Method

Point Cloud Residual Encoder For Object Detection And Tracking[la] [PointRes] [PointRes]
[Anonymous Submission]

Submitted on 30 Aug. 2020 17:24 by
[Anonymous Submission]

Running time:0.013 s
Environment:1 core @ 2.5 Ghz (Python + C/C++)

Method Description:
Point Cloud Residual Encoder For Object Detection
And Tracking
Parameters:
None
Latex Bibtex:
None

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) 96.69 % 93.47 % 90.46 %
Car (Orientation) 96.57 % 93.15 % 90.04 %
Car (3D Detection) 88.73 % 79.55 % 74.17 %
Car (Bird's Eye View) 93.17 % 89.42 % 84.25 %
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