Research map / Point clouds / Segmentation and detection
Adaptive 3D Perception for Small Aerial Targets Under Sparse Sampling via Reinforcement Learning
Shenghai Yuan, Wei Yihan, Jason Yee, Zhuoran Qiao, boyang lou, Enwen Hu
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 |
|---|---|---|---|
| add RL policy (adaptive parameters) (ablation) | A3PRL · MMAUD V2/V3 (train V1, 3D localization) | RMSE: 1.45 -> 1.17 | Improved |
| add tracking-continuity cue to policy observation (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 2.80 -> 2.55 | Improved |
| add acceptance ratio to policy observation (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 2.55 -> 2.33 | Improved |
| add foreground density to observation (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 2.33 -> 2.12 | Improved |
| let policy also adapt detection threshold (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 2.12 -> 1.95 | Improved |
| let policy also adapt association gate (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 1.95 -> 1.72 | Improved |
| use full 5D action space (ablation) | A3PRL · MMAUD V2/V3 (observation/action ablation) | RMSE: 1.72 -> 1.17 | Improved |
| increase policy depth from 3 to 4 layers (ablation) | A3PRL · MMAUD V2/V3 (policy depth) | RMSE: 1.171 -> 1.174 | No clear effect |
| add TDS head (ablation) | A3PRL · MMAUD V2/V3 (module ablation) | RMSE: 2.84 -> 1.17 | Improved |
| add VC head (ablation) | A3PRL · MMAUD V2/V3 (module ablation) | RMSE: 1.92 -> 1.17 | Improved |
| replace RL with heuristic voxel adaptation (ablation) | A3PRL · MMAUD V2/V3 (train V1, 3D localization) | RMSE: 1.17 -> 1.41 | Hurt |
| replace RL with MLP imitation of oracle actions (ablation) | A3PRL · MMAUD V2/V3 (train V1, 3D localization) | RMSE: 1.17 -> 1.35 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PointPillars — PointPillars: Fast Encoders for Object Detection From Point Clouds (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.88)
- Fully Sparse 3D Object Detection (NeurIPS 2022 · similarity 0.87)
- LSK3DNet: Towards Effective and Efficient 3D Perception with Large Sparse Kernels (CVPR 2024 · similarity 0.86)
- LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detection (ICLR 2024 · similarity 0.86)
- MOS: Model Synergy for Test-Time Adaptation on LiDAR-Based 3D Object Detection (ICLR 2025 · similarity 0.86)
- SwiftPillars: High-Efficiency Pillar Encoder for Lidar-Based 3D Detection (AAAI 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.
