page TrajectoryReconnection vers Trajectory Reconnection

← Version précédente	Version du 22 mai 2016 à 21:04
(Aucune différence)

NicolasSaunier : Page créée avec « As identified by Paul St-Aubin, there are three cases of split trajectories: »

2016-05-22T21:04:22Z

Page créée avec « As identified by Paul St-Aubin, there are three cases of split trajectories: »

Nouvelle page

As identified by Paul St-Aubin, there are three cases of split trajectories: