Geometric Feature Embedding for Effective 3D Few-Shot Class Incremental Learning

Xiangqi Li, Libo Huang, Zhulin An, Weilun Feng, Chuanguang Yang, Boyu Diao, Fei Wang, Yongjun Xu
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:34674-34687, 2025.

Abstract

3D few-shot class incremental learning (FSCIL) aims to learn new point cloud categories from limited samples while preventing the forgetting of previously learned categories. This research area significantly enhances the capabilities of self-driving vehicles and computer vision systems. Existing 3D FSCIL approaches primarily utilize multimodal pre-trained models to extract the semantic features, heavily dependent on meticulously designed high-quality prompts and fine-tuning strategies. To reduce this dependence, this paper proposes a novel method for 3D FSCIL with Embedded Geometric features (3D-FLEG). Specifically, 3D-FLEG develops a point cloud geometric feature extraction module to capture category-related geometric characteristics. To address the modality heterogeneity issues that arise from integrating geometric and text features, 3D-FLEG introduces a geometric feature embedding module. By augmenting text prompts with spatial geometric features through these modules, 3D-FLEG can learn robust representations of new categories even with limited samples, while mitigating forgetting of the previously learned categories. Experiments conducted on several publicly available 3D point cloud datasets, including ModelNet, ShapeNet, ScanObjectNN, and CO3D, demonstrate 3D-FLEG’s superiority over existing state-of-the-art 3D FSCIL methods. Code is available at https://github.com/lixiangqi707/3D-FLEG.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-li25ad, title = {Geometric Feature Embedding for Effective 3{D} Few-Shot Class Incremental Learning}, author = {Li, Xiangqi and Huang, Libo and An, Zhulin and Feng, Weilun and Yang, Chuanguang and Diao, Boyu and Wang, Fei and Xu, Yongjun}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {34674--34687}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/li25ad/li25ad.pdf}, url = {https://proceedings.mlr.press/v267/li25ad.html}, abstract = {3D few-shot class incremental learning (FSCIL) aims to learn new point cloud categories from limited samples while preventing the forgetting of previously learned categories. This research area significantly enhances the capabilities of self-driving vehicles and computer vision systems. Existing 3D FSCIL approaches primarily utilize multimodal pre-trained models to extract the semantic features, heavily dependent on meticulously designed high-quality prompts and fine-tuning strategies. To reduce this dependence, this paper proposes a novel method for 3D FSCIL with Embedded Geometric features (3D-FLEG). Specifically, 3D-FLEG develops a point cloud geometric feature extraction module to capture category-related geometric characteristics. To address the modality heterogeneity issues that arise from integrating geometric and text features, 3D-FLEG introduces a geometric feature embedding module. By augmenting text prompts with spatial geometric features through these modules, 3D-FLEG can learn robust representations of new categories even with limited samples, while mitigating forgetting of the previously learned categories. Experiments conducted on several publicly available 3D point cloud datasets, including ModelNet, ShapeNet, ScanObjectNN, and CO3D, demonstrate 3D-FLEG’s superiority over existing state-of-the-art 3D FSCIL methods. Code is available at https://github.com/lixiangqi707/3D-FLEG.} }
Endnote
%0 Conference Paper %T Geometric Feature Embedding for Effective 3D Few-Shot Class Incremental Learning %A Xiangqi Li %A Libo Huang %A Zhulin An %A Weilun Feng %A Chuanguang Yang %A Boyu Diao %A Fei Wang %A Yongjun Xu %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-li25ad %I PMLR %P 34674--34687 %U https://proceedings.mlr.press/v267/li25ad.html %V 267 %X 3D few-shot class incremental learning (FSCIL) aims to learn new point cloud categories from limited samples while preventing the forgetting of previously learned categories. This research area significantly enhances the capabilities of self-driving vehicles and computer vision systems. Existing 3D FSCIL approaches primarily utilize multimodal pre-trained models to extract the semantic features, heavily dependent on meticulously designed high-quality prompts and fine-tuning strategies. To reduce this dependence, this paper proposes a novel method for 3D FSCIL with Embedded Geometric features (3D-FLEG). Specifically, 3D-FLEG develops a point cloud geometric feature extraction module to capture category-related geometric characteristics. To address the modality heterogeneity issues that arise from integrating geometric and text features, 3D-FLEG introduces a geometric feature embedding module. By augmenting text prompts with spatial geometric features through these modules, 3D-FLEG can learn robust representations of new categories even with limited samples, while mitigating forgetting of the previously learned categories. Experiments conducted on several publicly available 3D point cloud datasets, including ModelNet, ShapeNet, ScanObjectNN, and CO3D, demonstrate 3D-FLEG’s superiority over existing state-of-the-art 3D FSCIL methods. Code is available at https://github.com/lixiangqi707/3D-FLEG.
APA
Li, X., Huang, L., An, Z., Feng, W., Yang, C., Diao, B., Wang, F. & Xu, Y.. (2025). Geometric Feature Embedding for Effective 3D Few-Shot Class Incremental Learning. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:34674-34687 Available from https://proceedings.mlr.press/v267/li25ad.html.

Related Material