Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understanding
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2024-12 · 46 (12) · 8517-8533
DOI: 10.1109/tpami.2024.3410324 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024-12. Verfasst von Runyu Ding und 5 weiteren Autor:innen. Die Arbeit wurde bisher 40-mal zitiert (Crossref) und nennt 69 Literaturangaben. Gefördert von: Hong Kong Research Grant Council, General Research Fund Scheme, RGC Theme-based research.
Abstract (Originalsprache)
Open-world instance-level scene understanding aims to locate and recognize unseen object categories that are not present in the annotated dataset. This task is challenging because the model needs to both localize novel 3D objects and infer their semantic categories. A key factor for the recent progress in 2D open-world perception is the availability of large-scale image-text pairs from the Internet, which cover a wide range of vocabulary concepts. However, this success is hard to replicate in 3D scenarios due to the scarcity of 3D-text pairs. To address this challenge, we propose to harness pre-trained vision-language (VL) foundation models that encode extensive knowledge from image-text pairs to generate captions for multi-view images of 3D scenes. This allows us to establish explicit associations between 3D shapes and semantic-rich captions. Moreover, to enhance the fine-grained visual-semantic representation learning from captions for object-level categorization, we design hierarchical point-caption association methods to learn semantic-aware embeddings that exploit the 3D geometry between 3D points and multi-view images. In addition, to tackle the localization challenge for novel classes in the open-world setting, we develop debiased instance localization, which involves training object grouping modules on unlabeled data using instance-level pseudo supervision. This significantly improves the generalization capabilities of instance grouping and, thus, the ability to accurately locate novel objects. We conduct extensive experiments on 3D semantic, instance, and panoptic segmentation tasks, covering indoor and outdoor scenes across three datasets. Our method outperforms baseline methods by a significant margin in semantic segmentation (e.g. 34.5% ∼ 65.3%), instance segmentation (e.g. 21.8% ∼ 54.0%), and panoptic segmentation (e.g. 14.7% ∼ 43.3%). Code will be available.
Förderung
- Hong Kong Research Grant Council
- General Research Fund Scheme
- RGC Theme-based research
- RGC Matching Fund Scheme
- JC STEM Lab of Robotics for Soft Materials
- The Hong Kong Jockey Club Charities Trust
Zitieren
APA
Ding, R., Yang, J., Xue, C., Zhang, W., Bai, S., & Qi, X. (2024). Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12), 8517-8533. https://doi.org/10.1109/tpami.2024.3410324
Vancouver
Ding R, Yang J, Xue C, Zhang W, Bai S, Qi X. Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2024;46(12):8517-8533. doi:10.1109/tpami.2024.3410324
BibTeX
@article{Ding2024,
author = {Ding, Runyu and Yang, Jihan and Xue, Chuhui and Zhang, Wenqing and Bai, Song and Qi, Xiaojuan},
title = {Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understanding},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2024},
volume = {46},
number = {12},
pages = {8517--8533},
doi = {10.1109/tpami.2024.3410324}
}
EUVIMED