Recognizing activities with cluster-trees of tracklets - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2012

Recognizing activities with cluster-trees of tracklets

Résumé

We address the problem of recognizing complex activities, such as pole vaulting, which are characterized by the composition of a large and variable number of different spatio-temporal parts. We represent a video as a hierarchy of mid-level motion components. This hierarchy is a data-driven decomposition specific to each video. We introduce a divisive clustering algorithm that can efficiently extract a hierarchy over a large number of local trajectories. We use this structure to represent a video as an unordered binary tree. This tree is modeled by nested histograms of local motion features. We provide an efficient positive definite kernel that computes the structural and visual similarity of two tree decompositions by relying on models of their edges. Contrary to most approaches based on action decompositions, we propose to use the full hierarchical action structure instead of selecting a small fixed number of parts. We present experimental results on two recent challenging benchmarks that focus on complex activities and show that our kernel on per-video hierarchies allows to efficiently discriminate between complex activities sharing common action parts. Our approach improves over the state of the art, including unstructured activity models, baselines using other motion decomposition algorithms, graph matching, and latent models explicitly selecting a fixed number of parts.
Fichier principal
Vignette du fichier
gaidon_tracklets_bmvc2012.pdf (547.54 Ko) Télécharger le fichier
Vignette du fichier
tracklets_clustertree_small.jpg (490.61 Ko) Télécharger le fichier
onepager_bmvc2012_tracklets.pdf (1.95 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Format : Figure, Image
Format : Autre
Loading...

Dates et versions

hal-00722955 , version 1 (06-08-2012)
hal-00722955 , version 2 (07-08-2012)

Identifiants

Citer

Adrien Gaidon, Zaid Harchaoui, Cordelia Schmid. Recognizing activities with cluster-trees of tracklets. BMVC 2012 - British Machine Vision Conference, Sep 2012, Guildford, United Kingdom. pp.30.1-30.13, ⟨10.5244/C.26.30⟩. ⟨hal-00722955v2⟩
1015 Consultations
2249 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More