Temporal Attentive Alignment for Large-Scale Video Domain Adaptation

By Min-Hung Chen, Zsolt Kira, Ghassan Al-Regib, Jaekwon Yoo, Ruxin Chen, Jian Zheng. 2019

209 citations recorded by the source corpus.

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 two large-scale video DA datasets with much larger domain discrepancy: UCF-HMDB_full and Kinetics-Gameplay. 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 four video DA datasets (e.g. 7.9% accuracy gain over “Source only” from 73.9% to 81.8% on “HMDB --> UCF”, and 10.3% gain on “Kinetics --> Gameplay”). The code and data are released at http://github.com/cmhungsteve/TA3N.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Dima Damen
2. Jonathan Munro
3. Barbara Caputo
4. Giovanni Maria Farinella
5. Antonino Furnari
6. Chiara Plizzari
7. Emanuele Alberti
8. Mirco Planamente

In Computer Vision and Pattern Recognition

1. Kristen Grauman
2. Luc Van Gool
3. Giovanni Maria Farinella
4. Antonino Furnari
5. Jitendra Malik
6. Dima Damen
7. Andrew Zisserman
8. Cordelia Schmid

In Representation & Self-Supervised Learning

1. Giovanni Maria Farinella
2. Antonino Furnari
3. Barbara Caputo
4. Emanuele Alberti
5. Chiara Plizzari
6. Mirco Planamente
7. Dima Damen
8. Jonathan Munro

Citations in the corpus

• Rescaling Egocentric Vision: Collection, Pipeline and Challenges for EPIC-KITCHENS-100 (2020, methodology): “8 2019 Real IEMOCAP→ AFEW [92] 6611 795 N/A 4 2 2 N/A 2018 Real Kinetics↔ Gameplay [91] 43378 2625 749 30 2 1 N/A 2019 Real/Syn EPIC-KITCHENS-100 16115 26115 5909 3369 16 3 2.”
• Multi-Modal Domain Adaptation for Fine-Grained Action Recognition (2019, methodology): “However, very few works have attempted deep UDA for video data [7, 19].”
• Multi-Modal Domain Adaptation for Fine-Grained Action Recognition (2020, methodology): “However, very few works have attempted deep UDA for video data [7, 19].”
• Team VI-I2R Technical Report on EPIC-KITCHENS-100 Unsupervised Domain Adaptation Challenge for Action Recognition 2021 (2022, methodology): “Subsequently, we train the modified TA3N [1] for domain-adaptive action recognition.”
• Cross-Domain First Person Audio-Visual Action Recognition through Relative Norm Alignment (2021, background): “Only few works analyzed UDA for video understanding [10, 41, 11, 25].”
• Omni-sourced Webly-supervised Learning for Video Recognition (2020, background): “Previous efforts [7,46,4] in domain adaptation focus on mitigating the data shift [35] in terms of data distributions.”
• Adversarial Cross-Domain Action Recognition with Co-Attention (2019, methodology): “Previous cross-domain action recognition meth-ods directly match segment feature distributions (Jamal et al. 2018) or with weights based on an attention mechanism (Chen et al. 2019).”