Method

Progressive Coordinate Transforms for Monocular 3D Object Detection [PCT]
https://github.com/amazon-research/progressive-coordinate-transforms

Submitted on 4 Aug. 2021 05:46 by
Tonglin Chen (Fudan University)

Running time:0.045 s
Environment:1 core @ 2.5 Ghz (C/C++)

Method Description:
https://arxiv.org/abs/2108.05793
Parameters:
TBD
Latex Bibtex:
@article{wang2021pct,
title={Progressive Coordinate Transforms for
Monocular 3D Object Detection},
author={Wang, Li and Zhang, Li and Zhu, Yi and
Zhang, Zhi and He, Tong and Li, Mu and Xue,
Xiangyang},
booktitle={NeurIPS},
year={2021}
}

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) 96.45 % 88.78 % 78.85 %
Car (3D Detection) 21.00 % 13.37 % 11.31 %
Car (Bird's Eye View) 29.65 % 19.03 % 15.92 %
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