TY  - CONF
AU  - Swathikiran Sudhakaran
AU  - Sergio Escalera
AU  - Oswald Lanz
A2  - CVPR
PY  - 2019//
TI  - LSTA: Long Short-Term Attention for Egocentric Action Recognition
BT  - 32nd IEEE Conference on Computer Vision and Pattern Recognition
SP  - 9946
EP  - 9955
N2  - Egocentric activity recognition is one of the most challenging tasks in video analysis. It requires a fine-grained discrimination of small objects and their manipulation. While some methods base on strong supervision and attention mechanisms, they are either annotation consuming or do not take spatio-temporal patterns into account. In this paper we propose LSTA as a mechanism to focus on features from spatial relevant parts while attention is being tracked smoothly across the video sequence. We demonstrate the effectiveness of LSTA on egocentric activity recognition with an end-to-end trainable two-stream architecture, achieving state-of-the-art performance on four standard benchmarks.
UR  - https://ieeexplore.ieee.org/document/8954401
L1  - http://refbase.cvc.uab.es/files/SEL2019.pdf
UR  - http://dx.doi.org/10.1109/CVPR.2019.01019
N1  - HuPBA; no proj
ID  - Swathikiran Sudhakaran2019
ER  -