Learning Domain Adaptive Object Detection with Probabilistic Teacher

Meilin Chen, Weijie Chen, Shicai Yang, Jie Song, Xinchao Wang, Lei Zhang, Yunfeng Yan, Donglian Qi, Yueting Zhuang, Di Xie, Shiliang Pu
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:3040-3055, 2022.

Abstract

Self-training for unsupervised domain adaptive object detection is a challenging task, of which the performance depends heavily on the quality of pseudo boxes. Despite the promising results, prior works have largely overlooked the uncertainty of pseudo boxes during self-training. In this paper, we present a simple yet effective framework, termed as Probabilistic Teacher (PT), which aims to capture the uncertainty of unlabeled target data from a gradually evolving teacher and guides the learning of a student in a mutually beneficial manner. Specifically, we propose to leverage the uncertainty-guided consistency training to promote classification adaptation and localization adaptation, rather than filtering pseudo boxes via an elaborate confidence threshold. In addition, we conduct anchor adaptation in parallel with localization adaptation, since anchor can be regarded as a learnable parameter. Together with this framework, we also present a novel Entropy Focal Loss (EFL) to further facilitate the uncertainty-guided self-training. Equipped with EFL, PT outperforms all previous baselines by a large margin and achieve new state-of-the-arts.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-chen22b, title = {Learning Domain Adaptive Object Detection with Probabilistic Teacher}, author = {Chen, Meilin and Chen, Weijie and Yang, Shicai and Song, Jie and Wang, Xinchao and Zhang, Lei and Yan, Yunfeng and Qi, Donglian and Zhuang, Yueting and Xie, Di and Pu, Shiliang}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {3040--3055}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/chen22b/chen22b.pdf}, url = {https://proceedings.mlr.press/v162/chen22b.html}, abstract = {Self-training for unsupervised domain adaptive object detection is a challenging task, of which the performance depends heavily on the quality of pseudo boxes. Despite the promising results, prior works have largely overlooked the uncertainty of pseudo boxes during self-training. In this paper, we present a simple yet effective framework, termed as Probabilistic Teacher (PT), which aims to capture the uncertainty of unlabeled target data from a gradually evolving teacher and guides the learning of a student in a mutually beneficial manner. Specifically, we propose to leverage the uncertainty-guided consistency training to promote classification adaptation and localization adaptation, rather than filtering pseudo boxes via an elaborate confidence threshold. In addition, we conduct anchor adaptation in parallel with localization adaptation, since anchor can be regarded as a learnable parameter. Together with this framework, we also present a novel Entropy Focal Loss (EFL) to further facilitate the uncertainty-guided self-training. Equipped with EFL, PT outperforms all previous baselines by a large margin and achieve new state-of-the-arts.} }
Endnote
%0 Conference Paper %T Learning Domain Adaptive Object Detection with Probabilistic Teacher %A Meilin Chen %A Weijie Chen %A Shicai Yang %A Jie Song %A Xinchao Wang %A Lei Zhang %A Yunfeng Yan %A Donglian Qi %A Yueting Zhuang %A Di Xie %A Shiliang Pu %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-chen22b %I PMLR %P 3040--3055 %U https://proceedings.mlr.press/v162/chen22b.html %V 162 %X Self-training for unsupervised domain adaptive object detection is a challenging task, of which the performance depends heavily on the quality of pseudo boxes. Despite the promising results, prior works have largely overlooked the uncertainty of pseudo boxes during self-training. In this paper, we present a simple yet effective framework, termed as Probabilistic Teacher (PT), which aims to capture the uncertainty of unlabeled target data from a gradually evolving teacher and guides the learning of a student in a mutually beneficial manner. Specifically, we propose to leverage the uncertainty-guided consistency training to promote classification adaptation and localization adaptation, rather than filtering pseudo boxes via an elaborate confidence threshold. In addition, we conduct anchor adaptation in parallel with localization adaptation, since anchor can be regarded as a learnable parameter. Together with this framework, we also present a novel Entropy Focal Loss (EFL) to further facilitate the uncertainty-guided self-training. Equipped with EFL, PT outperforms all previous baselines by a large margin and achieve new state-of-the-arts.
APA
Chen, M., Chen, W., Yang, S., Song, J., Wang, X., Zhang, L., Yan, Y., Qi, D., Zhuang, Y., Xie, D. & Pu, S.. (2022). Learning Domain Adaptive Object Detection with Probabilistic Teacher. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:3040-3055 Available from https://proceedings.mlr.press/v162/chen22b.html.

Related Material