Submission details of F-PointNet

Name F-PointNet
Paper Link https://arxiv.org/abs/1711.08488
Code Link https://github.com/charlesq34/frustum-pointnets
AP 0.382051
Input 2d3d
Runtime 0.17 s
Environment 1 GPU (Titan X)
Abstract In this work, we study 3D object detection from RGB-D data in both indoor and outdoor scenes. While previous methods focus on images or 3D voxels, often obscuring natural 3D patterns and invariances of 3D data, we directly operate on raw point clouds by popping up RGB-D scans. However, a key challenge of this approach is how to efficiently localize objects in point clouds of large-scale scenes (region proposal). Instead of solely relying on 3D proposals, our method leverages both mature 2D object detectors and advanced 3D deep learning for object localization, achieving efficiency as well as high recall for even small objects. Benefited from learning directly in raw point clouds, our method is also able to precisely estimate 3D bounding boxes even under strong occlusion or with very sparse points. Evaluated on KITTI and SUN RGB-D 3D detection benchmarks, our method outperforms the state of the art by remarkable margins while having real-time capability.

Detailed results

Overall precision/recall curve

Per-sequence results

sequence name AP
cubberly-auditorium-2019-04-22_1 0.450084
discovery-walk-2019-02-28_0 0.54314
discovery-walk-2019-02-28_1 0.60324
food-trucks-2019-02-12_0 0.442772
gates-ai-lab-2019-04-17_0 0.553188
gates-basement-elevators-2019-01-17_0 0.477009
gates-foyer-2019-01-17_0 0.6986
gates-to-clark-2019-02-28_0 0.666941
hewlett-class-2019-01-23_0 0.577646
hewlett-class-2019-01-23_1 0.842756
huang-2-2019-01-25_1 0.533702
huang-intersection-2019-01-22_0 0.531303
indoor-coupa-cafe-2019-02-06_0 0.303315
lomita-serra-intersection-2019-01-30_0 0.710519
meyer-green-2019-03-16_1 0.314786
nvidia-aud-2019-01-25_0 0.345765
nvidia-aud-2019-04-18_1 0.529262
nvidia-aud-2019-04-18_2 0.570361
outdoor-coupa-cafe-2019-02-06_0 0.20002
quarry-road-2019-02-28_0 0.627778
serra-street-2019-01-30_0 0.517037
stlc-111-2019-04-19_1 0.665362
stlc-111-2019-04-19_2 0.700961
tressider-2019-03-16_2 0.554056
tressider-2019-04-26_0 0.224149
tressider-2019-04-26_1 0.386433
tressider-2019-04-26_3 0.257069
total 0.382051