Method

Sparse Voxel based 3D Object Detection using RGBD Data [SparVox3D]
[Anonymous Submission]

Submitted on 5 Dec. 2020 14:49 by
[Anonymous Submission]

Running time:0.05 s
Environment:GPU @ 2.0 Ghz (Python)

Method Description:
We explore Sparse Voxel Representation for
Monocular 3D Object Detection
Parameters:
|alpha=0.2
Latex Bibtex:

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) 83.76 % 67.88 % 52.56 %
Car (3D Detection) 5.27 % 3.20 % 2.56 %
Car (Bird's Eye View) 10.20 % 6.39 % 5.06 %
Pedestrian (Detection) 69.33 % 52.84 % 48.49 %
Pedestrian (3D Detection) 1.93 % 1.35 % 1.04 %
Pedestrian (Bird's Eye View) 2.90 % 2.05 % 1.69 %
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



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