Method

DD3D [DD3D]
https://github.com/TRI-ML/dd3d

Submitted on 8 Dec. 2021 10:20 by
Dennis Park (TRI)

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

Method Description:
A fully convolutional network that is pre-trained on
large-scale depth data with no human annotations.
paper: https://arxiv.org/abs/2108.06417
code:https://github.com/TRI-ML/dd3d
Parameters:
n/a
Latex Bibtex:
@inproceedings{dd3d,
author = {Dennis Park and Rares Ambrus and Vitor
Guizilini and Jie Li and Adrien Gaidon},
title = {Is Pseudo-Lidar needed for Monocular 3D
Object detection?},
booktitle = {IEEE/CVF International Conference on
Computer Vision (ICCV)},
primaryClass = {cs.CV}
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) 94.69 % 93.99 % 89.37 %
Car (Orientation) 94.67 % 93.78 % 88.99 %
Car (3D Detection) 23.19 % 16.87 % 14.36 %
Car (Bird's Eye View) 32.35 % 23.41 % 20.42 %
Pedestrian (Detection) 85.71 % 73.09 % 68.54 %
Pedestrian (Orientation) 77.09 % 63.92 % 59.41 %
Pedestrian (3D Detection) 16.64 % 11.04 % 9.38 %
Pedestrian (Bird's Eye View) 18.58 % 12.51 % 10.65 %
Cyclist (Detection) 81.13 % 65.98 % 58.86 %
Cyclist (Orientation) 73.60 % 57.42 % 50.90 %
Cyclist (3D Detection) 7.52 % 4.79 % 4.22 %
Cyclist (Bird's Eye View) 9.20 % 5.69 % 5.20 %
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



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



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