DAPP: double adversarial point cloud prediction
Year: 2025
Type: article
Abstract: 3D point cloud completion plays a crucial role in real-world scene understanding. With the increasing popularity of depth cameras, point clouds now carry high-resolution depth and structural information of objects, and semantic voxels can be extracted by further learning of the point cloud. On the other hand, adversarial Networks have unparalleled advantages in generating tasks, and are often utilized in image style transfer, music generating, image inpainting task, etc. In this paper, we introduce a new approach to reconstruct the missing parts of an incomplete point cloud. Our model, called DAPP (Double Adversarial Point-cloud Prediction), employs two sets of adversarial neural network structures. Sharing the same generator, the two discriminators in DAPP work together to guide the point cloud generation in different resolutions. We further propose a feature distribution block to extract distributed region features to generate seed predict points, with an improved 3D Transformers to understand the geometry shapes and a 3D Expansion Unit to bring diversity in the generation of precise structures. We also employ PAConv blocks with the Euclidean distances between 3D coordinates in the network to guide the generation of point clouds with different densities, and at the same time, multi-dimensional Fourier transforms are used in multiple places on the network to remind high-dimensional features of the overall point cloud shape where they are located. Combining the above modules, the DAPP learns 3D shapes well and is able to understand long-term semantic geometry relationships. With the discriminators designed differently, the network can focus on different features of the input point cloud in the coarse prediction and fine prediction phases. Furthermore, we introduce a new dataset called MultiView57, which better simulates real-world scenarios when using depth cameras to detect objects. The results show that DAPP outperforms existing methods on both our dataset and commonly used ones.
Source: Optics Continuum
Authors: Shuyu Sun, Jianqiang Huang, Shuai Zhao, Tengchao Huang
Institutions: Zhejiang University, Shanghai Advanced Research Institute
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,372
OpenAlex ID: W4411256670