Human activity recognition via optical flow: decomposing activities into basic actions
- PDF / 1,793,385 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 4 Downloads / 161 Views
(012 3456789(). ,- volV)(0123456789().,-volV)
IAPR-MEDPRAI
Human activity recognition via optical flow: decomposing activities into basic actions Ammar Ladjailia1 • Imed Bouchrika2 • Hayet Farida Merouani1 • Nouzha Harrati2 • Zohra Mahfouf2 Received: 24 July 2018 / Accepted: 18 December 2018 Springer-Verlag London Ltd., part of Springer Nature 2019
Abstract Recognizing human activities using automated methods has emerged recently as a pivotal research theme for securityrelated applications. In this research paper, an optical flow descriptor is proposed for the recognition of human actions by considering only features derived from the motion. The signature for the human action is composed as a histogram containing kinematic features which include the local and global traits. Experimental results performed on the Weizmann and UCF101 databases confirmed the potentials of the proposed approach with attained classification rates of 98.76% and 70%, respectively, to distinguish between different human actions. For comparative and performance analysis, different types of classifiers including Knn, decision tree, SVM and deep learning are applied to the proposed descriptors. Further analysis is performed to assess the proposed descriptors under different resolutions and frame rates. The obtained results are in alignment with the early psychological studies reporting that human motion is adequate for the perception of human activities. Keywords Action recognition Motion descriptor Optical flow Decomposing activities
1 Introduction Much scientific research in computer vision is dedicated to the arena of human motion analysis. These studies are supported by the large number of applications where automated analysis of human motion is deemed very crucial including biometrics, smart automated surveillance, sports arbitration and human–machine interaction. As we are becoming more digital natives in such a modern era, the recognition of human activities is becoming an interesting research area with the potency to be integrated within various realistic human-centric contexts [1, 6]. Additionally, because of the unprecedented increase in multimedia data produced continuously from security cameras, movie & Ammar Ladjailia [email protected] Imed Bouchrika [email protected] 1
Department of Computer Science, University of Annaba, 23000 Annaba, Algeria
2
Faculty of Science and Technology, University of Souk Ahras, 41000 Souk Ahras, Algeria
production and Web uploads, it is now becoming an important necessity to analyse such video content semantically via automated methods. This would be a major milestone to facilitate the process of indexing, search and retrieval of multimedia content. The deployment of automated vision systems to recognize human activities can stand as an innovative solution to increase the adoption and usability for such smart visual applications. The process of extracting and recognizing human actions via automated marker-less methods are two separate tasks that are affirmed to be cumbersome and
Data Loading...
