Research map / Point clouds / Point cloud classification
Learning to Assign Orientations to Feature Points
Check 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 learned SIFT-based orientations instead of the descriptor's original orientation assignment | EF-VGG · Viewpoints, Webcam, EF, Strecha, DTU (average) | mAP: 0.35 -> 0.45 | Helped |
| Use learned SIFT-based orientations instead of the descriptor's original orientation assignment | EF-Daisy · Viewpoints, Webcam, EF, Strecha, DTU (average) | mAP: 0.30 -> 0.40 | Helped |
| Use the proposed learned orientation assignments | Rotation-sensitive descriptors, including SIFT, SURF, Daisy, and VGG · All tested datasets | matching performance | Helped |
| Learn orientations using SIFT descriptor distances rather than the respective descriptor | EF-Daisy and EF-VGG · Test datasets | matching performance | Helped |
| Use GHH activation instead of ReLU (ablation) | EF-SIFT+ orientation network · Descriptor matching evaluation | mAP | Helped |
| Use GHH activation instead of Tanh (ablation) | EF-SIFT+ orientation network · Descriptor matching evaluation | mAP | Helped |
| Use GHH activation instead of maxout (ablation) | EF-SIFT+ orientation network · Descriptor matching evaluation | mAP | Helped |
| Use GHH activation instead of PReLU (ablation) | EF-SIFT+ orientation network · Descriptor matching evaluation | mAP | Helped |
| Assign all feature points zero orientation (ablation) | EF-Daisy and EF-VGG · Non-upright datasets | matching performance | Hurt |
| Use learned orientation assignments instead of zero orientations (ablation) | EF-Daisy and EF-VGG · Non-upright datasets | matching performance | Helped |
| Use learned orientation assignments instead of original orientation estimates (ablation) | EF-VGG · Upright datasets | matching performance | No clear effect |
| Use the proposed orientation estimator with EF-VGG | EF-VGG · Strecha fountain | reconstruction vertices: 323979 -> 365261 | Helped |
Closest papers by meaning
- Patch Reordering: A NovelWay to Achieve Rotation and Translation Invariance in Convolutional Neural Networks (AAAI 2017 · similarity 0.86)
- Self-Supervised Representation Learning by Rotation Feature Decoupling (CVPR 2019 · similarity 0.85)
- RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews From Unsupervised Viewpoints (CVPR 2018 · similarity 0.85)
- Learning Rotation-Equivariant Features for Visual Correspondence (CVPR 2023 · similarity 0.85)
- Learning to Orient Surfaces by Self-supervised Spherical CNNs (NeurIPS 2020 · similarity 0.85)
- Learning with 3D rotations, a hitchhiker's guide to SO(3) (ICML 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.