From Synthetic to Real: Unsupervised Domain Adaptation for Animal Pose Estimation

By Chen Li, Gim Hee Lee. 2021

100 citations recorded by the source corpus.

Animal pose estimation is an important field that has received increasing attention in the recent years. The main challenge for this task is the lack of labeled data. Existing works circumvent this problem with pseudo labels generated from data of other easily accessible domains such as synthetic data. However, these pseudo labels are noisy even with consistency check or confidence-based filtering due to the domain shift in the data. To solve this problem, we design a multi-scale domain adaptation module (MDAM) to reduce the domain gap between the synthetic and real data. We further introduce an online coarse-to-fine pseudo label updating strategy. Specifically, we propose a self-distillation module in an inner coarse-update loop and a mean-teacher in an outer fine-update loop to generate new pseudo labels that gradually replace the old ones. Consequently, our model is able to learn from the old pseudo labels at the early stage, and gradually switch to the new pseudo labels to prevent overfitting in the later stage. We evaluate our approach on the TigDog and VisDA 2019 datasets, where we outperform existing approaches by a large margin. We also demonstrate the generalization ability of our model by testing extensively on both unseen domains and unseen animal categories. Our code is available at the project website1.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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. Mingkui Tan
2. Shuaicheng Niu
3. Yifan Zhang
4. Chuang Gan
5. Zhen Hu
6. Sankaran Mahadevan
7. Hongbin Lin
8. Zhen Qiu

In Computer Vision and Pattern Recognition

1. Mingkui Tan
2. Shuaicheng Niu
3. Yifan Zhang
4. Chuang Gan
5. Hongbin Lin
6. Zhen Qiu
7. Yanxia Liu
8. Byeng Dong Youn

Citations in the corpus

• Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation (2022, methodology): “Due to the integration of these challenges, existing UDA methods [10,14,23,34,44,52] are incapable of handling CI-UDA.”
• A comprehensive review of digital twin — part 1: modeling and twinning enabling technologies (2022, background): “Further to the two scenarios of domain shifts caused by differences in operating conditions or by differences between different units of a fleet, a third scenario, particularly relevant to digital twins is the synthetic to real gap existing between synthetically generated data stemming potentially f”