Research map / Detection and segmentation / CNN features
E²I-VRWKV: Explicit EPI-Representation and Interaction-Aware Vision-RWKV for Light Field Semantic Segmentation
Wei Zhang, Chen Jia, Xu Cheng, Fan Shi, Hui Liu, Shengyong Chen
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 GC-Gate geometric gating instead of naive concatenation (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 84.68 -> 86.46 | Improved |
| add VRWKV mixer (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 83.37 -> 86.46 | Improved |
| use MHSA-based EPI geometric prior instead of EPI Bias (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 83.48 -> 86.46 | Improved |
| add both VRWKV and EPI Bias (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 80.20 -> 86.46 | Improved |
| add Dual-Modal Stream (DMS) rectification (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 85.32 -> 86.46 | Improved |
| add Gated Fusion (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 84.98 -> 86.46 | Improved |
| use q shift token shift instead of single shift (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 84.69 -> 86.46 | Improved |
| use q shift instead of double shift (ablation) | E2I-VRWKV · UrbanLF-Syn | mIoU: 85.78 -> 86.46 | Improved |
| scale backbone from MiT-B3 to MiT-B4 (ablation) | E2I-VRWKV · UrbanLF | mIoU: 83.08 -> 86.46 | Improved |
| scale backbone from MiT-B1 to MiT-B4 (ablation) | E2I-VRWKV · UrbanLF | mIoU: 70.61 -> 86.46 | Improved |
| add explicit EPI geometry-aware tokenization to flattened linear baseline (ablation) | E2I-VRWKV · UrbanLF-Syn | BT mIoU: 66.52 -> 83.01 | Improved |
| use E2I-VRWKV instead of OAFuser | E2I-VRWKV · UrbanLF-Syn | latency (ms): 52.4 -> 44.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegFormer — SegFormer: A Topic Segmentation Model with Controllable Range of Attention (AAAI 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
- EPMF: Efficient Perception-Aware Multi-Sensor Fusion for 3D Semantic Segmentation (TPAMI 2024 · similarity 0.86)
- Epipolar Consistent Attention Aggregation Network for Unsupervised Light Field Disparity Estimation (ICCV 2025 · similarity 0.86)
- EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth From Light Field Images (CVPR 2018 · similarity 0.86)
- Neural EPI-Volume Networks for Shape From Light Field (ICCV 2017 · similarity 0.86)
- From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy (NeurIPS 2025 · similarity 0.84)
- NeLF-Pro: Neural Light Field Probes for Multi-Scale Novel View Synthesis (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.
