LAUSR

In this paper, we propose a novel method for assessing 3D model representation via CNN and RNN networks. First, a visual tool developed with OpenGL is utilized to extract virtual views of each 3D model from different angles. These views are extracted by 10-degree wrap around the model. Second, a CNN model is used to extract the feature vectors of these virtual images. Then, these feature vectors as the input of an RNN are fused into a new feature to represent the 3D model. Finally, the Euclidean distance is used to obtain the similarity measure between two different models for the retrieval problem. In the experimental section, NTU, PSB and ShapeNet datasets are utilized to evaluate the performance of the proposed method. Several classic 3D model retrieval and classification methods are leveraged as comparison methods in this paper. The corresponding experiments also demonstrate the superiority of our approach.