Nearest Neighbors Meet Deep Neural Networks for Point Cloud Analysis
This work addresses the problem of oversized models and complex designs in point cloud analysis for researchers and practitioners, offering an incremental improvement through a non-parametric adapter.
The paper tackles the plateau in performance on 3D point cloud benchmarks by introducing the Spatial-Neighbor Adapter (SN-Adapter), a plug-and-play module that enhances existing deep neural networks without redesign or extra parameters, achieving improved performance across tasks like shape classification, part segmentation, and 3D object detection.
Performances on standard 3D point cloud benchmarks have plateaued, resulting in oversized models and complex network design to make a fractional improvement. We present an alternative to enhance existing deep neural networks without any redesigning or extra parameters, termed as Spatial-Neighbor Adapter (SN-Adapter). Building on any trained 3D network, we utilize its learned encoding capability to extract features of the training dataset and summarize them as prototypical spatial knowledge. For a test point cloud, the SN-Adapter retrieves k nearest neighbors (k-NN) from the pre-constructed spatial prototypes and linearly interpolates the k-NN prediction with that of the original 3D network. By providing complementary characteristics, the proposed SN-Adapter serves as a plug-and-play module to economically improve performance in a non-parametric manner. More importantly, our SN-Adapter can be effectively generalized to various 3D tasks, including shape classification, part segmentation, and 3D object detection, demonstrating its superiority and robustness. We hope our approach could show a new perspective for point cloud analysis and facilitate future research.