Problem
Research questions and friction points this paper is trying to address.
Long-tailed 3D object detection
LiDAR
Geometric evidence
Class-frequency imbalance
Object observability
GA-EIRFS: A Geometry-Augmented Repeat-Factor Sampling Method for Long-Tailed LiDAR 3D Object Detection
🤖 AI Summary
This study addresses the mismatch between category frequency and geometric evidence in LiDAR-based 3D detection by proposing Geometry-Augmented Exponentially Weighted Instance Repeat Factor Sampling (GA-EIRFS). The method introduces a fixed geometry score to enhance frequency-driven sampling, modulating repeat factors through point count, normal entropy, and coverage rate. By adjusting only frame-level sampling probabilities without modifying detector architectures, GA-EIRFS achieves plug-and-play, geometry-aware data rebalancing. Experimental results demonstrate that GA-EIRFS effectively improves mAP and NDS on the nuScenes dataset, yielding significant AP gains for tail categories such as bicycles, although performance on KITTI exhibits fluctuations attributable to random seed sensitivity.
📝 Abstract
Long-tailed 3D object detection is treated as a class-frequency problem, but LiDAR supervision quality depends on object observability: similar frequencies can hide different geometric evidence. We introduce Geometry-Augmented Exponentially Weighted Instance-Aware Repeat Factor Sampling (GA-EIRFS), a detector-agnostic method that modulates a frequency-based repeat factor with a fixed geometry score combining point count, surface-normal entropy, and surface coverage. GA-EIRFS changes only frame-sampling probabilities, leaving the detector and inference unchanged. On nuScenes it improves mean average precision (mAP) and the nuScenes detection score (NDS) in four converged experiments with CenterPoint and PointPillars over two seeds; for CenterPoint at seed 666, mAP rises from 0.552 to 0.563 and bicycle AP from 0.306 to 0.359. Per-class gains correlate with the class sampling-weight increase (Spearman rho=0.70, p=0.025) but not with geometry score alone (rho=0.32, p=0.37), so geometry amplifies frequency-driven need. KITTI results vary across seeds, most for the rarest class. Code: https://github.com/Multimodal-Sensing-Lab/GA-EIRFS.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Long-tailed 3D object detection
Geometry-augmented sampling
Repeat-factor sampling
LiDAR point cloud
Detector-agnostic
🔎 Similar Papers
No similar papers found.
