A Boosting Method Based on SVM for Relevance Feedback in Content-Based 3D Model Retrieval

Tao Wei,Zheng Qin,Xiaoman Cao,Biao Leng
2010-01-01
Abstract:The technique of relevance feedback has been introduced to content-based 3D model retrieval. Support Vector Machine as a learner is one of the classical approaches in relevance feedback. And the Boosting method, as one of the ensemble methods, can establish a strong leaner by combing the component learners. In this paper, a novel relevance feedback mechanism, which makes use of the main idea of boosting and the component SVM, is presented and applied to the content-based 3D model retrieval. The experiments, based on the 3D model database Princeton Shape Benchmark, show that the relevance feedback algorithm can improve the retrieval performance of traditional SVM in 3D model retrieval.
What problem does this paper attempt to address?