Research map / Detection and segmentation / CNN features
CycleBEV: Regularizing View Transformation Networks via View Cycle Consistency for Bird's-Eye-View Semantic Segmentation
Jeongbin Hong, Dooseop Choi, Taeg-Hyun An, Kyounghwan An, Kyoung-Wook Min
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 view cycle consistency via IVT network (ablation) | CVT · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 39.70 -> 40.55 | Improved |
| add height-aware geometric regularization (ablation) | CVT + VCC · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 40.55 -> 41.40 | Improved |
| add cross-view latent consistency (Align) (ablation) | CVT + VCC + Height · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 41.40 -> 41.78 | Improved |
| add view cycle consistency via IVT network (ablation) | BEVFormer · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 41.00 -> 41.71 | Improved |
| add height-aware geometric regularization (ablation) | BEVFormer + VCC · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 41.71 -> 41.80 | Improved |
| add cross-view latent consistency (Align) (ablation) | BEVFormer + VCC + Height · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 41.80 -> 42.02 | Improved |
| use dual-branch instead of single-branch IVT (ablation) | CVT + CycleBEV · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 41.67 -> 41.78 | Improved |
| use IVT guidance instead of BEV auto-encoder guidance (ablation) | CVT · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 39.88 -> 40.47 | Improved |
| add semantic-aware backbone training SA-II (ablation) | CVT · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 39.70 -> 38.52 | Hurt |
| add semantic-aware backbone training SA-I (ablation) | CVT · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 39.70 -> 39.86 | Mixed |
| add image augmentation on top of CycleBEV | BEVFormer · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 42.02 -> 43.77 | Improved |
| add CycleBEV to the baseline | LSS · nuScenes val BEV segmentation (average of 3 classes) | mIoU: 28.59 -> 31.55 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PETRv2 — PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images (ICCV 2023) · better on 3 of 3 shared results
- BEVFormer — BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (ECCV 2022) · better on 3 of 3 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
- Improving Bird's Eye View Semantic Segmentation by Task Decomposition (CVPR 2024 · similarity 0.89)
- BEV-CAR: Enhancing Monocular Bird's Eye View Segmentation with Context-Aware Rasterization (CVPR 2026 · similarity 0.89)
- Open-Vocabulary BEV Segmentation with 3D-Aware Geometric Constraints (ECCV 2026 · similarity 0.88)
- VQ-Map: Bird's-Eye-View Map Layout Estimation in Tokenized Discrete Space via Vector Quantization (NeurIPS 2024 · similarity 0.85)
- Predictive Uncertainty Quantification for Bird's Eye View Segmentation: A Benchmark and Novel Loss Function (ICLR 2025 · similarity 0.85)
- Weakly Supervised Semantic Segmentation for Driving Scenes (AAAI 2024 · similarity 0.85)
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.
