NeuralBF: Neural Bilateral Filtering for Top-down Instance Segmentation on Point Clouds
This addresses a critical bottleneck in 3D instance segmentation for applications like robotics and autonomous driving, though it appears incremental as it builds on bilateral filtering concepts.
The paper tackles inaccurate instance proposal generation in 3D point clouds by introducing NeuralBF, an iterative bilateral filtering method with learned kernels that considers both deep feature embeddings and spatial locations. It achieves state-of-the-art instance segmentation performance on the ScanNet benchmark among top-down methods.
We introduce a method for instance proposal generation for 3D point clouds. Existing techniques typically directly regress proposals in a single feed-forward step, leading to inaccurate estimation. We show that this serves as a critical bottleneck, and propose a method based on iterative bilateral filtering with learned kernels. Following the spirit of bilateral filtering, we consider both the deep feature embeddings of each point, as well as their locations in the 3D space. We show via synthetic experiments that our method brings drastic improvements when generating instance proposals for a given point of interest. We further validate our method on the challenging ScanNet benchmark, achieving the best instance segmentation performance amongst the sub-category of top-down methods.