Research map / 3D vision / Depth and stereo
Geometry Aware Convolutional Filters for Omnidirectional Images Representation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use geometry-aware anisotropic filters instead of standard ConvNets | CNN · Spherical (S) | accuracy (%): 94.2 -> 96.9 | Helped |
| use geometry-aware filters instead of standard ConvNets | CNN · Mod-Spherical MS1 | accuracy (%): 91.3 -> 95.1 | Helped |
| use geometry-aware filters instead of standard ConvNets | CNN · Mod-Spherical MS3 | accuracy (%): 90.5 -> 94.9 | Helped |
| use geometry-aware filters instead of standard ConvNets | CNN · Cube-Map (CM) | accuracy (%): 79.4 -> 84.3 | Helped |
| use geometry-aware filters instead of standard ConvNets | CNN · Fish-eye (F) | accuracy (%): 93.4 -> 95.7 | Helped |
| anisotropic GA filters (I-GA) instead of isotropic equal-weight regular grid graph filters, both graph-based (ablation) | graph-based convolution · Cube-Map (CM) | accuracy (%): 40.0 -> 44.2 | Helped |
| add geometric edge weights from Euclidean distances (RGE) to regular grid graph (ablation) | graph-based convolution · Spherical (S) | accuracy (%): 69.4 -> 69.8 | No clear effect |
| full anisotropic GA filters on a regular-structure approach instead of isotropic graph filters | graph-based convolution · Spherical (S) | accuracy (%): 70.2 -> 96.9 | Helped |
| use proposed filters instead of SphereNet | CNN · Spherical (S) | accuracy (%): 94.8 -> 96.9 | Helped |
| use proposed filters instead of Spherical CNN (SCNN) | CNN · Mod-Spherical MS2 | accuracy (%): 83.3 -> 95.3 | Helped |
| use proposed filters instead of Spherical CNN (SCNN) | CNN · Spherical (S) | accuracy (%): 95.2 -> 96.9 | Helped |
| replace isotropic graph filters with anisotropic GA filters in compression network (ablation) | Balle et al. 2017 compression network · Cube-map omnidirectional images | PSNR vs bpp | Helped |
Closest papers by meaning
- SpherePHD: Applying CNNs on a Spherical PolyHeDron Representation of 360deg Images (CVPR 2019 · similarity 0.88)
- OmniFusion: 360 Monocular Depth Estimation via Geometry-Aware Fusion (CVPR 2022 · similarity 0.88)
- OmniVGGT: Omni-Modality Driven Visual Geometry Grounded Transformer (CVPR 2026 · similarity 0.87)
- OA-CNNs: Omni-Adaptive Sparse CNNs for 3D Semantic Segmentation (CVPR 2024 · similarity 0.86)
- OmniSDF: Scene Reconstruction using Omnidirectional Signed Distance Functions and Adaptive Binoctrees (CVPR 2024 · similarity 0.84)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (AAAI 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.