Research map / Detection and segmentation / 3D detection and driving
Deep Sliding Shapes for Amodal 3D Object Detection in RGB-D Images
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 |
|---|---|---|---|
| Add joint 2D image features to depth features (ablation) | Deep Sliding Shapes · NYUv2 test set, five-category comparison | mAP: 67.8 -> 72.3 | Helped |
| Add object-size pruning (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: 31.5 -> 32.3 | Helped |
| Use SVM scores instead of softmax scores (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: +0.5 | Helped |
| Add 3D bounding-box regression (ablation) | Deep Sliding Shapes with RPN proposals · NYUv2 test set | mAP: +4.1 | Helped |
| Add 3D bounding-box regression (ablation) | Deep Sliding Shapes with 3D Selective Search proposals · NYUv2 test set | mAP: +4.4 | Helped |
| Use directional TSDF rather than single-distance TSDF (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: 30.2 -> 32.3 | Helped |
| Use accurate TSDF rather than projective TSDF (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: 32.2 -> 36.4 | Helped |
| Use 2D image features rather than encoding RGB in 3D voxels (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: 30.1 -> 36.4 | Helped |
| Add HHA features (ablation) | Deep Sliding Shapes · NYUv2 test set | mAP: 36.4 -> 36.2 | Hurt |
| Use a multi-scale RPN instead of a single-scale RPN (ablation) | Deep Sliding Shapes · NYUv2 object proposal evaluation | Recall | Helped |
| Use RPN proposals instead of 3D Selective Search proposals | Deep Sliding Shapes · NYUv2 test set | mAP: 27.4 -> 32.3 | Helped |
| Add RGB color to the 3D TSDF proposal encoding (ablation) | Deep Sliding Shapes RPN · NYUv2 object proposal evaluation | Recall | Helped |
Closest papers by meaning
- Amodal Detection of 3D Objects: Inferring 3D Bounding Boxes From 2D Ones in RGB-Depth Images (CVPR 2017 · similarity 0.90)
- PerspectiveNet: 3D Object Detection from a Single RGB Image via Perspective Points (NeurIPS 2019 · similarity 0.88)
- 3D Bounding Box Estimation Using Deep Learning and Geometry (CVPR 2017 · similarity 0.87)
- In-Hand 3D Object Reconstruction from a Monocular RGB Video (AAAI 2024 · similarity 0.86)
- Training an Open-Vocabulary Monocular 3D Detection Model without 3D Data (NeurIPS 2024 · similarity 0.85)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.85)
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.