Abstract : We propose a new method for activity recognition based on a view independent representation of human motion. Robust 3D volume motion templates (VMTs) are calculated from tracklets. View independence is achieved through a rotation with respect to a canonical orientation. From this volumes, features based on 3D gradients are extracted, projected to a codebook and pooled into a bags-of-words model classified with an SVM classifier. Experiments show that the method outperforms the original HoG3D method.
Type de document :
Communication dans un congrès
International Conference on Image Processing, Sep 2015, Quebec City, Canada. IEEE, 2015
https://hal.archives-ouvertes.fr/hal-01147957
Contributeur : Christian Wolf
<>
Soumis le : dimanche 3 mai 2015 - 22:46:30
Dernière modification le : jeudi 1 novembre 2018 - 01:19:54
Emre Dogan, Gonen Eren, Christian Wolf, Atilla Baskurt. Activity Recognition with Volume Motion Templates and Histograms of 3D Gradients. International Conference on Image Processing, Sep 2015, Quebec City, Canada. IEEE, 2015. 〈hal-01147957〉
