Method

Stereo R-CNN [st] [Stereo R-CNN]
https://github.com/HKUST-Aerial-Robotics/Stereo-RCNN

Submitted on 6 Feb. 2019 13:22 by
Peiliang Li (Robotics Institute, HKUST)

Running time:0.3 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
Stereo R-CNN is a 3D object detector utilizing raw
stereo
images. It requires neither 3D position supervision
nor
depth input.
Parameters:
Please see the paper.
Latex Bibtex:
@inproceedings{licvpr2019,
title={Stereo R-CNN based 3D Object Detection
for
Autonomous Driving},
author={Li, Peiliang and Chen, Xiaozhi and Shen,
Shaojie},
booktitle={CVPR},
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) 93.98 % 85.98 % 71.25 %
Car (3D Detection) 47.58 % 30.23 % 23.72 %
Car (Bird's Eye View) 61.92 % 41.31 % 33.42 %
This table as LaTeX

2D object detection results.
This figure as: png eps pdf txt gnuplot



3D object detection results.
This figure as: png eps pdf txt gnuplot



Bird's eye view results.
This figure as: png eps pdf txt gnuplot




eXTReMe Tracker