Sphererpn: Learning Spheres For High-Quality Region Proposals On 3D Point Clouds Object Detection

In 2D object detection, a bounding box commonly serves as the proxy for detecting an object. However, extending the bounding box to 3D detection makes the accuracy sensitive to localization error. This problem is more severe on flat objects since little localization error may lead to low overlaps between prediction and ground truth. To address this problem, this paper proposes Sphere Region Proposal Network (SphereRPN) which detects objects by learning spheres as opposed to bounding boxes. We demonstrate that spherical proposals are more robust to localization error compared to bounding box proposals. The proposed SphereRPN is not only accurate but also fast. Experiment results on the standard ScanNet dataset show that the proposed SphereRPN outperforms the previous state-of-the-art methods by a large margin while being 2x to 7x faster. The code will be made publicly available.