%0 Conference Proceedings %T A spatio-temporal approach for multiple object detection in videos using graphs and probability maps %+ Image, Modélisation, Analyse, GEométrie, Synthèse (IMAGES) %+ Département Traitement du Signal et des Images (TSI) %A Morimitsu, Henrique %A Cesar, R. M. %A Bloch, Isabelle %< avec comité de lecture %Z HM:ICIAR-14 %( International Conference on Image Analysis and Recognition (ICIAR) %B International Conference on Image Analysis and Recognition (ICIAR) %C Vilamoura, Portugal %V LNCS 8815 %P 421-428 %8 2014-11 %D 2014 %K Structural information %K Graph %K Tracking %K Video %K Object detection %Z Computer Science [cs]/Image Processing [eess.IV]Conference papers %X This paper presents a novel framework for object detection invideos that considers both structural and temporal information. Detec-tion is performed by first applying low-level feature extraction techniquesin each frame of the video. Then, additional robustness is obtained byconsidering the temporal stability of videos, using particle filters andprobability maps, which encode information about the expected locationof each object. Lastly, structural information of the scene is describedusing graphs, which allows us to further improve the results. As a prac-tical application, we evaluate our approach on table tennis sport videosdatabases: the UCF101 table tennis shots and an in-house one. The ob-served results indicate that the proposed approach is robust, showing ahigh hit rate on the two databases. %G English %L hal-02286919 %U https://telecom-paris.hal.science/hal-02286919 %~ INSTITUT-TELECOM %~ CNRS %~ ENST %~ TELECOM-PARISTECH %~ PARISTECH %~ LTCI %~ IDS %~ IMAGES %~ INSTITUTS-TELECOM