Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Unsupervised Video Domain Adaptation for Action Recognition: A Disentanglement Perspective

About

Unsupervised video domain adaptation is a practical yet challenging task. In this work, for the first time, we tackle it from a disentanglement view. Our key idea is to handle the spatial and temporal domain divergence separately through disentanglement. Specifically, we consider the generation of cross-domain videos from two sets of latent factors, one encoding the static information and another encoding the dynamic information. A Transfer Sequential VAE (TranSVAE) framework is then developed to model such generation. To better serve for adaptation, we propose several objectives to constrain the latent factors. With these constraints, the spatial divergence can be readily removed by disentangling the static domain-specific information out, and the temporal divergence is further reduced from both frame- and video-levels through adversarial learning. Extensive experiments on the UCF-HMDB, Jester, and Epic-Kitchens datasets verify the effectiveness and superiority of TranSVAE compared with several state-of-the-art approaches. Code is publicly available.

Pengfei Wei, Lingdong Kong, Xinghua Qu, Yi Ren, Zhiqiang Xu, Jing Jiang, Xiang Yin• 2022

Related benchmarks

TaskDatasetResultRank
Action RecognitionUCF-HMDB
Accuracy (UCF -> HMDB)87.78
46
Open-set Unsupervised Video Domain AdaptationEpic-Kitchens
Average Performance52.6
31
Unsupervised Domain AdaptationUCF-HMDB
Accuracy (U -> H)87.8
9
Action RecognitionJester Js -> Jt
Top-1 Acc66.1
7
Action RecognitionEpic-Kitchens D1 -> D2
Top-1 Acc50.5
7
Action RecognitionEpic-Kitchens D1 -> D3
Top-1 Acc50.3
7
Action RecognitionEpic-Kitchens D2 -> D1
Top-1 Acc50.3
7
Action RecognitionEpic-Kitchens D2 -> D3
Top-1 Accuracy58.6
7
Action RecognitionEpic-Kitchens D3 -> D1
Top-1 Accuracy48
7
Action RecognitionEpic-Kitchens D3 -> D2
Top-1 Accuracy58
7
Showing 10 of 10 rows

Other info

Code

Follow for update