一种半监督三维模型语义自动标注方法

Method of semi-supervised 3D objects semantic automatic annotation

三维模型语义自动标注的目标是自动给出最适合描述模型的标注词集合,是基于文本的三维模型检索的重要环节。语义鸿沟的存在使得相似匹配技术得到的标注效果有待提高。为了在用户提供的有限模型数量和对应的标注词信息下,在自动标注过程中利用大量的未标注样本改善三维模型的标注性能,提出了一种半监督测度学习方法完成三维模型语义自动标注。该方法首先使用基于图的半监督学习方法扩展已标注模型集合,并给出扩展集合中语义标签表征模型的语义置信度,使用改进的相关成分分析方法学习马氏距离度量,依据学习到的距离和语义置信度形成多语义标注策略。在PSB(Princeton Shape Benchmark)数据集上的测试表明,该方法利用了大量未标注样本参与标注过程,取得了比较好的标注效果。

In order to improve the 3D model annotation performance using a large number of unlabeled samples, this paper pro- poses a semi-supervised measure learning method to realize the 3D models multiple semantic annotation. A graph-based semi-su- pervised learning is firstly used to expand the training set, and the semantic words confidence of the models in the extension set is proposed. An improved relevant component analysis method is introduced in this paper to learn a distance measure based on the extended training set. The approach is introduced to complete multiple semantic annotation task based on the learned dis- tance measure and label confidence. The test result on the PSB data set has shown that the method making use of the unlabeled samples has achieved a better annotation result when a small amount of labels are given.