BROWSE
Detail
Conference Paper
Part-Aware Embedding for Fine-Grained 3D Shape Retrieval
Cited 0 time in 
Share 
Share
- Authors
- Suwoong Lee, Da-un Jung, Juwon Lee, Seungjae Lee, Junmo Kim
- Issue Date
- 2025-12
- Citation
- ACM SIGGRAPH Asia (SA) 2025, pp.1-3
- Language
- English
- Type
- Conference Paper
- Abstract
- We propose a part-aware retrieval framework for 3D shapes that enables fine-grained similarity analysis by embedding both whole objects and their constituent parts into a unified feature space. Unlike previous methods that rely on global object-level embeddings and often overlook locally distinctive structures, our approach incorporates part segmentation and part-level embeddings to enhance retrieval accuracy, particularly for queries targeting specific components. Evaluations on a custom dataset derived from Objaverse V1 demonstrate that our method outperforms whole-object-only retrieval in both quantitative metrics and qualitative results.
- KSP Keywords
- 3D shape retrieval, Feature space, Fine grained(FG), Object-level, Part-aware, Quantitative Metrics, Unified feature, retrieval accuracy, similarity analysis, specific components