Research map / 3D vision / 3D generation
Monocular Semantic Scene Completion via Masked Recurrent Networks
Xuzhi Wang, Xinran Wu, Song Wang, Lingdong Kong, Ziping Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MS-GRU recurrent refinement to coarse MSSC (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 27.47 -> 29.16 (+1.69) | Improved |
| add Distance Attention Projection (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 29.16 -> 30.11 (+0.95) | Improved |
| add mask updating module (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 30.11 -> 30.73 (+0.62) | Improved |
| replace surface projection with distance attention projection (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 29.36 -> 30.73 | Improved |
| replace sight projection with distance attention projection (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 30.12 -> 30.73 | Improved |
| use sparse masked GRU instead of dense GRU (ablation) | MonoMRN · NYUv2 test | SSC mIoU and MACs: 29.67 -> 30.73; 171.99 -> 52.44 MACs (G) | Improved |
| add sparse convolution to submanifold convolution mixture in MS-GRU (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 30.12 -> 30.73 | Improved |
| tie weights across iterations rather than untied weights (ablation) | MonoMRN · NYUv2 test | SSC mIoU and params: 30.34 -> 30.73; 2.66 -> 1.33 params (M) | Improved |
| increase iterations from 2 to 3 (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 30.73 -> 30.86 | No clear effect |
| increase iterations from 3 to 4 (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 30.86 -> 30.17 | Hurt |
| add mask initialization from coarse stage (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 29.62 -> 30.73 | Improved |
| add mask loss supervision (ablation) | MonoMRN · NYUv2 test | SSC mIoU: 29.67 -> 30.73 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- MonoScene — MonoScene: Monocular 3D Semantic Scene Completion (CVPR 2022) · better on 5 of 5 shared results
- NDC-Scene — NDC-Scene: Boost Monocular 3D Semantic Scene Completion in Normalized Device Coordinates Space (ICCV 2023) · better on 4 of 4 shared results
- OccFormer — OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Prediction (ICCV 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Global-Aware Monocular Semantic Scene Completion with State Space Models (ICCV 2025 · similarity 0.91)
- Sparsity-Aware Voxel Attention and Foreground Modulation for 3D Semantic Scene Completion (CVPR 2026 · similarity 0.90)
- AdaSFormer: Adaptive Serialized Transformers for Monocular Semantic Scene Completion from Indoor Environments (CVPR 2026 · similarity 0.89)
- Semantic Scene Completion From a Single Depth Image (CVPR 2017 · similarity 0.88)
- Unleashing Network Potentials for Semantic Scene Completion (CVPR 2024 · similarity 0.87)
- Memory-Augmented Re-Completion for 3D Semantic Scene Completion (AAAI 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
