Research map / Point clouds / Segmentation and detection
PointFusion: Deep Sensor Fusion for 3D Bounding Box Estimation
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 dense point-relative box offsets instead of direct absolute corner regression (ablation) | PointFusion · KITTI val, car | AP3D: 43.29 -> 71.53 (easy) | Helped |
| Add image input to the point-cloud model (ablation) | PointFusion dense architecture · KITTI val, car | AP3D: 62.13 -> 71.53 (easy) | Helped |
| Use unsupervised learned point scoring rather than supervised inside-box scoring (ablation) | PointFusion dense architecture · KITTI val, car | AP3D: 71.53 -> 74.71 (easy) | Helped |
| Train the final model on all three KITTI categories rather than cars only (ablation) | PointFusion · KITTI val, car | AP3D: 74.71 -> 77.92 (easy) | Helped |
| Replace raw-point PointNet with a CNN on RGB-depth input (ablation) | PointFusion · KITTI val, car | AP3D: 74.71 -> 7.43 (easy) | Hurt |
| Replace raw-point PointNet with a CNN on RGB-depth input (ablation) | PointFusion · SUN-RGBD test, 10 classes | mAP: 45.38 -> 38.17 | Hurt |
| Add image input to the point-cloud model (ablation) | PointFusion dense architecture · SUN-RGBD test, 10 classes | mAP: 39.24 -> 45.38 | Helped |
| Vary the maximum input points per region of interest (ablation) | PointFusion final · KITTI val | AP3D | Mixed |
| Use PointFusion rather than DSS | SUN-RGBD 3D detection · SUN-RGBD test, 10 classes | mAP: 42.1 -> 45.38 | Helped |
| Use PointFusion rather than COG | SUN-RGBD 3D detection · SUN-RGBD test, 10 classes | mAP: 47.63 -> 45.38 | Hurt |
| Use PointFusion rather than MV3D | KITTI car detection · KITTI val, car, easy | AP3D: 71.29 -> 74.71 | Helped |
| Use PointFusion rather than MV3D | KITTI car detection · KITTI val, car, moderate | AP3D: 62.68 -> 61.24 | Hurt |
Closest papers by meaning
- DeepFusion: Lidar-Camera Deep Fusion for Multi-Modal 3D Object Detection (CVPR 2022 · similarity 0.90)
- Frustum PointNets for 3D Object Detection From RGB-D Data (CVPR 2018 · similarity 0.89)
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection (CVPR 2018 · similarity 0.89)
- Not All Points Are Equal: Learning Highly Efficient Point-Based Detectors for 3D LiDAR Point Clouds (CVPR 2022 · similarity 0.89)
- Voxel or Pillar: Exploring Efficient Point Cloud Representation for 3D Object Detection (AAAI 2024 · similarity 0.88)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · 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.