PU-EdgeFormer: Edge Transformer for Dense Prediction in Point Cloud Upsampling
This work addresses a problem in computer vision for 3D data processing, offering an incremental improvement over prior methods.
The paper tackles the challenge of simultaneously training local and global structures in point cloud upsampling by proposing PU-EdgeFormer, which combines graph convolution and transformer modules, and reports better performance than existing state-of-the-art methods in both subjective and objective aspects.
Despite the recent development of deep learning-based point cloud upsampling, most MLP-based point cloud upsampling methods have limitations in that it is difficult to train the local and global structure of the point cloud at the same time. To solve this problem, we present a combined graph convolution and transformer for point cloud upsampling, denoted by PU-EdgeFormer. The proposed method constructs EdgeFormer unit that consists of graph convolution and multi-head self-attention modules. We employ graph convolution using EdgeConv, which learns the local geometry and global structure of point cloud better than existing point-to-feature method. Through in-depth experiments, we confirmed that the proposed method has better point cloud upsampling performance than the existing state-of-the-art method in both subjective and objective aspects. The code is available at https://github.com/dohoon2045/PU-EdgeFormer.