LMSCNet: Lightweight Multiscale 3D Semantic Completion

24 Aug 2020  ·  Luis Roldão, Raoul de Charette, Anne Verroust-Blondet ·

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.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
3D Semantic Scene Completion NYUv2 LMSCNet-SS mIoU 28.4 # 20
3D Semantic Scene Completion from a single RGB image NYUv2 LMSCNet (rgb input - reported in MonoScene paper) mIoU 15.88 # 4
3D Semantic Scene Completion from a single RGB image SemanticKITTI LMSCNet (rgb input - reported in MonoScene paper) mIoU 7.09 # 3
3D Semantic Scene Completion SemanticKITTI LMSCNet mIoU 17 # 9
3D Semantic Scene Completion SemanticKITTI LMSCNet-SS mIoU 17.6 # 7

Methods


No methods listed for this paper. Add relevant methods here