Research map / Detection and segmentation / CNN features
Gaussian Label Distribution Learning for Spherical Image Object Detection
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 GLDL-ATSS and GLDL regression loss | Spherical object detectors · 360-Indoor and PANDORA | AP: +4.6 AP; +6.0 AP | Helped |
| Use GLDL-ATSS and GLDL regression loss | Sphere-SSD · 360-Indoor and PANDORA | AP: +4.9 AP; +5.7 AP | Helped |
| Use GLDL-ATSS and GLDL regression loss | Reprojection R-CNN · 360-Indoor and PANDORA | AP: +5.8 AP; +6.9 AP | Helped |
| Replace Smooth L1 regression loss with GLDL loss | Sphere-CenterNet and R-CenterNet · 360-Indoor and PANDORA | AP: +1.1 AP; +1.4 AP | Helped |
| Replace L1 regression loss with GLDL loss (ablation) | RetinaNet with fixed SphIoU sample selection · 360-Indoor and PANDORA | AP50: +3.1 AP50; +4.2 AP50 | Helped |
| Replace fixed SphIoU sample selection with fixed GLDL selection (ablation) | RetinaNet with GLDL regression loss · 360-Indoor and PANDORA | AP50: +5.2 AP50; +5.5 AP50 | Helped |
| Replace fixed SphIoU selection with SphIoU-ATSS (ablation) | RetinaNet with L1 regression loss · 360-Indoor and PANDORA | AP50 | Helped |
| Replace L1 regression loss with GLDL loss under SphIoU-ATSS (ablation) | RetinaNet · 360-Indoor and PANDORA | AP50: +2.2 AP50; +3.8 AP50 | Helped |
| Use GLDL-ATSS with GLDL loss rather than SphIoU-ATSS with GLDL loss (ablation) | RetinaNet · 360-Indoor and PANDORA | AP50 | Helped |
| Use GLDL loss instead of Smooth L1 loss (ablation) | RetinaNet · 360-Indoor and PANDORA | AP50: 17.6 -> 13.7; 17.2 -> 12.9 | Hurt |
| Set GLDL loss parameter τ to 1, 2, 3, 4, or 5 (ablation) | RetinaNet with GLDL loss · 360-Indoor and PANDORA | AP: 20.3, 20.5, 20.4, 20.0, 19.4; 20.1, 20.3, 20.1, 19.9, 19.7 | No clear effect |
| Set GLDL-ATSS normalization parameter c to 1, 2, 3, 4, or 5 (ablation) | RetinaNet with GLDL-ATSS · 360-Indoor and PANDORA | AP: 21.5, 21.8, 21.7, 21.4, 21.2; 20.9, 21.3, 21.2, 21.1, 22.8 | Mixed |
Closest papers by meaning
- Unbiased IoU for Spherical Image Object Detection (AAAI 2022 · similarity 0.89)
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.84)
- GauCho: Gaussian Distributions with Cholesky Decomposition for Oriented Object Detection (CVPR 2025 · similarity 0.83)
- Open-World Objectness Modeling Unifies Novel Object Detection (CVPR 2025 · similarity 0.83)
- Exploring Orthogonality in Open World Object Detection (CVPR 2024 · similarity 0.83)
- UNION: Unsupervised 3D Object Detection using Object Appearance-based Pseudo-Classes (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.