Convolutional Shape-Aware Representation for 3D Object Classification

摘要

Deep learning has recently emerged as one of the most popular and powerful paradigms for learning tasks. In this paper, we present a deep learning approach to 3D shape classification using convolutional neural networks. The proposed framework takes a multi-stage approach that first represents each 3D shape in the dataset as a 2D image using the bag-of-features model in conjunction with intrinsic spatial pyramid matching that leverages the spatial relationship between features. These 2D images are then fed into a pre-trained convolutional neural network to learn deep convolutional shape-aware descriptors from the penultimate fully-connected layer of the network. Finally, a multiclass support vector machine classifier is trained on the deep descriptors, and the classification accuracy is subsequently computed. The effectiveness of our approach is demonstrated on three standard 3D shape benchmarks, yielding higher classification accuracy rates compared to existing methods.