Temporal Attentive Alignment for Video Domain Adaptation
It addresses domain shift in videos, a less explored area compared to images, with incremental improvements in dataset scale and method integration.
The paper tackles video domain adaptation by proposing a new dataset, UCF-HMDB_full, and a method called Temporal Attentive Adversarial Adaptation Network (TA3N) that aligns temporal dynamics, achieving state-of-the-art performance on three datasets.
Although various image-based domain adaptation (DA) techniques have been proposed in recent years, domain shift in videos is still not well-explored. Most previous works only evaluate performance on small-scale datasets which are saturated. Therefore, we first propose a larger-scale dataset with larger domain discrepancy: UCF-HMDB_full. Second, we investigate different DA integration methods for videos, and show that simultaneously aligning and learning temporal dynamics achieves effective alignment even without sophisticated DA methods. Finally, we propose Temporal Attentive Adversarial Adaptation Network (TA3N), which explicitly attends to the temporal dynamics using domain discrepancy for more effective domain alignment, achieving state-of-the-art performance on three video DA datasets. The code and data are released at http://github.com/cmhungsteve/TA3N.