Method

Improving Monocular 3D Object Detection with Ground-Guide Model and Adaptive Convolution [GAC3D]


Submitted on 26 Mar. 2021 04:00 by
Khanh Nguyen (Ho Chi Minh University of Science)

Running time:0.25 s
Environment:1 core @ 2.5 Ghz (Python)

Method Description:
We propose a novel keypoint-based method for
monocular 3D object detection. Our approach
leverages depth estimation and geometric
constraints to regress 3D information.
Parameters:
none
Latex Bibtex:
@article{gac3d2021,
author = {Minh-Quan Viet Bui and Duc Tuan
Ngo and Hoang-Anh Pham and Duc Dung Nguyen},
title = "{GAC3D: improving monocular 3D
object detection with
ground-guide model and adaptive convolution}",
URL = {https://peerj.com/articles/cs-686}
journal = {PeerJ Comput. Sci},
year = {2021},
doi = {10.7717/peerj-cs.686}
}

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) 83.30 % 70.73 % 52.23 %
Car (Orientation) 83.27 % 70.49 % 52.04 %
Car (3D Detection) 17.75 % 12.00 % 9.15 %
Car (Bird's Eye View) 25.80 % 16.93 % 12.50 %
This table as LaTeX


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



Orientation estimation 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