Space-time robust representation for action recognition

Abstract : We address the problem of action recognition in unconstrained videos. We propose a novel content driven pooling that leverages space-time context while being robust toward global space-time transformations. Being robust to such transformations is of primary importance in unconstrained videos where the action localizations can drastically shift between frames. Our pooling identifies regions of interest using video structural cues estimated by differ ent saliency functions. To combine the different structural information, we introduce an iterative structure learning algorithm, WSVM (weighted SVM), that determines the optimal saliency layout of an action model through a sparse regularizer. A new optimization method is proposed to solve the WSVM' highly non-smooth objective function. We evaluate our approach on standard action datasets (KTH, UCF50 and HMDB). Most noticeably, the accuracy of our algorithm reaches 51.8% on the challenging HMDB dataset which outperforms the state-of-the-art of 7.3% relatively.
Document type :
Conference papers
Complete list of metadatas

https://hal-cea.archives-ouvertes.fr/cea-01841174
Contributor : Léna Le Roy <>
Submitted on : Tuesday, July 17, 2018 - 10:16:03 AM
Last modification on : Wednesday, January 23, 2019 - 2:39:26 PM

Identifiers

Collections

Citation

N. Ballas, Y. Yang, Z.-Z. Lan, B. Delezoide, F. Preteux, et al.. Space-time robust representation for action recognition. 2013 IEEE International Conference on Computer Vision, Dec 2013, Sydney, NSW, Australia. pp.2704-2711, ⟨10.1109/ICCV.2013.336⟩. ⟨cea-01841174⟩

Share

Metrics

Record views

60