Jing Liu's Website
Jing Liu's Website
Home
Publications
Research
Posts
Light
Dark
Automatic
3
Domain Generalization with Semi-Supervised Learning for People-Centric Activity Recognition
The proposed DGSSL method effectively enhances people-centric activity recognition by addressing the domain shift problem in semi-supervised learning manner, with autoregressive discriminator for adversarial training and two reconstruction tasks, resulting in better performance and generalization than three state-of-the-art methods.
Jing Liu
,
Wei Zhu
,
Di Li
,
Xing Hu
,
Liang Song
Cite
Distributional and Spatial-Temporal Robust Representation Learning for Transportation Activity Recognition
This paper presents a parallel model called DSTRR for Transportation Activity Recognition (TAR), which combines automatic learning of statistical, spatial, and temporal features as a robust representation specific to TAR.
Jing Liu
,
Yang Liu
,
Wei Zhu
,
Xiaoguang Zhu
,
Liang Song
Cite
Cite
×