Abstract
The rapid development of 3D technique has led to the dramatic increase in 3D data. The scalable and effective 3D object retrieval and classification algorithms become mandatory for large-scale 3D object management. One critical problem of view-based 3D object retrieval and classification is how to exploit the relevance and discrimination among multiple views. In this paper, we propose a multi-view hierarchical fusion network (MVHFN) for these two tasks. This method mainly contains two key modules. First, the module of visual feature learning applies the 2D CNNs to extract the visual feature of multiple views rendered around the specific 3D object. Then, the multi-view hierarchical fusion module we proposed is employed to fuse the multiple view features into a compact descriptor. This module can not only fully exploit the relevance among multiple views by intra-cluster multi-view fusion mechanism, but also discover the content discrimination by inter-cluster multi-view fusion mechanism. Experimental results on two public datasets, i.e., ModelNet40 and ShapeNetCore55, show that our proposed MVHFN outperforms the current state-of-the-art methods in both the 3D object retrieval and classification tasks.
Highlights
With the explosive growth of 3D data, the rapid development of 3D reconstruction technology and the wide use of 3D equipment, the importance of 3D object retrieval and classification has been increasing in recent years [1]–[3]. 3D objects have been widely applied in medical diagnosis, intelligent robot, self-driving car and some other fields
Experimental results on ModelNet40 and ShapeNetCore55 show that our proposed multi-view hierarchical fusion network (MVHFN) outperforms the current state-of-the-art methods in both the 3D object retrieval and classification tasks
We propose a novel method named MVHFN for 3D object retrieval and classification, which can hierarchically and adaptively fuse the multi-view information according to the intrinsic correlation and discrimination among multiple views
Summary
With the explosive growth of 3D data, the rapid development of 3D reconstruction technology and the wide use of 3D equipment, the importance of 3D object retrieval and classification has been increasing in recent years [1]–[3]. 3D objects have been widely applied in medical diagnosis, intelligent robot, self-driving car and some other fields. Liu et al.: MVHFN for 3D Object Retrieval and Classification views with the maximum values and may neglect some valuable information Another way is to develop a strategy to determine the importance of each view and fuse the multiview features with different weights. Several methods [5] exploit a LSTM-based attention mechanism to aggregate view features into a 3D object descriptor Such methods are superior to MVCNN in the 3D object retrieval and classification tasks, but are not robust enough to determine the weight of each view. To overcome this problem, we develop a hierarchical and adaptive strategy to firstly split the views into different clusters according to their discrimination values, and appropriately fuse multi-view features. Experimental results on ModelNet and ShapeNetCore show that our proposed MVHFN outperforms the current state-of-the-art methods in both the 3D object retrieval and classification tasks
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
