Classification of Human Actions in Videos with a Large-Scale Photonic Reservoir Computer
Affiliation auteurs | !!!! Error affiliation !!!! |
Titre | Classification of Human Actions in Videos with a Large-Scale Photonic Reservoir Computer |
Type de publication | Conference Paper |
Year of Publication | 2019 |
Auteurs | Antonik P, Marsal N, Brunner D, Rontani D |
Editor | , Kurkova V, Karpov P, Theis F |
Conference Name | ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING - ICANN 2019: WORKSHOP AND SPECIAL SESSIONS |
Publisher | SPRINGER INTERNATIONAL PUBLISHING AG |
Conference Location | GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND |
ISBN Number | 978-3-030-30493-5; 978-3-030-30492-8 |
Mots-clés | Computer vision, Human action classification, Photonic reservoir computing |
Résumé | The identification of different types of human actions in videos is a major computer vision task, with capital applications in e.g. surveillance, control, and analysis. Deep learning achieved remarkable results, but remains hard to train in practice. Here, we propose a photonic reservoir computer for recognition of video-based human actions. Our experiment comprises off-the-shelf components and implements an easy-to-train neural network, scalable up to 16,384 nodes, and performing with a near state-of-the-art accuracy. Our findings pave the way towards photonic information processing systems for real-time video processing. |
DOI | 10.1007/978-3-030-30493-5_15 |