Research map / Detection and segmentation / Open-set and few-shot detection
360(o) Camera Alignment via Segmentation
Benjamin Davidson, Mohsan S. Alvi, João F. Henriques
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 vanishing point stream (ablation) | segmentation network · Sun360 Test Flat 2 deg | accuracy within 2 deg (%): 1.3 -> 78.4 | Improved |
| add vanishing point stream (ablation) | segmentation network · Sun360 Test Deep360Up 5 deg | accuracy within 5 deg (%): 68.2 -> 92.6 | Improved |
| add vanishing point stream (ablation) | segmentation network · Sun360 Val Flat 5 deg | accuracy within 5 deg (%): 97.0 -> 97.9 | Improved |
| segmentation formulation with VP vs Deep360Up regression | alignment method · Sun360 Test Flat 5 deg | accuracy within 5 deg (%): 60.1 -> 97.8 | Improved |
| segmentation formulation vs Coarse2Fine at 1 deg threshold | alignment method · Sun360 Test Deep360Up 1 deg | accuracy within 1 deg (%): 30.9 -> 19.7 | Hurt |
| rotate input images away from level | GSCNN trained on original images · Sun360 segmentation | mean IoU (%): 40.1 -> 26.7 | Hurt |
| level rotated images with proposed method before segmentation | GSCNN trained on original images · Sun360 segmentation | mean IoU (%): 26.7 -> 40.0 | Improved |
| train segmentation on rotated images instead of levelled | GSCNN evaluated on rotated val · Sun360 segmentation | mean IoU (%): 31.4 -> 39.1 | Improved |
| train on synthetically rotated noise images | Deep360Up · random noise | average angular error: around 5 degrees | Mixed |
Closest papers by meaning
- Learning Spherical Convolution for Fast Features from 360° Imagery (NeurIPS 2017 · similarity 0.85)
- UprightNet: Geometry-Aware Camera Orientation Estimation From Single Images (ICCV 2019 · similarity 0.85)
- On the Robustness of Language Guidance for Low-Level Vision Tasks: Findings from Depth Estimation (CVPR 2024 · similarity 0.84)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · similarity 0.84)
- Depth Anywhere: Enhancing 360 Monocular Depth Estimation via Perspective Distillation and Unlabeled Data Augmentation (NeurIPS 2024 · similarity 0.83)
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · similarity 0.83)
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.
