Method

multi-task MMCOM [MMCOM]


Submitted on 16 Aug. 2020 06:12 by
Chen Zhang (Illinois Institute of Technology)

Running time:0.04 s
Environment:1 core @ 2.5 Ghz (Python)

Method Description:
Model#8 with object detection, lane detection and road segmentation
Parameters:
learning rate 0.03
max ite=100000
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) 95.08 % 87.58 % 77.48 %
Car (Orientation) 39.58 % 36.08 % 32.06 %
Pedestrian (Detection) 86.01 % 73.08 % 68.38 %
Pedestrian (Orientation) 46.12 % 39.20 % 36.81 %
Cyclist (Detection) 85.83 % 77.43 % 68.34 %
Cyclist (Orientation) 35.29 % 32.52 % 28.89 %
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



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



Orientation estimation 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




eXTReMe Tracker