The KITTI Vision Benchmark Suite

Method

PointRCNN: 3D Object Proposal Generation and Detection from Point Cloud (LiDAR only) [la] [MMLab-PointRCNN]
https://github.com/sshaoshuai/PointRCNN

Submitted on 7 Feb. 2019 18:32 by
Shaoshuai Shi (CUHK)

Running time:		0.1 s
Environment:		GPU @ 2.5 Ghz (Python + C/C++)

Method Description:

See the paper, trained from scratch with 50/50
split.

Update: fix little bug, add results of pedestrian
and cyclist.

V1.0
submitted on 17 Nov. 2018
Car (Detection) 90.77% 89.75% 80.98%
Car (Orientation) 90.76% 89.55% 80.76%
Car (3D Detection) 84.32% 75.42% 67.86%
Car (Bird's Eye View) 89.28 % 86.04 % 79.02 %

Parameters:

See the paper and wait for the code release.

Latex Bibtex:

@inproceedings{shi2019pointrcnn,
title={Pointrcnn: 3d object proposal generation
and
detection from point cloud},
author={Shi, Shaoshuai and Wang, Xiaogang and
Li,
Hongsheng},
booktitle={Proceedings of the IEEE Conference
on
Computer Vision and Pattern Recognition},
pages={770--779},
year={2019}
}

Detailed Results

Object detection and orientation estimation results. Results for object detection are given in terms of average precision (AP) and results for joint object detection and orientation estimation are provided in terms of average orientation similarity (AOS).

Benchmark	Easy	Moderate	Hard
Car (Detection)	95.92 %	91.90 %	87.11 %
Car (Orientation)	95.90 %	91.77 %	86.92 %
Car (3D Detection)	86.96 %	75.64 %	70.70 %
Car (Bird's Eye View)	92.13 %	87.39 %	82.72 %
Pedestrian (Detection)	58.93 %	49.41 %	46.44 %
Pedestrian (Orientation)	57.19 %	47.33 %	44.31 %
Pedestrian (3D Detection)	47.98 %	39.37 %	36.01 %
Pedestrian (Bird's Eye View)	54.77 %	46.13 %	42.84 %
Cyclist (Detection)	86.21 %	73.42 %	66.45 %
Cyclist (Orientation)	85.94 %	72.81 %	65.84 %
Cyclist (3D Detection)	74.96 %	58.82 %	52.53 %
Cyclist (Bird's Eye View)	82.56 %	67.24 %	60.28 %