Computer Science · cs.CV
Class-balanced Grouping and Sampling for Point Cloud 3D Object Detection
Benjin Zhu, Zhengkai Jiang, Xiangxin Zhou, Zeming Li, Gang Yu
See on the mapAbstract
This report presents our method which wins the nuScenes3D Detection Challenge [17] held in Workshop on Autonomous Driving(WAD, CVPR 2019). Generally, we utilize sparse 3D convolution to extract rich semantic features, which are then fed into a class-balanced multi-head network to perform 3D object detection. To handle the severe class imbalance problem inherent in the autonomous driving scenarios, we design a class-balanced sampling and augmentation strategy to generate a more balanced data distribution. Furthermore, we propose a balanced group-ing head to boost the performance for the categories withsimilar shapes. Based on the Challenge results, our methodoutperforms the PointPillars [14] baseline by a large mar-gin across all metrics, achieving state-of-the-art detection performance on the nuScenes dataset. Code will be released at CBGS.
References
The reference list for this paper has not been collected yet.
Cited by 7
7 papers in the index cite this (Semantic Scholar counts 598 in total).
The cited-by list is still filling in while reference lists are collected.