Research map / 3D vision / Depth and stereo
Distortion-Aware Convolutional Filters for Dense Prediction in Panoramic Images
Keisuke Tateno, Nassir Navab, Federico Tombari
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 standard convolution with distortion-aware convolution on equirectangular input | FCRN depth prediction · Stanford 2D-3D-S depth | rel error: 0.201 -> 0.176 | Improved |
| replace cube map rectification with distortion-aware convolution | FCRN depth prediction · Stanford 2D-3D-S depth | rel error: 0.220 -> 0.176 | Improved |
| distortion-aware convolution versus cube map on rms and log10 | FCRN depth prediction · Stanford 2D-3D-S depth | rms / log10 error: 0.371 -> 0.369 rms; 0.0818 -> 0.0829 log10 | No clear effect |
| replace standard convolution with distortion-aware convolution when trained on NYU v2 perspective images | FCRN depth prediction · Stanford 2D-3D-S depth (trained on NYU v2) | rel error: 0.604 -> 0.517 | Improved |
| replace cube map with distortion-aware convolution when trained on NYU v2 | FCRN depth prediction · Stanford 2D-3D-S depth (trained on NYU v2) | rms error: 0.692 -> 0.578 | Improved |
| replace standard convolution with distortion-aware convolution | FCRN segmentation · Stanford 2D-3D-S semantic segmentation | mIoU (%): 32.63 -> 34.56 | Improved |
| replace cube map with distortion-aware convolution | FCRN segmentation · Stanford 2D-3D-S semantic segmentation | mIoU (%): 33.82 -> 34.56 | Improved |
| replace standard convolution with distortion-aware convolution on floor class | FCRN segmentation · Stanford 2D-3D-S semantic segmentation (floor class) | IoU (%): 78.01 -> 83.40 | Improved |
| distortion-aware convolution versus standard convolution on clutter class | FCRN segmentation · Stanford 2D-3D-S semantic segmentation (clutter class) | IoU (%): 18.84 -> 19.52 | No clear effect |
| distortion-aware convolution versus cube map on wall class | FCRN segmentation · Stanford 2D-3D-S semantic segmentation (wall class) | IoU (%): 61.77 -> 57.17 | Hurt |
Closest papers by meaning
- Bending Reality: Distortion-Aware Transformers for Adapting to Panoramic Semantic Segmentation (CVPR 2022 · similarity 0.89)
- Depth Any Panoramas: A Foundation Model for Panoramic Depth Estimation (CVPR 2026 · similarity 0.89)
- Behind Every Domain There is a Shift: Adapting Distortion-Aware Vision Transformers for Panoramic Semantic Segmentation (TPAMI 2024 · similarity 0.88)
- DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images (NeurIPS 2026 · similarity 0.88)
- PanDA: Towards Panoramic Depth Anything with Unlabeled Panoramas and Mobius Spatial Augmentation (CVPR 2025 · similarity 0.86)
- GoodSAM: Bridging Domain and Capacity Gaps via Segment Anything Model for Distortion-aware Panoramic Semantic Segmentation (CVPR 2024 · similarity 0.86)
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.
