TY - GEN
T1 - Foundation Model-Powered 3D Few-Shot Class Incremental Learning via Training-Free Adaptor
AU - Ahmadi, Sahar
AU - Cheraghian, Ali
AU - Saberi, Morteza
AU - Abir, Md Towsif
AU - Dastmalchi, Hamidreza
AU - Hussain, Farookh
AU - Rahman, Shafin
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
PY - 2025
Y1 - 2025
N2 - Recent advances in deep learning for processing point clouds hold increased interest in Few-Shot Class Incremental Learning (FSCIL) for 3D computer vision. This paper introduces a new method to tackle the Few-Shot Continual Incremental Learning (FSCIL) problem in 3D point cloud environments. We leverage a foundational 3D model trained extensively on point cloud data. Drawing from recent improvements in foundation models, known for their ability to work well across different tasks, we propose a novel strategy that does not require additional training to adapt to new tasks. Our approach uses a dual cache system: first, it uses previous test samples based on how confident the model was in its predictions to prevent forgetting, and second, it includes a small number of new task samples to prevent overfitting. This dynamic adaptation ensures strong performance across different learning tasks without needing lots of fine-tuning. We tested our approach on datasets like ModelNet, ShapeNet, ScanObjectNN, and CO3D, showing that it outperforms other FSCIL methods and demonstrating its effectiveness and versatility. The code is available at https://github.com/ahmadisahar/ACCV_FCIL3D.
AB - Recent advances in deep learning for processing point clouds hold increased interest in Few-Shot Class Incremental Learning (FSCIL) for 3D computer vision. This paper introduces a new method to tackle the Few-Shot Continual Incremental Learning (FSCIL) problem in 3D point cloud environments. We leverage a foundational 3D model trained extensively on point cloud data. Drawing from recent improvements in foundation models, known for their ability to work well across different tasks, we propose a novel strategy that does not require additional training to adapt to new tasks. Our approach uses a dual cache system: first, it uses previous test samples based on how confident the model was in its predictions to prevent forgetting, and second, it includes a small number of new task samples to prevent overfitting. This dynamic adaptation ensures strong performance across different learning tasks without needing lots of fine-tuning. We tested our approach on datasets like ModelNet, ShapeNet, ScanObjectNN, and CO3D, showing that it outperforms other FSCIL methods and demonstrating its effectiveness and versatility. The code is available at https://github.com/ahmadisahar/ACCV_FCIL3D.
KW - 3D point cloud
KW - Few-shot learning
KW - Incremental learning
UR - http://www.scopus.com/inward/record.url?scp=85213344031&partnerID=8YFLogxK
U2 - 10.1007/978-981-96-0972-7_11
DO - 10.1007/978-981-96-0972-7_11
M3 - Conference contribution
AN - SCOPUS:85213344031
SN - 9789819609710
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 178
EP - 195
BT - Computer Vision – ACCV 2024 - 17th Asian Conference on Computer Vision, Proceedings
A2 - Cho, Minsu
A2 - Laptev, Ivan
A2 - Tran, Du
A2 - Yao, Angela
A2 - Zha, Hongbin
PB - Springer Science and Business Media Deutschland GmbH
T2 - 17th Asian Conference on Computer Vision, ACCV 2024
Y2 - 8 December 2024 through 12 December 2024
ER -