Shape-Based Autotagging of 3D Models for Retrieval

Abstract

Abstract. This paper describes an automatic annotation, or autotagging, algorithm that attaches textual tags to 3D models based on their shape and semantic classes. The proposed method employs Manifold Ranking by Zhou et al, an algorithm that takes into account both local and global distributions of feature points, for tag relevance computation. Using Manifold Ranking, our method propagates multiple tags attached to a training subset of models in a database to the other tag-less models. After the relevance values for multiple tags are computed for tag-less points, the method selects, based on the distribution of feature points for each tag, the threshold at which the tag is selected or discarded for the points. Experimental evaluation of the method using a text-based 3D model retrieval setting showed that the proposed method is effective in autotagging 3D shape models.

Ranking on Cross-Domain Manifold for Sketch-Based 3D Model Retrieval

“…Query-by-3D model has been the most popular modality so far, but a user often does not have a 3D model to be used as a query. A possible alternative is to use a set of keywords as a query [18,19]. But most 3D model lacks text metadata.…”

SimRank similarity preserving projection for shape‐based 3D model auto‐annotation

“…However, their text‐based approach was not able to obtain sufficient search performance because 3D models are not necessarily tagged with keywords expressing their shapes. This research triggered interest in the investigation of the techniques, called shape‐based 3D model auto‐annotation (3DMA) , that aim at predicting the appropriate keywords representing the shape features of a 3D model. In this letter, we propose a new dimensionality reduction method that finds a subspace preserving the relationship between 3D models and keywords.…”