What question did this study set out to answer?

The aim is to enhance semantic scene completion by integrating a refinement module into existing models.

March 8, 2026Open Access

Enhancing 3D semantic scene completion with refinement module

Key Points

The aim is to enhance semantic scene completion by integrating a refinement module into existing models.
Developed a plug-and-play refinement framework called ESSC-RM.
Implemented a 3D U-Net based Prediction Noise-Aware Module (PNAM).
Employed a Voxel-level Local Geometry Module (VLGM) under multiscale supervision.
Conducted experiments using the SemanticKITTI dataset.
ESSC-RM improved mean IoU from 16.87% to 17.27% when applied to CGFormer.
For MonoScene, mean IoU increased from 11.08% to 11.51%.
Demonstrated the refinement module's applicability across various semantic scene completion models.

Abstract

We propose ESSC-RM, a plug-and-play Enhancing framework for Semantic Scene Completion with a Refinement Module, which can be seamlessly integrated into existing semantic scene completion (SSC) models. ESSC-RM operates in two phases: a baseline SSC network first produces a coarse voxel prediction, which is subsequently refined by a 3D U-Net–based Prediction Noise-Aware Module (PNAM) and Voxel-level Local Geometry Module (VLGM) under multiscale supervision. Experiments on SemanticKITTI show that ESSC-RM consistently improves semantic prediction performance. When integrated into CGFormer and MonoScene, the mean IoU increases from 16.87 to 17.27% and from 11.08 to 11.51%, respectively. These results demonstrate that ESSC-RM serves as a general refinement framework applicable to a wide range of SSC models. Project page: https://github.com/LuckyMax0722/ESSC-RM and https://github.com/LuckyMax0722/VLGSSC .

Bookmark

View Full Paper