Normalized Cuts and Image Segmentation
IEEE Transactions on Pattern Analysis and Machine Intelligence
X-means: Extending K-means with Efficient Estimation of the Number of Clusters
ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
The Princeton Shape Benchmark (Figures 1 and 2)
SMI '04 Proceedings of the Shape Modeling International 2004
Automatic Selection and Combination of Descriptors for Effective 3D Similarity Search
ISMSE '04 Proceedings of the IEEE Sixth International Symposium on Multimedia Software Engineering
Feature Combination and Relevance Feedback for 3D Model Retrieval
MMM '05 Proceedings of the 11th International Multimedia Modelling Conference
Regularized query classification using search click information
Pattern Recognition
Hidden annotation for image retrieval with long-term relevance feedback learning
Pattern Recognition
Researches on Semantic Annotation and Retrieval of 3D Models Based on User Feedback
SKG '10 Proceedings of the 2010 Sixth International Conference on Semantics, Knowledge and Grids
Hi-index | 0.00 |
The performance of 3D model retrieval can be greatly improved by adopting precise semantics. Since manual annotation of semantics is too time-consuming, it is necessary to explore an automatic or semi-automatic way. Although it is widely accepted that users' feedbacks contain semantics, previous researches usually utilize relevance feedbacks in computing similarity of 3D models. The paper proposes a strategy for semantics clustering, annotation and retrieval of 3D models, which adopts not only relevance feedbacks but also noisy user operations. The strategy first converts implicit feedbacks into a weighted semantics network of 3D models. After analyzing this semantics network, this paper proposes an agglomerative hierarchical clustering method based on a novel concept of semantics core to obtain the semantics communities under different granularity. Finally, this paper shows an automatic semantics annotation method using the semantics of only a few 3D models. The proposed method is verified by simulated feedbacks with strong noise and real feedbacks of the Princeton Shape Benchmark. Our experiments show that the strategy achieve good performance not only in semantics clustering but also in semantics annotation.