Research map / Detection and segmentation / 3D detection and driving
Deep Learning of Local RGB-D Patches for 3D Object Detection and 6D Pose Estimation
Wadim Kehl, Fausto Milletari, Federico Tombari, Slobodan Ilic, Nassir Navab
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 |
|---|---|---|---|
| use convolutional autoencoder instead of PCA for patch descriptors (F=128) (ablation) | patch descriptor · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.46 -> 0.70 | Improved |
| use convolutional autoencoder instead of PCA for patch descriptors (F=32) (ablation) | patch descriptor · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.33 -> 0.32 | No clear effect |
| use CAE instead of AE at F=128 (ablation) | patch descriptor · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.65 -> 0.70 | Improved |
| use CAE instead of AE at F=32 (ablation) | patch descriptor · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.43 -> 0.32 | Hurt |
| increase CAE descriptor dimension from 64 to 128 (ablation) | CAE · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.58 -> 0.70 | Improved |
| increase CAE descriptor dimension from 128 to 256 (ablation) | CAE · Tejani multi-instance (re-annotated, k=3, tau=inf) | F1-score: 0.70 -> 0.69 | No clear effect |
| increase number of retrieved neighbors k from 1 to 3 (ablation) | Proposed · Tejani multi-instance (re-annotated) | F1-score | Improved |
| increase k beyond 3 (ablation) | Proposed · Tejani multi-instance (re-annotated) | runtime | Hurt |
| add distance threshold tau (constrained voting), CAE-128 optimum near tau=10 (ablation) | CAE-128 · Tejani multi-instance (re-annotated) | F1-score | Mixed |
| use denser scene sampling (ablation) | Proposed · Tejani multi-instance (re-annotated) | F1-score | Improved |
Closest papers by meaning
- Implicit 3D Orientation Learning for 6D Object Detection from RGB Images (ECCV 2018 · similarity 0.89)
- Augmented Autoencoders: Implicit 3D Orientation Learning for 6D Object Detection (IJCV 2019 · similarity 0.88)
- RGB-D Salient Object Detection via 3D Convolutional Neural Networks (AAAI 2021 · similarity 0.88)
- 2D-Driven 3D Object Detection in RGB-D Images (ICCV 2017 · similarity 0.87)
- SSD-6D: Making RGB-Based 3D Detection and 6D Pose Estimation Great Again (ICCV 2017 · similarity 0.87)
- StablePose: Learning 6D Object Poses From Geometrically Stable Patches (CVPR 2021 · similarity 0.87)
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.
