Skip to Main content Skip to Navigation
Conference papers

LMSCNet: Lightweight Multiscale 3D Semantic Completion

Abstract : We introduce a new approach for multiscale 3Dsemantic scene completion from voxelized sparse 3D LiDAR scans. As opposed to the literature, we use a 2D UNet backbone with comprehensive multiscale skip connections to enhance feature flow, along with 3D segmentation heads. On the SemanticKITTI benchmark, our method performs on par on semantic completion and better on occupancy completion than all other published methods -- while being significantly lighter and faster. As such it provides a great performance/speed trade-off for mobile-robotics applications. The ablation studies demonstrate our method is robust to lower density inputs, and that it enables very high speed semantic completion at the coarsest level. Our code is available at https://github.com/cv-rits/LMSCNet.
Document type :
Conference papers
Complete list of metadatas

https://hal.inria.fr/hal-02979521
Contributor : Luis Guillermo Roldao Jimenez <>
Submitted on : Tuesday, October 27, 2020 - 10:53:55 AM
Last modification on : Thursday, December 3, 2020 - 3:33:07 PM

Links full text

Identifiers

  • HAL Id : hal-02979521, version 1
  • ARXIV : 2008.10559

Collections

Citation

Luis Roldão, Raoul de Charette, Anne Verroust-Blondet. LMSCNet: Lightweight Multiscale 3D Semantic Completion. 3DV 2020 - International Virtual Conference on 3D Vision, Nov 2020, Fukuoka / Virtual, Japan. ⟨hal-02979521⟩

Share

Metrics

Record views

19