PointAtrousGraph: Deep Hierarchical Encoder-Decoder with Point Atrous Convolution for Unorganized 3D Points
This work addresses the challenge of 3D semantic perception in point clouds, which is crucial for fields like robotics and autonomous driving, but it appears incremental as it builds on existing multi-scale encoding ideas from image analysis.
The paper tackles the problem of efficiently exploiting multi-scale edge features in unorganized 3D point clouds for semantic perception, proposing the PointAtrousGraph (PAG) with novel modules like Point Atrous Convolution, and it outperforms previous state-of-the-art methods on various 3D applications.
Motivated by the success of encoding multi-scale contextual information for image analysis, we propose our PointAtrousGraph (PAG) - a deep permutation-invariant hierarchical encoder-decoder for efficiently exploiting multi-scale edge features in point clouds. Our PAG is constructed by several novel modules, such as Point Atrous Convolution (PAC), Edge-preserved Pooling (EP) and Edge-preserved Unpooling (EU). Similar with atrous convolution, our PAC can effectively enlarge receptive fields of filters and thus densely learn multi-scale point features. Following the idea of non-overlapping max-pooling operations, we propose our EP to preserve critical edge features during subsampling. Correspondingly, our EU modules gradually recover spatial information for edge features. In addition, we introduce chained skip subsampling/upsampling modules that directly propagate edge features to the final stage. Particularly, our proposed auxiliary loss functions can further improve our performance. Experimental results show that our PAG outperform previous state-of-the-art methods on various 3D semantic perception applications.