%Aigaion2 BibTeX export from Idiap Publications %Friday 11 October 2024 01:25:14 AM @INPROCEEDINGS{odobez-gretsi-03, author = {Odobez, Jean-Marc and Ba, Sil{\`{e}}ye O.}, projects = {Idiap}, title = {Mod{\'{e}}lisation implicite du mouvement en suivi par filtrage de Monte Carlo s{\'{e}}quentiel}, booktitle = {GRETSI conference, Signal and Image Processing,}, year = {2003}, address = {Paris, France}, crossref = {odobez-rr-03-15}, abstract = {Le filtrage par m{\'{e}}thode de Monte-Carlo s{\'{e}}quentiel (MCS) est l'une des m{\'{e}}thodes les plus populaires pour effectuer du suivi visuel. Dans ce contexte, il est g{\'{e}}n{\'{e}}ralement fait l'hypoth{\`{e}}se que, {\'{e}}tant donn{\'{e}}e la position d'un objet dans des images successives, les observations extraites des images de cet objet sont ind{\'{e}}pendantes. Dans cet article, nous soutenons que, au contraire, ces observation sont fortement corr{\'{e}}l{\'{e}}es. Pour prendre en compte cette correlation, nous proposons un nouveau mod{\`{e}}le qui peut s'interpr{\'{e}}ter comme l'ajout d'un terme de vraisemblance mod{\'{e}}lisant implicitement des mesures de mouvement. Le nouveau mod{\`{e}}le permet de lever des ambigu{\"{\i}}t{\'{e}}s visuelles tout en gardant des mod{\`{e}}les d'objet simples, comme le montrent les r{\'{e}}sultats obtenus sur plusieurs s{\'{e}}quences et mod{\`{e}}les d'objets diff{\'{e}}rents (contour ou distribution de couleurs).}, pdf = {https://publications.idiap.ch/attachments/reports/2003/odobez_2003_gretsi.pdf}, postscript = {ftp://ftp.idiap.ch/pub/reports/2003/odobez_2003_gretsi.ps.gz}, ipdinar={2003}, ipdmembership={vision}, ipdpriority={5}, } crossreferenced publications: @TECHREPORT{odobez-rr-03-15, author = {Odobez, Jean-Marc and Ba, Sil{\`{e}}ye O. and Gatica-Perez, Daniel}, projects = {Idiap}, title = {An Implicit {M}otion {L}ikelihood for {T}racking with {P}article {F}ilters}, type = {Idiap-RR}, number = {Idiap-RR-15-2003}, year = {2003}, institution = {IDIAP}, address = {Martigny, Switzerland}, note = {Published in British Machine Vision Conference (BMVC,',','), Norwich, 2003.}, abstract = {Particle filters are now established as the most popular method for visual tracking. Within this framework, it is generally assumed that the data are temporally independent given the sequence of object states. In this paper, we argue that in general the data are correlated, and that modeling such dependency should improve tracking robustness. To take data correlation into account, we propose a new model which can be interpreted as introducing a likelihood on implicit motion measurements. The proposed model allows to filter out visual distractors when tracking objects with generic models based on shape or color distribution representations, as shown by the reported experiments.}, pdf = {https://publications.idiap.ch/attachments/reports/2003/rr03-15.pdf}, postscript = {ftp://ftp.idiap.ch/pub/reports/2003/rr03-15.ps.gz}, ipdinar={2003}, ipdmembership={vision}, ipdpriority={8}, language={English}, }