Research map / 3D vision / 3D generation
REL-SF4PASS: Panoramic Semantic Segmentation with REL Depth Representation and Spherical Fusion
Xuewei Li, Xinghan Bao, Zhimin Chen, Xi Li
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 |
|---|---|---|---|
| replace HHA depth representation with REL (ablation) | CMX + SMMF · Stanford2D3D Panoramic (fold 1) | mIoU (%): 65.43 -> 67.37 | Improved |
| replace original depth with REL representation (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 59.03 -> 64.47 | Improved |
| add SMMF instead of DyMM fusion (ablation) | CMX + REL · Stanford2D3D Panoramic (fold 1) | mIoU (%): 66.73 -> 67.37 | Improved |
| add DyMM dynamic fusion (ablation) | CMX + REL · Stanford2D3D Panoramic (fold 1) | mIoU (%): 64.47 -> 66.73 | Improved |
| replace original depth with ReD channel only (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 59.03 -> 62.02 | Improved |
| replace original depth with EGVIA channel only (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 59.03 -> 63.35 | Improved |
| replace original depth with LOA channel only (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 59.03 -> 61.28 | Improved |
| use only normal-gravity angle as depth channel (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 59.03 -> 59.76 | Improved |
| fuse height with angle into EGVIA instead of height alone (ablation) | CMX · Stanford2D3D Panoramic (fold 1) | mIoU (%): 62.47 -> 63.35 | Improved |
| use REL instead of HHA under 3D disturbance (SGA) (ablation) | CMX + SMMF · Stanford2D3D Panoramic fold 1 (SGA validation) | variance of mIoU: 5.316 -> 1.607 | Improved |
| use REL instead of HHA under 3D disturbance (SGA) (ablation) | CMX + SMMF · Stanford2D3D Panoramic fold 1 (SGA validation) | mean mIoU (%): 62.237 -> 65.380 | Improved |
| compare REL-SF4PASS-Reproj with SGAT4PASS under SGA validation | REL-SF4PASS · Stanford2D3D Panoramic (SGA validation) | mean mIoU (%): 55.984 -> 65.040 | Improved |
Closest papers by meaning
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · similarity 0.88)
- AdapToPASS: Ambiguity-aware Adaptive Spherical Transformer for Panoramic Semantic Segmentation (NeurIPS 2026 · similarity 0.87)
- SGFormer: Semantic-Geometry Fusion Transformer for Multi-modal 3D Panoptic Segmentation (AAAI 2025 · similarity 0.85)
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (NeurIPS 2026 · similarity 0.84)
- Semantics Distortion and Style Matter: Towards Source-free UDA for Panoramic Segmentation (CVPR 2024 · similarity 0.84)
- GoodSAM: Bridging Domain and Capacity Gaps via Segment Anything Model for Distortion-aware Panoramic Semantic Segmentation (CVPR 2024 · similarity 0.84)
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.
