域適應式平均教師於類別層級之物體位姿估計

Abstract

類別層級之物體位姿估計致力於估測未見過之物體的六自由度位姿，而現有的方法大多倚賴像是物體位姿及 CAD 模型的標記。然而，在真實世界中以人工取得這些標記相當費時且容易出錯。因此，我們提出一種方法來解決類別層級之物體位姿估計中無監督領域自適應的問題。我們採用了一個師生共同學習的架構來同時利用有標記的合成資料與無標記的真實世界資料，學生模型與教師模型被訓練成在不同的干擾下做出一致的預測。此外，我們引入了領域對抗訓練來縮短合成資料與真實世界資料之間的差距，為了避免領域間錯誤的特徵對齊，我們使用了多個領域識別器，以在已知類別的情況下進行特徵對齊。實驗結果顯示我們的方法在公開的真實世界資料集上，取得了無監督方法中最好的結果。透過消融研究，我們也證明了我們的方法不受特定網路架構的限制，並可以作為一種於類別層級物體位姿估計的通用無監督領域自適應方法。

Category-level object pose estimation aims at predicting 6-DoF object poses for previously unseen objects. Current methods mostly rely on ground-truth labels such as object poses and CAD models. However, manually annotating these labels is time-consuming and error-prone in the real-world scenario. Hence, we propose a method to solve unsupervised domain adaptation (UDA) for category-level object pose estimation. We adopt a teacher-student joint learning framework to utilize both labeled synthetic data and unlabeled real-world data. The student model and the teacher model are trained to make consistent predictions under different perturbations. Furthermore, we introduce domain adversarial training to bridge the domain gap between synthetic and real-world data. To prevent false feature alignment between domains, we adopt multiple domain discriminators instead of a single one and perform category-aware alignments. Extensive experiments show that our method achieves state-of-the-art results among unsupervised methods on a public real-world dataset. Through ablation studies, we also demonstrate that our method is not restricted to certain network architectures and can serve as a general UDA method for category-level object pose estimation.