Temporal Attentive Alignment for Video Domain Adaptation

26 May 2019  ·  Min-Hung Chen, Zsolt Kira, Ghassan AlRegib ·

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.

PDF Abstract


Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Domain Adaptation HMDBfull-to-UCF TA3N Accuracy 81.79 # 1
Domain Adaptation HMDBsmall-to-UCF TA3N Accuracy 99.47 # 1
Domain Adaptation Olympic-to-HMDBsmall TA3N Accuracy 92.92 # 1
Domain Adaptation UCF-to-HMDBfull TA3N Accuracy 78.33 # 1
Domain Adaptation UCF-to-HMDBsmall TA3N Accuracy 99.33 # 1
Domain Adaptation UCF-to-Olympic TA3N Accuracy 98.15 # 1


No methods listed for this paper. Add relevant methods here