Research map / Point clouds / Segmentation and detection
PointPillars: Fast Encoders for Object Detection From Point Clouds
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 PointPillars encoding instead of fixed encoders (ablation) | PointPillars detection pipeline · KITTI val BEV mAP across 0.16^2–0.28^2 resolutions | mAP: 72.0 vs 67.6 at 0.28^2 | Helped |
| Add pillar-center offsets x_p and y_p to point decorations (ablation) | PointPillars encoder · KITTI detection | mAP: 0.5 mAP | Helped |
| Use minimal rather than extensive per-box augmentation (ablation) | PointPillars training · KITTI, especially pedestrians | detection performance | Helped |
| Increase pillar grid size (ablation) | PointPillars · KITTI val BEV detection | speed: 105 Hz at 0.28^2 | Helped |
| Increase pillar grid size (ablation) | PointPillars · KITTI val pedestrian and cyclist detection | detection performance | Hurt |
| Use one PointNet instead of two sequential PointNets (ablation) | PointPillars encoder · Inference runtime | runtime: 2.5 ms | Helped |
| Lower the first backbone block dimensions to match encoder output size (ablation) | PointPillars backbone · Inference runtime | runtime: 4.5 ms | Helped |
| Cut upsampled feature-layer output dimensions in half to 128 (ablation) | PointPillars backbone · Inference runtime | runtime: 3.9 ms | Helped |
| Slim the backbone dimensions and use one PointNet (ablation) | PointPillars · KITTI detection | detection performance | No clear effect |
| Switch inference kernels from PyTorch to TensorRT (ablation) | PointPillars inference pipeline · Inference speed | speedup: 45.5% | Helped |
Closest papers by meaning
- Not All Points Are Equal: Learning Highly Efficient Point-Based Detectors for 3D LiDAR Point Clouds (CVPR 2022 · similarity 0.90)
- Towards Unsupervised Object Detection From LiDAR Point Clouds (CVPR 2023 · similarity 0.90)
- Voxel or Pillar: Exploring Efficient Point Cloud Representation for 3D Object Detection (AAAI 2024 · similarity 0.89)
- SwiftPillars: High-Efficiency Pillar Encoder for Lidar-Based 3D Detection (AAAI 2024 · similarity 0.89)
- PointSR: Self-Regularized Point Supervision for Drone-View Object Detection (CVPR 2025 · similarity 0.87)
- LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detection (ICLR 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.