Method

End-to-end Pseudo-LiDAR [st] [Pseudo-LiDAR E2E]
[Anonymous Submission]

Submitted on 18 Nov. 2019 21:23 by
[Anonymous Submission]

Running time:0.4 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
End-to-end Pseudo-LiDAR
Parameters:
End-to-end Pseudo-LiDAR
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) 94.00 % 82.54 % 75.31 %
Car (Orientation) 93.74 % 81.56 % 74.23 %
Car (3D Detection) 64.75 % 43.92 % 38.14 %
Car (Bird's Eye View) 79.58 % 58.84 % 52.06 %
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