Abstract

Point cloud classification is a popular task in 3D vision. However, previous works, usually assume that point clouds at test time are obtained with the same procedure or sensor as those at training time. Unsupervised Domain Adaptation (UDA) instead, breaks this assumption and tries to solve the task on an unlabeled target domain, leveraging only on a supervised source domain. For point cloud classification, recent UDA methods try to align features across domains via auxiliary tasks such as point cloud reconstruction, which however do not optimize the discriminative power in the target domain in feature space. In contrast, in this work, we focus on obtaining a discriminative feature space for the target domain enforcing consistency between a point cloud and its augmented version. We then propose a novel iterative self-training methodology that exploits Graph Neural Networks in the UDA context to refine pseudo-labels. We perform extensive experiments and set the new state-of-the art in standard UDA benchmarks for point cloud classification. Finally, we show how our approach can be extended to more complex tasks such as part segmentation.

Method overview

Illustration of our framework:
Left: weakly and strongly augmented point clouds are generated with two transformation functions for both domains. The weakly augmented shapes are fed to an exponential moving average (EMA) encoder, the teacher, while the strongly augmented are processed by the student. A consistency loss is applied between the corresponding embeddings.
Right: the whole target dataset is processed by a GCN online during self-training to iteratively refine and update.

Citation

@inproceedings{cardace2023selfdistill,
    title={Self-Distillation for Unsupervised 3D Domain Adaptation},
    author={Adriano Cardace and Riccardo Spezialetti and Pierluigi Zama Ramirez and Samuele Salti and Luigi Di Stefano},
    booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
    year = {2023},
}