icece2020-E3020final.pdf (437.27 kB)
Download file

A Lightweight Action Recognition Method for Unmanned-Aerial-Vehicle video

Download (437.27 kB)
conference contribution
posted on 09.03.2021, 11:04 by M Ding, N Li, J Song, R Zhang, X Zhang, Huiyu Zhou
In recent year, due to motility and wide coverage, unmanned aerial vehicle (UAV) has been widely applied in surveillance system. Human action recognition in UAV video is essential for surveillance video understanding. However, existing action recognition methods suffer from heavy computing, which makes it hard to deploy in real applications. In this paper, a lightweight action recognition method for UAV video(LARMUV) is proposed. This method is based on TSN and adopt Mo-bileNetV3 as backbone, which greatly reduces amount of computing and parameters. Self-attention mechanism is adopted to capture temporal structure among different frames. For loss function, Focal Loss is used to putting more focus on hard, misclassified examples. Last but not least, knowledge distillation is employed to enhance the performance of our model, which transfer knowledge from a larger teacher model to student model. Experimental results on HMDB51, UCF101 and UAV dataset show that our method can achieve competitive performance compared to baseline methods while run in real-time mode.

History

Author affiliation

School of Informatics

Source

2020 IEEE 3rd International Conference on Electronics and Communication Engineering (ICECE), 14-16 Dec. 2020, Xi'An, China

Version

AM (Accepted Manuscript)

Published in

2020 IEEE 3rd International Conference on Electronics and Communication Engineering (ICECE)

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

isbn

978-1-7281-9258-1

Copyright date

2020

Available date

17/02/2021

Spatial coverage

Xi'An, China

Temporal coverage: start date

14/12/2020

Temporal coverage: end date

16/12/2020

Language

en

Usage metrics

Categories

Exports